WPCc  '}>wYhj:s>μ!BU9j-t& 4`kᇉd3.9gB'YI>ϳK)X{͖!e/q5;Q&T>x|25ÄQpCV w㞸I[{=@0(,@V74Aj/ZDx6yhf.>cXB/!6Mi9fsB*')gYd=NǴ ȴITm5$ $"m co((D?!F.uHRn{'ꌑ$VmQlLvVMƞ$WDM٢?$DȸjA(0et辳r߆wUiW<ad&ybX7SnEdυzVUePhb~6dGwH5^0gBn;7w`N8+ 2n^>R[@DZ#U]N % 0(  01^ w41@ mB 1uY 0c 721 0"lc 0 D 0w 0 0 0 0 0W 0 1 0U 0 f! 0! 0" 0B# 0# 0$ 0[% 0& 0& 0h' 09( 0( 0) 0@* 0* 0|+ 0, 0, 0`- 0.. 0. 0/ 0,0 00 0b1 01a~2f2a222f2a2;2/Xh33j3f?5aA5 0t@U5f5a5U @5U,@5+6@6 B*V6 BB6 B6N66f6a6f 7a7f"7a$7f87a:7fN7fP7 BuR7ao7 D57f7a7f7a7 C 7f7a88f)8a+8?8 B7P9fm9ao999UD9f9a9k: i[r;f;a;f;a;f;a;a<f#<a%<f9<a;<fO<aQ<U>e<f< i$8<a<f<a<f=a =f= C=a9=M=c=fy=a{==!=="=f=a=>f(>a*>f>>a@>fT>aV>fj> 0Al> 0D>a>!?&?f?a?f@a@f@a@f.@a0@!D@e@{@ 0D7CU :{C 0eTC D- D 0K6D ADE,E@.EodfDEp ASo pp0pEp[pqpppp 0Dxq 0Dqr 1t Aktu"uuev 0hw 0xxx 0y 0>y 0y 1z 0~{ 0`| 0B} 0$~ 0 0 0 02t*'@gf<  0 1]U@ 0:$^t:6P 0Ytߐ S_Vh 0_H?] B* 0 A 0K 0͞ D/Ӡ 0 D/ 07 1 06j 0S 0: 0^# 0< 1eU6"X fTJw 0<======U*yyyfMaOOOOOOOOOOOOOOOOOOBcccccc"ǵܵܵ  0N\\CENSUS\PRED-TR2-2-HP8000-3A.PRED.DECENNIAL.HQ.CENSUS0(9 Z6Times New Roman RegularX($<PP@:QuickFormat1XXX  cb  XXX  cb  (w;/ **/x.cBackup3|xSU3#37=CIQYag1.a.i.(1)(a)(i)1)a)(;3$2#  0  .3  0   i)((3$ !      0  (#$  0  (O;$0  2#  a  .3  0` (#(#(b$0  0` (#(#2#   .3  0 ` (#` (#({$0  0` (#(#0 ` (#` (#0 (# (#0h(#(#0h(#h(#0(#(#2#  a  )3  0p(#(#(xir$0  0` (#(#0 ` (#` (#2#(  0  )3  0 (# (#($0  0` (#(#0 ` (#` (#0 (# (#2#(  a  )3  0h(#(#(F$0  0` (#(#0 ` (#` (#0 (# (#0h(#(#2#(   )3  0h(#h(#($0  0` (#(#0 ` (#` (#0 (# (#0h(#(#0h(#h(#2#  0  )3  0(#(# p%Gaeimsw{Bullet ListBullets Lista.*+ (_2623  ..*G+G (_25   /%` ` hp x /23  ../%` ` hp x /   O*8+8 (_24  ," hp x ,23  ..," hp x ,  *5+5 (_23 ` ) hp x )23  ..) hp x ) ` *2+2 (_22  &hhp x &23  ..&hhp x &  */+/ (_21  #p x #23  ..#p x #  *,+, (_20 h  p x 23  .. p x h *)+) (_19  pp x 23  ..pp x   *&+& (_18   x 23  .. x   *AA (_17  5+ ` hp x 523  Ԁ5+ ` hp x 5  *GG (_16   /%` ` hp x /23  Ԁ/%` ` hp x /   *88 (_15  ," hp x ,23  Ԁ," hp x ,  *55 (_14 ` ) hp x )23  Ԁ) hp x ) ` *22 (_13  &hhp x &23  Ԁ&hhp x &  *// (_12  #p x #23  Ԁ#p x #  *,, (_11 h  p x 23  Ԁ p x h *)) (_10  pp x 23  Ԁpp x   (&& &_9   x 23  Ԁ x   (AA &_8  5+ ` hp x 523  5+ ` hp x 5  (GG &_7   /%` ` hp x /23  /%` ` hp x /   (88 &_6  ," hp x ,23  ," hp x ,  (55 &_5 ` ) hp x )23  ) hp x ) ` (22 &_4  &hhp x &23  &hhp x &  (// &_3  #p x #23  #p x #  (,, &_2 h  p x 23   p x h ()) &_1  pp x 23  pp x   &&& $_   x 23   x   TABLE AOTABLE A      OTABLE Apage page chapter volume footnote endnote VW najJDPrjn jones347delPi001neuge001     sociopolitical a auncodeable5b eds CPH HDFmailout HUsoversampledethnicitiesReinterview NDRcheckboxesSymens MAFHeraclitus ICM CATIOTABLE A(H$0   OTABLE A(E Z(Times New Roman (9 Z(Times New Roman       &0 d d>Thick Double Pd PdP Pd   &TABLE EQTABLE DQTABLE DTABLE F dTABLE NXXXX)!dxdxQTABLE DPTABLE Add TABLE B   TABLE B  $    28    _ԀAlthoughBerkowitz(2001:17)reportsstrongreactionstothe race/ethnicity  questions,mostofthereactionsshereportsseemdirectedsolelyatrace.  dTTABLE C      7\  `,TimesNewRoman,BoldETABLE A a $    29    _ԀMartin,Elizabeth,2002, QuestionnaireEffectsonReportingofRaceandHispanic  Origin:ResultsofaReplicationofthe1990MailShortForminCensus2000,Alternate  QuestionnaireExperiment.  XfffKCC$ TABLE DTABLE DTABLE BTABLE KTABLE BTABLE BTABLE B\  `&Times New RomanOXfff( c TABLE AQTABLE DOTABLE AOdTABLE A      QTABLE D                ѰTABLE A   ѰTABLE ATABLE BTABLE Bz(Vm$0   (3$ !  TABLE C     сLevel 1Level 2Level 3Level 4Level 5TABLE ATABLE ATABLE BTABLE B          $    13    _ԀJonesandSmith(2002:2425)ddfoundthatmorethan1.6millionadditionalchildren  couldhavebeenreportedasmorethanoneracebasedontheirinterracialparentage.Coupled  withtheactualnumberofchildren(2.1million)inthefourgroupsexaminedwhowerereported  asmorethanonerace,thetotalnumberofchildrenreportedasmorethanoneracecouldbe t nearly4millionorhigher.Theauthorsrefertothispopulationasthe potentialpoolof ` multiracialchildren.(3$ !  .\  `"TimesNewRoman(,!$0  0` (#(#  )!dxdx( $ Figure  1  &OLE 2.0 Box <=8C HKKKK*OLE*WPCaf(*(*"5139hD !(*   !!!"""###$$$%%%&&&'''((()))***+++,,,---...///000111222333444555666777888999:::;;;<<<===>>>???@@@AAABBBCCCDDDEEEFFFGGGHHHIIIJJJKKKLLLMMMNNNOOOPPPQQQRRRSSSTTTUUUVVVWWWXXXYYYZZZ[[[\\\]]]^^^___```aaabbbcccdddeeefffggghhhiiijjjkkklllmmmnnnooopppqqqrrrssstttuuuvvvwwwxxxyyyzzz{{{|||}}}~~~a}sϏO/?oL߯O? ?O??O??̅{3OD_/_3߬.7_/o.ߓ3o/_2;)ߜ?A?>2_?_O/+.? ?5_/o+_?1oo//_7) O:+?AO=?0/=?$o-+?._ O??4+0?/"4?)$_8+o@o=??O/;?'?.+?. ?3_,/?/$3?O*&_6+@=?O~pP00.`/? 0:p? O_/>?Z +@2__o/: . Oo? P/0$2pP00*/%_050m``<pP 0- O_?8_  _ .O*?.o1_? ,??- O 0&_1*O'5*? ?<￟? ߿_. ?7/ ??0Ͽ? ߂0o __2/ _-߿/ ?Ͽ. 2 O 1￟? ߿_.Ͽ? OO 6 ς?_ <￟? ߿_ ??0??6/ ?2 o3 /3/O0 0o4? _ 3 ??2  7? ?> o? ??~)p8P?0P`5/ /P@[E/0< b/8/Q0? 3)pT/ ` `j` ~`0@/O`  5/ _@?2 F0$`2@?a/ op04@ _ ?@=`0PM / _@6@ @D `Ip P~ W`:O _"020 3 $ 2/P p0 & /PU? = '2 @ >p 0'~@b/  90 ?@ZPO?P3  ?p9 ?`00 bP`V _1@ _W?P= O =07~.. ?`?8 z@p0B  9Pp`@6P/9 _ @\Q@? p5 @<_h ~ 0.@0 2p/ ?/6`j!?2`߿<p&P.@@￯`40 < 03p 50dp`H?~VP@ 1 o@0P2 [P?PT?  @`1P 0&.P@P QP 02[0P p?P= ,~0`PM$@RWa? o/p20?/0/ = 6~pMp pRP oppO t~pPL`/O@[40 @PPP t~p0L ` Z/0xp^p `t~pK / `?[px0`/00/Ps~fp/@JP O o@O@ `m0  ? s~f0 J   Pp0l@P`pOr~f ?H/ pp/O@ 0 o@k@O?`_P `r~f@ `Gp ? 0N P 0j/P` r~gp `F @0 NP 0 ? @h_/PP @@@q~g D@ /pN P pgP ? q~gp _@C/ @  M`/ @O  d0oo@PPP` _p~h/ @0 ?@oN O` ?Ѐb/`O _ `p~i op?p `?`M0 @  @_/p_ P  p~i  @=  Mp_ ? P  \@//P @@o~j0  FPON__Ѐ`?ߐX _O Op ` o~k0?߀90u`HOPNoP`opS @___`C@:P  _n~l0/`8p @jH N `@J߿/_p@_?D :?   O pn~m _0`?O`/?߀@?/Y_ߠ `/O] 9@  n~n?πJo@O0o@ _`V? @`? o]p9` P/0m~pP_0J  P/? T?0? _08_`` m~q/π`Kp/Q@O@ ?PUo`oP` Pl~s@П߃ L 0R_`BO/ PoߐIσ9/0 Opl~u`?0 @`M0 S?O_Ao А [0@p`80 l~vO0 /pT0_ ?/H@_ O0p`/0k~x /  -@  UP /@ /J@? 0 ?pPP` k~z@П 7@@oW_ _` /L 0@Po  p Pj~|`o /,?`W/ 0? _PMP/@ ?pj~} o +   X@   N0/ ?p0/ j~@/ +P 0POZ ?@p `NO@`?  ``? i~_ _ *POPZP 0/ O  ` i~/ P+P  Z0 O 0O@O@0o ?P`(`i~ + @p?[0 p8r  P  )`h~0s @+@ o@[p? 8Ps@_ @P0/P:0q  ?*:,~t` *0O p [ P 0QP09r`? OP+? g~p0@M` `+@ p/[@ ./0L OQ `/p0@K `?? /`0P~06  0[p. &M[ /0}0_ P `/ &~@M06@/@0 [8p0U0@L OPP ?`8P0~00N6 /Z 80p@M@00L  Op .0~`?0PN 6`?@ p c.P0`?0`Mp/ `? -`/0`~  PX`,p @oc .P0PXQ / 0P N OP`/ ` - 0` ~@@@O0 -?b`.0 @N 0QP 00@P@O@_ ` `.@P/@~p.``@ ?. @d0PC@0/@P@.P@`PP / P0pP p ?P-pO.0/ X 1?O?1?O///??o1  /1O4  5 )_! ?- O~ / OO4@0Wp" PS`@/ /p/@`4?pP1 p 1 / OOQp/  4"00(!P?@, / OO ~p/  _? 00 /# U`_?0SF?0P`N@p=@ P2pS ?0O?p'@?P"` -p~`@_?@/`02p_ /  < 1  p?^p`1?po/?@OO _ 2`@_?@5@_ />pFp# ?@,`@_?@~O`06`_?@3/ /@?2"7`F02 _0`6 2p 4Op06 @2"@,@?# O -Op0~`?`0o@6o04/ `W[ >@`O?`3@`6`o3P 0    4`?`0o@U 1PP/ $ O ,`?`0O@~p8_p` 5/ o` V@\/D@_4pOp8P> 0O ?6pWp/ <@00%@?%P -p  ?2 O _9/ ߏ/-Ͽo ? O-_? ?0 ? 7 ?/2 ? ? 1߿ ρ_ /8 ?4￟/ o ?(ϯ_ ?/ "/_߿/ o+ ?  ? 3 o_ _;_/oￏ_ -?,?,o ?0 /? 9_ ?ߏ2 O 2(/: ? 5?o_(-"O o+ ?  OO/ 4?? >?/-/,?,_ ?0/_o? ; /_O 3Oo? 3'o= OO/ 6?O(-"? ?, O_/  /5!@O$/-?,? ?0!O=?4!/3%?/7/?'-" /-/ o6oC? 0-?,__?0@o65#?Co7 '-O"/o _9ϏO_G__2-,??1o?C_o9?6/_F_9'-?!o0__/?ϟo?/Lo/?5﮿,ϐ?6O/_HOO?_/O9ߝN_/;'߭!ߒ3_/ϟO?/?_G߿?/?ORo??_Ƈ;o??_OϏ??oHϯo??OϟO?/?_lϟO?/?_AJPO?[?_o/O߁o?P/[_O ~P@cPO?/0   $.O]  TO)F 0ooo/?_No%?c o  O&/-] _ TO+G O/?/L??$c~  &//-Ph `O]*@R.?pPp@3/0J?P$Pc /'o._oWO,I0Ooϯ/H$c O(.O/`/_YO-I?0/ _G$Od ?(.a_o o?Z?-J0??O F%d ?(o /Z-??JO0o/ /)??O [oo.oJ/0O_?)? /\?O/o??-J/0_O?o(O] /o-I_0o__O ?/o _?߃?/o//__?/o? ߏ?/ _?o?? /o/?oO?? ߁?ߏ?O?O o_ o?? ߂o?__O?_ _o//?/O//o?ߏ?__/Oo? O?o _?_/O _?o___/O/_o o /??_ ?  ?_/  ?? ??__    OO_O__?/ _ O  o? ? ?_?/ ? ??? ??? _?O_ O /_o_? O? _ o _ // o  /  O?_/ oO   ?/ ??/  / ? ??? oo/ oo/ o_? __/?/ _ߟ/_? ?_?oO   O_/ o/ o o/o/  o _ OOo?? //_ _?߯_/o _/o __? _ ? _o /  ?oO  o / o_ ?/? ? /o ? ? ?o_ //? __ _//? _oO /o_? /? /oo/??  ? ?O/ _ /  oO ? O/  O o/ _  _/ o  _/? o//_? _ o??oo_? _ / Oo/_? /?O? ? /_O //? ?/ O o /?     _ ? _ ?OO_o   ?// ?_o_ /o_   ?/߿/?/_  / O??/? O  / /o _ o_O/ O//  ?? O_   !_/ / _ Oo/o   _O ?/ ? _O _ o/ ??// O _/ O?/_   o    /_?? O߿? /_ _ o/ ?_? O o" O ? ?/_? /? O/ o O/  /?   //O/  // O? __ o /  ?ooO _  ??#/O _ ? /o  _ /? ?_ O_   ?O_Oo_ ?  / _ ? O ?_O/?  /?O /O  #  ??? /? ?   _o   oo O OoO_? /O? O_O/ _/ o _o ?_ _ # / ?o O?o?O _  ?/OOo_ ?/??  _ _?o?/??__  ??_? ?? ? Oo _O? ?/  // / #o O  ?o   O?O_O//  _  //__ ? OO_o/_/_  ?? /_?_/o _ oo  ?/ O$   ?O? _o ? ?/  /?  /  _// _/ o/ oo  _ ooo/  ?/ _  _  ?$   o O  ?_ OO   _? OO _?ߏ _/ O_ o _ _/o ?? _ _oO  _? /? ? /O o$? O/ ?/ O_O  o_??   o  o OO?o _?_/o  ?o?? / /? O  /o /o   o  O ?$? ? O__??/O  ?/O? _"_  ?O   _  /O   ??_ _ ?// o /o o ???o? ?    _ # o _?oo o O/O%/ ?  _  ?? _ ?_O O_??oO/  /  ?? o?O o /  ? o$  /o?O?_  /+/   /?  _ _ /? _O/o/?/  O  o ?? ? _ _ _/O?$o  _?oo_? ?O ?+_ O    O  _/? oO   _ ~ 0(P ?P  p/   p@ P/@P$@0 _/`@ OP00@P_@P/@@ PO`>0 PP  @  `_ P`PP P %p`_ 0POP / p  P/p@    Pp/0/__o   _  ?_$? ? O O///_ ? o_  /O    _  /OOoO??//o??_ O_ / /? o/ / /  ??  o ?#_  o /__O ?_ O?/  o/   _ O  ? O  ?/_?/ ? oo_??o  ?  /?O ~@` / @/P  /? 0 `?/  @/  O@@#  @0/ P0/0@ p0 ?p@/0__`@ @0/ @ Po0/ P Pp0 `O`oP  @P @P @@ ?  0  @p P/0 Pp @Џ_0 `0@`P?o@ ?P ?0 @ /pO` 0 O _ ??  ? _ ? ?O/O/$? _O/ o O O ?_/  _?? /  _ /  o O?? ?/ OO/?o? oo? o ?o _ O/    _oo/_$ / ? ߿  ?_/? OOOߏ/ O ? _ oO  ? oOO_?ߟ?o ?/?O/oO// _O / ?_ _ ??????O?   /!o O_ߟ/O  oO/oO? _o/?? o/Ϗo / _ oO? /o?o??/o??__oo ? ?OO/      ? ?  _!O / ? ?O ?_  o _  ?  /  o /?_ _?/_   ?  /? / _   _ o_?/ Oo  ? oo O? _?  o ? /?/ ? O  ? ? //? o  O  ?  o  _   //_ ?O/?߄  O?/!ߊOO/oτ?o ?OO?_  _? o?/_ ߄  OO?o?oO _?O  //o ?? 4߿%_V _@aD\߿m߿`Ͽ*Ͽ)Ͽ@\bi\Ooj_\_Poj?\?Q /jo[oQ_? j[Qoko[oQ??l/[/Sn?Z?S_ /noZ߁X߯oqWPWin 6.0/OLE 1.0 Prefix Information MarkerPresentations.Drawing.10ࡱ> 9:8  !"#$%&'()*+,-./01234567<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~Root EntryPRP'U HHyOle WPG20dfCompObj}Й PRP'Corel Presentations 10 Drawing Embed SourcePresentations.Drawing.109qP ^Object(s) forward one5n  WPCaf(*(*"5139hD !(*   !!!"""###$$$%%%&&&'''((()))***+++,,,---...///000111222333444555666777888999:::;;;<<<===>>>???@@@AAABBBCCCDDDEEEFFFGGGHHHIIIJJJKKKLLLMMMNNNOOOPPPQQQRRRSSSTTTUUUVVVWWWXXXYYYZZZ[[[\\\]]]^^^___```aaabbbcccdddeeefffggghhhiiijjjkkklllmmmnnnooopppqqqrrrssstttuuuvvvwwwxxxyyyzzz{{{|||}}}~~~a}sϏO/?oL߯O? ?O??O??̅{3OD_/_3߬.7_/o.ߓ3o/_2;)ߜ?A?>2_?_O/+.? ?5_/o+_?1oo//_7) O:+?AO=?0/=?$o-+?._ O??4+0?/"4?)$_8+o@o=??O/;?'?.+?. ?3_,/?/$3?O*&_6+@=?O~pP00.`/? 0:p? O_/>?Z +@2__o/: . Oo? P/0$2pP00*/%_050m``<pP 0- O_?8_  _ .O*?.o1_? ,??- O 0&_1*O'5*? ?<￟? ߿_. ?7/ ??0Ͽ? ߂0o __2/ _-߿/ ?Ͽ. 2 O 1￟? ߿_.Ͽ? OO 6 ς?_ <￟? ߿_ ??0??6/ ?2 o3 /3/O0 0o4? _ 3 ??2  7? ?> o? ??~)p8P?0P`5/ /P@[E/0< b/8/Q0? 3)pT/ ` `j` ~`0@/O`  5/ _@?2 F0$`2@?a/ op04@ _ ?@=`0PM / _@6@ @D `Ip P~ W`:O _"020 3 $ 2/P p0 & /PU? = '2 @ >p 0'~@b/  90 ?@ZPO?P3  ?p9 ?`00 bP`V _1@ _W?P= O =07~.. ?`?8 z@p0B  9Pp`@6P/9 _ @\Q@? p5 @<_h ~ 0.@0 2p/ ?/6`j!?2`߿<p&P.@@￯`40 < 03p 50dp`H?~VP@ 1 o@0P2 [P?PT?  @`1P 0&.P@P QP 02[0P p?P= ,~0`PM$@RWa? o/p20?/0/ = 6~pMp pRP oppO t~pPL`/O@[40 @PPP t~p0L ` Z/0xp^p `t~pK / `?[px0`/00/Ps~fp/@JP O o@O@ `m0  ? s~f0 J   Pp0l@P`pOr~f ?H/ pp/O@ 0 o@k@O?`_P `r~f@ `Gp ? 0N P 0j/P` r~gp `F @0 NP 0 ? @h_/PP @@@q~g D@ /pN P pgP ? q~gp _@C/ @  M`/ @O  d0oo@PPP` _p~h/ @0 ?@oN O` ?Ѐb/`O _ `p~i op?p `?`M0 @  @_/p_ P  p~i  @=  Mp_ ? P  \@//P @@o~j0  FPON__Ѐ`?ߐX _O Op ` o~k0?߀90u`HOPNoP`opS @___`C@:P  _n~l0/`8p @jH N `@J߿/_p@_?D :?   O pn~m _0`?O`/?߀@?/Y_ߠ `/O] 9@  n~n?πJo@O0o@ _`V? @`? o]p9` P/0m~pP_0J  P/? T?0? _08_`` m~q/π`Kp/Q@O@ ?PUo`oP` Pl~s@П߃ L 0R_`BO/ PoߐIσ9/0 Opl~u`?0 @`M0 S?O_Ao А [0@p`80 l~vO0 /pT0_ ?/H@_ O0p`/0k~x /  -@  UP /@ /J@? 0 ?pPP` k~z@П 7@@oW_ _` /L 0@Po  p Pj~|`o /,?`W/ 0? _PMP/@ ?pj~} o +   X@   N0/ ?p0/ j~@/ +P 0POZ ?@p `NO@`?  ``? i~_ _ *POPZP 0/ O  ` i~/ P+P  Z0 O 0O@O@0o ?P`(`i~ + @p?[0 p8r  P  )`h~0s @+@ o@[p? 8Ps@_ @P0/P:0q  ?*:,~t` *0O p [ P 0QP09r`? OP+? g~p0@M` `+@ p/[@ ./0L OQ `/p0@K `?? /`0P~06  0[p. &M[ /0}0_ P `/ &~@M06@/@0 [8p0U0@L OPP ?`8P0~00N6 /Z 80p@M@00L  Op .0~`?0PN 6`?@ p c.P0`?0`Mp/ `? -`/0`~  PX`,p @oc .P0PXQ / 0P N OP`/ ` - 0` ~@@@O0 -?b`.0 @N 0QP 00@P@O@_ ` `.@P/@~p.``@ ?. @d0PC@0/@P@.P@`PP / P0pP p ?P-pO.0/ X 1?O?1?O///??o1  /1O4  5 )_! ?- O~ / OO4@0Wp" PS`@/ /p/@`4?pP1 p 1 / OOQp/  4"00(!P?@, / OO ~p/  _? 00 /# U`_?0SF?0P`N@p=@ P2pS ?0O?p'@?P"` -p~`@_?@/`02p_ /  < 1  p?^p`1?po/?@OO _ 2`@_?@5@_ />pFp# ?@,`@_?@~O`06`_?@3/ /@?2"7`F02 _0`6 2p 4Op06 @2"@,@?# O -Op0~`?`0o@6o04/ `W[ >@`O?`3@`6`o3P 0    4`?`0o@U 1PP/ $ O ,`?`0O@~p8_p` 5/ o` V@\/D@_4pOp8P> 0O ?6pWp/ <@00%@?%P -p  ?2 O _9/ ߏ/-Ͽo ? O-_? ?0 ? 7 ?/2 ? ? 1߿ ρ_ /8 ?4￟/ o ?(ϯ_ ?/ "/_߿/ o+ ?  ? 3 o_ _;_/oￏ_ -?,?,o ?0 /? 9_ ?ߏ2 O 2(/: ? 5?o_(-"O o+ ?  OO/ 4?? >?/-/,?,_ ?0/_o? ; /_O 3Oo? 3'o= OO/ 6?O(-"? ?, O_/  /5!@O$/-?,? ?0!O=?4!/3%?/7/?'-" /-/ o6oC? 0-?,__?0@o65#?Co7 '-O"/o _9ϏO_G__2-,??1o?C_o9?6/_F_9'-?!o0__/?ϟo?/Lo/?5﮿,ϐ?6O/_HOO?_/O9ߝN_/;'߭!ߒ3_/ϟO?/?_G߿?/?ORo??_Ƈ;o??_OϏ??oHϯo??OϟO?/?_lϟO?/?_AJPO?[?_o/O߁o?P/[_O ~P@cPO?/0   $.O]  TO)F 0ooo/?_No%?c o  O&/-] _ TO+G O/?/L??$c~  &//-Ph `O]*@R.?pPp@3/0J?P$Pc /'o._oWO,I0Ooϯ/H$c O(.O/`/_YO-I?0/ _G$Od ?(.a_o o?Z?-J0??O F%d ?(o /Z-??JO0o/ /)??O [oo.oJ/0O_?)? /\?O/o??-J/0_O?o(O] /o-I_0o__O ?/o _?߃?/o//__?/o? ߏ?/ _?o?? /o/?oO?? ߁?ߏ?O?O o_ o?? ߂o?__O?_ _o//?/O//o?ߏ?__/Oo? O?o _?_/O _?o___/O/_o o /??_ ?  ?_/  ?? ??__    OO_O__?/ _ O  o? ? ?_?/ ? ??? ??? _?O_ O /_o_? O? _ o _ // o  /  O?_/ oO   ?/ ??/  / ? ??? oo/ oo/ o_? __/?/ _ߟ/_? ?_?oO   O_/ o/ o o/o/  o _ OOo?? //_ _?߯_/o _/o __? _ ? _o /  ?oO  o / o_ ?/? ? /o ? ? ?o_ //? __ _//? _oO /o_? /? /oo/??  ? ?O/ _ /  oO ? O/  O o/ _  _/ o  _/? o//_? _ o??oo_? _ / Oo/_? /?O? ? /_O //? ?/ O o /?     _ ? _ ?OO_o   ?// ?_o_ /o_   ?/߿/?/_  / O??/? O  / /o _ o_O/ O//  ?? O_   !_/ / _ Oo/o   _O ?/ ? _O _ o/ ??// O _/ O?/_   o    /_?? O߿? /_ _ o/ ?_? O o" O ? ?/_? /? O/ o O/  /?   //O/  // O? __ o /  ?ooO _  ??#/O _ ? /o  _ /? ?_ O_   ?O_Oo_ ?  / _ ? O ?_O/?  /?O /O  #  ??? /? ?   _o   oo O OoO_? /O? O_O/ _/ o _o ?_ _ # / ?o O?o?O _  ?/OOo_ ?/??  _ _?o?/??__  ??_? ?? ? Oo _O? ?/  // / #o O  ?o   O?O_O//  _  //__ ? OO_o/_/_  ?? /_?_/o _ oo  ?/ O$   ?O? _o ? ?/  /?  /  _// _/ o/ oo  _ ooo/  ?/ _  _  ?$   o O  ?_ OO   _? OO _?ߏ _/ O_ o _ _/o ?? _ _oO  _? /? ? /O o$? O/ ?/ O_O  o_??   o  o OO?o _?_/o  ?o?? / /? O  /o /o   o  O ?$? ? O__??/O  ?/O? _"_  ?O   _  /O   ??_ _ ?// o /o o ???o? ?    _ # o _?oo o O/O%/ ?  _  ?? _ ?_O O_??oO/  /  ?? o?O o /  ? o$  /o?O?_  /+/   /?  _ _ /? _O/o/?/  O  o ?? ? _ _ _/O?$o  _?oo_? ?O ?+_ O    O  _/? oO   _ ~ 0(P ?P  p/   p@ P/@P$@0 _/`@ OP00@P_@P/@@ PO`>0 PP  @  `_ P`PP P %p`_ 0POP / p  P/p@    Pp/0/__o   _  ?_$? ? O O///_ ? o_  /O    _  /OOoO??//o??_ O_ / /? o/ / /  ??  o ?#_  o /__O ?_ O?/  o/   _ O  ? O  ?/_?/ ? oo_??o  ?  /?O ~@` / @/P  /? 0 `?/  @/  O@@#  @0/ P0/0@ p0 ?p@/0__`@ @0/ @ Po0/ P Pp0 `O`oP  @P @P @@ ?  0  @p P/0 Pp @Џ_0 `0@`P?o@ ?P ?0 @ /pO` 0 O _ ??  ? _ ? ?O/O/$? _O/ o O O ?_/  _?? /  _ /  o O?? ?/ OO/?o? oo? o ?o _ O/    _oo/_$ / ? ߿  ?_/? OOOߏ/ O ? _ oO  ? oOO_?ߟ?o ?/?O/oO// _O / ?_ _ ??????O?   /!o O_ߟ/O  oO/oO? _o/?? o/Ϗo / _ oO? /o?o??/o??__oo ? ?OO/      ? ?  _!O / ? ?O ?_  o _  ?  /  o /?_ _?/_   ?  /? / _   _ o_?/ Oo  ? oo O? _?  o ? /?/ ? O  ? ? //? o  O  ?  o  _   //_ ?O/?߄  O?/!ߊOO/oτ?o ?OO?_  _? o?/_ ߄  OO?o?oO _?O  //o ?? 4߿%_V _@aD\߿m߿`Ͽ*Ͽ)Ͽ@\bi\Ooj_\_Poj?\?Q /jo[oQ_? j[Qoko[oQ??l/[/Sn?Z?S_ /noZ߁X߯oq/  SpanishinCensus2000    AncestryQuestion 9/&x  0   ff 09dV0Columbian 0&!  X 0   @  @  @ ! 0&!  0X 0  3.6% :0'  0 00 X 0:Cuban 0&BOlePartOlePres000;     V h ~   V   H~C HH(H  !!!"""###$$$%%%&&&'''((()))***+++,,,---...///000111222333444555666777888999:::;;;<<<===>>>???@@@AAABBBCCCDDDEEEFFFGGGHHHIIIJJJKKKLLLMMMNNNOOOPPPQQQRRRSSSTTTUUUVVVWWWXXXYYYZZZ[[[\\\]]]^^^___```aaabbbcccdddeeefffggghhhiiijjjkkklllmmmnnnooopppqqqrrrssstttuuuvvvwwwxxxyyyzzz{{{|||}}}~~~?????o_?/_//?/?/Oo?????OO??__o/_OO???o?oO?O/o/_o_?/o?ooO?_??/?/??//o?o???????O?/oO/OoOO?o?o/Ϗoo??/o??o??/???__o?O?ooo_??_/OOO/OO?O_??o?/?/oO///?/_??//O/?/?/_OO/o?oOO_??//__o_?O/_?o_/O_OOo????__/?oo//_O//?Oo_?//?__o/O????/o?o/?o?_///_/?_/?O?o??oooO//?_??_??oO//??_oO_?/?o??/O_??OO?o_?_/o??//?Ooo//_?oOO_OO_?_/O_o__o?O??///O?_/_OO_o__o?/?oO?o?O_?O?/?_o??_??_???Oo/?oO_?/O_?/o/?_O_?Oo_/?O??/__o_?OO??_?/?OO/?/O//?_/o/_??Oo?//?o_o_?//?/_/O?O//_O/O?O/o???o_/?__//?_oOo?o??O/oo/o/o_/_/oOO_/oo//oO??//__?_o_/_??_//?_?_???_OO__o???_??__O???/_????o/_?/oߏ/???o???/o?oO???????o_??o???__??__//?//ߏ_Oo???Oo???__?o___/_?/?O/??_O/??ooO_o????OOO//O?/_/OoOo?_o?O??????/O???O?/??/????_o?_??o???O?/?O?/?_O__??oo?/__o_oo??_?o?ooOO/???????_o?/oOO/?O?O_/o___/o_?o???/??o_O_?/??_oO?__/__?o?/?OoOOO?_/??oO?_?O?O/_???/?//O?/O/?O??O/?o/O_?OOO?_?/OO_OO????O/o???o??/???OO????/O??o????oO??_????O/??/OO???oo??//??/?/o????oO???/_?_???_OoO?/???_??/O?o?o_??/__?O//?????OO??o_?_o??//?_oo????O__/??o?_o/??/?/??_??_???o?o/?o??o??///??O??????o????O?//???/?????ߟoo?_?/???_??o?_??/?????__///o????/_???/?o/Oo??o?O_?__?o_??OO?O??O/?_o/oOo_O/O?/??_O???_o?O??_o?o/_O?O/??O???O?NANI METAFILEPICTVpV  ~   V   H~C HH(H  !!!"""###$$$%%%&&&'''((()))***+++,,,---...///000111222333444555666777888999:::;;;<<<===>>>???@@@AAABBBCCCDDDEEEFFFGGGHHHIIIJJJKKKLLLMMMNNNOOOPPPQQQRRRSSSTTTUUUVVVWWWXXXYYYZZZ[[[\\\]]]^^^___```aaabbbcccdddeeefffggghhhiiijjjkkklllmmmnnnooopppqqqrrrssstttuuuvvvwwwxxxyyyzzz{{{|||}}}~~~?????o_?/_//?/?/Oo?????OO??__o/_OO???o?oO?O/o/_o_?/o?ooO?_??/?/??//o?o???????O?/oO/OoOO?o?o/Ϗoo??/o??o??/???__o?O?ooo_??_/OOO/OO?O_??o?/?/oO///?/_??//O/?/?/_OO/o?oOO_??//__o_?O/_?o_/O_OOo????__/?oo//_O//?Oo_?//?__o/O????/o?o/?o?_///_/?_/?O?o??oooO//?_??_??oO//??_oO_?/?o??/O_??OO?o_?_/o??//?Ooo//_?oOO_OO_?_/O_o__o?O??///O?_/_OO_o__o?/?oO?o?O_?O?/?_o??_??_???Oo/?oO_?/O_?/o/?_O_?Oo_/?O??/__o_?OO??_?/?OO/?/O//?_/o/_??Oo?//?o_o_?//?/_/O?O//_O/O?O/o???o_/?__//?_oOo?o??O/oo/o/o_/_/oOO_/oo//oO??//__?_o_/_??_//?_?_???_OO__o???_??__O???/_????o/_?/oߏ/???o???/o?oO???????o_??o???__??__//?//ߏ_Oo???Oo???__?o___/_?/?O/??_O/??ooO_o????OOO//O?/_/OoOo?_o?O??????/O???O?/??/????_o?_??o???O?/?O?/?_O__??oo?/__o_oo??_?o?ooOO/???????_o?/oOO/?O?O_/o___/o_?o???/??o_O_?/??_oO?__/__?o?/?OoOOO?_/??oO?_?O?O/_???/?//O?/O/?O??O/?o/O_?OOO?_?/OO_OO????O/o???o??/???OO????/O??o????oO??_????O/??/OO???oo??//??/?/o????oO???/_?_???_OoO?/???_??/O?o?o_??/__?O//?????OO??o_?_o??//?_oo????O__/??o?_o/??/?/??_??_???o?o/?o??o??///??O??????o????O?//???/?????ߟoo?_?/???_??o?_??/?????__///o????/_???/?o/Oo??o?O_?__?o_??OO?O??O/?_o/oOo_O/O?/??_O???_o?O??_o?o/_O?O/??O???O?ZE<<CBtables                     уLevel 1Level 2Level 3Level 4Level 5(3$ !  ($$   1    $    32    _Ԁ}X\XXXPublicLaw94171,enactedin1975,amendedsection141oftitle13,UnitedStates  Code,#X\XX}X\U#}X\XXX\whichdirectstheCensusBureautoprovideredistrictingdataneededbythe50statesfor  theiruseinredrawingdistrictsoftheUnitedStatesCongressandstatelegislatures.#X\XX}X\#(EGMSY_ekqAutoList521.1.1.1.1.1.1.1.]Inline Te&xt" <<CLevel 1Level 2Level 3Level 4Level 5         [ $    30    _ԀSinger,Phyllis,andSharonR.Ennis,2002, Census2000ContentReinterviewSurvey:  AccuracyofDataforSelectedPopulationandHousingCharacteristicsasMeasuredby  Reinterview,Census2000EvaluationB5.0AA.Normal  5+ ` hp x 55+ ` hp x 5  <AA:Definition T  5+ ` hp x 55+ ` hp x 5        <AA:Definition L , 5+ ` hp x 5  5+ ` hp x 586Definition(hh&H1  5+ ` hp x 5      5+ ` hp x 5  (5EGMSY_ekqAutoList561.1.1.1.1.1.1.1.(]]&H2  5+ ` hp x 5  5+ ` hp x 5  (]]&H3  5+ ` hp x 5  5+ ` hp x 5  (]]&H4  5+ ` hp x 5 XXX 5+ ` hp x 5  (]]&H5  5+ ` hp x 5  5+ ` hp x 5  (]]&H6  5+ ` hp x 5  5+ ` hp x 5  2DD0Address  5+ ` hp x 55+ ` hp x 5  8MM6Blockquote , , 5+ ` hp x 5   5+ ` hp x 5,*CITE z $    31    _ԀForsimplicityofexpression,thefollowingtermsusedintheCRSreportwere  modified:V  ԀW1)lowinconsistency=goodconsistency;2)moderateinconsistency=moderate  consistency;and3)highinconsistency=poorconsistency.     $    1    _ԀTheGAOreferencestheAccuracyandCoverageEvaluation(A.C.E.)andthe  IntegratedCoverageMeasurement(ICM)programs.dB  ԍ(20Highlightspage,R2)dC    $    2    _ԀThistableincludedbothnonHispanicsandrespondentswhodidnotanswerthe  Hispanicoriginquestion,whichmakessensebecauseourpreviousresearchsuggeststhatmostof  thenonrespondersarenotHispanic(McKenney,Bennett,Harrison,anddelPinal,1993:5).dB  ԍ(34  page5,R1)dC   \ $    3    _ԀGAO(2003:14)reported 93percentofHispanicsgiventhe1990styleformreporteda  specificsubgroup,comparedto81percentofHispanicsgiventhe2000styleform,butthatwas  basedonpreliminaryAQEfindings. $    4    _dGroupswiththeirownspecificcheckboxincluded:1)Mexican,MexicanAm.,Chicano;  2)PuertoRican;and3)Cuban.d $    5    _dGroupsgivenasspecificexamplesinthe1990styleformincluded:Argentinean,  Columbian,Dominican,Nicaraguan,Salvadoran,andSpaniard.d,dl*CODEK<6X9`(Courier NewKS}\  `&Times New RomanS  $    6    _ԀX\XXXԀForCRS,#X\XXX\T#X\XXX\acharacteristicisrarewhen5percentorlesscasesfallinthecategory.#X\XXX\#ԀdB  ԍ(2  page9,R2)dX\XXX\ԍ.C  #X\XXX\m#(%EGMSY_ekqAutoList611.1.1.1.1.1.1.1.(9 Z(Times New Roman 42Emphasis   тLevel 1Level 2Level 3Level 4Level 5  $    7    _ԀAfterremovingcensusnoninterviews,CRSnoninterviews,andnonmatches,CRShad  19,554samplepersonmatches.dB  ԍ(2page4,R1)dC   , $    8    _dInthisreport, proxyreferstoarespondentwhowasahouseholdmemberbutnotthe  sampleperson.ddB  ԍ(2page11,R3)C  ԚV  dW 64Hyperlink     j $    9    _ԀdForsimplicityofexpression,thefollowingtermsusedintheCRSreportweremodified:  1)lowinconsistency=goodconsistency;2)moderateinconsistency=moderateconsistency;and  3)highinconsistency=poorconsistency.d  $    10    _ԀThephrase moreconsistencyisusedinthisreportinsteadof lessinconsistency,  andsoonfromtheCRSreport,foreaseofexpression.  $    11    _ԀIn1990theaggregateindexwasreferredtoasanLfoldindexandwasdefinedas   aweightedaverageoftheindividualindexescomputedforeachcategoryofadistribution  (Thomas,Dingbaum,andWoltman1993:9). C dB  ԍ(40page9,R3)C  Ԛd L $    7    _ԀIn1990,thiswasreferredtoasanLfoldin1990andwasdefinedas aweighted  averageoftheindividualindexescomputedforeachcategoryofadistribution(Thomas,  Dingbaum,andWoltman1993:9). w $    12    _Thegeneraltermsusedincluded:Hispanic,Latino,Spanish,SpanishAmerican,Other  CentralAmerican,OtherSouthAmerican,OtherHispaniccheckboxwithnowritein,Spaniard  (includingspecificterms),andallothernonspecificnationalorigins.<:FollowedHype      $    14    _ԀItshouldbenotedthatobservationswerenotbasedonascientificallyselectedsample,B  d(new=HoughandBorsa6Bpage38,R1)(originally=Borsa6page47,R1)C  Ԛd  andwerebasedonsubjectivejudgmentsofindividualobservers.dB  ԍ(New=HoughandBorsa6Bpage38R2)  (originally=Borsa6page47,R2)C  Ԛd  $    15    _ԀHowever,thelackofconsistencymayberelatedtoswitchingtogeneralresponsessuch  as Hispanicor Spanish,dasshowninTables3.9and3.10.d $    16    _Ԁ Latinowasnottabulatedseparatelyandmaybetabulatedwith Othergroups.&P d d4go2Keyboard K<6X9`(Courier NewKS}\  `&Times New RomanS <:Preformatted  /%  ,Kk %#/K<6X9`(Courier NewKS}\  `&Times New RomanS/%  ,Kk %#/  <:zBottom of 7(X7    5+ ` hp x 5?%2A`Arial?  S}\  `&Times New RomanS7(X75+ ` hp x 5  I)1dxd<:zTop of For7(X7    5+ ` hp x 5?%2A`Arial?  S}\  `&Times New RomanS7(X75+ ` hp x 5  I)2dxd0KS.SampleK<6X9`(Courier NewKS}\  `&Times New RomanS(֓GIOU[agmsAutoList2081.1.1.1.1.1.1.1. $    17    _ԀForadditionalinformationaboutthesefourstrataseeBentley,Mattingly,Houghand  Bennett(2003:1516).0.Strong 8dl6TypewriterK<6X9`(Courier NewKS}\  `&Times New RomanS $    18    _ԀWhite;BlackorAfricanAmerican;AmericanIndianandAlaskaNative;Asian;and  NativeHawaiianandOtherPacificIslander. I $    19    _ԀMoverscreatedproblemswithsampleweightingbecauseofdifferentialsamplingof  racialcombinations.ForadditionalinformationaboutthisissueseeBentley,Mattingly,Hough  andBennett(2003:2728).42Variable: 8HTML MarkupB      2 0CommentB  #$%&'()* 35;AGMSY_11.1.1.1.1.1.1.1.<6X9`(Courier New $    20    _ԀNote,thesefiguresdonotincludeindividualswhodidnotreportaHispanicorigin.  $    21    _ԀMostofthesecombinationsarenumericallythelargestineachpanelandarealsoof  policyinterest,butwereselectedprimarilyforillustrativepurposes. \ $    22    _ԀTheracealonecategoriesrepresentrespondentswhoreportedonerace(plusacategory  withallrespondentswhoreportedTwoormoreraces).Racealoneorincombinationcategories  representrespondentswhoselectedaparticularraceregardlessofthenumberofotherraces  selected(i.e., thecombinationofpeoplewhoreportedoneraceandpeoplewhoreportedthat t sameraceinadditiontooneormoreotherraces). $    23    _ԀForacomprehensivelistofdifferences,seeTable2inLeslie,Raglin,andSchwede,  (2002:2064).  $    24    _ԀItshouldbenotedthatobservationswerenotbasedonascientificallyselectedsample,B  d(new=HoughandBorsa6Bpage38,R1)(originally=Borsa6page47,R1)C  Ԛd  andwerebasedonsubjectivejudgmentsofindividualobservers.dB  ԍ(new=HoughandBorsa6Bpage38R2)  (originally=Borsa6page47,R2)C  d $    25    _ԀFiftystatesandtheDistrictofColumbiaconstitutethestatesidedata. $    26    _ԀHispanicsoverallandPuertoRicanswhoseoriginwasnoteditedorimputed.   фLevel 1Level 2Level 3Level 4Level 5Level 1Level 2Level 3Level 4Level 5<:Default Para%2A`Arial $    27    _ԀHereafter,IwillrefertotheislandofPuertoRicoas theIsland.TABLE Rhttp://www.census.gov/pred/www/.              (O$ !  _d 8\XXdd8  Y1XX  *X'` XX*GPCensus2000Testing,Experimentation,  andEvaluationProgram0 h 0h(#h(#0(#(#0p(#(#0p(#p(#0 (#(##X\X Y1}#;XX\September26,2003#X\X;#Ԛd Y1XX\d: (# (# #X\X Y1"#AX))xdE)xA TopicReportSeries,No.9  ` IXX\#X\XI#L0XX\  RACEANDETHNICITY V  INCENSUS2000 T  Ѐ#X\XL0L# Y1XX\ R    #X\X Y1#K+ ` hp x X'` XK}X\XXX\}NCX}X\#}X\X}NC#Qualityassuranceprocedureswereappliedthroughoutthecreationofthisreport. t  #X\XX}X\#}X\XXX\Thistopicreportintegratesfindingsandprovidescontextandbackgroundforinterpretationof L resultsfromCensus2000evaluations,tests,andotherresearchundertakenbytheU.S.Census 8 Bureau.Itispartofabroadprogram,theCensus2000Testing,Experimentation,andEvaluation $t program,designedtoassessCensus2000andtoinform2010Censusplanning.}NCX}X\#NC}NC# ` #X\XNCE#   Y1XX\m9)%`|0' `(*ES},2,mPreparedby: n(# AX)*dE*AJorgedel_Pinal_,Ph.D. )$ PopulationDivision *2&   $(#$(#(#(##X\X Y1# (#(#$(#$(# ./~*$ GP G  G  TheCensus2000Test,Experimentation,andEvaluationProgramprovidesmeasuresof L  effectivenessfortheCensus2000design,operations,systems,andprocessesandprovides 8  informationonthevalueofnewordifferentmethodologies.Theresultsandrecommendations $ t fromtheseanalysesprovidevaluableinformationcrucialtoplanningthe2010Census.  ` ByprovidingmeasuresofhowwellCensus2000wasconducted,thisprogramfullysupports  L  theCensusBureausstrategytointegratethe2010earlyplanningprocesswithongoingMaster  8  AddressFile/TopologicallyIntegratedGeographicEncodingandReferencing(_MAF_/TIGER) $  enhancementsandtheannualAmericanCommunitySurvey.Thepurposeofthereportthat   followsistosynthesizeresultsfromCensus2000evaluations,experiments,andother   assessmentsandtomakerecommendationsforplanningthe2010Census.Census2000Testing,   Experimentation,andEvaluationreportsareavailableontheCensusBureausInternetsiteat:   4+ YO  5  http://www.census.gov/pred/www/. 6YO   7 ; p        `     h      p ̚    `     h      h+&*   TABLEOFCONTENTS  B   Vm   `     h   Vm`0     1.IntroductionandBackground#""J(#.)(#(##1Vm`݌  Ќ  !  !k0  0` (#(#    1.1.Relatedreports  #""J(#.` (#` (##1!k݌ t Ќ  !  !R0  0` (#(#    1.2.Pastresearch#""J(#. "` (#` (##1!Rm݌ ` Ќ  !  !(0  0` (#(#    1.3.Researchquestions#""J(#.'` (#` (##3!(C݌ L  Ќ  Vm  Vm0     2.Census2000AlternateQuestionnaireExperiment(_AQE_)#""J(#.ppA(#(##5Vm݌ $ t Ќ  !  ! 0  0` (#(#    2.1.Studydesign#""J(#. !` (#` (##5! (݌  ` Ќ  !  !0  0` (#(#    2.2.Limitations#""J(#.0 0 ` (#` (##5!݌  L  Ќ  !  !0  0` (#(#    2.3.Findingsinbrief#""J(#. &` (#` (##5!݌  8  Ќ  !  !0  0` (#(#    2.4.Overallracereporting#""J(#.vv+` (#` (##7!݌ $  Ќ  !  !r0  0` (#(#    2.5.OverallHispanicoriginreporting#`"`"I(#.6` (#` (##12!r݌   Ќ  !  !]0  0` (#(#    2.6.DetailedHispanicoriginreporting#`"`"I(#.^^7` (#` (##13!]x݌   Ќ  Vm  VmL0     3.Census2000ContentReinterviewSurvey(_CRS_)#`"`"I(#.9(#(##21VmLh݌   Ќ  !  !L0  0` (#(#    3.1.Studydesign#`"`"I(#. !` (#` (##21!Lg݌ p Ќ  !  !%0  0` (#(#    3.2.Limitations#`"`"I(#.0 0 ` (#` (##22!%@݌ \ Ќ  !  !0  0` (#(#    3.3.Findingsinbrief#`"`"I(#. &` (#` (##23!݌ H Ќ  !  !0  0` (#(#    3.4.Consistencyofracereporting#`"`"I(#.2` (#` (##29!݌ 4 Ќ  !  !0  0` (#(#    3.5.Consistencyofancestryreporting#`"`"I(#.  6` (#` (##32!݌  p Ќ  !  ! 0  0` (#(#    3.6.Consistencyofplaceofbirthreporting#`"`"I(#.<` (#` (##35 ! ݌  \ Ќ  Vm  Vm!0     4.CensusQualitySurvey(_CQS_)toEvaluateResponsestotheCensus2000Questionon 4 Race:AnIntroductiontotheData#`"`"I(#.tt0(#(##37Vm!!݌   Ќ  !  !"0  0` (#(#    4.1.Studydesign#`"`"I(#. !` (#` (##37!"#݌   Ќ  !  !#0  0` (#(#    4.2.Limitations#`"`"I(#.0 0 ` (#` (##39!##݌  Ќ  !  !$0  0` (#(#    4.3.Findingsinbrief#`"`"I(#. &` (#` (##39!$$݌  Ќ  !  !~%0  0` (#(#    4.4.Discussionof_CQS_Ԁfindings#`"`"I(#.VV/` (#` (##42!~%%݌  Ќ  Vm  Vm&0     5.ComparingtheRaceandHispanicOriginDatafromtheAmericanCommunitySurvey X  (ACS)andCensus2000#`"`"I(#.$(#(##47Vm&&݌ D! Ќ  !  !'0  0` (#(#    5.1.Studydesign#`"`"I(#. !` (#` (##47!''݌ 0"  Ќ  !  !(0  0` (#(#    5.2.Limitations#`"`"I(#.0 0 ` (#` (##47!((݌ #l! Ќ  !  !_)0  0` (#(#    5.3.Findingsinbrief#`"`"I(#. &` (#` (##48!_)z)݌ $X" Ќ  Vm  Vm:*0     6.PuertoRicoCensus2000RaceandEthnicityQuestions#`"`"I(#.A(#(##57Vm:*V*݌ %0!$ Ќ  !  !%+0  0` (#(#    6.1.Studydesignandlimitations#`"`"I(#.~~1` (#` (##57!%+@+݌ &"% Ќ  !  !,0  0` (#(#    6.2._Nonresponse_ԀtoraceandHispanicorigin#`"`"I(#.LL<` (#` (##58!,),݌ '#& Ќ  !  !-0  0` (#(#    6.3.Hispanicoriginreporting#`"`"I(#..` (#` (##58!-:-݌ (#' Ќ  !  !.0  0` (#(#    6.4.Racereporting#`"`"I(#.2 2 #` (#` (##60!..݌ )$( Ќ  Vm @-(,   Vm.0     d7.Conclusion#`"`"I(#.h h (#(##66Vm./݌  Ќ  !  !/0  0` (#(#    7.1.Effectsofquestionnairechanges#`"`"I(#.5` (#` (##66!//݌  Ќ  !  !00  0` (#(#    7.2.Consistencyinreporting#`"`"I(#.>>-` (#` (##68!00݌  Ќ  !  !10  0` (#(#    7.3.Sequencingand_nonresponse_#`"`"I(#./` (#` (##69!11݌ t Ќ  !  !20  0` (#(#    7.4.ComparingCensus2000tootherdatasources#`"`"I(#.@` (#` (##70!22݌ ` Ќ  !  !30  0` (#(#    7.5.ComparingCensus2000andthe1990census#`"`"I(#.>` (#` (##72!33݌ L  Ќ  !  !t40  0` (#(#    7.6.PuertoRico#`"`"I(#.N N ` (#` (##73!t44݌ 8  Ќ  !  !I50  0` (#(#    7.7.Futureresearch#`"`"I(#.h h $` (#` (##74 !I5d5݌ $ t Ќ  Vm  Vm%60     8.Recommendations#`"`"I(#. (#(##75Vm%6A6݌  L  Ќ  Vm  Vm60     9.References#`"`"I(#.L L (#(##80Vm67݌ $  Ќ  T (83 ! !   `        LISTOFTABLES   8NCXX\#X\XNC8#   BH  BHE90     Table2.1 AlternativeQuestionnaireExperiment(_AQE_)Item_Nonresponse_ԀforHispanic  OriginandRacebyHispanicOrigin #""J(#.$$1(#(##7BHE9`9݌  Ќ  BH  BH:0     Table2.2 AlternativeQuestionnaireExperiment(_AQE_)RaceResponsesbyHispanics #""J(#.4!4![(#(##9BH::݌ z  Ќ  BH  BH;0     Table2.3 AlternativeQuestionnaireExperiment(_AQE_)RaceResponsesbyNonHispanics V  andHispanicOriginNotAscertained #`"`"I(#.tt2(#(##11BH;<݌ F  Ќ  BH  BHO=0     Table2.4 AlternativeQuestionnaireExperiment(_AQE_)DetailedHispanicOrigin "r  ResponsesbyFormType #`"`"I(#.jj%(#(##14BHO=j=݌ b  Ќ  BH  BHݟ>0     Table2.5 ComparisonofSpecificHispanicOriginDistributionsFromCensus2000 >  LongFormsandSimulatedTotalsUsingSupplementalInformationonPlaceof .  BirthandAncestry #`"`"I(#.t t !(#(##17BHݟ>>݌  Ќ  BH  BH1@0     Table3.1 AggregateResponseVarianceMeasuresforHispanicOrigin(UneditedData) #`"`"I(#.!!](#(##24BH1@L@݌  Ќ  BH  BH;A0     Table3.2 ResponseVarianceMeasuresforHispanicOrigin(EditedData) #`"`"I(#.DDQ(#(##24BH;AVA݌  Ќ  BH  BH9B0     Table3.3 HispanicOriginIndexofInconsistency:2000and1990 #`"`"I(#.^^J(#(##26BH9BTB݌ ^ Ќ  BH  BH0C0     Table3.4 HispanicOriginQuestionbyQuestionnaireType #`"`"I(#.C(#(##27BH0CKC݌ : Ќ  BH  BH D0     Table3.5 OtherHispanicCategorybyQuestionnaireType #`"`"I(#.D(#(##27BH D;D݌ f Ќ  BH  BHE0     Table3.6 GeneralHispanicResponsesinCensus2000andContentReinterviewSurvey #`"`"I(#.!!](#(##28BHE2E݌ B Ќ  BH  BH!F0     Table3.7 ResponseVarianceMeasuresforRacebyHispanicOrigin(EditedData) #`"`"I(#. Y(#(##30BH!F@30.53Hz>@r  30.53 rhCt' Hz>@30.53 Hz>@X {Gz#-9.74{Gz#rЀ 9.74 rhCt( {Gz#-9.74 {Gz#X fffff?-31.9fffff?r  ӄ31.9 YOIt)  fffff?-31.9 40 fffff?X 4Y < Originmissing PF!*f \(\*@13.18\(\*@P  13.18 qgC+ \(\*@13.18 \(\*@f q= ףp#@9.72q= ףp#@q  9.72 pfB, q= ףp#@9.72 q= ףp#@f Gz @3.46Gz @pЀ 3.46 pfB- Gz @3.46 Gz @f A@35.6A@p   @ 35.6XNL.  A@35.6 40  A@f XQSource:DerivedfromMartin(2002a:7,11Table2andTable4)B  dd(1page7,R4)C  ddB  ԍ(1page11,R2)C   8.  h XX\Note: Bold numbersinColumn3indicatesignificantdifferencesatthep<.05level#X\X h#. $/ ' x2.3.3Conclusiononitem_nonresponse_ 1 Itisworthmentioningagainthatthepreviouslydiscussedresultsmayonlyapplytomail 3 responses.Thechangestothe2000questionnaireappeartohaveproducedaverysalutaryeffect p4 onHispanicorigin_nonresponse_,atleastinmailreturns.Althoughtheresponseratestothe \5 questiononHispanicoriginarevastlyimprovedinthe2000stylequestionnaire(3.3percent H6 comparedwith14.5percent),_nonresponse_ԀtoHispanicoriginremainsonthehighside. 4 7 _Nonresponse_ԀtotheracequestionisverylowfornonHispanics(0.6percentonthe2000style  !p8 form,and1.5percentonthe1990styleform).Ontheotherhand,race_nonresponse_Ԁremains  "\9 unacceptablyhighforHispanicsatover20percentdespiteasignificantimprovementinrace "H: reportingbyHispanicsin2000styleforms. Futureresearchisneededtoaddressthis #4; persistingissue . $$ <   2.4.OverallracereportingB  ԍ(1pageiv,R4)C  gl` &"> 2.4.1. ChangestotheCensus2000questionnairealsoaffectedracereporting(Martin, (#@ 2002a:iv,12). t)$A  0    ReportingofTwoormoreraces:IntheCensus2000stylequestionnaire 2.03percent L+&C  ofrespondentsreportedTwoormoreracescomparedwith 0.82percent inthe1990style <,'D questionnaire.B  ԍ(1page12,R1)dC  d,-|(E(#(#  0    ReportingofNativeHawaiianandOtherPacificIslander: IntheCensus2000style  questionnaire 0.17percent ofrespondentsreportedNativeHawaiianandOtherPacific  Islandercomparedwith 0.05percent inthe1990stylequestionnaireB  ԍ.(1page12,R1)C  .x(#(#  0    ReportingofSomeotherrace: IntheCensus2000stylequestionnaire 3.72percent P (#(# 0  ofrespondentsreportedSomeotherrace comparedwith 4.42percent inthe1990style @  questionnaire.B  ԍ(1page12,R1)dC  d0 (#(# 2.4.2.Discussionofoverallracereporting  X  Comparedtothe1990style,the2000styleformyieldsahigherproportionofresponsesofmore 0  thanonerace.Weexpectedthisfindingbecausethe2000styleformallowsreportingofmore   thanoneracebutthe1990styledoesnot.Thiswasalsooneofthechangescalledforbythenew   OMBstandards(OMB,1997:58789).B  ԍ(28page58789,R2)C  ԀWhatismoreinterestingisthatnearlyonepercentof   respondentstothe1990styleformalsogavemorethanonerace.Whileitiswellknownthat   peoplehaverespondedinthismannerinpastcensuses(_Edmonston_,Goldstein,and_Tamayo_ԀLott, | 1996:23),B  ԍ(29page23,R1)C  ourprocedureseditedmultipleraceresponsesintosingleresponses(_Cresce_,2003). h Anotherissueofconcernisthereportingof Someotherrace,whichisnotastandardOMB @ racecategory(OMB,1997:58789).B  ԍ(28page58789,R3)dC  dThe Someotherracecategorywasaddedforrespondents ,| whowereunabletoidentifywithoneormoreoftheOMBcategories(White;BlackorAfrican h American;AmericanIndianorAlaskaNative;Asian;andNativeHawaiianandOtherPacific T Islander).B  ԍ(30page2,R1)dC  dOnedifficultyisthat Someotherrace (_SOR_) hasbecomethethirdlargestcategory @ after Whiteand BlackorAfricanAmerican(_Grieco_ԀandCassidy,2001:23).B  ԍ.dd(30page3,R1)dC  dAnother , difficultyisthatforallotherfederalstatisticalpurposeswehavereclassifiedthe_SOR_Ԁresponses  intotheOMBcategories,andthereis nowaytoevaluatehowthisreclassificationcorresponds  topeoplesselfperception(_Edmonston_,Goldstein,and_Tamayo_ԀLott,1996:39).B  ԍ(29page39,R1)dC  dAsinprevious  censuses,thevastmajorityofpeopleinthe_SOR_ԀcategoryinCensus2000wereofHispanic  originB  ԍ(30page 0#! 11,R1)C  (_Grieco_ԀandCassidy,2001:11).Forallthesereasons,itisimportanttoexaminerace x reportingseparatelybyHispanicorigin(Martin,2002a:13,14). d  2.4.3.RacereportingbyHispanics <"   0    ReportingofTwoormoreracesbyHispanics: IntheCensus2000stylequestionnaire$`"(#(#  0  7.84percent ofHispanicsreportedTwoormoreracescomparedwith 4.59percent in %P # the1990stylequestionnaires.B  ԍ(1page13,R1)dC  d     p %@!$(#(#  0    ReportingofSomeotherracebyHispanics: IntheCensus2000stylequestionnaire'#&(#(#  0  39.03percent ofHispanicsreportedSomeotherracecomparedwith 51.47percent in ($' the1990stylequestionnaire.B  ԍ(1page13,R1)dC  d h )$((#(#  0    ReportingofWhitebyHispanics: IntheCensus2000stylequestionnaire 48.98percent x+&*  ofHispanicsreportedWhitecomparedwith 39.88percent inthe1990style h,'+ questionnaire.B  ԍ(1page13,R1)C  X-(,(#(#  &  tablesTable2.2 AlternativeQuestionnaireExperiment(_AQE_)RaceResponsesbyHispanics tables  *6X7dd44 44 44 F44 F44 X5(#(#,` dd ,Fdd ,dd ,Fdd ,dd +    Race 6,!@ ff 62000Style T (1) 6,!@" ff 61990Style T (2) 6,!@" ff 6Difference T (3=12) 6,!@ " ff 6Percent T  Difference @  (4=3/2) 6,&, "   ff 46White , |   PF! hX = ףp}H@48.98= ףp}H@P  48.98 rhC, | = ףp}H@48.98 = ףp}H@X q= ףC@39.88q= ףC@r  39.88 qgC, | q= ףC@39.88 q= ףC@X 333333"@9.10333333"@qЀ 9.10 pfB, | 333333"@9.10 333333"@X 6@22.86@pЀ22.8 XNH, | 6@22.8 40 6@X 4XBlack l   h   OE!T X (\@2.07(\@O  2.07 pfBh  (\@2.07 (\@X (\@2.32(\@p  2.32 qgBh  (\@2.32 (\@X п-0.25пqЀ0.25 rhCh  п-0.25 пX %-10.8%rЀ10.8 YOIh  %-10.8 40 %X 4YAmericanIndianand   AlaskaNative OE! X Gz?1.48Gz?O  1.48 pfB  Gz?1.48 Gz?X  ףp= ?0.72 ףp= ?p  0.72 pfB   ףp= ?0.72  ףp= ?X RQ?0.76RQ?pЀ0.76 qgB  RQ?0.76 RQ?X ffffffZ@105.6ffffffZ@qЀ105.6 YOI  ffffffZ@105.6 40 ffffffZ@X 4YAsian l  0   OE! X (\?0.58(\?O  0.58 pfB0 ! (\?0.58 (\?X )\(?0.88)\(?p  0.88 qgB0 " )\(?0.88 )\(?X 333333ӿ-0.30333333ӿqЀ0.30 rhC0 # 333333ӿ-0.30 333333ӿX  A-34.1 ArЀ34.1 YOI0 $  A-34.1 40  AX 4YNativeHawaiianand l % OtherPacificIslander OE!X&X {Gz?0.01{Gz?O  0.01 pfBl ' {Gz?0.01 {Gz?X 333333?0.15333333?p  0.15 qgBl ( 333333?0.15 333333?X Q-0.14QqЀ0.14 rhCl ) Q-0.14 QX 33333SW-93.333333SWrЀ93.3 YOIl * 33333SW-93.3 40 33333SWX 4YSomeotherrace   X+  PF!D,X p= ׃C@39.03p= ׃C@P  39.03 rhCX- p= ׃C@39.03 p= ׃C@X \(I@51.47\(I@r  51.474 siCX. \(I@51.47 \(I@X zG(-12.44zG(sЀ 12.44 siDX/ zG(-12.44 zG(X 3333338-24.23333338sЀ24.2 YOIX0 3333338-24.2 40 3333338X 4YTwoormoreraces   1  OE!2f \(\@7.84\(\@O  7.84 pfB3 \(\@7.84 \(\@f \(\@4.59\(\@p  4.59 pfB4 \(\@4.59 \(\@f  @3.25 @pЀ 3.25 pfB5  @3.25  @f 33333Q@70.833333Q@pЀ70.8XNL6 33333Q@70.8 40  33333Q@f XSource:DerivedfromMartin(2002a:13Table6) 6 h XX\Note: Bold numbersindicatesignificantdifferencesatthep<.05level.#X\X h# 7 '2.4.4.DiscussionofracereportingbyHispanics :9 ThereareseveralsignificantdifferencesinracereportingbyHispanicsinthe2000and1990 b; styleforms,ascanbeseeninTable2.2.First,Hispanicsweremuchlesslikelytoreportas_SOR_ N< (about24percentless),andmuchmorelikelytoreportasWhiteinthe2000styleforms(about := 23percentmore).Theywerealsomorelikelytoselectmorethanonerace(about71percent) &> thaninthe1990style,asexpected.Otherresearch(del_Pinal_,Martin,Bennet,and_Cresce_,  ? 2002:3)showsthatmuchoftheTwoormoreracesreportingbyHispanicsinvolves_SOR_Ԁin !@ combinationwithotherracesasoneoftheraces.Thus,eliminating_SOR_Ԁresponsesreduces "A HispanicreportingofTwoormoreracestoaboutthesamelevelasnonHispanics.B  ԍdd(31page3, )f$H R1)dC  d #B AnotherinterestingfindingisthatHispanicsaremoreV  WԀlikely(about106percentmore) ^% D toreportasAmericanIndianin2000stylethanin1990styleforms,andmuchlesslikely J&!E (about93percentless)toreportasNativeHawaiianandOtherPacificIslander.Butoverall 6'"F thesedifferencesarenotstatisticallysignificant.However,Martin(2002a:13)reportsthatthe "(r#G differenceissignificantinthelowcoveragearea(_LCA_)strata(2.08vs.0.79percent,orabout )^$H a163percentdifference)butnotinthehighcoveragearea(_HCA_)strata.The2000styleform )J%I capturedmoreNativeHawaiianandOtherPacificIslanderresponses,althoughthepercentages *6&J aresmall.Atthispointitmaybeworthremindingreadersthatsmallcategoriesare more +"'K vulnerabletoinaccuraciesduetobothsamplingandnonsamplingerror(_Edmonston_,Goldstein, ,(L and_Tamayo_ԀLott1996:24,39).B  ԍ(29page39,R1;page24R1)C  Indeed,Martin(2002a:5)prominentlylistsamong -(M thelimitationsofthisstudytherelativelysmallsamplesize! sostatisticalinferencesabout  smalldifferencesbetweenforms,orsmallpopulationgroupsmaynotbereliable.B  ԍ(1page5,R1)dC  dInviewof  thislimitationMartins(2002a)findingsareremarkableindeed.  2.4.5.RacereportingbynonHispanics `  0    ReportingofNativeHawaiianandOtherPacificIslanderbynonHispanics:4 (#(# 0  IntheCensus2000stylequestionnaire 0.18percent ofnonHispanicsreported $ t NativeHawaiianandOtherPacificIslandercomparedwith 0.04percent inthe  d 1990stylequestionnaire.B  ԍ(1page14,R1)C    T (#(# 0    ReportingofWhitebynonHispanics: IntheCensus2000stylequestionnaire( (#(#  0  81.15percent ofnonHispanicsreportedWhitecomparedwith 82.43percent 󀀀   inthe1990stylequestionnaire.B  ԍ(1page14,R1)dC  d  (#(# 0   ReportingofTwoormoreracesbynonHispanics: IntheCensus2000style   questionnaire 1.45percent ofnonHispanicsreportedTwoormoreraces |  comparedwith 0.48percent inthe1990stylequestionnaire.B  ԍ(1page14,R1)dC  d l(#(# &   tablesTable2.3 AlternativeQuestionnaireExperiment(_AQE_)RaceResponsesbyNonHispanics  andHispanicOriginNotAscertained tables" #  2        *:X;dd` dd ` Fdd Fdd Fdd Fdd 6X7(#(#,` dd ,Fdd ,dd ,Fdd ,dd +    Race 6,!, ff 62000Style @ (1) 6,!," ff 61990Style @ (2) 6,!," ff 6Difference @  (3=12) 6,!, " ff 6Percent @  Difference ,  (4=3/2) 6,& "   ff 46White@` ` `   h  PF! TX IT@81.15IT@P  81.15 rhC h IT@81.15 IT@X QT@82.43QT@r  82.43 rhC h QT@82.43 QT@X {Gz-1.28{GzrЀ 1.28  zpC h {Gz-1.28 {GzX -1.6zЀ1.6 XNH h" -1.6 40 X 4XBlack@` ` `  T   PF!@ X (\(@12.28(\(@P  12.28 rhCT  (\(@12.28 (\(@X  ףp= (@12.02 ףp= (@r  12.02 qgCT   ףp= (@12.02  ףp= (@X p= ף?0.26p= ף?qЀ0.26 xnBT  p= ף?0.26 p= ף?X @2.2@xЀ2.2 WMGT " @2.2 40 @X 4WAmericanIndianand   AlaskaNative OE!| X RQ?0.38RQ?O  0.38 pfB  RQ?0.38 RQ?X Q?0.48Q?p  0.48 qgB  Q?0.48 Q?X -0.10qЀ0.10 {qC  -0.10 X 4-20.84{Ѐ20.8 YOI " 4-20.8 40 4X 4YAsian    OE! !X (\@4.39(\@O  4.39 pfB " (\@4.39 (\@X \(\@4.34\(\@p  4.34 pfB # \(\@4.34 \(\@X ?0.05?pЀ0.05 xnB $ ?0.05 ?X 333333?1.2333333?xЀ1.2 WMG %" 333333?1.2 40 333333?X 4WNativeHawaiianand X& OtherPacificIslander OE!D'X  ףp= ?0.18 ףp= ?O  0.18 pfBX(  ףp= ?0.18  ףp= ?X {Gz?0.04{Gz?p  0.04 pfBX) {Gz?0.04 {Gz?X Q?0.14Q?pЀ 0.14 zpBX* Q?0.14 Q?X u@350.0u@zЀ350.0 YOIX+" u@350.0 40 u@X 4YSomeotherraceV   D, WG3 OE!0-X (\?0.17(\?O  0.17 pfBD. (\?0.17 (\?X ?0.20?p  0.20 qgBD/ ?0.20 ?X Q-0.03QqЀ0.03 {qCD0 Q-0.03 QX .-15.0.{Ѐ15.0 YOID1" .-15.0 40 .X 4YTwoormoreracesV     2 W5 OE!l3f 333333?1.45333333?O  1.45 pfB4 333333?1.45 333333?f Q?0.48Q?p  0.48 pfB5 Q?0.48 Q?f  ףp= ?0.97 ףp= ?pЀ 0.97 zpB6  ףp= ?0.97  ףp= ?f 33333Ci@202.133333Ci@zЀ202.1YOM7" 33333Ci@202.1 40  33333Ci@f YSource:DerivedfromMartin(2002a:14Table7) 7 h XX\Note: Bold numbersindicatesignificantdifferencesatthep<.05level; BoldItalic numberindicatessignificant t8 differencesatthep<.10level.#X\X h8# :9 'l"2.4.6.DiscussionofracereportingbynonHispanics :; ThereareseveralsignificantdifferencesinracereportingbynonHispanicsandrespondents = whodidnotreportaHispanicorigininthe2000and1990styleforms(seeTable2.3).First, > nonHispanicswereslightlylesslikelytoreportasWhite(about1.6percentless),andmuch  ? morelikelytoreportasPacificIslanderinthe2000styleforms(about350percentmore). !@ Asexpected,andsimilartoHispanics,nonHispanicswerealsomorelikelytoselectmore r"A thanonerace(about202percentmore)inthe2000styleform. ^#B Martin(2002a:14)explainstheslightlylowerreportingofWhiteamongnonHispanicsin 6% D 2000styleformsasaneffectoftheoptionofreportingmorethanonerace,yettherewasno "&r!E measurabledownwardeffectonothercategories.B  ԍ(1page14,R2)C  ԀIfthispropositionistrue,itsuggeststhat '^"F  peopleofmorethanoneracetendtoreportasWhitewhenonlyoneraceresponseisallowed, 'J#G butreportasTwoormoreraceswhenmultipleraceresponsesareallowed.InalatersectionI   examinethepropensitytoreportWhiteamongrespondentswhoreportmorethanonerace,  whichmayshedlightonthisissue.  Anotherinterestingfindingisthat contrarytowhatmighthavebeenexpected,thereislittle ` evidencethatallowingrespondentstoreportmorethanoneracereducedthesingleracereporting L  inthe5V  ԀW@majorracecategories(Martin,2002a:iv).B  ԍ(1pageiv,R5)C  ԀThismayallaysomefearsamongthosewho 8  thoughtthatthereportedsizeofsomeminoritycategoriesmaybesmallerbecauseofthe $ t reportingofmorethanonerace.However,onereasonthatthenonWhitecategoriesappearnot  ` tobeasaffectedisthatthe1990styleformsalsohadsome(0.82percent)respondentsreport  L  morethanoneracedespitetheinstructiontoreportone.B  ԍ(1page12,R1)C  ԀThesemultipleresponseswouldhave  8  beeneditedintoasingleracecategoryin1990.Inaddition,almostonethird(30.5percent)of $  Hispanicsdidnotreportarace,soitisunknownhowtheirresponseswouldhaveimpactedthe   results.B  ԍ(1page11,R1)dC  dTheactualeffectinpublishedracedatamaybeaffectedbyhowtheseresponsesare   allocated.Inaddition,asMartin(2002a:5)remindsus,thesefindingsare_generalizable_Ԁonlyto   theCensus2000_mailout_Ԅ_mailback_Ԁuniverse.B  ԍ(1page5,R2)dC  d     2.5.OverallHispanicoriginreporting  F \ AccordingtoMartin(2002a:v), despitethereversedsequenceofHispanicoriginandrace 4 questionwordingdifferences,thesamepercentage(slightlyover11.1percent)reportedas  p Hispanicinbothforms.B  ԍ(1pagev,R1)C  Ԁ  \ Martin(2002a:7)reportsthatboththe2000and1990styleformsyieldednearlyidentical 4 proportionsofHispanicrespondents!about11percent.However,thehighratesofmissing   datacreateuncertaintyabouttheoverallpercentageofHispanicsidentifiedbyeachform.   Ontheotherhand,theproportionofnonHispanicsinthe2000styleformwasabout85percent  comparedtoabout74percentin1990styleforms.Theremainingdifferenceisduetopeople  whodidnotrespond!about3percentdidnotrespondin2000styleformscomparedtoabout  14percentin1990styleforms.B  ԍ(1pagev,R1)C   l 2.5.1.DiscussionofoverallHispanicoriginreporting D! Asdiscussedabove,ourpreviousresearchsuggeststhatinthepastnonHispanicsweremuch #l! morelikelytoomitansweringtheHispanicoriginquestion.Martin(2002a:7)concludesthat $X"  underthisassumption,theresultssuggestthe2000stylequestionnairedidnotaffectreporting $D # asHispanic,excepttoreducethenumberofnonHispanicswhowouldhavelefttheitemblankin %0!$ a1990stylequestionnaire.Theultimatedistributionaleffectwoulddependonhowthemissing &"% dataareeditedandimputed.B  ԍ(1page7,R2)dC  dMartin(2002a:7)notesthatthe differenceinratesofmissingdata '#& isverylarge,andwasexpectedbasedonprevioustestsofeffectsofitemsequenceandanadded (#' instruction.B  ԍ(1page7,R3)C  Ԛ )$( ThisfindingisveryimportantbecauseoftheconcernsthatsequencingHispanicoriginaheadof h+&* racemighthavetheeffectofartificiallyinflatingthenumberofHispanics(Petersen,1987:207).B  ԍ  Ўdd(25page207,R1)dC  d T,'+ TheequalproportionsofHispanicsinthe2000styleand1990styleforms(about11percent) @-(, stronglysuggestthatthiswasnotthecase.ThissupportsMartins(2002:v)conclusionthat   anychangesfrom1990to2000inthefractionofthepopulationidentifyingasHispanicarenot  duetochangesinthedesignofmailquestionnaire.However,tFX\XXX\herewerequestionnaireeffects  onreportingadetailedHispanicorigin,asIdiscussnext.B  ԍ(1pagev,R2)dC  dWeshouldnotethatthehighratesof t missingdatacreateuncertaintyabouttheoverallfractionofHispanicsthatwouldbeidentifiedby ` eachformafterthedatawerefullyeditedandimputed.#X\XXFX\Q# L    2.6.DetailedHispanicoriginreporting eS $ t AccordingtoMartin(2002a:v), the2000stylequestionnaireselicitedfewerreportsofspecific  L  Hispanicgroups,andmorereportsofgeneralHispanicidentity(e.g.,Hispanic,Latino,Spanish)  8  thanthe1990stylequestionnaires.B  ԍ(1pagev,R3)dC  d $  Martins(2002a:10)_AQE_Ԁfindingspointoutthat about92percentofHispanicsreporteda   specificgroupin1990styleforms,comparedwith80percentwhofilledout2000styleforms. #  3      ׀   Martin(2002a:10)brokeouttheseresponsesintofivecategories(showninTable2.4below),   andreportedthatthe2000styleformsproducedmoregeneralornonspecificHispanicresponses p (e.g., Hispanic, Latino, Spanish;or OtherHispanicwithoutprovidingawritein \ response)andfewerspecificgroups( checkboxgroups,T #  4      ׀ examplegroups, #  5      ׀andallother H specificnationalorigingroups).B  ԍdd(1page10,R1)dC  dTable2.4summarizesthefinal_AQE_ԀresultsregardingHispanic 4 subgroupreporting.  p    \  & f  tablesTable2.4 AlternativeQuestionnaireExperiment(_AQE_)DetailedHispanicOriginResponses  byFormType tablesZ  iZ*@XAdd` dd ` Fdd Fdd Fdd Fdd :X;(#(#, dd ,d ,Fd ,Fd ,d +    Race l 6,!, ff 62000Style @ (1) 6,!," ff 61990Style @ (2) 6,!," ff 6Difference @  (3=12) 6,!, " ff 6Percent @  Difference ,  (4=3/2) ?5& "   ff 4?Totalpeopleidentified  h asHispanic(percent) ZP! TX Y@100.00Y@ZЀ100.00 }sD h3 Y@100.00 Y@X Y@100.00Y@}Ѐ100.00 YOD h3 Y@100.00 Y@X Y  -#! h3X - 7-' h 40 X 47 Checkboxgroups T   YO!@ X Q@70.25Q@Y @ 70.25 {qCT 3 Q@70.25 Q@X QNR@73.23QNR@{Ѐ73.23 {qCT 3 QNR@73.23 QNR@X ףp= -2.98ףp= {Ѐ2.98 zpCT 3 ףp= -2.98 ףp= X ffffff-4.1ffffffzЀ4.1 XNHT 3 ffffff-4.1 40 ffffffX 4XMexican,Mexican   American,Chicano YO!| X zG!K@54.26zG!K@Y @ 54.26 {qC 3 zG!K@54.26 zG!K@X ףp= WM@58.68ףp= WM@{ @1 58.68 {qC 3 ףp= WM@58.68 ףp= WM@X Gz-4.42Gz{Ѐ 4.42 zpC 3 Gz-4.42 GzX -7.5zЀ7.5 XNH 3 -7.5 40 X 4X Examplegroups XN! X p= ף@6.41p= ף@X @ 6.41 zpB !3 p= ף@6.41 p= ף@X RQ&@11.16RQ&@z @1 11.16 {qC "3 RQ&@11.16 RQ&@X -4.75{Ѐ 4.75   # @@: {qC $3 -4.75 X LE-42.6LE{Ѐ42.6 YOI %3 LE-42.6 40 LEX 4YAllotherspecific  \& Hispanicgroups XN!H'X @4.20@X @ 4.20 yoB \(3 @4.20 @X \(\!@8.68\(\!@y @1 8.68 zpB \)3 \(\!@8.68 \(\!@X Q-4.48QzЀ 4.48 {qC \*3 Q-4.48 QX I-51.6I{Ѐ51.6 YOI \+3 I-51.6 40 IX 4YWriteingeneraldescriptor H, h XX\( Hispanic/ Latino/ Spanish)#X\X hak# YO!4-X '@11.90'@Y @ 11.90 zpCH.3 '@11.90 '@X ffffff?1.90ffffff?z @1 1.90 zpBH/3 ffffff?1.90 ffffff?X $@10.00$@zЀ 10.00 {qCH03 $@10.00 $@X fffffr@526.3fffffr@{Ѐ526.3 YOIH13 fffffr@526.3 40 fffffr@X 4YOtherHispanic,nowritein \2  XN!H3f @7.25@X @ 7.25 yoB\43 @7.25 @f Q@5.03Q@y @1 5.03 yoB\53 Q@5.03 Q@f (\@2.22(\@yЀ 2.22 yoB\63 (\@2.22 (\@f  F@44.1 F@yЀ44.1XNL\73  F@44.1 40   F@f XSource:DerivedfromMartin(2002a:9Table9) d7 h XX\Note: Bold numbersindicatesignificantdifferencesatthep<.05level .#X\X hq#'fzZ P8 Thelargestdifferencebetweenthe2000styleand1990styleformsistheproportionofgeneral R: Hispanicresponses( Hispanic, Latino,and Spanish).The2000styleformsproduced >; 10percentagepointsor526percentmoreoftheseresponsesthandidthe1990styleforms. *< Similarly,the2000styleformalsoproducedanother2.22percentagepointsor44percentmore =  OtherHispanicresponseswithnowritein(seeTable2.4).Ontheotherhand,the2000style > formsproducedfewerspecificHispanicgroupsthanthe1990styleforms.The2000styleform  ? hadabout43percent(4.75percentagepoints)feweroftheexamplegroupsandabout52percent !@ (4.48percentagepoints)fewerofthespecificnonexamplegroups.Although2000styleforms v"A had4percent(2.98percentagepoints)fewerspecificcheckboxgroupsoverallthandidthe b#B 1990styleform,thatdifferencewasnotstatisticallysignificant.However,whencompared N$C separately,theMexicanorigincheckboxgroupwas7.5percent(4.42percentagepoints)lower :% D inthe2000styleforms,andthatdifferenceisstatisticallysignificant. &&v!E Martin(2002a:10)concludes:B  ԍ(1page16,R1)C  Ԛ 'N#G 0  ...theexperimentdoesofferevidencethatthequestionnaireaffectedreportingofdetailed )&%I Hispanicorigin.Hispanicswhofilledout2000stylemailquestionnaireswerelesslikely *&J toreportaspecificHispanicgroupandmorelikelytoreportageneraldescriptor(suchas +&K  Hispanic,Latino,orSpanish)thanthosewhofilledout1990stylequestionnaires. ,'L Althoughthecauseoftheeffectisuncertain,itisprobablyduetothecombinedeffectof  questionwordingandtheeliminationofexamplesintheCensus2000questionnaire.(#(#  2.6.1.DiscussionofdetailedHispanicoriginreporting t Arethe_AQE_Ԁresultsjustaflukeoristhereotherevidenceofdifferencesinreporting?Ibelieve L  the_AQE_ԀresultsfordetailedHispanicreportingdo,infact,V  ԀW{explainmuchofwhatwasnoticed 8  fromtheCensus2000data.InareportabouttheHispanicpopulationfromCensus2000, $ t _Guzmn_Ԁ(2001:2)alsonotedthat 17.3percent(6.1million)ofthetotalHispanicpopulation  ` didnotgiveaspecificnationalorigingroup;andtheseresponses weresecondinsizebehind  L  thepopulationthatreportedMexicanorigin.B  ԍ(33page2,R1)C    8  AsadditionalinformationfromCensus2000becameavailableatmorelocallevelsduringthe   summerof2001,communityadvocates,journalistsandresearchersnotedunexpectedlylow   numbersofspecificHispanicgroups.Accordingto_Suro_Ԁ(2002:3),twocompetingexplanations   emerged: eitheralargenumberofpeoplehadchosentoidentifythemselveswithabroadethnic   designation,suchasHispanicorLatino,ratherthanaspecificnationalorigin,suchasDominican p orB  ԍdd(36page3,R1)dC  dSalvadoran,ortheseresultswereaproductofchangesinthewaythecensusquestionnaire \ askedaboutHispanicorigin.B  ԍ(36page4,R1)dC  d H eAfterexaminingtheHispanicorigindata,Logan(2002:3,4)concludedthat Census2000didan  p excellentjobofcountingHispanics,butperformedpoorlyinidentifyingtheirorigin.B  dd(35page3, H R1)C  ddAmongthe  \ likelycauses,Logannotedthat noexamplesanda changeinwordingofthequestionitself H inCensus2000B  ԍ(35page3,R2,R3)C  resultedin asevereunderestimateofthenumbersofspecificHispanicgroupsin 4 2000.B  ԍ(35page4,R1)C  Loganalsonotedamoredramaticeffectinstatesandmetropolitanareaswithlarge   concentrationsofspecificHispanicB  ԍ(35page H 4,R2)dC  dgroups.AnotherreasonLogan(2001:4)hasforfindingthe   Census2000resultsimplausibleisacomparisonwiththeCensus2000SupplementarySurvey  (C2SS).IntheC2SS(whichwastakenthesameyear),about9.6percentofHispanicsdidnot  reportaspecificnationalorigin,comparedwithabout17.6percentinCensus2000.B  ԍ(37page4,R1)dC  dSimilarly,  _Suro_Ԁ(2002:8)findsthedistributionofspecificHispanicgroupsmoreplausibleinC2SS.B  ԍdd(36page8,  R1)dC  d  l    `  eBRespondingtocomplaintsfromcommunitygroups,localgovernmentofficials,andresearchers, D! membersofCongressaskedtheU.S.GeneralAccountingOffice(GAO)tolookintotheissue.B  ԍ D! (22page1,R1)C  Ԛ 0"  TheGAO(2003a:1)reportexpressesconcernsthatthe deletionofHispanicsubgroup #l! examplesfromtheCensus2000questionnairewasthecauseoflowerthanexpected $X"  countsofDominicansandotherHispanicsubgroups.B  ԍ(22pagei,R1)dC  dGAOconcludedthatfor Census2000, $D # the[Census]Bureauremovedthesubgroupexamplesaspartofabroaderefforttosimplifythe %0!$ questionnaireandhelpimproveresponserates.B  ԍ(22pagei,R1)dC  dGAO(2003a:14),asnotedabove,foundthat &"% early_AQE_ԀresultsandC2SSdataseemedtoindicateaproblemwiththeCensus2000detailed '#& Hispanicdistribution.B  ԍ(22page14,R1;R2)dC  d    h    (#' WhilethedebateabouthowtoidentifytheHispanicpopulationdatesbacktothe1960s |*%) (see_Choldin_,1986:403),B  ԍ(22page403,R1)dC  dasseenabove,theissueinCensus2000isthedistributionby h+&* specificnationalorigingroups._Choldin_Ԁ(1986:404)notedthat nationalstatisticsmustchange T,'+ inresponsetosociopoliticalchangesandthat theroleofthestatisticianisnotsimplyscientific, @-(, butisalsoconditionedbyeventsinthepoliticalenvironment.B  ԍ(22page  404,R1)dC  dHowever,upuntilCensus2000,  thedifferentialundercountwascontroversial,notthedistributionofspecificgroups.  2.6.2.AnalysisofgeneralHispanicresponsesinCensus2000 `   AttherequestofmembersofCongress,B  ԍ(38page1,R1)dC  dtheCensusBureauundertookthetaskofusing 8  informationonplaceofbirthandancestryfromtheCensus2000longformtosupplementthe $ t generalHispanicoriginresponses(_Cresce_ԀandRamirez,2003).Thesenewestimates donot  ` fullyreflectselfidentificationandarenotmeanttoreplacetheofficialCensus2000figures.  L  B  ԍ(8  8  page1,R1)dC  dStill,this simulationproducedinterestingresults:ofanestimated5.7millionindividualswho  8  providedageneralHispanicresponse,54percent(3.1million)alsoprovidedmoreinformation $  abouttheirspecificoriginineitherplaceofbirthorancestry.Thatleftabout2.6million   individualswhogavenoadditionalinformationabouttheirspecificHispanicorigin(_Cresce_Ԁ   andRamirez,2003:9).B  ԍ(8page9,R1)dC  d   Thissimulationsuggeststhateverysinglespecificcategory(checkboxandspecificwritein p groups)couldbeincreasedusingadditionalinformationfromplaceofbirthandancestry \ (seeTable2.5).Thesimulationincreasestheproportioninallspecificgroupsfrom84percent H to93percentofallHispanics,anincreaseofabout10percent.However,examplegroups 4 increasedby35percent(2.4percentagepoints)andotherspecificgroupsincreasedby  p 28percent(1.4percentagepoints).Andbydesign,thegeneralresponsesdeclinedby  \ 52percent(5.1percentagepoints)andtheothernonspecificresponsesdeclinedby57percent H (3.3percentagepoints).TheseresultsareverysimilartothoseofMartin(2002),Logan(2001), 4 and_Suro_Ԁ(2002),whichalluseslightlydifferentdatasourcesandmethods._Cresce_ԀandRamirez   (2003:19)specificallycomparethesimulationtotaltoLogan(2001)and_Suro_Ԁ(2002),andfind   thattheformerovershootsandthelatterundershootsthesimulationtotals.      `    &   tablesTable2.5 ComparisonofSpecificHispanicOriginDistributionsFromCensus2000  LongFormsandSimulatedTotalsUsingSupplementalInformationonPlaceofBirth  andAncestry tablesɚ  *CXDdd dd d Fd FFd Fd @XA(#(#," 44 ,F44 ,44 ,e44 ,44 +  xx  Race 6,! ff 6Census2000 , LongForm  (1) 6,! " ff 6Simulated ,  Totals   (2) 6,!  " ff 6Difference ,  (3=21) 6,! " ff 6Percent , Difference  (4=3/1) ?5& "   ff 4?Totalpeopleidentified  T asHispanic(percent) YO! @X Y@100.0Y@YЀ100.0 {qC T Y@100.0 Y@X Y@100.0Y@{Ѐ100.0 OEC T Y@100.0 Y@X O  -#! TX - 7-' T 40 X 47 Checkboxgroups!total @   OE!,| X  R@72.5 R@OЀ72.5 pfB@   R@72.5  R@X LS@77.2LS@pЀ77.2 oeB@  LS@77.2 LS@X ffffff@4.6ffffff@oЀ4.6 ndA@  ffffff@4.6 ffffff@X @6.4@nЀ6.4 `VG@  @6.4 40 @X 4`'lA  XX'Mexican,Mexican |  American,Chicano OE!h "X fffffM@59.3fffffM@OЀ59.3 pfB|  fffffM@59.3 fffffM@X 33333O@63.433333O@pЀ63.4 oeB|  33333O@63.4 33333O@X ffffff@4.1ffffff@oЀ4.1 ndA| ! ffffff@4.1 ffffff@X @6.9@nЀ6.9 WMG| " @6.9 40 @X 4W Examplegroups ND! #X @6.9@N ^ 6.9 ndA $ @6.9 @X ffffff"@9.2ffffff"@n  9.2 ndA % ffffff"@9.2 ffffff"@X 333333@2.4333333@nЀ2.4  & @EE<E oeA ' 333333@2.4 333333@X YA@34.7YA@oЀ34.7 XNH ( YA@34.7 40 YA@X 4XAllotherspecificHispanicgroups D)  ND!0*X @4.9@N ^ 4.9 ndAD+ @4.9 @X 333333@6.3333333@n  6.3 ndAD, 333333@6.3 333333@X ffffff?1.4ffffff?nЀ1.4 oeAD- ffffff?1.4 ffffff?X ffffff<@28.4ffffff<@oЀ28.4 XNHD. ffffff<@28.4 40 ffffff<@X 4XWriteingeneraldescriptor 0/ ( Hispanic/Latino/Spanish) ND!l0X #@9.9#@NЀ9.9 ndA01 #@9.9 #@X 333333@4.8333333@n  4.8 oeA02 333333@4.8 333333@X ffffff-5.1ffffffoЀ5.1 qgB03 ffffff-5.1 ffffffX fffffI-51.8fffffIqЀ51.8 YOI04 fffffI-51.8 40 fffffIX 4YOtherHispanicnospecific l5  ND!X6f 333333@5.8333333@N ^ 5.8 ndAl7 333333@5.8 333333@f @2.5@n  2.5 oeAl8 @2.5 @f ffffff -3.3ffffff oЀ3.3 qgBl9 ffffff -3.3 ffffff f @L-56.5@LqЀ56.5YOMl: @L-56.5 40  @Lf YSource:Derivedfrom_Cresce_ԀandRamirez(2003:11Table6) t: '       h 2.6.3.DiscussionofgeneralHispanicresponsesinCensus2000 L< _Cresce_ԀandRamirez(2003:7)B  ԍ(8page8,R1)dC  dlistseverallimitationstotheirsimulationanalysis,someofwhich $t> alsoapplytotheresearchbyLogan(2001)and_Suro_Ԁ(2002).Theseanalysesonlyaddtospecific `? groupsbysubtractingfromthegeneralgroups,anddontusecontradictoryinformationtoreduce L@ specificgroups.AllthreeanalysesassumethetotalHispanicpopulationiscorrectanddonot  8A addorsubtractfromthattotal.TheMartin(2002)analysisdoesnothavethislimitation,butis !$B limitedtoarelativelysmallsampleofmailreturns.While_Cresce_ԀandRamirez(2003)use "C Census2000longformdata,Logan(2001)modelsthedistributionofspecificgroupswith #D CurrentPopulationSurvey(CPS)data,and_Suro_Ԁ(2002)usesC2SSdata.Foramoredetailed $E discussionofthesedifferences,see_Cresce_ԀandRamirez(2003:78,1920).B  ԍ(8page % F 7,8,19,20,R1)C  ԚAllofthesestudies % F seemtoindicatethattheobservedchangesinthedistributionofCensus2000detailedHispanic p&!G groupscomparedwithchangesseeninothersourceswerenotdueentirelytoashiftinhow \'"H peopleofHispanicorigindefinethemselves,butrathertosomeproductofthechangesinthe H(#I wayweaskedtheHispanicoriginquestion.Weareleftwiththequestionofwhetherthe 4)$J eliminationofexampleswastheprobablecauseofthereportingdifferencesindetailedHispanic  *p%K groups.  +\&L   +H'M [    &   2.6.4.ConclusionsaboutdetailedHispanicresponsesinCensus2000  IndiscussingthereportingchangesinHispanicgroups,Martin(2002a:16)speculates:   0  Althoughthecauseoftheeffectisuncertain,itisprobablyduetothecombinedeffect ` of'questionwordingandtheeliminationofexamplesintheCensus2000questionnaire. L  Theexamplesnexttothewriteinboxprovidedcuesaboutthetypeofanswerintended 8  bythequestioninthe1990styleform.IntheCensus2000questionnaire,theinstruction $ t to printgrouprightafterthe Yes,otherSpanish/Hispanic/Latinoresponsecategory  ` mayhavesuggestedtosomerespondentsthattheyshouldprintwhicheverofthesethree  L  termstheypreferred. 8 (#(# Althoughtheeliminationofexamplesiscommonlyassumedtobethemaincauseofthis   problem(seeGAO2003:2forexample),Martin(2002a:16)arguesthat thehypothesisof   exampleeffectsdoesnotaccountforthehigherreportingofMexicansinthe1990styleform.   Thisdifferencerequiresadifferentexplanation,becausethespecificexamples(Mexican,   MexicanAm.,Chicano)areidenticalinbothforms.B  ԍ(1page16,R2)dC  dSimilarly,theanalysisby_Cresce_Ԁand p Ramirez(2003:11)suggeststhatallcheckboxgroups(Mexican,PuertoRican,andCuban)may \ havebeenaffected,whichalsoarguesthatsomethingotherthanremovingexampleswasatwork. H Martin(2002a:16)goesontoarguethat:B  ԍ(1page16,R3)C  Ԛ  p 0  Thewordingchangefrom IsthispersonofSpanish/Hispanicorigin?to Isthisperson H Spanish/Hispanic/Latino?mayhavecontributedtothereportingdifference.TheCensus 4 2000questionappearsdirectedtoanoverarchingidentificationasHispanic(orSpanishor   Latino),andtheabsenceofspecificHispanicexampleswouldreinforcethiswording   effect.(#(# ReflectiveoftheissueofexamplesastheymayhaveaffectedbothHispanicandracereporting,  Martin(2002b:4)notes:B  ԍ(7page4,R4)C  Ԛ l EXXXX\0  TheapparentcontrastbetweentheeffectsofexamplesintheHispanicoriginandrace D! itemsmeritsfurtheranalysisandconsideration.Theexamplesinthe1990Hispanic 0"  originquestionmayhaveservedtoclarifythattheintentofthequestionwastocollect #l! detailedHispanicorigin,whiletheracequestionmaynothavesufferedfromthesame $X" ambiguity,hencemaynotneedexamples.Inaddition,theexampleswereplaced $D # differentlyinthetwoquestions.Inthe1990form,theHispanicexampleswere %0!$ prominentlyplaced,justbelowthe otherSpanish/Hispanicresponseoption,abovethe &"% writeinspace.Theraceexampleswereofftotheleft,belowthequestionandremote '#& fromthewriteinspace,wheretheywerelesslikelytobeseenthantheHispanic (#' exampleswere....Thisdifferenceinplacementwouldlikelyreducetheirimpactinthe )$( raceitemcomparedtotheHispanicoriginitem.#X\XXEX=#|*%)(#(# Ѐ  T,'+ Unfortunately,asMartin(2002a:16)suggests, theexperimentwasdesignedtoevaluatethe   effectsofallthewordinganddesigndifferencesbetweenthe1990and2000mailquestionnaires,  itisnotwellsuitedtoisolatingthecausesforthisorotherdifferences.B  ԍ(1page16,R4)dC  dWespeculatethatin  effectwechangedthe senseoftheHispanicoriginquestionbyremovingexamples,dropping t  originfromthequestion,usingthreegeneraltermsseparatedbyslashmarks ` (Spanish/Hispanic/Latino),andusingawriteininstruction( Yes,otherSpanish/Hispanic/Latino L  !Printgroup)thatseemstorequestonetermshouldbeprinted.Allofthesecombinedchanges 8  mayhavecausedrespondentstoselectamongthetermslisted(orevenrejecttheseterms)rather $ t thanreporttheirspecificorigin.Inlatersections,Iwillpresentotherevidencetosupportthis  ` contention.Inanycase,the_AQE_Ԁprovidesthemostimportantandtellingevidencetodateon  L  theeffectofquestionnairechangesinCensus2000.  8  AsreflectedintheGAO(2003a:10)report,neitherthe1997OMBrevisionstoDirectiveNo.15   norPublicLaw94311requireustocollectdataondetailedHispanicgroupsbutwehavedone   sointhebestefforttogetanaccurateoverallcountoftheHispanicpopulation.B  ԍ(22page10,R1)dC  dAllevidence   pointstotheachievementofthisgoalinCensus2000.B  ԍ(22page11,R1)dC  dHowever,thefactwepublishdataon   detailedHispanicorigingroupsindicatestodatausersthatwehavesomeconfidenceinthe p accuracyofthereporteddata.AsGAO(2003a:3)summarizedtheissue, whilethe[Census] \ Bureaureported whatrespondentsmarked ontheirquestionnaires,becauseof confusion over H thewordingofthequestion,thesubgroupdatacouldbemisleading[ emphasisadded ].B  ԍ(22page3,R1)dC  dItmay 4 nolongerbepossibleforusmerelytopublishwhatrespondentsprovidedwithoutathorough  p assessmentofthedataandadecisionprocessaboutwhethertopublishornot.However,the  \ publicdemandforcensusdata,nomatterhowflawedorinconclusive,maygiveusnorecourse H buttomakethedataavailable.Thispointiswellillustratedbythedemandforgroupquarters 4 (seeGAO,2003b)B  ԍ(21   page3,R1)dC  dandadjustmentdata(seeGAO,2003c)fromCensus2000.B  ԍ(21page3,R1)C  Ԛ   WhilerespondentconfusionmayplayaroleinproducingdifferencesindetailedHispanic  reporting,itisalso  .4  likelythatourinstructionswerenotclearincommunicatingwhatwewanted  fromtherespondent.AsMartin(2002c:592)remindsus, questionnairechangesthatseem  minorcanhaveimportanteffectsonourdata.Therefore,weneedto pretestandevaluateall l questionnairechanges,andalthoughwedidconducttestspriortothechangesinthecensus X  questionnaire, perhapsthetestdesignand samplesize werenotadequatetodetectanyeffects D! thatwouldilluminatethesecomplexandimportantissues.ItseemsthatX  ԍYԀaninadequateandsmall 0"  samplesize,inparticular,maylimitourabilitytodetecttheeffectofchangesV  .W4ԀIwilladdress #l! theseconcernsinsubsequentsectionsofthisreport.V  W TheGAOreportemphasizestheneedfor $X" furtherimprovementsinthequalityofdetailedHispanicdata,andhighlightstheneedfor $D # consistencyamongdatasetsinthisregard. %0!$ 2.6.5.ReportingofdetailedAsianandPacificIslanderresponsesinCensus2000 '#& GiventheconcernabouttheeffectsofdroppingexamplesonthereportingofspecificHispanic )$( groups,Martin(2002b:1)undertookanexaminationofthe_AQE_Ԁdatatoseehowthechangesin |*%) thequestionnaireaffectedthereportingofspecificracegroups.B  ԍdd(7page1,R1)dC  dLookingfirstattheraceexample h+&* groups(_Hmong_,Fijian,Laotian,Thai,Tongan,Pakistani,andCambodian)takenasawhole, T,'+ Martin(2002b:2)foundastatisticallysignificantdifferenceinthereportingofthesespecific @-(, groups.However,the2000styleform, whichdidnotlistexamples ,showedahigherproportion  oftheseexamplegroupsthanthe1990styleform(0.356vs0.106percent).B  ԍ.dd(7  page2,R1)dC  dMartin(2002b:3)  alsonotesthat inEXXXX\general,the2000styleformelicitedmorereportsofboththeAsianandthe  PacificIslanderexamplegroups,althoughonlytheoveralldifferencesforAsiansandforPacific t Islandersarestatisticallysignificantatthe.05level.Onedifficultywiththeanalysiswasthat ` therewerenoresponsesofspecificPacificIslandergroupsinthe1990styleforms(seeMartin L  2002b:3,Table2),B  dd(7page3,R3)C  ddindicatingthatthissamplemayhavebeentoosmalltoconcludeanything 8  aboutexampleeffectsinthiscase.#X\XXEXH#Martin(2002b:3)alsonotesthatEXXXX\alargersampleisneeded, $ t butpointsoutthat thedifferenceisconsistentforallthegroups,andmarginallysignificantfor  ` several(t>1.645issignificantatp<.10witha2tailedtest),despiteverysmallcellfrequencies.  L  AdditionalresearchontheuseofexamplesisaddressedbyMartin,Gerber,andRedline(2003).#X\XXEXD#EXXXX\Ԁ  8  #X\XXEX#EXXXX\ #X\XXEXs#AmongMartins(2002:34)otherfindingswasthediscoverythattherewasnodifferencein   overallreportingoftheAsiancategory(4.04percentin2000styleand4.06percentinthe   1990styleforms),B  ԍdd(7page3,R2)dC  dbutthereweresignificantlymorePacificIslandersinthe2000styleforms   (0.17percentvs0.05percent).B  ԍ.dd(7page4,R3)dC  dMartin(2002b:4)concludesthatthe EXXXX\resultsdonotindicatethat   droppingtheexampleshadanynegativeeffectsonreportingofthe[AsianandPacificIslander] p examplegroupsin2000styleforms,#X\XXEX#EXXXX\Ԁbutthat differencesinreportingprobablyarisefromother \ designfeaturesofthequestionnaire,andareprobablynota(perverse)effectofexamples.B  ԍ#X\XXEX#dd(7page4,R1)C  ddEXXXX\ H Martins(2002b:4)preliminaryconclusionsareasfollows:#X\XXEX/#B  ԍ(7page4,R2)C  ԚEXXXX\  p #X\XXEX#EXXXX\0  OtherquestionnairefeaturesareprobablyinfluencingtheresultsforPacificIslanders, H inparticular,splittingtheAPI#X\XXEX@#EXXXX\Ԁ[AsianandPacificIslander]categoryintotwoseparate 4 categories[ Asian; NativeHawaiianandOtherPacificIslander].ThePacificIslander   categoryisprobablymorepopulatedin2000styleformsbecauseitiseasierforPacific   IslanderstoreportwhenthePacificIslanderboxesaregroupedtogetherratherthan  interspersedamongAsianboxes,astheyareinthe1990styleform,andwhenthey  havetheirown OtherPacificIslanderresponseboxassociatedwithawriteinspace.(#(# 0  BothAsianandPacificIslanderrespondentsmayhavebeenconfusedbythelabel X   OtherAPIusedinthe1990styleform,whichrequirescloseattentionandskilled D! readingtodecode,andwhichmayhavecontributedtothedifferenceinwriteinsof 0"  examplegroups.Ihavenotyetexaminedwhethertherearealsoformdifferencesin #l! writeinsofnonexampleAsiangroups,whichmightshedlightonwhethertherevisions $X" madetothe2000styleformsledtoageneralincreaseinwriteinsofspecificAsian $D # groups.#X\XXEX!#%0!$(#(#   '#&  ;XX\Dg[Rr rttr rP>P4P1TQ  3.Census2000ContentReinterviewSurvey(_CRS_)#X\X;#          h     Contentreinterviewsurveysconductedduringdecennialcensuseshavetraditionallybeenan  importanttoolinassessingthequalityofcensusdata(Thomas,_Dingbaum_,and_Woltman_,  1993:5).B  ԍ(40page5,R1)dC  dTheCensus2000ContentReinterviewSurvey(_CRS_)isnoexception(Singerand r _Ennis_,2002).ThepurposeoftheCensus2000_CRS_Ԁwastoevaluatetheconsistencyofresponses ^  toCensus2000throughareinterviewofasampleofrespondents.B  ԍdd(2pagexxi,R1)dC  dAsummaryoftheCensus J  2000_CRS_Ԁfindingsfollows(Singerand_Ennis_,2002). 6    3.1.Studydesign d  ^  The_CRS_Ԁrandomlyselected30,000householdsthatwerescheduledtoreceivetheCensus2000 6  longform.Uponreceiptofthelongformfromthesehouseholds,theybecameeligiblefora "  reinterview.The_CRS_Ԁrandomlychoseonesamplepersonfromeachhouseholdtobe   _reinterviewed_Ԁviaphone(fromtherostercollectedatthebeginningofthe_CRS_)byan   experiencedcensusfieldrepresentative.Ifarespondentcouldnotbereachedbyphone,   apersonalvisitinterviewwasattempted.B  ԍ(2pagexxi,R2)dC  dAbout78.2percentofinterviewswereconducted  bytelephoneand21.5percentbypersonalvisit;theremaininginterviewsutilizedbothmodes, n orthemodecouldnotbedetermined(Singerand_Ennis_,2002:3).B  ԍ(2page3,R1)dC  d Z Theprimarygoalofthe_CRS_ԀwastoevaluatethequalityofdatacollectedinCensus2000using 2 simpleresponsevarianceasmeasuredbytheindexofinconsistency(Singerand_Ennis_,2002:1).B  ԍ n Ўdd(2page1,R2)dC  d n Adiscussionofinterpretingtheindexofinconsistencyappearsbelowinsection3.3.2.  Z Whiletheindexofinconsistencyisapointestimate,thelevelofinconsistencywasconsidered F lowiftheindexwaslessthan20,moderateifbetween20and50,andhighifgreaterthan50. 2 Alowlevelofinconsistencyforanitemwasinterpretedasmeaningthatthereis usuallynota  majorproblem,amoderatelevelas somewhatproblematic,andhighas veryproblematic   (Singerand_Ennis_,2002:9).B  ԍ(2page9,R1)dC  d  Singerand_Ennis_Ԁ(2002:9)pointoutthat:B  dd(2page9,R2)dC  d ~ 8  FX\XXX\Theindexofinconsistencymaybesubstantiallyhigherforrarecategories #  6      ׀ V! whenonlyafewindividualsamongthesmallnumberreportingthecharacteristic B"  changetheirresponse(interviewvs.reinterview).Thismayalsobeaproblemfor .#~! smallsamplesizes,evenwhentheydonthaverarecharacteristics....Wemay $j" observehighindexesforrarecategoriesinadistributioneventhoughthegross %V # differencerate(theproportionofindividualsinthesamplechangingtheirminds) %B!$ maybesmall.#X\XXFX\,#&."%     '#& Dg      dUltimately,the_CRS_Ԁanalyzeddataforabout20,000 #  7      ׀preselectedhouseholds(Singerand_Ennis_,   2002:4).The_CRS_ԀusededitproceduressimilartoCensus2000forrace,Hispanicorigin,and  ancestry,butdidnotgoasfarasimputingfor_nonresponse_Ԁ(Singerand_Ennis_,2002:4).  &   3.2.Limitations # ` Thisstudydoesnotaddressresponsebiasbecause,unlikepreviouscensus_CRS_Ԁstudies,no'  8  probingquestionswereasked.Thetestretestresponseevaluationusedinthisstudymeasures $ t simpleresponsevariance(Singerand_Ennis_,2002:10).dB  ԍ(2page9,R1)dC  dThefactthatnoprobingquestionswere  ` askedisnotnecessarilyalimitationbecauseinordertomeasurebiasonemustknowthe true  L  valueofthecharacteristicbeingmeasured.Thepresumptionhadbeenthattheprobeorthe_CRS_  8  answerwastrue.The_CRS_Ԁquestionnairecloselyfollowedtheenumeratorquestionnairefor $  Census2000,but,unlikeCensus2000,mostinterviewswereconductedbytelephone.   The_mailback_Ԁuniversewasoverrepresentedinthe2000_CRS_Ԁ!aboutthreequartersofthecases   analyzedin_CRS_Ԁcompleted_mailback_ԀformsinCensus2000comparedwith58percentofthepre   selectedhouseholds.Foramajorityofcases,then,thereisadifferenceinthemodeofcollection p betweenthecensusandthe_CRS_.Asconsequencethisstudymay overestimateinconsistency in \ Census2000because datacollectedby_mailback_Ԁmaybe lessinconsistent thandatacollectedby H enumerators(Singerand_Ennis_,2002:xxi,1011).B  ԍ(2pagexxi,R3;page10,R2;page11,R1)d  p ЎC  dAdditionally,therespondentansweringthe 4 _CRS_Ԁwasnotalwaysthecensusrespondent.About68.4percentoftherespondentswerethesame  p in_CRS_Ԁandcensus,although48.2answeredforthemselvesinbothand20.2wereproxyh #  8        \ respondentsinboth.About22percentweredifferentrespondentson_CRS_Ԁandcensus,andwe H werenotabletodeterminetherespondentinabout9.6percentofthecases(Singerand_Ennis_, 4 2002:11).B  ԍ(2page11,R2)dC  dThedatainthisreportareselfweightedandnotweighteduptonationalestimates.   Eachhousingunithadthesameweightbecausethesamplewasselectedwithasinglestage   systematicsample.Thesampledpersonwasselectedatrandomwithineachhousehold,soeach  personhadanequalprobabilityofselectionwithinthehousehold.So samplepersonswithin  householdsofthesamesizehadthesameweight(Singerand_Ennis_,2002:11).B  ԍ(2page11,R4)dC  d  U The_CRS_Ԁstudycompares_CRS_ԀandCensus2000databeforeconsistencyeditsandimputations. X  Race,Hispanicorigin,andancestrywereeditedbasedonlyontheinformationofthesampled D! person.Amongthepossiblecontributorstoresponseerrorarethequestionnairedesign,question 0"  wording,interviewmode,interviewereffects,inadequateinstructions,scanningerrors,and #l! deliberatefalsification(Singerand_Ennis_,2002:11).B  dd(2page12,R1)dC  dThe_CRS_Ԁquestionnairemimickedthe $X" censusenumeratorquestionnaire.B  ԍdd(2page12,R2)dC  dCollectionofinformationonraceandHispanicoriginmay $D # havebeenaffectedbyadministrationmodebecauseresponsesmayhavebeenaffectedbythe %0!$ presenceorabsenceoftheflashcard(Singerand_Ennis_,2002:1112).B  ԍ(2page12,R3;page  13,R1)dC  d   &"% [    3.3.Findingsinbrief @*        Inthistopicreport,IfocusprimarilyontheconsistencyofraceandHispanicoriginreporting,  andtoalesserextentonplaceofbirthandancestry,whicharealsoofinterestinracialandethnic t research.TheremainingpopulationandhousingitemsarecoveredinthecompanionContent ` andDataQualityTopicReport.Ofthe58populationcharacteristicsevaluatedbythe_CRS_, L  16showedgoodconsistency,26moderateconsistency,and16poorconsistency. #  9      ׀B  ԍ(2pagexxii,  L  R1)dC  dThe_CRS_ 8  reportconsideredHispanicoriginandplaceofbirthreportingtobeofgoodconsistency,and $ t raceandancestryreportingtobeofmoderateconsistency.  ` B  V  ԍ(2page19,R1;page20,R1)WdV  ԍWC  dOver95percentofrespondentsansweredboththeraceandHispanicoriginquestionin  8  Census2000and_CRS_.Whenanswering28ofthe58populationquestions,includingancestry, $  householdswithnonHispanicsamplepersonsshowedmoreconsistencyT #  10      ׀thanhouseholds   withHispanicsamplepersons.Frommostconsistenttoleastconsistent,householdswith   WhitesamddplepersonsshowedmoreconsistencythanhouseholdswithAsiansamplepersons,   householdswithsamplepersonsreportingTwoormoreraces,householdswithBlacksample   persons,andhouseholdswithsamplepersonsreportingothersingleraces.However,households p withHispanicsamplepersonsweremoreconsistentinreportingplaceofbirththanhouseholds \ withnonHispanicsamplepersons(Singerand_Ennis_,2002:1920).B  ԍ(2page51,R1)dC  d H 3.3.1.ConsistencyofHispanicoriginreporting  p AccordingtoSingerand_Ennis_Ԁ(2002:xxiixxiii),FX\XXX\theediteddatafortheHispanicoriginquestion H displayedgoodconsistency .#X\XXFX\."#ԀB  ԍ(2pagexxii,R2)dC  dFX\XXX\ButthelackofinstructionsforHispanicoriginmayhavecaused 4 somerespondentsto #X\XXFX\##FX\XXX\choosemultiplecategoriesalthough#X\XXFX\##FX\XXX\Ԁtheintentwastogetonlyone   category.#X\XXFX\&$#B  ԍdd(2pagexxiii,R1)dC  d   Singerand_Ennis_Ԁ(2002:5253)notethatthechangesintheHispanicoriginquestion,including  sequencingitaheadofrace,thedroppingofexamples,changingthequestionwordingandadding   Latino,andthenewinstructionstoanswerbothHispanicoriginandracemayhaveinfluenced l consistency.B  ԍ(2page52,R1)dC  dTheyanalyzedtheHispanicoriginresponsesintwoways.First,theytreatedeach X  responsecategoryasa Yes/Noquestion,usingtheunediteddata.B  ԍ(2page52, $D # R2)dC  dSecond,theygroupedthe D! responses,includingwriteinentries,intoeightcategories,usingtheediteddata.B  ԍ(2page53,R1)dC  d 0"    #l! R      &   FX\XXX\ tablesddTable3.1#X\XXFX\)#FX\XXX\Ԁ AggregateResponseVarianceMeasuresforHispanicOrigin(UneditedData) #X\XXFX\)# tables1)  *Xdd" 44 " F44 F44 e44 e44 CXD(#(#,m dd ,F$$ ,$$ ,dd ,dd +  2f%  2 l  A DfX3! f DNetdifference  rate UfXD23ffX UConsistency  level QfD23ffX  QIndexofinconsistency QX@*"   f QReinterviewclassification ;X*!X ; >X-X > :XX)X :Estimate @XX/! 3XX @90percent   confidence   interval >4'3  XX >NotHispanic 0&  0*0.2 9/! 3 9Good [Q! 3 ffffff$@10.2ffffff$@[10.2 ZPB 3 ffffff$@10.2 ffffff$@ ZЀ9.3to11.1 AX4' 3   AMexican 3X& X 3*0.9 <X/! 3X <Good \XO! 3X  2@182@\18.0 [XN@ 3  2@18 2@X [16.6to19.5 EXX4' 3  X EPuertoRican 7XX& XX 7*0.3 @XX/! 3XX @Moderate bXXQ! 3XX 333336@22.7333336@b22.7 aXXPB 3 333336@22.7 333336@XX a19.4to26.6 EX4' 3  XX ECuban 7X& *X 7*0.3 @X/! *3X @Moderate bXQ! * 3X D@41.7D@b41.7 aXPB *!3 D@41.7 D@X a34.6to50.3 Ef4' *"3  X EOtherHispanic Kf: : #f # K0.0 SfB4 : $3 # f SModerate bfQ! : %3f E@42.2E@b42.2 afPB : &3 E@42.2 E@f a39.0to45.77-+ : '3   f 7*NDRsignificantlydifferentfromzero 8 ' Source:AdaptedfromSingerand_Ennis_Ԁ(2002:53Table33).ddB  ԍ(2page53,R2;andmodifiedby  * Singers7.30.03e.mail)d' (C  d $ ( Thefirstanalysissuggestedgoodconsistencyforthe nonHispanicandthe Mexican  * categories,butonlymoderateconsistencyforthe PuertoRican, Cuban,and OtherHispanic  + categories(seeTable3.1).Thesecondanalysiswitheightcategories(seeTable3.2)alsoshowed , goodconsistencywithonlyabout3.3percentofrespondentschangingtheiranswers,andan p- aggregateindexofinconsistencyof17.2.However,asSingerand_Ennis_Ԁ(2002:5354)remindus, \. allcategorieswere rareexceptthe NonHispanicand Mexicancategories.B  ԍ(2page53,R3)dC  dTheyalsonoted H/ thatabout20percentofthosewhochangedanswerswentfromnonHispanicinthecensustoa 40 mixofnonHispanicandHispanicin_CRS_,B  ԍ(2page53,R4)dC  dandabout53percentofthosechose nonHispanic  p1 and Mexican.About16percentofthosewhochangedanswerswere OtherHispanicin  \2 censusand Mexicanin_CRS_.B  ԍ(2page54,R2)dC  dWhatisclearisthatmostofthedinconsistencyarisesinthe H3  OtherHispaniccategoryandthemultiplereports,ascanbeseeninTable3.2. 44  FX\XXX\&  tablesTable3.2#X\XXFX\>#FX\XXX\ ResponseVarianceMeasuresforHispanicOrigin(EditedData) tables>  6 *VXvddm dd m F$$ F$$ dd dd X(#(#,m $$ , $$#, $$#,$$#,$$#+  ;fX*7 h; @,  @  @  @4 @fX/!8"fX @ @fX/!9"fX @ DfX3!:"fX DIndexofinconsistency OE/;"  h fX OHispanicorigincategories 9/!&v< 9Netdifference &v= rate#X\XXFX\?#FX\XXX\ 9/! b>3 9Consistency &v? level 9/! b@3 9Estimate 9/!&vA3 990percent &vB confidence  bC interval AX4' ND3   ANonHispanic 3X&"^EX 3*0.6 <X/!"^F3X <Good ^XQ!"^G3X 333333$@10.1333333$@^10.1 ]XPB"^H3 333333$@10.1 333333$@X ]9.2to11.0 >4'"^I3  X >Mexican 0&#nJ 0*0.3 9/!#nK3 9Good#X\XXFX\*C#FX\XXX\ [Q!#nL3 *@13.4*@[13.4 ZPB#nM3 *@13.4 *@ Z12.2to14.8 >4'#nN3   >PuertoRican D:.$~O # D0.0 LB4.$~P3 #  LGood#X\XXFX\F#FX\XXX\ [Q!.$~Q3 ffffff,@14.2ffffff,@[14.2 ZPB.$~R3 ffffff,@14.2 ffffff,@ Z11.5to17.6 >4'.$~S3   >Cuban 0&>% T 0*0.1 9/!>% U3 9Good#X\XXFX\H#FX\XXX\ [Q!>% V3 ffffff+@13.7ffffff+@[13.7 ZPB>% W3 ffffff+@13.7 ffffff+@ Z9.3to20.1 >4'>% X3   >OtherHispanic 0&N&!Y 0*0.4 9/!N&!Z3 9Moderate [Q!N&![3 fffff@@33.8fffff@@[33.8 ZPBN&!\3 fffff@@33.8 fffff@@ Z30.7to37.3 >4'N&!]3   >MultiplenonHispanic D:^'"^ # D0.0 LB4^'"_3 #  LPoor \R!^'"`3 Y@100.0Y@\100.0 [QC^'"a3 Y@100.0 Y@ [42.5to100.0 >4'^'"b3   >MultipleHispanic 0&n(#c 0*0.1 9/!n(#d3 9Poor#X\XXFX\zJ#FX\XXX\ [Q!n(#e3  T@80.5 T@[80.5 ZPBn(#f3  T@80.5  T@ Z62.4to100.0 >4'n(#g3   >MixednonHispanicandHispanic 0&~)$h 0*0.6 9/!~)$i3 9Poor#X\XXFX\~O#FX\XXX\ [Q!~)$j3 fffffX@98.6fffffX@[98.6 ZPB~)$k3 fffffX@98.6 fffffX@ Z88.0to100.0 SXfB'~)$l3   hS Aggregate   _  ԍ *%m Ў_____  _ AXf0"*%mXf A @Xf/!*%n3Xf @ Good bXfQ!*%o3Xf 3333331@17.23333331@b 17.2 aXfPB*%p3 3333331@17.2 3333331@Xf a 16.1to18.4 <20*%q3  h Xf <*NDRsignificantlydifferentfromzero#X\XXFX\WQ#FX\XXX\ +4'q ЀSource:Singerand_Ennis_Ԁ(2002:55Table36).d#X\XXFX\ , (r ЇOnecautionnotedbySingerand_Ennis_Ԁ(2002:55)wasthatthe FX\XXX\netdifferenceratesforall  categoriesexceptPuertoRicanandMultiplenonHispanicwerestatisticallydifferentfrom  zerosuggestingthatthe_CRS_Ԁwasnotindependentofthecensusand/ordidnotreplicatethe  censusconditionsaswellasdesired.Netdifferencerates(_NDRs_)givethedifferencebetween t theoriginalpercentinaspecificanswercategoryandthereinterviewpercentinthesame ` category.AnNDRthatisstatisticallydifferentfromzerosuggeststhattheassumptionof L  replicationisnotsatisfied.B  dd(perPhyllisSinger7.30.03e.mail)C  dd 8  #X\XXFX\V#FX\XXX\B  ԍ#X\XXFX\dZ#dd(2page55,R2)dC  d&  AmongSingerand_Ennis_(2002:5556)otherfindingsaboutHispanicoriginreportingwere:  ` FX\XXX\0  householdswithforeignbornsamplepersonsshowedgoodconsistencycomparedwith' Z  8  moderateconsistencyofhouseholdswithnativebornsamplepersons.#X\XXFX\[#B  ԍ(2page55,R3)dC  dFX\XXX\$ (#(# 0  bothrespondentswhoreportedon_mailback_Ԁformsand#X\XXFX\\#FX\XXX\toenumeratorsalsoshowedddgood   consistencyandwerenotstatisticallydifferent(withindexof17.6and16.9respectively)#X\XXFX\]#FX\XXX\.   #X\XXFX\k^#B  ԍ(2page55,R4)d\% N(#(# ЎC  dFX\XXX\ (#(# 0  whenthedatawereanalyzedassingleresponseversusmultipleresponse,theyshowed \ poorconsistency.Givingmultipleresponseswasa rarecategory,whichasstated H above,canaffecttheindexofinconsistency.Onlyabout1.4percentofresponseswere 4 multiple.#X\XXFX\_#B  ԍ(2page56,R1)d)H%S(#(# ЎC  dFX\XXX\ p(#(# 0  about77percent#X\XXFX\a#FX\XXX\ofthosewhochangedtheiranswersreportedasingleresponseinthe H censusandmultipleresponsesinthe_CRS_;andabout23percentreportedmultiple 4 responsesinthecensusandasingleresponseinthe_CRS_.B  ԍ#X\XXFX\a#ԍ(2page56,R2)dC  dFX\XXX\ (#(# #X\XXFX\Oc#3.3.2.DiscussionofHispanicoriginreporting  AccordingtoThomas,_Dingbaum_,and_Woltman_Ԁ(1993:89),thereareseveralwaystointerpret  theindexofinconsistency,dependingonthemethodologyusedtocollectreinterviewdata.B  ԍ(40 8  page8,R1)dC  d l 0  1.Ifeachofthetwoobservations(thecensusandthereinterviewinthiscase)isregarded D! asanindependentrepetitionofthesamesurveyprocedureunderthesamegeneral 0"  conditions,theindexofinconsistencyestimatestheratioofsimpleresponsevarianceto #l! thesumofsamplingvarianceandsimplevariance.Inthiscase,asnotedby_Biemer_ $X" (1985),the indexofinconsistencymeasurestheimpactofmisclassificationerrorson $D # totalvarianceofanobservation (emphasisadded).B  ԍ(40page8,R12)dC  d%0!$(#(# 0  2. Theindexofinconsistencymayalsobeinterpretedasacomplementofameasure '#& ofagreementbetweenthecensusandthereinterviewresponses .Viewedinthisway, (#' theindexistheratiooftheobservednumberofresponsedifferencestothenumberthat )$( wouldoccurifthecellcountswereformedbyarandomagreementmechanismbased|*%)(#(#   ontheobservedmarginaldistributions(censusandreinterview).B  ԍdd(40page8,R2)dC  d h+&*   @-(, 4P! 4  4  eSo  whenthesecondobservationisnotanattempttorepeattheoriginalinterviewprocedure   butmayrepresentanimproveddatasource,thefirstinterpretationoftheindexofinconsistency  maybequestionable.B  ԍ(40page9,R1)dC  dThesecondinterpretationisappropriate  evenwhenthesecond  observationisnotanattempttorepeattheoriginalinterviewprocedureidentically (Thomas, t _Dingbaum_,and_Woltman_Ԁ1993:9).B  ԍ(40page9,R2)dC  dInthisregard,itmaybemoreappropriatetoregardthe2000 ` _CRS_Ԁindexesofinconsistencyinthisfashionratherthanassimpleresponsevarianceestimators. L  ejHowdoesthe2000_CRS_Ԁcomparetothe1990_CRS_?Lookingfirstattheaggregateindexof $ t inconsistency #  11      ׀inTable3.3,the2000index(17.2)isgreaterthanthe1990index(12.2),although  ` botharestilllow.  L   &  tablesTable3.3 HispanicOriginIndexofInconsistency:2000and1990 tablesmp $  *NXOd dm $$ m $$#  $$# $$#$$#VXv(#(#,N dd ,` dw,dd +      Hispanicorigincategory -#!(x  ff -  2000_CRS_ -#!(x # ff -  1990_CRS_ 6,&(x     ff 46NotHispanic XN!x X 333333$@10.1333333$@X3bbpjXlA  X3  Ӏ10.1 oeBx # 333333$@10.1 333333$@X "@9.3"@o7lA  XbbpjXs7  Ӏ9.3 WMGx   "@9.3 40 "@X 4WMexican XN!X *@13.4*@X3bbpjXlA  X3  Ӏ13.4 oeB# *@13.4 *@X !@8.5!@o7lA  XbbpjX^u7  Ӏ8.5 WMG  !@8.5 40 !@X 4WPuertoRican XN!hX ffffff,@14.2ffffff,@X3bbpjXlA  X3  Ӏ14.2 oeBh# ffffff,@14.2 ffffff,@X 333333!@8.6333333!@o7lA  XbbpjXw7  Ӏ8.6 WMGh  333333!@8.6 40 333333!@X 4WCuban XN!hX ffffff+@13.7ffffff+@X3bbpjXlA  X3  Ӏ13.7 pfBh# ffffff+@13.7 ffffff+@X 333333+@13.6333333+@p7lA  XbbpjXx7  13.6 XNHh  333333+@13.6 40 333333+@X 4XOtherHispanic XN!X fffff@@33.8fffff@@X3bbpjXlA  X3  Ӏ33.8 pfB# fffff@@33.8 fffff@@X  A@34.1 A@p7lA  XbbpjX}z7  34.1 XNH   A@34.1 40  A@X 4XMultiplenonHispanic YO!XX Y@100.0Y@Y3bbpjXlA  X3  100.0 OECX # Y@100.0 Y@X Oh XX\7lA  XbbpjX?|7  (X) 7-'X!  40 X 47#X\X h|#MultipleHispanic XN!X"X  T@80.5 T@X3bbpjXlA  X3  Ӏ80.5 NDBX##  T@80.5  T@X Nh XX\7lA  XbbpjX}7  (X) 7-'X$  40 X 47#X\X h~#MixednonHispanicandHispanic XN!%X fffffX@98.6fffffX@X3bbpjXlA  X3  Ӏ98.6 NDB&# fffffX@98.6 fffffX@X Nh XX\7lA  XbbpjX7  (X) @6''  40 X 4@#X\X hT#Aggregate XN!H(f 3333331@17.23333331@X3bbpjXlA  X3  17.2 yoBH)# 3333331@17.2 3333331@f ffffff(@12.2ffffff(@y7lA  XbbpjX}7  12.2XNLH*  ffffff(@12.2 40 ffffff(@f Xh XX\(X)#X\X h؂#ԀNotapplicable  d* Source:AdaptedfromSingerand_Ennis_Ԁ(2002:55Table36)andThomas,_Dingbaum_,and P+ _Woltman_Ԁ(1993:36Table3;17Table4.1)B  ԍ(2page55Table36;40page36,Table1;40page17 (- R1)d  . ЎC  d'(^p <, Onereasonforthedifferenceinindexesisthatmorecategorieswereusedinthecalculationin  . 2000thanin1990.AsThomas,_Dingbaum_,and_Woltman_Ԁ(1993:9)remindus, thelevelof !/ indexissensitivetothe numberanddetail ofcategoriesinaclassificationsystemaswellas "0 tothe distributionofthepopulation overthesecategories[ emphasisadded ].B  ԍ(40page9,R2)d L&!4 ЎC  dSimilarly,as #1 discussedpreviously,Singerand_Ennis_Ԁ(2002:5354)remindusthatallcategorieswere rare t$2 excepttheNonHispanicandMexicancategories.B  ԍ(2page53,R3)dC  dAlthoughthetotalsamplesizeshould `% 3 havenoeffectonthedifferenceinindexes,thetotalsamplesizeinthe1990(n=29,647)was L&!4 about52percentlargerthanin2000(n=19,554)(seeThomas,_Dingbaum_Ԁand_Woltman_,1993:30; 8'"5 Singerand_Ennis_,2002:4).Alargersamplein2000mayhaveyieldedagreaternumberof $(t#6 observationsintherarercategories. )`$7 ЇTurningtoindividualcategories,weseeinTable3.3thattherewasmuchmoreconsistent  reportingin1990inthe Mexicanand PuertoRicancategories,butaboutthesame  consistencyinreportingforthe Cubanand OtherHispaniccategories.Oneexplanationfor  thedifferenceinreportingconsistencyisthatthe1990_CRS_Ԁusedexactlythesamequestionin t censusand_CRS_Ԁ(Thomas,_Dingbaum_,and_Woltman_,1993:6),B  ԍ(40page6,R1)dC  dbutthe2000_CRS_Ԁdidnot,aswe ` willseebelow.Anotherreasonisthatthe2000_CRS_Ԁusedtelephoneinterviews(78percent; L  seeSingerand_Ennis_,2002:3)toamuchgreaterextentthanwasprobablythecasein1990.B  ԍ(2page3,R1)dC  d 8  Unlikethe1990_CRS_,thequestionsaskedinthe2000_CRS_Ԁdifferedfromtheonesusedinthe  ` census.IncaseofHispanicorigininparticular,the_CRS_Ԁquestionisquitedifferentfromthemail  L  form,butmoresimilartotheCensus2000interviewerform(seeTable3.4).  8   & b tablesTable3.4 HispanicOriginQuestionbyQuestionnaireType tables̐   *PXQd dN dd N ` dw` dd NXO(#(#,Kd ,hdd +      &/%XX\Census2000questionnaire -#!d  ff -Hispanicoriginquestion 6,&d    ff 46 Census2000  P  FormD2(_mailback_Ԁlongform) -#!(xX - IsthispersonSpanish/Hispanic/Latino? MarkX P  the No boxif not Spanish/Hispanic/Latino. 7-',| 40 X 47 0 XlA  X0EnumeratorQuestionnaire  [  j 4lA  X XZ4FormD2(E) -#!BX - AreanyofthepersonsthatIhavelistedMexican, j PuertoRican,Cuban,orofanotherHispanicor B Latinogroup? 7-'j 40 X 47 ContentReinterviewSurvey V   V W&FormD1010(5102000) -#!.~f - (Areyou/Is...)Mexican,PuertoRican,Cuban,or V ofanotherHispanicorLatinogroup? 7-+.~ 40  f 7#X\X%&/P#'b        h Areasonablepersonmightconcludethatthe_mailback_ԀHispanicoriginquestionisreallyasking &v ifapersonis Spanish/Hispanic/Latino,whereastheenumeratorand_CRS_Ԁquestionsareasking f aboutspecificgroups(e.g., Mexican, PuertoRican, Cuban,orofanotherHispanicor V Latinogroup).Allhadverysimilarresponsecategories,withthepossibleexceptionofthe F   OtherHispaniccategory(seeTable3.5).Furthermore,the_mailback_Ԁquestioncouldalsobe 6! seenasaskingapersontoselectamongthechoices Spanish/Hispanic/Latino.The print &" groupinstructiononthemailformmayhavereinforcedthisbecausenoexampleswerelisted. # w &  tablesTable3.5 OtherHispanicCategorybyQuestionnaireType tables !% *RXSddKd hdd hPXQ(#(#, 44 ,44 +  "&"  Census2000questionnaire -#!"J' ff -Hispanicoriginquestion 2(&"J(   ff 2 Census2000  z$) FormD2(_mailback_Ԁlongform) -#!f% *X -Yes,otherSpanish/Hispanic/Latino ! Printgroup . -#!z$+X - EnumeratorQuestionnaire  &6", FormD2(E) -#!'"#-X -Yes,otherSpanish/Hispanic/LatinoFX\XXX\Ԁ! Whatisthis &6". group?#X\XXFX\*#FX\XXX\ -#!'&#/X - ContentReinterviewSurvey  V)$0 FormD1010(5102000) -#!B*%1f -Yes,otherSpanish/Hispanic/Latino#X\XXFX\#FX\XXX\Ԁ! Whatisthis V)$2 group?' ! #X\XXFX\#FX\XXX\1'%F*%3  f 1#X\XXFX\1#Inaddition,theinstruction MarkXthe No boxif not Spanish/Hispanic/Latinoonthe b+&3  _mailback_Ԁformmaybeinterpretedasinstructingtherespondenttomark No ifhe/shedoesnot N,'4 1TQ# 1 1  identifywithanyoralloftheterms.Eitheroftheseinterpretationscouldhaveledtosomeofthe  multipleresponsesandthe switchingobservedin_CRS_.   ConsiderahypotheticalexampleofarespondentofMexicanoriginwhomighthavereasonably t concludedthatthe properanswertothe_mailback_Ԁformwasoneofthefollowing: ` 0  1. No,notSpanish/Hispanic/Latinobecausehe/shedidnotidentifywithanyorallof 8  theterms;or$ t(#(# 0  2. No... and  Yes,Mexican,MexicanAm.,Chicanobecausehe/she didnot identify  L  withanyorallofthegeneralterms,butdoesidentifyasMexican!oritcouldbecause  8  he/sheisofmixedheritage;or$ (#(# 0  3.   Yes,otherSpanish/Hispanic/Latinoandawriteinof Spanish, Hispanic,or    Latino becausehe/sheidentifieswithoneofthegeneralterms;or (#(# 0  4.  Yes,Mexican... and Yes,other...andawriteinof Spanish, Hispanic,or p  Latino becausehe/sheidentifiesasMexicanandalsoidentifieswithoneofthegeneral \ terms(inessencevotesforafavoriterubric).H(#(# Yet,duringthereinterviewtherespondentmayhaveselectedthe Yes,Mexican...category  p oranotherinconsistentchoice.DroppingexamplesinCensus2000mayhavealsoledtothe  \ impressionwewereaskingrespondentstoselect,orevenreject,thegeneralresponses(see H Martin,2002:16). 4  &  tablesTable3.6 GeneralHispanicResponsesinCensus2000andContentReinterviewSurvey tables<   *TXUdd 44 44 RXS(#(#,d ,dd ,edd ,rdd ,dd +    h XX\Hispanicorigincategory -#!` ff -  NumberofGeneral ` Hispanicresponsesin " thecensusquestionby  _CRS_Ԁresponse -#!V  ff -  Weighted `  distribution -#!"!  ff -  NumberofGeneral `" Hispanicresponses "# inthe_CRS_Ԁquestion $ bycensusresponse -#!V%  ff -  Weighted `& distribution "'  ?5&(    ff 4?#X\X h#h XX\Total WM!| )X i@205i@W#X\X hJ#  205 VLA| *  i@205 i@X V  100.0% WM!| + X @a@138@a@W  138 VLA| ,  @a@138 @a@X V  100.0% 7-'| -  40 X 47h XX\NotHispanic MC!!.X  @@32@@M#X\X hd#  Ӏ32 LB@!/   @@32 @@X L  Ӏ12.3% MC!!0 X  ;@27;@M  Ӏ27 LB@!1   ;@27 ;@X L  Ӏ16.7% 7-'!2  40 X 47h XX\Mexican MC!#l3X  R@72R@M#X\X h]#  72 LB@#l4   R@72 R@X L  35.8% MC!#l5 X  A@35A@M  35 LB@#l6   A@35 A@X L  28.5% 7-'#l7  40 X 47h XX\PuertoRican LB!l$8X  @7@L#X\X hM#  Ӏ7 KA?l$9   @7 @X K  14.6% LB!l$: X  @7@L  7 KA?l$;   @7 @X K  4.4% 7-'l$<  40 X 47h XX\Cuban LB!% !=X  @5@L#X\X h<#  Ӏ5 KA?% !>   @5 @X K  6.7% -#!% !? X -  ! -#!% !@ X -  ! 7-'% !A  40 X 47h XX\CentralandSouthAmerican MC! '\"BX  E@42E@M#X\X h#  42 LB@ '\"C   E@42 E@X L  6.9% MC! '\"D X  6@226@M  22 LB@ '\"E   6@22 6@X L  15.2% <2' '\"F  40 X U4<h XX\Generalresponses MC!\(#Gf  G@47G@M#X\X h#  47 LB@\(#H   G@47 G@f L  23.8% MC!\(#I f  G@47G@M  47 LB@\(#J   G@47 G@f L  35.1%<20\(#K  40 U  f <!Representszero })$K Source:Specialtabulationofthe2000_CRS_Ԁmicrodata.B  ԍ(41page1,R1)C   i*%L  '- U+&M   X  Usingtheediteddatafromthe2000_CRS_Ԁstudy,Table3.6showsthedistributionofgeneral  Hispanicresponses(suchasSpanish,Hispanic,orLatino). #  12      ׀Thesedatasuggestthatsome   portionofrespondentsshiftbetweenageneralHispanicresponseandspecificresponses.  MostofthetimetheshiftistowardsaspecificHispanicnationalorigin.Forexample,about t 205respondentsin_CRS_ԀgaveageneralresponseinCensus2000.Ofthose,about24percent ` (weighted)gaveageneralresponse,12percentswitchedtononHispanic,and64percenttoa L  specificHispanicnationalorigin.Fromtheoppositeperspective,therewere138respondentsin 8  _CRS_ԀthatgavegeneralHispanicresponses.Ofthose,about35percentgavegeneralresponsesin $ t thecensus,17percentnonHispanic,and48percentspecificHispanicnationaloriginresponses.  ` Inanycase,anyconfusionarisingfromtheissuesdiscussedabovewouldleadtoamuchpoorer  L  consistency,asmeasuredbythe_CRS_.Manyofthesetypesofresponsesaretreatedasa change  8  inresponsealthoughtheymayreflectunintendedeffectsofquestiondesignchangesand $  methodologicaldifferencesratherthaninaccuraciesofreporting.      3.4.Consistencyofracereporting +   Theracequestionschangedsubstantiallybetween1990and2000.Amongthemostsignificant p changeswerethatinCensus2000respondentswereallowedtoselectmorethanonerace, \ whereasin1990theywereonlyallowedtoselectone;inCensus2000Hispanicoriginwas H sequencedaheadofrace,whileinthe1990censusitfollowed,withtwootherquestionsin 4 betweenthetwo;the1990category AsianandPacificIslanderwassplitintoseparate  p  Asianand NativeHawaiianandOtherPacificIslandercategories;the1990categories  \  AmericanIndian, Eskimo,and Aleutwerecombinedintoan AmericanIndianand H AlaskaNativecategory;andthe1990examplesforAsianandPacificIslandergroupswere 4 removed(seeSingerand_Ennis_,2002:56,andMartin,2002:2).   AswithHispanicorigin,Singerand_Ennis_Ԁ(2002:56)analyzetheracedataintwoways.Inthe  firstanalysis,Singerand_Ennis_Ԁ(2002:57)examineonlythecheckboxentriesandtreatthemas   Yes/Noresponses.Theynotethat allcategorieswererareexceptWhite,BlackorAfrican  Am.orNegroandSomeotherrace,andthat thenetdifferenceratesforelevenofthefifteen l categorieswerestatisticallydifferentfromzero,suggestingthatthe_CRS_Ԁwasnotindependent X  and/ordidnotreplicatethecensusconditionsverywell.B  ԍ(2page57,R1)dC  dOnlythe White, Black, Filipino, D! and Koreancategorieshavegoodconsistency.B  ԍ(2page57,R2)dC  dNext,Singerand_Ennis_Ԁ(2002:57)lookatedited 0"  (butnotimputed)racedatagroupedintosevencategories.B  ԍ(2page57,R3)dC  dTheediteddatashowedmoderate #l! consistency,with7.6percentofrespondentschangingtheirraceandanaggregateindexof23.1. $X"  AmericanIndianandAlaskaNative(_AIAN_), NativeHawaiianandOtherPacificIslander $D # (_NHPI_),and Twoormoreracescategorieswereconsideredrare.Inaddition,thenet %0!$ differenceratesforthe White, Someotherrace,and Twoormoreracescategoriesare &"% statisticallydifferentfromzero,meaningatleastoneofthemodelassumptionsofindependence '#& orreplicationwasnotmet.B  ԍ(2 (#' page57,R4)dC  d (#' r%     About14percentoftherespondentswhochangedtheirracebetweenthecensusandthe_CRS_  reportedas Whiteinthecensusand Someotherracein_CRS_.About32percentreported  justtheopposite! Someotherraceincensusand Whitein_CRS_.Analysisofthese t responsesindicatedthatthe majorityofthepersonsinthesetwoinconsistentcategories ` wereofHispanicorigin(Singerand_Ennis_,2002:58).B  ԍdd(2page58,R1)dC  d L  Singerand_Ennis_Ԁ(2002:59)thenanalyzedthedatabyHispanicoriginandfoundthathouseholds $ t withnonHispanicsamplepersonsshowedmoreconsistency(good)thanhouseholdswith  ` Hispanicsamplepersons(poor).Therefore,Singerand_Ennis_Ԁ(2002:59)concludethat this  L  suggeststhattheHispanicpopulationarecontributinggreatlytothevariabilityintheracedata.B  ԍdd(2page $  59,R1)dC  d  8  *WXYddd dd edd erdd rdd TXU(#(#, $$ ,$$#,$$#,$$#,B$$#,$$#,$$#,e$$#+  /%   /&  tablesTable3.7 ResponseVarianceMeasuresforRacebyHispanicOrigin(EditedData) XX\ tables QfD*  0  Q#X\X#h XX\Racecategories QfD2 0f Q @  @ NonHispanic ?f2$B "f  ? @f3!B " f @ @  @ Hispanic B8*B "  f B $    $ =3!* 3  =  Indexofinconsistency <2$* #  < 9/!* 3   9 =3!* 3  =  Indexofinconsistency ND** #   N 3)  3Consistency#X\X h#h XX\  level 9/!3 9Estimate 9/!3 990percent  confidence  interval 9/!D3 9 9/!3 9Consistency#X\X hA#h XX\  level 9/! 3 9Estimate 9/!!3 990percent " confidence # interval >4'D$3   >White 3X&*z% 3Good ]XP!*z&3X 333333"@9.1333333"@]9.1 \XOA*z'3 333333"@9.1 333333"@X \8.4to9.8 <X/!*z(3X < <X/!*z)3X <Poor ^XQ!*z*3X fffff&V@88.6fffff&V@^88.6 ]XPB*z+3 fffff&V@88.6 fffff&V@X ]84.8to92.8 >4'*z,3  X >Black,AfricanAm.,orNegro 0&b- 0Good ZP!b.3 333333@3.9333333@Z3.9 YOAb/3 333333@3.9 333333@ Y3.3to4.5 9/!b03 9 9/!b13 9Moderate [Q!b23 fffffG@47.8fffffG@[47.8 ZPBb33 fffffG@47.8 fffffG@ Z36.6to62.4 >4'b43   >Am.IndianorAlaskaNative 0&J5 0Moderate [Q!J63  @@32.1 @@[32.1 ZPBJ73  @@32.1  @@ Z26.1to39.5 9/!J83 9 9/!J93 9Poor YO!J:3  R@72R@Y72.0 XN@J;3  R@72 R@ X50.5to100.0 >4'J<3   >Asian 0&2= 0Good ZP!2>3 ffffff@7.1ffffff@Z7.1 YOA2?3 ffffff@7.1 ffffff@ Y5.9to8.6 9/!2@3 9 9/!2A3 9Moderate [Q!2B3 >@30.5>@[30.5 ZPB2C3 >@30.5 >@ Z11.7to79.8 >4'2D3   >NativeHawaiianorPacificIslander 0&E 0Moderate [Q!F3 @C@38.5@C@[38.5 ZPBG3 @C@38.5 @C@ Z26.0to57.0 9/!H3 9 9/!I3 9Poor ZP!J3 Y@100Y@Z100.0 YOAK3 Y@100 Y@ Y44.4to100.0 >4'L3   >Someotherrace 0&M 0Poor [Q!N3 V@90.5V@[90.5 ZPBO3 V@90.5 V@ Z74.5to100.0 9/!P3 9 9/!Q3 9Poor [Q!R3 V@90.5V@[90.5 ZPBS3 V@90.5 V@ Z86.2to95.2 >4'T3   >Twoormoreraces 0&U 0Poor [Q!V3 9R@72.99R@[72.9 ZPBW3 9R@72.9 9R@ Z67.5to78.7 9/!X3 9 9/!Y3 9Poor [Q!Z3 `U@85.5`U@[85.5 ZPB[3 `U@85.5 `U@ Z74.5to98.2 EXf4'\3   E Aggregate 7Xf&]Xf 7 Poor bXfQ!^3Xf 333333)@12.6333333)@b 12.6 aXfPB_3 333333)@12.6 333333)@Xf a 11.8to13.5 @Xf/!`3Xf @ @Xf/!a3Xf @ Poor bXfQ!b3Xf U@86.9U@b 86.9 aXfPBc3 U@86.9 U@Xf a 83.4to90.6 7-+d3  Xf 7#X\X h#Source:AdaptedfromSingerand_Ennis_Ԁ(2002:59Table41).B  ԍ(2page59,R2)dC  d'      x   Xd   3.4.1.Discussionofracereporting 0f Althoughtheconsistencyofreportingraceleavesmuchtobedesired,itisquiteclearthat  Xh respondentsofHispanicoriginarelesslikelytoreportconsistentlythannonHispanics.  Di However,amongnonHispanics,onlyBlacks,Asians,andWhitesshowedgoodconsistency, !0j whileAmericanIndiansandPacificIslandersshowedonlymoderatereportingconsistency "k (seeTable3.7).The Someotherraceand Twoormoreracescategoriesshowedpoor #l reportingconsistency.AswasdiscussedextensivelyintheHispanicoriginreportingsection, $m therearemanyreasonswhyweseesuchinconsistentreportinginrace. % n ResearchbyJonesandSmith(2003:4)foundthatthepotentialnumberofchildrenininterracial h'"p familieswhocouldhavebeenreportedasmorethanoneraceapproachesthenumberofchildren T(#q  whowereactuallyreportedasmorethanoneraceinCensus2000.Thus,Census2000doesnot @)$r reflectthepotentialnumberof multiracialchildren. #  13      ׀Thissuggeststhereis,andwillbe,a   substantialproportionofrespondentswhoatanyonetimemaymoveinandoutofthemultiple  racepopulation,makingtheexactmeasurementofthisgroupchallengingindeed.  HowdoesracereportinginCensus2000compareto1990?Thomas,_Dingbaum_,and_Woltman_ ` (1993:21)reportedgoodconsistencyforWhites(13.5indexofinconsistency),Blacks(3.9), L  andAsianandPacificIslanders(9.4);moderateconsistencyforAmericanIndian,Eskimo, 8  andAleut(41.2);andpoorconsistencyfor OtherAsianandPacificIslander(82.9)and $ t  Otherrace(70.3).B  ԍ(40page21,R1)dC  d  ` &  SimilartoSingerand_Ennis_Ԁ(1993),Thomas,_Dingbaum_,and_Woltman_Ԁ(1993:21)reportedthat  8  themajorityoftherespondentsswitchingbetween Whiteand Otherraceandviceversawere $  'Hispanic.B  ԍ(40page21,R2)dC  dUnliketheHispanicquestionin1990,theracedatawere evaluatedusingaresponse   bias(probing)typereinterview,and the_CRS_Ԁmaybeviewedasthepreferredmeasurement   technique(Thomas,_Dingbaum_,and_Woltman_,1993:6).B  ԍdd(40page6,   R2)dC  dGiventheassumptionthatthe_CRS_Ԁ   isthepreferredmeasureofrace,Thomas,_Dingbaum_,and_Woltman_Ԁ(1993:21)concludedthat    theHispanicpopulationarecontributingmostofthebiasintheracedatainthecensusB  ԍ(40page21,R3)dC  dby p overreportingas Otherandunderreportingas White.Thismayhavebeentheresultof \  respondentconfusionB  ԍ(44page554, H R1)dC  dor interviewerbehaviorinthereinterviewsurvey.B  ԍ(44page554,R2)dC  dInanycase,itis H clearinbothstudiesthatHispanicrespondentshadtroubleansweringtheracequestion. 4 Whataccountsforthedifferenceinthereportingofrace?One,thereissomeevidencebased  \ onobservations #  14      ׀B  ԍ(6page46,R1)C  Ԛfrom_nonresponse_Ԁfollowup(_NRFU_)interviewsthat asignificantnumberof H enumeratorsdidnotalwaysreadquestionnaireitemsaswritten,B  ԍ(new=Houghand   Borsa6Bpage39,R1)(original=Borsa6page48,R1;page55,R1)C  Ԁandoftendidnotusethe 4 _flashcards_Ԁprovided,particularlyintheraceandHispanicoriginquestions(Houghand_Borsa_,   2003:39).Two,similartotheHispanicquestion,theracequestionwasdifferentinthe_mailback_   and_CRS_Ԁforms,ascanbeseeninTable3.8.Additionally,the_CRS_Ԁ(andenumerator)formsmay  beperceivedbysomerespondentsassuggestingorencouragingreportingofmorethanonerace.          &   tablesTable3.8 RaceQuestionbyQuestionnaireType tables    *[X\d d $$ $$#$$#$$#B$$#B$$#$$#e$$#eWXY(#(#, dd ,dd +    Census2000Questionnaire -#!T ff -Racequestion 6,&T   ff 46 Census2000  T FormD2(_mailback_Ԁlongform) -#!@X - Whatisthispersonsrace?MarkXoneormore T races toindicatewhatthispersonconsiders @ himself/herselftobe. 7-', | 40 X 47 EnumeratorQuestionnaire  |   FormD2(E) -#!h  X - Nowchooseoneormoreracesforeachperson. |   Whichraceorracesdoeseachpersonconsider h   himself/herselftobe? 7-'T   40 X 47 ContentReinterviewSurvey    FormD1010(5102000) -#! f - Nowchooseoneormoreracesfor(yourself/...).   Whichraceorraces(doyou/does...)consider   yourself/himself/herself)tobe? 7-+|  40  f 7'  Three,thesamplesizeof_CRS_Ԁmaybetoosmalltoproperlymeasuredifferencesinreporting   patternsbothbecauseofrarecategoriesand/orbecausethenumberofrespondentsansweringa p  particularquestionissmall(Singerand_Ennis_,2002:9).B  ԍ(2page9,R3)dC  dFour,assuggestedbySingerand_Ennis_ \ (2002:56,59),_CRS_ԀmethodsdonotreplicatecensusmethodologywellforraceandHispanic H origin.B  ԍ(2page56,R3;page59,R3)dC  dFurthermore,asMartin(2002c:592)remindsus,evensmallquestionnairechangescan, 4 anddo,affectstudyresults,and testdesignandsamplesizemaynotbeadequatetodetectthese  p effects.Itisquitepossiblethatdifferencesinmodesofdatacollectionandinterviewereffects  \ mayaccountforsomeofthesedifferencesaswell.B  ԍdd(43page592,R1)C  dd H   3.5.Consistencyofancestryreporting  E   OneofthechangestotheancestryquestioninCensus2000wastherestructuringofthelistof  examplesfrom21to16exampleancestries.German,Croatian,Ecuadoran,Cajun,Irish,Thai,  andSlovakweredroppedfromthe1990list,andCambodianandNigerianwereaddedfor2000.B  dd(2,page27,R1)dC  d   Inordertoanalyzethesedata,wecollapsedtheancestryresponsesinto58categories.Single l! ancestryresponseswerereportedwithmoderateconsistency(about29percentofrespondents X" changedtheiranswersin_CRS_;theaggregateindexofinconsistencywas30.7).B  dd(2, 0!$ page28,R1)dC  dSomeofthekey D # findingsare: 0!$ 0  respondentswhoreportedon_mailback_Ԁformsshowedmoreconsistencythanthosewho #X& reportedtoenumerators,althoughbothweremoderate;B  dd(2,page27,R2)dC  d0(#(##D'(#(#   0  householdswithforeignbornsamplepeopleshowedmoreconsistencythanthosewith %!) nativebornsamplepeople(moderate);&"*(#(# B  dd(2,page27,R2)dC  d0  householdswithnonHispanicsamplepeopleshowedmoreconsistencythanthosewith (#, Hispanicsamplepeople(bothmoderate).B  dd(2,page27,R3)dC  d|)$-(#(# Oneofthedifficultieswithancestrydataisthatmanyrespondentsleavetheitemblank,butthe T+&/ questionwasmorelikelytobeunansweredinCensus2000(n=4,159orabout21.3percent) @,'0 comparedwith_CRS_Ԁ(n=1,603orabout8.2percent).Leavingancestryblankmaybearesultof ,-|(1 r(     perceivedredundancybymanyrespondentswhofelttheyhadalreadyprovidedthis  informationwhentheyansweredtheraceandHispanicoriginquestions(Martin,_Demaio_,  and_Campanelli_,1990:555556).B  dd(44page555,R1;page556,R1)dC  d  3.5.1.Discussionofancestryreporting ` Althoughancestrywasreportedwithmoderateconsistency,itwaslessconsistentlyreportedin 8  householdswithHispanicsamplepeople,butalsomoreconsistentlyinhouseholdswithforeign $ t bornrespondents.YetitisalsotruethatproportionatelymoreHispanichouseholdshaveforeign  ` bornpeoplethannonHispanichouseholds.Howcanthisbereconciled?  L  9& h tablesTable3.9 SingleAncestryResponsesfromContentReinterviewSurvey(_CRS_) $  04 XlA  X0 4 andCensus2000 tables#    9#4lA  X4 Xt$4*]X^dd dd dd [X\(#(#,dd ,dd ,Fdd ,dd ,dd ,dd +  h XX\ $   $#X\X hl&#_CRS_ d  Ancestry P  Response@@::: 6,!<" ff 6Number 6,!P " ff 6Same d  countryin P  Census2000 6,!<" ff 6Hispanicor d  Spanishin P  Census2000 6,!<" ff 6Allother d  responsesin P  Census2000 6,!<" ff 6Levelof d  consistency 1'!P " ff 41Colombian 6,!<X 628(100%) 6,!<3X 685.7% 6,!< "X 6! 6,!<!"X 614.3% 6,!<""X 6Good 7-'<#" 40 X 47Cuban 6,!$X 643(100%) 6,!%3X 695.3% 6,!&"X 62.3% 6,!'"X 62.3% 6,!("X 6Good 7-')" 40 X 47Dominican 6,!,*X 645(100%) 6,!,+3X 684.4% 6,!,,"X 68.9% 6,!,-"X 66.7% 6,!,."X 6Good 7-',/" 40 X 47Ecuadorian 6,!,|0X 622(100%) 6,!,|13X 695.5% 6,!,|2"X 6! 6,!,|3"X 64.5% 6,!,|4"X 6Good 7-',|5" 40 X 47Guatemala 6,!|6X 632(100%) 6,!|73X 668.8% 6,!|8"X 631.2% 6,!|9"X 6! 6,!|:"X 6Moderate 7-'|;" 40 X 47Honduran 6,!<X 622(100%) 6,!=3X 677.3% 6,!>"X 613.6% 6,!?"X 69.1% 6,!@"X 6Good 7-'A" 40 X 47Mexican 6,!lBX 6901(100%) 6,!lC3X 692.1% 6,!lD"X 65.8% 6,!lE"X 62.1% 6,!lF"X 6Good 7-'lG" 40 X 47PuertoRican 6,!lHX 6144(100%) 6,!lI3X 680.6% 6,!lJ"X 615.3% 6,!lK"X 64.2% 6,!lL"X 6Good 7-'lM" 40 X 47Salvadoran 6,! NX 636(100%) 6,! O3X 672.2% 6,! P"X 611.1% 6,! Q"X 616.7% 6,! R"X 6Moderate EX_0' S" 40 X 4EHispanic ;X_& \TX_ ;62(100%) DX_/! \U3X_ D8.1% DX_/! \V"X_ D25.8% DX_/! \W"X_ D66.1% DX_/! \X"X_ DPoor Ef_0' \Y" 40 X_ 4ESpanish ;f_&\!Zf_ ;102(100%) Df_/!\![3f_ D29.4% Df_/!\!\"f_ D8.8% Df_/!\!]"f_ D61.8% Df_/!\!^"f_ DPoor7-+\!_" 40 f_ 7!Representszero x"_ Source:AdaptedfromSingerand_Ennis_Ԁ(2002:E11E15TableE.29;C15C16TableC.29). d#` 'h#Table3.9showsninespecificsingleancestryHispanicnationaloriginentries,and onlytwo <% b ( Guatemalanand Salvadoran)hadmoderatelevelsofconsistency. #  15      ׀Ontheotherhand, (&x!c twogeneralsingleancestries( Hispanicand Spanish)showedevenpoorerlevelsof 'd"d consistency,meaningthatrespondentsanswereddifferentlyinthecensusand_CRS_. (P#e   )(%g Of62respondentswhoreportedas Hispanicin_CRS_,only8.1percentalsodidsoinCensus  2000;25.8percenthadreportedas SpanishinCensus2000,and66.1percentreportedother  responses(someofwhichcouldbeotherspecificHispanicorigincategories).Similarly,of  102respondentswhoreportedas Spanishin_CRS_,29.4percentalsodidsoinCensus2000; t 8.8percenthadreportedas HispanicinCensus2000,and61.8percentreportedother ` responses.Clearly, Hispanicand Spanisharenotconsistentlyreported. L   &  tablesTable3.10 Hispanicand SpanishSingleAncestryResponsesinCensus2000and $ t ContentReinterviewResponses tables?  d *`Xad ddd dd Fdd Fdd dd dd ]X^(#(#,( dd ,( dd ,( dd +   P  P  _CRS_ԀAncestryResponse -#!d   ff -  HispanicinCensus2000 d   AncestryQuestion -#!P 0 ff -dB  SpanishinCensus2000 d   AncestryQuestion 9/&P 0   ff 09BColumbian 0&! X 0   @  @  @ ! 0&! 0X 0  3.6% :0' 0 00 X 0:Cuban 0&!8 X 0  ! 0&!8 0X 0  8.3% :0'8 0 00 X 0:Dominican 0&!4 X 0  2.6% 0&!4 0X 0   @  @  @  @ ! :0'4 0 00 X 0:Mexican 0&!X 0  62.9% 0&!0X 0  25.0% :0'0 00 X 0:PuertoRican 0&!X 0  7.8% 0&!0X 0  3.6% :0'0 00 X 0:Salvadoran 0&!hX 0  2.6% 0&!h0X 0  1.2% :0'h 0 00 X 0:U.S.orAmerican 0&!d!X 0  1.7% 0&!d"0X 0  1.2% :0'd#0 00 X 0:Hispanic 0&!$X 0  4.3% 0&!%0X 0  10.7% :0'&0 00 X 0:Spanish 0&!L'X 0  13.8% 0&!L(0X 0  35.7% Af0'L)0 00 X 0AOthergroups .fH*f .  4.3% 7f&!H+0f 7  10.7% JX9'H,0 00 f 0JЀTotal @X/!-X @  100.0% @X/!.0X @  100% @6'/0 00 X 0@ЀNumber WM!00f ]@116]@W  116 vlA010 ]@116 ]@f  U@84U@v  84VLJ020  U@84 00 U@f V!Representszero L2 Source:AdaptedfromSingerand_Ennis_Ԁ(2002:E11E15). 83 '$ ?       h      p Table3.10showshowrespondentswhoreported Hispanicand SpanishintheCensus2000 !5 ancestryquestionreportedinthe_CRS_Ԁancestryquestion.Of116 HispanicentriesinCensus "6 2000,only4.3percentreported Hispanicin_CRS_.Nearlytwothirds(62.9percent)reported #7  Mexicanin_CRS_Ԁand13.8percentreported Spanish.About1.7percentreported U.S.or $8 American,andonly4.3reported othergroups(someofwhichcouldbeotherspecific p% 9 Hispanicorigincategories).Among84whoreportedas SpanishinCensus2000,35.7percent \&!: asreported Spanishin_CRS_,10.7percentreported Hispanic,and25.0percent Mexican. H'";    )p$= TheseresultssuggestthatmuchoftheinconsistencyinthereportingofHispanicancestriesis  relatedtoshiftingbetweengeneralterms( Hispanicor Spanish) #  16      ׀andspecificterms  ( Mexicanor PuertoRican),andbetweengeneraltermsthemselves.Table3.11shows  similarresultswhencomparing Hispanicand Spanishresponsesinthe_CRS_Ԁancestry t questionwiththematchedCensus2000ancestryquestionresponses.Clearly,somerespondents ` switchbetweenspecificandgeneralHispanicgroupterms,butrelativelyfewswitchbetween L  HispanicandnonHispanicancestries. 8   & 0 tablesTable3.11 Hispanicand SpanishSingleAncestryResponsesinContentReinterview  ` andCensus2000Responses tablesRV  P  *cXdd d( dd ( ( dd ( ( dd ( `Xa(#(#,dd ,dd ,dd +   <  <   Census2000 P  AncestryResponse -#!<  ff -  Hispanicin_CRS_ P  AncestryQuestion -#!< 0 ff -X  Spanishin_CRS_ P  AncestryQuestion 6,&< 0   ff 46Cuban 9/! X 9  0.9% 0&! 2X 0  ! :0' 0 40 X 4:Dominican 9/!, X 9  ! 0&!, 2X 0  3.9% :0', 0 40 X 4:Guatemalan 9/!,|X 9  1.7% 0&!,|2X 0  7.8% :0',|0 40 X 4:Honduran 9/!|X 9  0.9% 0&!|2X 0  2.0% :0'|0 40 X 4:Mexican 9/!X 9  25.9% 0&!2X 0  21.6% :0'0 40 X 4:PuertoRican 9/!l X 9  10.3% 0&!l!2X 0  9.8% :0'l"0 40 X 4:Salvadoran 9/!l#X 9  ! 0&!l$2X 0  3.9% :0'l%0 40 X 4:U.S.orAmerican 9/! &X 9  ! 0&! '2X 0  1.0% :0' (0 40 X 4:Hispanic 9/! \)X 9  4.3% 0&! \*2X 0  15.7% :0' \+0 40 X 4:Spanish 9/!\,X 9  7.8% 0&!\-2X 0  29.4% Af0'\.0 40 X 4AOthergroups 7f&/f 7  1.7% 7f&!02f 7  6.0% JX9'10 40 f 4JЀTotal @X/!L2X @  100.0% @X/!L32X @  100.0% C9'L40 40 X 4CЀNumber YO!L 5f  O@62O@Y  62 yo@L 62  O@62 O@f Y@102Y@y  102WMKL 70 Y@102 40 Y@f W!Representszero h!7 Source:AdaptedfromSingerand_Ennis_Ԁ(2002:E11E15).'0CV T"8   3.6.Consistencyofplaceofbirthreporting He ,$|: TheCensus2000questiononplaceofbirthincluded:1)checkboxesforrespondentstoindicate &T!< whethertheywerebornintheUnitedStatesoroutsidetheUnitedStates,and2)writeinspaces &@"= toreporttheirstateofbirthorcountryofbirth.Withrespecttothecheckboxresponses,placeof ',#> birthwasreportedveryconsistently(onlyabout0.5percentofrespondentsreportedadifferent ($? placeofbirthforthesampleperson,foranindexofinconsistencyof2.7).B  ԍ(2,page31,R1)C  ԀAmongthefindings )%@ are: *%A Ї0  respondentswhoreportedon_mailback_Ԁformsshowedmoreconsistencythanthosewho  reportedtoenumerators,althoughbothwerelow;B  dd(2,page31,R2)dC  d(#(# 0  FX\XXX\householdswithnativebornsamplepersons(asidentifiedbythecheckboxonthe t citizenshipquestion)showedmoreconsistencythanhouseholdswithforeignbornsample ` persons;#X\XXFX\i#B  dd(2,page31,R2)dC  d    h L (#(# 0  FX\XXX\householdswithHispanicsamplepersonsshowedmoreconsistencythanhouseholds $ t withnonHispanicsamplepersons#X\XXFX\k#.B  dd(2,page32,R1)dC  d `(#(# 3.6.1.Discussionofplaceofbirthreporting  8  Generallyspeaking,theconsistencyofplaceofbirthreporting(asidentifiedbythewritein   response)isquitegood(Singerand_Ennis_,2002:32).SFX\XXX\ampleindividualsbornoutsideofthe   UnitedStateswereaskedtoreportthecountryofbirth.Allresponsestoplaceofbirthwere   groupedinto68categories,whichincludedthe50states,theDistrictofColumbia,UnitedStates   territories,andothercountriesandregions.Approximately3percentof_CRS_Ԁrespondents p changedanswersduringthe_CRS_,yieldingan#X\XXFX\m#FX\XXX\Ԁaggregateindexof3.2#X\XXFX\p#.B  dd(2,page32,R1)dC  d \  &  tablesTable3.12 ContentReinterviewSurvey(_CRS_)PlaceofBirthReportingforCentraland 4 SouthAmerica tablesp $t *gXhdd( dd ( dd ( dd cXd(#(#,( dd ,( dd ,( dd +  ``  Area 6,!t ff 6ConsistencyLevel 6,!t" ff 6IndexofInconsistency 6,&t"   ff 46PuertoRico -#!X -  High ND! X ffffff@3.8ffffff@N  3.8 WMG  ffffff@3.8 40 ffffff@X 4WMexico -#!dX -  High ND!d X 333333?1.2333333?N  1.2 WMGd  333333?1.2 40 333333?X 4WOtherCentralAmerica -#!dX -  High ND!d  X ?1.5?N  1.5 WMGd!  ?1.5 40 ?X 4WCaribbean -#!"X -  High ND!# X @5.0@N  5.0 WMG$  @5.0 40 @X 4WSouthAmerica -#!T%f -  High ND!T& f @2.1@N  2.1WMKT'  @2.1 40  @f WSource:AdaptedfromSingerand_Ennis_Ԁ(2002:C23TableC.34).  p' ' Hp       h AsshowninTable3.12,placeofbirthreportingfromCentralandSouthAmericaappearstobe !H) quiteconsistent.Theseresultsforplaceofbirthandthepreviouslydiscussedresultsforancestry "4* suggestthat,atleastforHispanicgroups,thesequestionsmaybeconsideredreliablesupplements # + totheHispanicorigindata,asshownby_Cresce_ԀandRamirez(2003).However,theirusefor $ , supplementingracedataneedstobeexploredfurther.V   p W{  % - >P, >  >  ;XX\rP8[3  4.CensusQualitySurvey(_CQS_)toEvaluateResponsestotheCensus2000Question  onRace:AnIntroductiontotheData r|#X\X;|#   @*ThemainobjectiveoftheCensusQualitySurvey(_CQS_)wastoassistdatausersincomparing  racedataobtainedbyaskingrespondentsto markoneormoreraceswithdataobtainedby  askingrespondentsto markonerace.The_CQS_Ԁcollectedracedatausingbothmethodsfrom p  thesamepeople,sopotentiallyitcouldbeusedtoevaluatehowrespondentsreportingmultiple \  racesrespondwhenaskedtoreportasinglerace.V  ԀWForexample,thedatacouldbeuseddetermine H  theproportionofpeoplewhoreportasBlackwhenaskedtoreportonlyoneracebutreportas 4  White and Blackwhenaskedtoreportoneormoreraces.Thisinformationcouldbeused  p   tobridgethetwomethodsbyconstructingstatisticaladjustmentstoracedistributionsobtained  \  usingonemethodtomakethemmorecomparabletoracedistributionsobtainedusingtheother H  (Bentley,_Mattingly_,HoughandBennett,2003:1).B  ԍ(3page1,R1)dC  d 4    4.1.Studydesign     AccordingtoBentley,_Mattingly_,HoughandBennett(2003:11)samplehouseholdswere  contactedtwiceduringthe_CQS_Ԁsurveytoprovideinformationonrace.Botha markonerace  1990censusinstructionanda markoneormoreracesCensus2000instructionwere l administeredinasplitpaneldesign.Atotalsampleof55,000addresseswereselected.B  ԍ.dd(3page X 10,R1;page11,R1)dC  d X ThesamplehouseholdsreceivedamailedinitialquestionnaireinJune2001.Householdsthat D didnotreturntheinitialquestionnaireweremailedasecondquestionnaireinearlyJuly2001. 0 Householdsthatdidnotrespondtothefirstorsecondmailingswerecontactedwith_nonresponse_ l followup(_NRFU_)proceduressimilartothoseusedforCensus2000.V   X WTThesampleuniversewassplitintotwopanels(AandB).PanelA,consistingofrespondents 0 fromabout27,500housingunits(HUs),wereaskedtheCensus2000racequestion.PanelB,  consistingofrespondentsfromabout27,500housingunits,receivedasimilarquestionnaire  buttheinstructiontothequestiononracewasto markonerace.Duringtheinitialcontact,  about54percentofhouseholdsinbothpanelsrespondedbymailandtheremainderwere  interviewedin_NRFU_Ԁpersonalvisits.AsinCensus2000,enumeratorsused_flashcards_Ԁshowing |  theinstructionsandthecategoriesforthequestionsonraceandethnicityin_CQS_Ԁinitialcontact h! _NRFU_Ԁvisits.B  ԍdd(3page11,R2)dC  d T"  RespondentswerealsoaskedwhetheraCensus2000formhadbeenfilledoutforthehousehold ,$|" and,ifso,whocompletedtheform.Thisinformationwasusedtoassessconsistencyofreporting %h # whenracewasreportedbythesameoradifferentrespondent.B  ԍ(3page11,R3)dC  dThe_CQS_Ԁalsocollected &T!$ informationontheaddresswhereeachpersoninthehouseholdwaslivingonApril1,2000 &@"% toassistinmatching_CQS_ԀrespondentstotheirCensus2000data.B  ԍ(3page11,R4)dC  dFourtosixweeksafterthe ',#& secondmailout,householdsrespondingtotheinitialcontactphaseofthedatacollectionwere ($' thenrecontactedbytelephonetocollectdataonracefromthealternateracequestionaswellas )%( otherdata,suchaseducationandincome.B  ԍ(3page11,R5)dC  d *%)   +&* &   Inthe recontactphaseofdatacollection,PanelAhouseholdsthatreceivedthe markoneor  moreracesinstructionintheinitialdatacollectionwereaskedto chooseoneraceinthe  recontactinterview.Conversely,PanelBhouseholdsthatreceivedthe markonerace  instructionintheinitialcontactwereaskedto chooseoneormoreracesintherecontact t interview. ` 'Morethan70percentoftherecontactinterviewswereconductedbytelephone.B  ԍdd(3page12,R1)dC  dPersonal 8  interviewswereconductedtocollecttherecontactinformationforhouseholdsthatwerenot $ t contactedbytelephone.Inbothcases,everyeffortwasmadetospeakwiththeindividualwho  ` completedtheinitialquestionnaire.B  ԍ(3page12,R2)dC  dThePanelAquestionnaireincludedaprobeforadditional  L  informationininstanceswhererespondentswerereluctanttoreportasingleracewhenaskedto  8  doso.Respondentsinbothpanelswereaskedtoprovideadditionalsocialanddemographic $  information,suchasrelationship,veteransstatus,educationalattainment,householdincome,   andlanguagespokenathome,whichmightberelevanttotheissueofdifferentialracereporting.   B  ԍ(3page12,R2)C  Thefinalsamplesizeofthe_CQS_Ԁwasapproximately50,000interviewedhousingunitsand   155,000respondents. About25percentofthesamplewasallocatedtoeachofthefourcells p createdbycrossingpanel(AorB)bycensusformtype(shortorlong).Eachstatewastreated \ asanindependentsamplingstratumandfourdistinctsamplingstratawereidentifiedwithineach H state. #  17      ׀B  ԍ(3page13,R1)dC  dInordertomaximizethelikelihoodofcontactinghouseholdsin_CQS_Ԁwithindividuals 4 reportingmorethanonerace, 90percentoftheinitialsamplewasselectedfromamong  p householdscontainingatleastoneindividualwhoreportedmorethanoneraceinCensus2000.B  ԍ(3page14,R1)dC  d  \ Becausemostoftheresponsesthatarecodedas Someotherrace(_SOR_)inCensus2000are 4 Hispanicethnicities,the_CQS_ԀfocusedprimarilyontheOMBracecombinationsh #  18      .Inorderto   producegreaterreliabilityforthecombinationsoftwoOMBracecategories,combinations   including_SOR_Ԁweresampledatonethirdtherateoftheothercombinations.Asaresult,  18percentofthe_CQS_Ԁsampleconsistedof_SOR_Ԁcombinations,comparedwith42percentin  Census2000.B  ԍdd(3page15,R1)dC  dFinally,Census2000recordswerelinkedto_CQS_Ԁrecordsinordertofacilitate  comparisonsbetween_CQS_ԀandCensus2000racedata.Thislinkingprocessmatchedarecord l inthe100percentCensusUneditedFile(_HCUF_)torecordsinthe_CQS_Ԁfilebycomparingfields X  suchasfirstname,lastname,middleinitial,suffix,sex,dateofbirth,age,streetname, D! andzipcode. 0"      #l! P.     kXXXX\& H  tablesTable4.1#X\XXkX#kXXXX\ CensusQualitySurveyDataCollectionSequence:RaceInstructionbyPanel #X\XXkX:# tables  *lXmdd( dd ( ( dd ( ( dd ( gXh(#(#,dd ,dd ,dd , dd +    LY  _CQS_ԀPanel -#!T  ff -V  >WCensus2000 -#!T ff -_CQS_ԀInitialContact -#!T ff -_CQS_ԀRecontact ;1&T   ff ; A -#!"X -&/%XX\ markoneormoreraces -#!X - markoneormoreraces -#!X - chooseonerace 6,! X 6#X\X%&/# B  @l -#! T "f -&/%XX\ markoneormoreraces -#! T f - markonerace -#! T f - chooseoneormoreraces1'% T   f 1#X\X%&/f#Source:Bentley,_Mattingly_,HoughandBennett(2003:10,Table1).B  ԍ(3page10,R2)dC  d  p  'H۠Thismatchalsoprovidesanothersetofobservationswhichcanbeusedtoestimate bridging  H parameters,B  ԍ(3page16,R1)dC  dascanbeseeninTable4.1.Forexample,inPanelA,onewouldcomparethe  4  markoneormoreracesresponseinthe_CQS_Ԁinitialcontactwiththesingleraceresponse   inthe_CQS_Ԁrecontact.     4.2.Limitations N   AccordingtoBentley,_Mattingly_,HoughandBennett(2003:2122),thereareoperationaland l  qualitativelimitationstothisevaluation:1)thedesignofthe_CQS_ԀcouldnotrepeattheCensus X 2000environment;B  ԍ.dd(3page21,R1)C  dd2)differentcollectionmethodswereusedinthe_CQS_Ԁinitialcontactand D recontact;3)theresponsetoasubsequentquestiononracecanbeinfluencedorconditioned 0 bytheresponsetothepreviousquestion;B  ԍ(3page21,R3)dC  d4)proxyreporting;5)effectsofmoversonthe l sample; #  19      ׀B  ԍ(3page22,R2)dC  dand6)possibleerrorassociatedwithlinkingCensus2000data.B  ԍdd(3page22,R3)dC  d X   4.3.Findingsinbrief ngخ 0 4.3.1.Whatweretheresponseratesforeachpanel?  Afterexcludingvacanthousingunits,Bentley,_Mattingly_,HoughandBennett(2003:2324) ! reportthatresponserateswereabout97percentfortheinitialcontact.B  ԍ(3page23,R1)dC  dIntherecontact,about |" 87percentofPanelAhousingunitsresponded,comparedwithabout94percentofPanelB.B  ԍdd(3 @ % page24,R1)dC  d h# 4.3.2.Wasthe_CQS_ԀrepresentativeofCensus2000data? @ % Because analyticalresultscanbebiasediftheinterviewedsampleisnotrepresentativeofthe "h' populationofinterest,Bentley,_Mattingly_,HoughandBennett(2003:24)comparedaggregate #T( _CQS_ԀdistributionswithCensus2000reportingforeachpanelandconcluded(2003:vi):B  ԍdd(3page &", vi,R1)dC  d #@) 8  Theresultsfromthequestiononracesuggestthat eachpanelappears %!+ toberepresentativeofCensus2000. Aggregatedreportingofraceamong &", nonHispanicrespondentstothe markoneormoreracesinstructionclosely '"- resemblesCensus2000reportingofraceforeachpanel.Noracegroupappears (#. tt/      tobesignificantlydifferentfromCensus2000(p<0.1level)ineitherpanel,  includingtheTwoormoreracespopulation.ReportingofraceforHispanic  respondentsisalsosimilartothatinCensus2000,thoughinPanelAasmaller  proportionofHispanicschose Whiteasasingleraceandalargerproportion t chose SomeotherracecomparedwithCensus2000data.`          h      p 4.3.3.Persistenceofmorethanoneracereporting 8  TheeffectoftheprobequestioninPanelAreducedreportingofmorethanoneracefrom  ` 1.4percentto0.4percent.Totheauthorsthisindicated thatthereisasizeableportionof  L  peoplewhowillpersistentlyreportTwoormoreraceswhenaskedtoreportonlyone(Bentley,  8  _Mattingly_,HoughandBennett,2003:25).B  ԍdd(3page25,R1)dC  dTheauthorsalsonotethat ingeneral,unlessa $  probingquestionisasked,itappearsthatabouthalfofallTwoormoreracerespondentsdonot   giveasingleraceresponse.Nonetheless,thedatasuggestthattheracedistributionsdonot   changemuchwiththefollowupproberesults(Bentley,_Mattingly_,HoughandBennett,   2003:27).B  dd(3page27,R1)dC  d   4.3.4.Consistencyofracereportingbetweenthe_CQS_ԀandCensus2000data \ Bentley,_Mattingly_,HoughandBennett(2003:vi)reporta generallylowconsistencyof 4 reportingmorethanoneracebetweenCensus2000andthe_CQS_:B  ԍ(3pagevi,R2)dC  d  p 8  Only40percentofthenonHispanicrespondentsinPanelAwhoreportedmore H thanoneraceinCensus2000alsoreportedmorethanoneraceintheinitial 4 contact( markoneormoreracesinstruction).Similarly,only41percentof   thoseinPanelBwhoreportedmorethanoneraceinthecensusalsoreported   morethanoneraceintherecontact.Theother60percentreportedasinglerace.   8     Incontrast,97percentto98percentofthosewhoreportedasingleraceofWhite,  Black,orAsianinCensus2000reportedthesameraceintheCensusQuality  Survey.ForAmericanIndianorAlaskaNatives,NativeHawaiianorOther l PacificIslanders,andSomeotherracerespondents,thereportingofrace X  consistencyrangesfrom55percentto58percentinPanelA,and72percent D! to78percentinPanelB.0"    Tables4.2and4.3(Bentley,_Mattingly_,HoughandBennett,2003:28,30)showthelack $X"  ofconsistencyamongnonHispanics. $D # r0     & x  tablesTable4.2 OverallConsistencyofRaceReportingforNonHispanicsforPanelA tables*   *pXqdddd dd dd  dd  lXm(#(#V  ,44 ,44 ,444 ,444 W+  -# -Census2000Race @ @444 KA2, DmVff K  _CQS_ԀInitialContact( markoneormoreraces) >4)T   ff > 7f&LX 7]  SingleRace @  @D =f,!3f =TwoorMoreRaces =f,!3f =TOTAL ;1&3  f ;SingleRace 6,!\  X 696,987,813 \   n=34,839 6,!H  3X 61,286,746B  C  Ԛ \   n=1,978B  C  Ԛ 6,!H  3X 698,274,559B  ԍ \  ЎC   \  n=36,817B  C  Ԛ 6,!H 3X 6TwoorMoreRaces 6,! X 61,089,924B  C  Ԛ   n=9,089B  C  Ԛ A,! 3X A 724,686B  C  Ԛ   n=8,035B  C  Ԛ 6,! 3X 61,814,610B  C  Ԛ   n=17,124B  C  Ԛ 6,! 3X 6TOTAL 6,!4 f 698,077,737B  C   4  n=43,928B  C  Ԛ 6,! p 3f 62,011,432B  C   4  n=10,013B  C  Ԛ 6,! p 3f 6100,089,169B  C   4  n=53,941B  C  Ԛ1'% p 3  f 1h XX\*ThedatainTable4.2wererestrictedtomatchedpeoplewhodidnothaveanimputedraceinCensus2000thatis, <  onlythosecaseswherethefinaleditedracewas asreported,orwherethecodewaschanged throughconsistency P  edit.The_CQS_ԀinitialcontactHispanicoriginresponsewasused.Additionally,theweighteddatawereobtained   usingtheinverseoftheoriginalsamplingprobabilitieswithnoadjustment(Z_WGT1).#X\X h#   Source:Bentley,_Mattingly_,HoughandBennett(2003:28,Table9). B  ԍ(3page28,R1)dC  d  L! 'x & t tablesTable4.3 OverallConsistencyofRaceReportingforNonHispanicsforPanelB tables*  d$ *rXsdd44 44 444 4444 4pXq(#(#V  ,4d ,4d ,44d ,44d WS+  -#T%T -Census2000Race T' @ KA2@(  ~ff K  _CQS_ԀRecontact( markoneormoreraces) >4)h)   ff > 7f&)X 7  SingleRace @  @D =f,!d*3f =v  TwoorMoreRaces =f,!d+3f =  TOTAL@@B ;1&d,3  f ;\SingleRace 6,!-X 6b  89,881,179b B  C   . n=32,848B  C  Ԛ 6,!/3X 6935,610B  C   0 n=1,476B  C  Ԛ 6,!13X 690,816,789B  C   2 n=34,324B  C  Ԛ 6,!33X 6TwoorMoreRaces 6,!,|4X 6825,761B  C   ,|5 n=8,994B  C  Ԛ A,!h63X A 565,422B  C   ,|7 n=7,148 B  C  Ԛ 6,!h83X 61,391,183B  C   ,|9 n=16,142B  C  Ԛ 6,!h:3X 6TOTAL 6,!;f 690,706,940B  C   < n=41,842B  C  Ԛ 6,! =3f 61,501,032B  C   > n=8,624B  C  Ԛ 6,! ?3f 692,207,972B  C   @ n=50,466B  C  Ԛ1'% A3  f 1h XX\*ThedatainTable4.3wererestrictedtomatchedpeoplewhodidnothaveanimputedraceinCensus2000thatis, !A onlythosecaseswherethefinaleditedracewas asreported,orwherethecodewaschanged throughconsistency d"B edit.The_CQS_ԀinitialcontactHispanicoriginresponsewasused.Additionally,theweighteddatawereobtained (#xC usingtheinverseoftheoriginalsamplingprobabilitieswithnoadjustment(Z_WGT1).#X\X he# #<D Source:Bentley,_Mattingly_,HoughandBennett(2003:30,Table11). $ E B  ԍ(3page30,R1)dC  d'tAmongtheconsequencesofthelowlevelofconsistencyinthereportingofmorethanonerace, &!G theauthorslist:B  ԍ(3pagevi,R2)dC  d t'"H   The effectivesamplesize forcomputingbridgingparametersisreducedandthe L)$J   parametersaresensitivetowhichdataareusedtocomputethem. 8*%K   The stabilityofbridgingparameters maybeuncleargiventheobservedinstabilityin ,`'M   reportingmorethanonerace. ,L(N        h      p      x   -8)O  r131       4.3.5.Tabulating markoneraceresponsesbyspecificcombinationsof markoneormoreraces  %Bentley,_Mattingly_,HoughandBennett(2003:vii,32)findthat evenwiththemarkonerace  instruction,asignificantportionofrespondentsreportTwoormoreraces,B  ԍ(3pagevii,R1)dC  dand evenwitha t followup,asignificantportionofrespondentsreportTwoormoreraces.B  ԍdd(3page L  32,R1)dC  dDatausersmustin ` theenddecidehowtodealwiththe reluctantcaseswhencomputingbridgingparameters L  whichmayinturndepend ontheparticularpurposeanduses.B  ԍ(3pagevii,R1)dC  d 8    4.4.Discussionof_CQS_Ԁfindings   ` The_CQS_Ԁisveryimpressiveinfourrespects:  8  0   largesamplesize about25,000housingunitsperpaneland155,000respondents. (#(# ̀0   veryhighhousingunitresponserates about97percentfortheinitialcontactinboth   panels,andrecontactresponseratesof87percentinPanelAand94percentinPanelB. (#(# ̀0    representativeness eachpanelappearstoberepresentativeofCensus2000. \ AggregatedreportingofracebynonHispanicscloselyresemblesCensus2000reporting H inbothpanels.RacereportingbyHispanicsisalsosimilartoCensus2000,butinPanelA 4 asmallerproportionchoseWhiteandalargerproportionchose_SOR_Ԁcomparedwith  p Census2000data. \(#(# 0   highmatchingrate about86percentof_CQS_Ԁpersonrecordswerematchedtotheir 4 respectiveCensus2000record.B  ԍ(3page16,R2)C  Ԛ (#(# ЀDespitetheenviablesurveyexecutiondescribedabove,forthepurposesofstudyingpossible  bridgingparameters,the_CQS_Ԁhasseverallimitations:  0   toofewcasesreportingmorethanonerace despiteveryhighhousingunitresponse l rates,andahighrateofoversamplingofhouseholdswhoreportedmorethanonerace, X  thenumberofcaseswhoreportedmorethanoneracein_CQS_Ԁisquitelow.Among D! HispanicsandnonHispanicstherewereabout21,501cases #  20      ׀(orabout17.8percent 0"  of120,522totalcases)reportingmorethanoneracein_CQS_Ԁ(PanelsAandB)B  ԍ(3page59 $D # 60,R1)(3page28,30,31aR1)C  anditis #l! thosecasesthatareofmostinterestforcomputingbridgingparameters.$X"(#(# 0   fewercasesofTwoormoreracesduetoinconsistentracereporting asmentioned %0!$ intheresultssection,thereisadditionalattritiontothecasesofmajorinterestdueto &"% inconsistentracereporting(Bentley,_Mattingly_,HoughandBennett,2003:2830).B  ԍ(3page |*%) 27,R2)C  Ԁ '#& JonesandSmith(2002)alsonotethatthereisasubstantialpoolofchildrenwho (#' couldhavebeenreportedasmultiracialbutwerenot,suggestingthattheremaybe )$(  someinstabilityassociatedwithmeasuringthispopulation.B  ԍ(59page??,R1)C  ԀHowever,itmaybe |*%) 8[2 8 8  possibletoovercomethislimitationbyselectingportionsoftheinconsistentresponses  andpoolingdatafrombothpanels.(#(#  0   fewercasesduetoreluctancetoselectonerace inPanelAabout2.0percentof t nonHispanicsreportedmorethanoneraceintheinitialcontact.Aftertherecontact ` (whichaskedforonerace)therewerestill1.4percentreportingmorethanonerace. L  Evenafterprobingforonerace,0.4percentremained.8 (#(# 0   fewercasesduetosplitpaneldesign unlessthereissomestatisticallyvalidmethod  ` topoolPanelsAandB,theeffectivesamplesizeisreducedtotheobservationsavailable  L  ineachpanel.Anamelioratingfactoristhatagoodportionofthe_CQS_Ԁcaseswere  8  successfullymatchedtotheirrespectiveCensus2000records.$ (#(# 0   complexmethodologyandmultiplemodesofdatacollection inselectingthe_CQS_   methodology,apaneldesignandcontact/recontactmethodologywasselectedovera   methodofoneinstrumentwithtwoquestions.Studydesignerswereworriedaboutthe   lackofindependenceandtheconditioningeffectsofthelattermethod(seeAttachment3 p inBentley,_Mattingly_,HoughandBennett,2003:5456forthesixoptionsconsidered). \ Theybelieved thatsubstantial,butunmeasurable,interactionswilltakeplacebetween H thecollecteddataforbothmeasurementswithbothracequestionsinthesame 4 instrument(Bentley,_Mattingly_,HoughandBennett,2003:56).B  ԍ(3page27,R2)dC   p(#(# Inretrospect,itseemsthatthe_CQS_Ԁmethodologymayhaveintroducedmanymoresources H ofbias,suchastimelag,movergainsandlosses,interviewereffects,modedifferences,proxy 4 reporting,andpossiblymatchingproblems(allofwhichmaygiverisetoapparentlyinconsistent   reporting)withoutentirelyeliminatingconditioningeffectsorensuringtheindependenceof   observations.  Tables4.4and4.5show_CQS_Ԁrespondentsreportingselectedcombinationsofraces #  21      ׀   andhowtheyreportedonthealternativemeasurement. l  &    tablesTable4.4 NonHispanicsReportingSelectedCombinationsofTwoRacesinPanelA  InitialInterviewbyRecontactResponseIncludingProbe tables  *tXudd4d 4d 44d 444d 4rXs(#(#,Fdd ,dd#,dd ,dd ,Fdd ,edd ,dd +    _CQS_ԀInitial @ Contact -#!, ff -  Number -#!,# ff -  First @ Race -#!,  ff -  Second @  Race -#!,   ff -  Same @  Combination -#!,   ff -  Different @  Response -#!,  ff -  No @ Response 2(&,    ff 2WhiteBlack ZP!\ X `@105222`@Z  105,222 PFD\   `@105222 `@X PP  11.9% -#!\  X -  33.8% -#!\  X -  20.5% -#!\  X -  19.3% -#!\  X -  14.5% -#!\  X -White_AIAN_ ZP! ,X Є@129101Є@Z  129,101 PFD ,  Є@129101 Є@X P  50.1% -#! , X -  26.7% -#! , X -  Ӏ8.1% -#! , X -  Ӏ2.9% -#! , X -  12.2% -#! , X -WhiteAsian ZP!\ X ]A175034]AZ  175,034 PFD\    ]A175034 ]AX P  36.9% -#!\ ! X -  24.3% -#!\ " X -  18.5% -#!\ # X -  Ӏ6.7% -#!\ $ X -  13.7% -#!\ % X -White_SOR_ YO! , &X @32634@Y  Ӏ32,634 OEC , '  @32634 @X O  69.7% -#! , ( X -  10.1% -#! , ) X -  Ӏ3.6% -#! , * X -  Ӏ3.2% -#! , + X -  13.4% -#! , , X -Black_AIAN_ YO!\ -X d@20880d@Y  Ӏ20,880 OEC\ .  d@20880 d@X O  56.2% -#!\ / X -  10.1% -#!\ 0 X -  11.5% -#!\ 1 X -  Ӏ9.0% -#!\ 2 X -  13.2% -#!\ 3 X -Asian_NHPI_ YO!, 4f Q@24900Q@Y  24,900 OEC, 5  Q@24900 Q@f O  25.4% -#!, 6 f -  47.0% -#!, 7 f -  10.0% -#!, 8 f -  Ӏ5.5% -#!, 9 f -  12.2%1'%, :   f 1Source:DerivedfromBentley,_Mattingly_,HoughandBennett(2003:32,Table13).B  ԍ(3page32, H : R2)C  d H : '  &         h tablesTable4.5 NonHispanicsReportingSelectedCombinationsofTwoRacesinPanelB  = RecontactInterviewbyInitialContactResponse tables > *wXxddFdd Fdd#dd dd Fdd Fedd edd tXu(#(#,Fdd ,dd#,dd ,dd ,Fdd ,edd ,dd +  ?  _CQS_ L@ Recontact -#!8A ff -  Number -#!8C# ff -  First LD Race -#!8E  ff -  Second LF Race -#!8G  ff -  Same LH Combination -#!8I  ff -  Different LJ Response -#!8K  ff -  No LL Response 2(&8M    ff 2WhiteBlack QG!hNX 0A1371260AQ  137,126 PFDhO# 0A137126 0AX P*  13.3% -#!hP X -  35.6% -#!hQ X -  29.1% -#!hR X -  20.6% -#!hS X -  Ӏ1.4% -#!hT X -White_AIAN_ QG!8UX 0% A2305660% AQ  230,566 PFD8V# 0% A230566 0% AX P  58.0% -#!8W X -  23.8% -#!8X X -  14.3% -#!8Y X -  Ӏ2.9% -#!8Z X -  Ӏ0.9% -#!8[ X -WhiteAsian QG!h\X  A211546 AQ  211,546 PFDh]#  A211546  AX P  25.2% -#!h^ X -  31.5% -#!h_ X -  31.0% -#!h` X -  Ӏ11.8% -#!ha X -  Ӏ0.4% -#!hb X -White_SOR_ QG!8cX A171512AQ  171,512 PFD8d# A171512 AX P  76.1% -#!8e X -  14.7% -#!8f X -  Ӏ2.4% -#!8g X -  Ӏ6.3% -#!8h X -  Ӏ0.4% -#!8i X -Black_AIAN_ PF!hjX @37927@P  Ӏ37,927 OEChk# @37927 @X O  65.8% -#!hl X -  13.4% -#!hm X -  11.8% -#!hn X -  Ӏ6.5% -#!ho X -  Ӏ2.5% -#!hp X -Asian_NHPI_ PF! 8qf Z@35543Z@P  35,543 OEC 8r# Z@35543 Z@f O  34.9% -#! 8s f -  26.6% -#! 8t f -  25.3% -#! 8u f -  12.2% -#! 8v f -  0.9%1'% 8w   f 1Source:DerivedfromBentley,_Mattingly_,HoughandBennett(2003:33,Table14).B  ԍdd(3page33, "@x R1)dC  d "Tw 'TFirst,wecanseethatmanymorerespondentsdidnotanswerinPanelA(morethan12.0percent) $ z !wheretheinitialcontactasked markoneormoreracesandtherecontact(andprobequestion) %!{ asked markonerace!thaninPanelB(nomorethan2.5percent).Thisisnotsurprisingbecause &!| _CQS_Ԁdeliberatelyoversampledthe Twoormoreracespopulation,soitisreasonabletoexpect '"} thatinPanelArespondentsmayhavebeenreluctanttoreportonlyonerace.Ontheotherhand, x(#~ inPanelB,onemighthaveexpectedthat,havingbeenrestrictedtooneraceinitially,these d)$ respondentswouldhavebeeneagertoreportmorethanonerace. P*% $ -d( P4     Inbothpanels,theproportionsgivingthe sameresponse inbothmeasurementswas10percent  orhigher(exceptfor White and Someotherraceand White and AmericanIndianand  AlaskaNative). White and Blackand White and Asianweremostlikelytoprovidethe  sameresponse(about20percentinPanelAtoabout30percentinPanelB).Fairlysubstantial t proportionsinbothpanelsgavedifferentorinconsistentresponses(rangingfrom2.9to ` 20.6percent). White and Blackrespondentswereparticularlysusceptibletothis(19.3percent L  inPanelAand20.6percentinPanelB),while White and AmericanIndianandAlaskaNative 8  respondentswereamongleastsusceptible(2.9percent)inbothpanels.Often,whenaskedto $ t reportmorethanonerace,respondentsmayreporttheirraceas multiracial, mixed,or  `  biracial,whichincensusproceduresgetcodedas Someotherrace.Additionalanalysis  L  oftheseresponsesshouldbedone.  8     &  tablesTable4.6 PercentofNonHispanicsReportingSelectedCombinationsofTwoRaces   ProvidingorNotProvidingOneConsistentRacebyPanel tables;-   *yXzddFdd Fdd#dd dd Fdd Fedd edd wXx(#(#,dd ,dd ,dd ,dd , dd +      Combination P  Reportedin_CQS_ -#!< ff -  PanelA!one P  consistentrace -#!<  ff -  PanelA!no P  consistentrace -#!<  ff -  PanelB!one P  consistentrace -#!<  ff -  PanelB!no P  consistentrace 6,&<    ff 46WhiteBlack -#!<X -  45.7% -#!< X -  54.3% -#!< X -  48.9% -#!< X -  51.1% 7-'<  40 X 47White_AIAN_ -#!X -  76.8% -#! X -  23.2% -#!  X -  81.8% -#!! X -  18.2% 7-'"  40 X 47WhiteAsian -#!,#X -  61.1% -#!,$ X -  38.9% -#!,% X -  Ӏ56.7% -#!,& X -  43.3% 7-','  40 X 47White_SOR_ -#!,|(X -  79.8% -#!,|) X -  20.2% -#!,|* X -  Ӏ90.9% -#!,|+ X -  Ӏ9.1% 7-',|,  40 X 47Black_AIAN_ -#!|-X -  66.3% -#!|. X -  33.7% -#!|/ X -  79.2% -#!|0 X -  20.8% 7-'|1  40 X 47Asian_NHPI_ -#!2f -  72.4% -#!3 f -  27.6% -#!4 f -  61.6% -#!5 f -  38.4%7-+6  40  f 7Source:derivedfromBentley,_Mattingly_,HoughandBennett(2003:3233,Tables13and14).V  B  W[:ԍ(3 8 page32,R2;dd3page33,R1)V  C  dW:Ԛ' ,- 86 Table4.6shows_CQS_Ԁrespondentsreportingselectedcombinationsofracesandwhetherthey 8 reportedoneconsistentraceinthealternativemeasurement!forexample,someonereporting 9  White and Asianinonequestionand Whiteor Asianintheotherisaconsistentanswer. : Althoughsomerespondentsdidreportoneraceinthealternatequestion,sometimesthatrace  ; wasnotconsistent(e.g.,someonereporting White and Asianinonequestionand Blackin p!< theotherisaninconsistentanswer).Additionalresearchontheseinconsistentresponsesneeds \"= tobedone. H#> Ingeneral, White and BlackorAfricanAmericanrespondentsinbothpanelsweremost  %p @ resistanttoselectingoneconsistentrace(54.3and51.1percentinPanelAandBrespectively),  &\!A while White and Someotherracerespondentswereleastresistant(20.2and9.1percent &H"B respectively).Thesignificanceofthesefindingsisthatsubstantialproportionsofrespondents '4#C refusedorwereunabletogiveustheinformationweneedtocalculate bridgingparameters, ( $D andtherebyfurtherreducethenumberofusefulcases. ) %E   *%F  &   tablesTable4.7 Example BridgingParametersforNonHispanicsReporting  SelectedCombinationsofTwoRacesandOneConsistentRacebyPanel tables(A  *{X|dddd dd dd dd  dd yXz(#(#,dd ,odd ,dd ,dd , dd +    Combination @ Reportedin_CQS_ -#!, ff -  PanelA! @ firstrace -#!,  ff -  PanelA! @ secondrace -#!,  ff -  PanelB! @  firstrace -#!,   ff -  PanelB! @  secondrace 6,&,     ff 46WhiteBlack -#!, | X -  26.0% -#!, | X -  74.0% -#!, | X -  27.3% -#!, | X -  72.7% 7-', |  40 X 47White_AIAN_ -#!| X -  65.2% -#!|  X -  34.8% -#!|  X -  70.9% -#!|  X -  29.1% 7-'|   40 X 47WhiteAsian -#! X -  60.3% -#!  X -  39.7% -#!  X -  Ӏ44.4% -#!  X -  55.6% 7-'   40 X 47White_SOR_ -#! lX -  87.3% -#! l X -  12.7% -#! l X -  Ӏ83.8% -#! l X -  16.2% 7-' l   40 X 47Black_AIAN_ -#!l !X -  84.8% -#!l " X -  15.2% -#!l # X -  83.0% -#!l $ X -  17.0% 7-'l %  40 X 47Asian_NHPI_ -#! &f -  35.1% -#! ' f -  64.9% -#! ( f -  56.8% -#! ) f -  43.2%7-+ *  40  f 7Source:DerivedfromBentley,_Mattingly_,HoughandBennett(2003:3233,Tables13and14).B  ԍ(3 ( * page32,R2;d3page33,R1)dC  d' A ( * Consideringonlythosecaseswhichprovidethenecessaryinformationforcomputingbridging , parameters(thatis,racequestionsareansweredinbothinstruments,amultipleraceresponseis - providedinoneinstrument,anda consistentsingleraceresponseisprovidedintheother . instrument),whatproportionofselectedcombinationsselectoneraceovertheother?Table4.7 t/ showssomeexamplebridgingparameterscomputedbyignoringallcasesthatdidnotreport `0 oneconsistentrace.ForexampleinPanelA,among White and BlackorAfricanAmerican L1 respondents whodo selectoneconsistentrace,26.0percentselect Whiteand74.0percent 82 select BlackorAfricanAmerican.Despitethedifferentmethodologies,PanelBshowsvery $t3 similarproportions!27.3percentselect Whiteand72.7percentselect BlackorAfrican `4 American.However,thesecalculationsignore overhalf ofthe White and BlackorAfrican L5 Americanrespondents,asseeninTable4.6above.Weseesimilarconsistencybetweenpanels 86 for BlackorAfricanAmerican and AmericanIndianandAlaskaNative.About84.8percent $7 select BlackorAfricanAmericaninPanelAand83.0percentinPanelB.Among Whiteand 8 AmericanIndianandAlaskaNativerespondents,65.2percentselectedWhiteinPanelA,and 9 70.9percentinPanelB.About87.3percent(PanelA)and83.8percent(PanelB)of White and  : Someotherracerespondentsselect White.Inthecaseof White and Asianand Asian and   ; NativeHawaiianandOtherPacificIslanderPanelAandBproducecontradictoryparameters. p!< InPanelA,39.7percentof White and Asianselect Asian,whileinPanelBthatproportion \"= is55.6percent.SimilarlyinPanelA,35.1percentof Asian and NativeHawaiianandOther H#> PacificIslanderselect Asian,comparedwith56.8percentinPanelB. 4$? Althoughmuchmoreanalysisneedstoconducted,aquestionthatneedstobeanswerediswhich  &\!A bridgingparametershouldbeusedforanyracecombination.ShoulditcomefromPanelAor &H"B PanelB,orfromapooledsampleofAandB?Inaddition,matchingCensus2000recordsto '4#C _CQS_Ԁrecordsaffordsusatleasttwomorepossiblesourcesofbridgingparameters(Census2000 ( $D toPanelArecontactandCensus2000toPanelBinitialcontact).Itisunknownwhetherthese ) %E mayyieldeitherdifferentparametersor,worse,inconsistentparameters.Unfortunately,atthis *%F stagethereisnoaprioriwaytodecidewhichapproachyieldsthebestbridgingparameters. +&G Inanyevent,welosecasesbecausesignificantproportionsofrespondentsdonotprovideone ,'H consistentraceinthealternatequestion.     p      x   l-(I r6     ;XX\rP  RPr9P  5.ComparingtheRaceandHispanicOriginDatafromtheAmericanCommunity  Survey(ACS)andCensus2000 #X\X;\# -q] h       OneofthemainobjectivesoftheAmericanCommunitySurvey(ACS)istoserveasa  replacementforthelongforminthe2010Census.Anotheristoprovideacontinuouslyupdated  sourceofdemographic,socioeconomic,andhousingdataforsmallareasandpopulationgroups, p  eitherassingleyearestimatesormultiyearaverages(BennettandGriffin2002:206).B  ԍ(45page \  206,R1)dC  dInthis \  chapter,IwillconcentrateonhowraceandHispanicorigindifferinCensus2000andtheCensus H  2000SupplementarySurvey(C2SS)basedontheworkofBennettandGriffin(2002);Leslie, 4  _Raglin_,and_Schwede_Ԁ(2002);_Raglin_ԀandLeslie(2002);and_Schwede_,Leslie,andGriffin(2002).B  ԍ(originally=5page1,R1)dC  d  p    5.1.Studydesign -b H  TheprimaryobjectiveofC2SS wastoevaluatethefeasibilityofcollectinglongformdata   outsidethedecennialcensusduringCensus2000(BennettandGriffin,2002:206).B  ԍdd(45page206,   R2)dC  dTheC2SS   wasasurveyofabout700,000housingunitsusingtheACSmethodology.Itwasanoperational   feasibilitytesttolearnhowtocollectlongformdataatthesametimeas,butseparatelyfrom,  Census2000.TheC2SSwasthefirstlargescalenationaldatacollectionusingtheACSmethods  (_Raglin_ԀandLeslie,2002:2826).B  ԍ(47page2826,R1)dC  dTheC2SSusedthequestionnaireandmethodsdevelopedfor l theACStocollectdemographic,social,economic,andhousingdatafromanationalsampleof X households.C2SSdatacollectionbeganinJanuary2000andranthroughDecember2000. D TheC2SSwasconductedin1,203counties,andwhentheoriginal31siteswereadded,thefull 0 samplesizewaslargeenoughtoproducedataforeverystate,andmostcountiesandmetropolitan l areaswithpopulationsof250,000ormore(BennettandGriffin,2002:206).B  ԍ(45page206,R1)dC  d X Datawerecollectedinthreephases.First,aprenoticeletterwassenttoeachsampledunit, 0 followedbyaquestionnaireinthemailaweeklater.Ifnecessary,theinitialmailquestionnaire  wasfollowedbyaremindercard,andafterthreeweeks,ar_eplacement_Ԁquestionnairewassent.  Second,atelephonefollowupwasattemptedtoobtaininformationfromhouseholdsthatdid  notreturnthereplacementquestionnaire.Third,asampleof_nonrespondents_Ԁwasselectedfor  apersonalvisitinterview._Nonresponse_Ԁfollowup(_NRFU_)interviewswereconductedby |  permanentsurveyfieldrepresentativesusingcomputerassistedtechnology(BennettandGriffin, h! 2002:207).B  ԍ(45page207,R1)dC  d T"    5.2.Limitations -l ,$|" }X\XXX\OnemighthaveexpecteddifferencesbetweenCensus2000andtheC2SSbecausetheyhad &T!$ differentpurposesand,therefore,haddifferentdesignandimplementationmethods.TheC2SS &@"% collecteddatacontinuouslythroughouttheyearusingacombinationofmail,telephone,and ',#& personalvisitfollowupwhichlastedoverathreemonthperiod.Census2000,ontheother ($' hand,wasasinglemassivedatacollectionoveraveryshortperiodfromlateMarch2000to )%( July2000thatincludedaninitialmailoutmodeandsubsequentpersonalvisit_NRFU_Ԁinterviews *%) inasmanynonrespondinghouseholdsaspossible#X\XX}X\l#}X\XXX\.Asafinalresort,Census2000allowed +&* proxyresponsesfromnonhouseholdrespondents,suchasneighbors,whiletheC2SSdidnot. x,'+  d-(, Therewereseveralotherimportantdifferences#X\XX}X\p#}X\XXX\ԀbetweenCensus2000andtheC2SS:theC2SS  hadfollowupproceduresformissingitemsonmailreturns,whileCensus2000didnot;  #X\XX}X\Kq#FX\XXX\questionnairesdiffered,residencerulesandreferenceperiodsdiffered,andsomeeditingand  allocationproceduresvaried.Additionally,followupdatawerecollectedinpersonusingpaper t questionnairesinCensus2000,butbyphoneorinpersonusingautomatedinstrumentsinC2SS.B  ԍ(perBetsyMartin7.22.03e.mail)dC  d ` Inaddition,censusenumeratorswer#X\XXFX\:r#}X\XXX\etemporaryworkersandwerenotaswelltrainedoras L  experiencedasC2SSfieldrepresentatives(_FRs_).#X\XX}X\?t#}X\XXX\ԀFinally,theC2SSestimatesaresubjectto 8  samplingerrorbecausetheyarebasedonasampleofthepopulation,#X\XX}X\u#}X\XXX\whiletheshortformcensus $ t totalsarenot#X\XX}X\u#(BennettandGriffin,2002:208).B  ԍ(45 $  page208,R1)(originally=5page2,R1,page3,R1)C  }X\XXX\ԀMoreover,comparisonsbetweenCensus2000  ` andC2SSarelimitedtothehouseholdpopulationbecausebydesigntheC2SSdidnotinclude  L  thepopulationlivingingroupquarters.  8  Ѐ#X\XX}X\v# $    5.3.Findingsinbrief th-x          h      p   Althoughother100percentitemsareavailableforcomparisonbetweenCensus2000andC2SS,   thediscussioninthischapterfocusesontheHispanicoriginandracevariables.   5.3.1.ReportingofHispanicorigin \  &  tablesTable5.1 Census2000andCensus2000SupplementarySurvey(C2SS)HispanicResponses 4 (HouseholdPopulationOnly) tablesz $t  *~Xdddd odd odd dd  dd {X|(#(#,` dd ,Fdd ,dd ,Fdd ,dd +  ``  HispanicOrigin 6,!` ff 6Census2000 t (1) 6,!`" ff 6C2SS t (2) 6,!`" ff 6Difference t (3=21) 6,!`" ff 6Percent t Difference ` (4=3/1) 6,&L "   ff 46HispanicorLatino: -#!!X -  12.6% -#!" X -  12.6% -#!# X -  Ӏ! -#!$ X -  Ӏ! 7-'%  40 X 47ЀMexican -#!<&X -  Ӏ7.4% -#!<' X -  Ӏ7.9% WM!<( X ?0.5?W  Ӏ0.5 MCA<)  ?0.5 ?X M  Ӏ6.8% 7-'<*  40 X 47ЀPuertoRican -#!<+X -  1.2% -#!<, X -  1.3% WM!<- X ?0.1?WЀ0.1 MCA<. ?0.1 ?X MЀ8.3% 7-'</ 40 X 47ЀCuban -#!0X -  0.4% -#!1 X -  0.5% ND!2 X ?0.1?N  Ӏ0.1 MCA3  ?0.1 ?X M  25.0% 7-'4  40 X 47ЀOtherHispanicorLatino -#! ,5f -  3.6% -#! ,6 f -  3.0% OE! ,7 f 333333-0.6333333O  Ӏ 0.6 NDB ,8  333333-0.6 333333f N  ӄ16.7%7-+ ,9  40  f 7h XX\Note: Bold numbersinColumn3indicatesignificantdifferencesatthep<.10level#X\X hr#. !L9 Source:AdaptedfromBennettandGriffin(2002:210Table6)'Hz "8: BennettandGriffin(2002:210)foundnodiscernibledifferencesintheproportionofHispanic $ < originresponses,althoughthereweresignificantdifferencesinthedetailedHispanicorigin % = responses.B  dd(45page210,R1)C  ddB  dd(Originally=5page51,R1)C  ddTable5.1showsthat,comparedwithCensus2000,C2SSproducedabout &!> 6.8percentmoreMexicans.Ontheotherhand,the OtherHispaniccategorywasabout '"? 16.7percentless.TheproportionofCubansandPuertoRicanswerenotstatisticallydifferent.  p(#@ R8      5.3.2.DiscussionofHispanicorigin  Presumablythelowerproportioninthe OtherHispaniccategoryinC2SSreportedbyBennett  andGriffin(2002:210)reflectsfewergeneralHispanicresponses( Hispanic, Spanish,and t  Latino),asshowninotherresearch(see_Cresce_ԀandRamirez,2003;Logan2002;and_Suro_ ` 2002).BennettandGriffin(2002:210)speculatethattheobserveddifferencesareduetotheuse L  ofexamplesintheC2SS.Duringtelephoneandpersonalvisitinterviewing,respondentswere 8  readorshownexamplesforthe OtherSpanish/Hispanic/Latinocategorysimilartothoseused $ t inthe1990census.B  dd(45page210,R1)C  ddB  C  ԀB  ԍ(originally=5page52,R1)dC  dTheseaidswerenotprovidedduringCensus2000operations,althoughone  ` couldarguethatthepresenceoftheHispanicorigincheckboxgroupsactasexamples.B  dd(perBetsyMartins7.22.03e.mailcomment)C  ddThisdoes  L  notexplainwhytheMexicanpercentageisalsolowerinCensus2000!thesecategorieswere  8  presentinalldatacollections.ThePuertoRicanandCubanproportionsalsoshowsthesame $  patternbutwerenotstatisticallysignificant.    &  tablesTable5.2 HispanicOriginQuestionbyQuestionnaireType tables   *Xdd` dd ` Fdd Fdd Fdd Fdd ~X(#(#, dd ,dd +      Questionnaire -#!< ff -Hispanicoriginquestion 6,&<   ff 46 Census2000  < FormD2(_mailback_Ԁlongform) (x !personbasedorlinearlayout -#!dX - IsthispersonSpanish/Hispanic/Latino? MarkX < the No boxif not Spanish/Hispanic/Latino. 7-',| 40 X 47 AmericanCommunitySurvey d  FormACS1(2000)!matrixlayout -#!PX - IsthispersonSpanish/Hispanic/Latino? MarkX d the No boxif not Spanish/Hispanic/Latino. 7-'T 40 X 47 EnumeratorQuestionnaire   FormD2(E) -#!X - AreanyofthepersonsthatIhavelisted  Mexican,PuertoRican,Cuban,orofanother  HispanicorLatinogroup? 7-'| 40 X 47 AmericanCommunitySurvey    CATI/_CAPI_Ԁinstrument -#!!f -Part1. Is/AreyouSpanish,Hispanic, " orLatino?     9   #  Part2. Is/AreyouofMexicanorigin, % PuertoRican,Cubanorsomeother  & Spanish/Hispanic/Latinogroup? 7-+p!' 40  f 7'AlthoughtheformatandwordingoftheHispanicoriginquestiononthemailquestionnaireusedin d$) C2SSandCensus2000weresimilar,thereweredifferencesintheotherinstruments(seeTable5.2). P% * TheACS_CATI_/_CAPI_ԀinstrumentshadexamplesfortheotherSpanish/Hispanic/Latinocategory <&!+ (e.g.,Argentinean,Columbian,Dominican,Nicaraguan,Salvadoran,Spaniard),butthedecennial ('x", mailandenumeratorinstrumentsdidnothaveexamples.Thebasicresponsecategoriesweresimilar, (d#- buttheCensus2000mailquestionnairecategoriesweredoublebankedV  ԀWB(BennettandGriffin, )P$. 2002:207).B  ԍ(45page207,R2)dC  dHavingtheonequestionsplitintotwoseparatequestionsin_CATI_/_CAPI_Ԁwould )<%/ presumablymakeiteasiertoaskandanswerininterviewsituations.Thiseffectivelyreducesthe *(&0 doublenegativestatement Mark[X] No boxif not Spanish...foundonthemailquestionnaires +'1 (_Schwede_,2003personalcommunication).Itmakessensethattheuseofexperiencedinterviewers ,(2 P9     toprobeforresponsesinotherdatacollectionsmayhavecontributedtogettingmoredetailinC2SS  thanCensus2000.Thisargumentisexploredmorevigorouslyinexplainingthedifferencesinrace  reporting.  `    5.3.3.Racereporting ` BothCensus2000andtheC2SSallowedrespondentstoreportoneormoreraces.Bennett 8  andGriffin(2002:208210) foundsignificantdifferences betweenC2SSandCensus2000 $ t  distributionsforboththeracealoneandracealoneorincombinationcategories. #  22      ׀While  ` theauthorsfoundanumberofdifferencesintheracedistributions,thepercentofrespondents  L  reporting Whitealone and Someotherracealoneshowedthegreatestdifferenceinthe  8  distributions.Inaddition,theC2SSdistributionhadasignificantlylowerproportionof (  respondentsreporting Twoormoreraces.Smallbutsignificantdifferencesalsoexistfor   }X\XXX\ BlackorAfricanAmericanalone AmericanIndianorAlaskaNativealone,and    Asianalone#X\XX}X\k#(BennettandGriffin2002:208).B  dd(45page208,R2)C  ddB  ԍ(originally=5page5,R1)dC  d}X\XXX\   #X\XX}X\#}X\XXX\#X\XX}X\A#}X\XXX\ItisimportanttocomparetheracedistributionsforHispanicsandnonHispanicsbecause t reportingpatternstendtobequitedifferentforHispanicrespondents.Table5.3showsthat ` theracedistributionfornonHispanicsinC2SSisnotverydifferentfromthatofCensus2000. L Thereweresignificantdifferencesforalloftheracegroups,exceptfor NativeHawaiianand 8 OtherPacificIslander.ThelargestdifferencebetweenCensus2000andC2SSwasforthe $t  Someotherracealonepopulation.ComparedwithCensus2000,C2SShadslightlymore ` reportsof Whitealone(0.4percent)and Asianalone(3.9percent),andfewerreportsof L  AmericanIndianandAlaskaNative(9.5percent)and Twoormoreraces(7.9percent). 8 When Twoormoreraces#X\XX}X\#}X\XXX\isbrokeninto TworaceswhichincludeSomeotherrace󀀀 $ and Allotherracecombinations,weseethatCensus2000hadproportionatelymorerace  combinationsthatincluded Someotherrace#X\XX}X\#}X\XXX\ԀasoneoftheracesthandidC2SS(0.50versus  0.15percent).Ontheotherhand,C2SShadproportionatelymorereportsofallotherrace  combinationsthandidCensus2000(1.59#X\XX}X\#}X\XXX\Ԁand1.39percent,respectively).  #X\XX}X\1#}X\XXX\ #X\XX}X\# \   & (  tablesTable5.3 Census2000andCensus2000SupplementarySurvey(C2SS)  SelectedRaceResponsesbyNonHispanics tables'Ԁ(HouseholdPopulationOnly)   *Xdd dd dd X(#(#, dd ,dd ,dd ,dK ,dd +    Race l  A -#!, ff -  Census @ 2000 , (1) -#!  ff -  C2SS @ (2) 6,!,   ff 6Difference @  (3=21) -#!, " ff -  Percent @  Difference ,  (4=3/1)  ǰ  2(&     ff 2:White -#!4 X -  79.30% -#!4  X -  79.58% XN!4  X Q?0.28Q?X0.28 NDB4 " Q?0.28 Q?X N@ D 0.4% -#!4 X -BlackorAfricanAmerican -#! X -  13.49% -#!  X -  13.21% YO!  X Qѿ-0.28QѿYЄ0.28 OEC " Qѿ-0.28 QѿX O  ӄ2.1% -#!  X -AmericanIndianand 4  AlaskaNative -#! p X -  Ӏ0.84% 4    -#! p X -  Ӏ0.76% YO!4  X {Gz-0.08{GzYЄ0.08 4    XNC p  {Gz-0.08 {GzX XЄ9.5% -#!4 !"X -Asian -#! "X -  Ӏ4.15% -#! # X -  Ӏ4.31% XN! $ X {Gz?0.16{Gz?X0.16 WMB %" {Gz?0.16 {Gz?X W3.9% -#! &"X -NativeHawaiianand  p ' OtherPacificIslander  -#! \(X -  Ӏ0.14% -#! p ) X -  Ӏ0.16% XN! p * X {Gz?0.02{Gz?X0.02 WMB p +" {Gz?0.02 {Gz?X W14.3% -#! p ,"X -Someotherrace -#!-X -  0.19% -#!. X -  0.25% XN!/ X Q?0.06Q?X0.06 WMB0" Q?0.06 Q?X W31.6% -#!1"X -Twoormoreraces -#! \2X -  1.89% -#! \3 X -  1.74% YO! \4 X 333333ÿ-0.15333333ÿYЄ0.15 XNC \5" 333333ÿ-0.15 333333ÿX XЄ7.9% 6,! \6"X 6}X\XXX\󀀀Tworaceswhichinclude 7 ЀSomeotherrace#X\XX}X\a# 6,!x8X 60.50% 6,!9"X 60.15% YO!:"X ffffffֿ-0.35ffffffֿYЄ0.35 XNC;" ffffffֿ-0.35 ffffffֿX XЄ70.0% 6,!<"X 6}X\XXX\ԀAllotherracecombinations#X\XX}X\h# 6,!H=f 61.39% 6,!H>"f 61.59% XN!H?"f ?0.20?X0.20 WMBH@" ?0.20 ?f W14.4%1'%HA" f 1h XX\Note: Bold numbersinColumn3indicatesignificantdifferencesatthep<.10level#X\X h;#. dA Source:AdaptedfromBennettandGriffin(2002:208209Table2andTable4) PB '(Table5.4showstheracedistributionforHispanics.ComparedwithCensus2000,C2SShas (D about31percentmorereportsof}X\XXX\ Whitealone,about30percentfewer Someotherrace#X\XX}X\#}X\XXX\ E reports,andabout24percentfewerreportsof Twoormoreraces#X\XX}X\#}X\XXX\ԀamongHispanics. F When Twoormoreraces#X\XX}X\5#}X\XXX\werebrokeninto TworaceswhichincludeSomeotherrace󀀀  G and Allotherracecombinations,Census2000hadproportionatelymoretworace !H combinationsthatincluded Someotherrace#X\XX}X\#ԀasoneoftheracesthandidC2SS(5percent t"I versus3percent),andproportionatelyfewerof Allotherracecombinations. `#J   L$K 9P; 9  9  &   tablesTable5.4 Census2000andCensus2000SupplementarySurvey(C2SS)  SelectedRaceResponsesbyHispanics(HouseholdPopulationOnly) tablesC  *Xd d dd dd dd dK dd X(#(#, dd ,dd ,dd ,edd ,dd +    Race" 6,!, ff 6Census @ 2000 , (1) 6,!" ff 6C2SS @ (2) 6,!, " ff 6Difference @  (3=21) 6,!, " ff 6Percent @  Difference ,  (4=3/1) 2(&"   ff 2HispanicorLatino: -#!H X -  100.00% -#!H  X -  100.00% -#!H  X -  ӄ -#!H  X -  ӄ -#!H  X -ЀWhite -#! X -  Ӏ47.89% -#!  X -  Ӏ62.91% YO!  X  ףp= .@15.02 ףp= .@Y @ 15.02 XNC "  ףp= .@15.02  ףp= .@X X31.4% -#! "X -ЀSomeotherrace -#!H X -  42.21% -#!H  X -  29.39% ZP!H  X p= ף)-12.82p= ף)ZЄ12.82 YODH " p= ף)-12.82 p= ף)X YЄ30.4% -#!H "X -ЀTwoormoreraces -#! X -  6.31% -#!  X -  4.79% YO!  X RQ-1.52RQYЄ1.52 XNC !" RQ-1.52 RQX XЄ24.1% -#! ""X -ЀTworaceswhichinclude H # ЀSomeotherrace -#!4 $X -  5.09% -#!4 & X -  3.35% YO!4 ( X ףp= -1.74ףp= Ȳ1.74 XNC4 *" ףp= -1.74 ףp= X X̄34.2% -#!4 ,"X -ЀAllotherracecombinations -#!-f -  1.22% -#!. f -  1.45% XN!/ f q= ףp?0.23q= ףp?X0.23 WMB0" q= ףp?0.23 q= ףp?f W18.9%1'%1"  f 1h XX\Note: Bold numbersinColumn3indicatesignificantdifferencesatthep<.10level#X\X he#.  1 Source:AdaptedfromBennettandGriffin(2002:210Tables3and4)B  ԍ(45pages209,R4R5)dC  d  2 '*5.3.4.Discussionofracereporting 4 QuestionWording.Whilethewordingandresponsecategoriesofthemailquestionnairesfor l6 Census2000andC2SSwereidentical(seeTable5.5),thereweredifferencesintheformatof X7 thequestionnaires.Withtheexceptionofthe_nonresponse_Ԁfollowupquestionnaire,Census2000 D8 questionnaireswerepersonbased(severalquestionsaskedofeachindividual),whileC2SSwas 09 matrixbased(characteristicsofallrespondentsinahouseholdwerecollectedinacolumn l: format).ThewordingoftheracequestionsusedintelephoneandpersonalvisitsinC2SSand X; Census2000differedfromthemailversionsandfromeachother.Someofthedifferences D< wereneededtoaccommodatethedatacollectionmode,butotherdifferencesdidnotappearto 0= benecessary.Oneofthemostnotabledifferenceswasthatboththemailandtheenumerator > decennialquestionnairesaskedfortheraceorracesthatarespondent considers himself/herselfto  ? be,whiletheC2SS_CATI_/_CAPI_Ԁquestionnaireaskedthecategoryorcategoriesthat bestindicate  !@ therespondentsrace,whichmaybemeasuringdifferentcognitivedomains.TheC2SS "A _CATI_/_CAPI_Ԁinstrumentsalsohadexamplesforthe OtherAsian(e.g.,Cambodian,_Hmong_, |#B Thai,Indonesian)and OtherPacificIslander(e.g.,Tahitian,Fijian)categories,whiletheother h$C  threedidnot(BennettandGriffin,2002:207). #  23      B  ԍ(45page207,R3)dC  dB  ԍ(49page2064,R1)dC  d T% D  &   tablesTable5.5 RaceQuestionbyQuestionnaireType tablesK  *Xdd dd dd dd edd edd X(#(#, 44 ,44 +    Questionnaire -#!T ff -Racequestion 2(&T   ff 2 Census2000   FormD2(_mailback_Ԁlongform)! p personbasedorlinearlayout -#!\ X - Whatisthispersonsrace?MarkXoneormore  races toindicatewhatthispersonconsiders p himself/herselftobe. -#!\  X - AmericanCommunitySurvey  ,   FormACS1(2000)!matrixlayout -#!  X - Whatisthispersonsrace?MarkXoneormore  ,  races toindicatewhatthispersonconsiders    himself/herselftobe. -#! X - EnumeratorQuestionnaire 4   FormD2(E)    D   -#! p X - Nowchooseoneormoreracesforeachperson. 4  Whichraceorracesdoeseachpersonconsider  p  himself/herselftobe? -#! \ X - AmericanCommunitySurvey    CATI/_CAPI_Ԁinstrument -#!x f - {Showrespondent_flashcard_ԀB}Imgoingto   readyoualistofracecategories.Please/Using x  thislist,please/chooseoneormorecategories d thatbestindicate{s/yourrace. 1'%P  f 1      '2Despitethesubtledifferencesinthemethodologies,_Schwede_,Leslie,andGriffin(2002:3136) D notetheseracequestionsshareacommoncharacteristic: 0 0  TheresponsecategoriesforraceonthecensusandACSpresentastrangepasticheX(#(# 0  ofskincolor(whiteandblack),internalindigenousethnicgroups(e.g.,American D Indian/AlaskaNative),U.S.IslandAreas(e.g.,Samoa),nationality(e.g.,Japanese), 0  andgeographicalregionformanycountries(otherAsian).B  ԍ(48page3136,R1)dC  d!(#(# InterviewerEffects.}X\XXX\InexaminingdatafromCensus2000andC2SS#X\XX}X\#}X\XXX\,_Schwede_#X\XX}X\R#}X\XXX\,Leslie,and # Griffin(2002:3134)foundunexpectedlylargedifferencesinthedistributionofrace,particularly $ amongHispanicsinintervieweradministereddatacollections.B  ԍ#X\XX}X\#dd(48page3134,R1)d}X\XXX\ԍC  dTheynotethat#X\XX}X\ #}X\XXX\aboutthesame | % percentage(46percent)ofHispanicsreportedaraceof Whiteasreported Someotherrace h!& inenumeratorcollecteddatainCensus2000.Ontheotherhand,#X\XX}X\n#}X\XXX\morethantwiceasmany T"' Hispanicsreportedas White(64percent)asreported Someotherrace#X\XX}X\#}X\XXX\Ԁ#X\XX}X\m#}X\XXX\(30percent)inthe @#( C2SSdatacollectedbyinterviewers.#X\XX}X\#B  ԍ(48page3134,R2)dC  d}X\XXX\ ,$|) Basedonthatfinding,theCensusBureauconductedtwostudies.Thefirstwasasemistructured &T!+ debriefingstudyofACSinterviewers(Leslie,_Raglin_,and_Schwede_,2002)#X\XX}X\#}X\XXX\.#X\XX}X\#Theauthors &@", hypothesizethattheracereportingdifferencesmaybedueto interviewerbehaviorB  ԍ(49page ',#- 2065,R3)dC  dcaused ',#- bydifferencesinexperienceandtraining: ($. 0  C2SSinterviewersareexperienced,welltrained,andlongterminterviewerswhowork *%0 onotherdemographicsurveys,B  ԍ(49page2065,R3)dC  dbutCensus2000interviewerswerehiredjustforCensus +&1 2000.B  ԍ(49page2065,R4)C  x,'2(#(#  d-(3 r=     0  MostCensusBureaudemographicsurveysaskpreCensus2000raceandHispanic  originquestionswhichdonotaskformorethanoneraceanddonotallowreporting(#(#   of Someotherrace.B  ԍ(49page2065,R5)C          0  Insomesurveys,interviewers havebeentrainedtomarkracebyobservationifthe ` respondentsrefuseincertainsituations.L (#(# 0  UnliketheCensus2000,theC2SS_flashcard_Ԁdoesnotincludeaninstructionthat $ t respondentsmayselectmorethanonerace.B  ԍ(49page2065,R7)C  Ԛ `(#(# }X\XXX\AlthoughthisstudyoccurredwellafterCensus2000andisbasedonreportednotobserved  8  behavior,itsuggeststhepossibilitythatsomeinterviewersmayhaveusedactiveprobeswhich $  mighthaveinfluencedreportingofspecificracesresponses(Leslie,_Raglin_,and_Schwede_,   2002:2068)#X\XX}X\O#}X\XXX\.#X\XX}X\1#}X\XXX\Theauthorshypothesizethattheracereportingdifferencesmaybedueto    interviewerbehavior.B  ԍ(49page2068,R1)d#X\XX}X\u#ԍC  dInanotherstudyofthedebriefingdata,_Schwede_,Leslie,andGriffin   (2002)foundthatfeweryearsofexperience,regionofthecountry,andinterviewerinterpretation   ofwhattheracequestionwasaskingwereassociatedwith_FRs_Ԁ(fieldrepresentatives)accepting p andrecording Hispanicasaresponsein otherrace. \ Whatisparticularlyinterestingaboutthisstudyisthat widedifferencesin_FRs_interpretations 4 ofwhattheracequestionisaskingforsuggestinterviewersinterpretationsoftheracequestion  p maydifferfromregiontoregionaswell(_Schwede_,Leslie,andGriffin,2002:3136).B  ԍ(48page  \ 3136,R2)dC  dInfact,  \ infocusgroups,_FRs_Ԁ pressed...researchershardtoexplainjustwhatitisheadquarterswants H tocollectwiththeracequestion.B  ԍ(48page3136,R3)dC  dThequestionsthemselvesleavesomedoubtastowhatis 4 wanted:V  ԀWinmailquestionnairestheracequestionasksfortheraceorracestherespondent    considers him/herselftobe,whiletheACSCATIand_CAPI_Ԁaskforoneormorecategories   that bestindicate therespondentsrace.B  ԍ(48page3136,R3)C  Ԛ  }X\XXX\ThesecondstudyexaminedamatchedsampleofCensus2000#X\XX}X\/#}X\XXX\andC2SSrecords.#X\XX}X\#Ԛ}X\XXX\_Raglin_Ԁand  Leslie(2002:2827)matchedrespondentsinterviewedintheC2SSinMarch,April,andMay2000 l totheirrespectiveCensus2000recordsandcomparedresponseswiththeracequestion.#X\XX}X\#B  ԍ(47page X  2827,R1)d}X\XXX\ԍC  d X  Theadvantageofthisstudyisthe abilitytocomparethepairedresponsesforpeopleas D! opposedtolookingattotals(_Raglin_ԀandLeslie,2002:2829).#X\XX}X\#B  ԍ(47page2829,R1)d}X\XXX\ԍC  dTheauthorsfoundmuchmore 0"  consistentraceresponsesamongrespondents!bothHispanicsandnonHispanics#X\XX}X\#}X\XXX\!who #l! answeredCensus2000andC2SSviamail,thanthosewhowereinterviewedineachdata $X" collection(_Raglin_ԀandLeslie,2002:2831).#X\XX}X\#Ԛ}X\XXX\Ԁ#X\XX}X\#}X\XXX\Inexplainingthefinding,_Raglin_ԀandLeslie $D # (2002:2831)notethathouseholdswerenotassignedrandomlytomailversusinterview,but %0!$ ratherwereinterviewedbecausetheydidnotrespondtothemailquestionnaire. Therefore, &"% thesepeoplearethehardesttocollectdatafrom.#X\XX}X\#B  ԍ(47page2831,R1)d}X\XXX\ԍC  d_Raglin_ԀandLeslie(2002:2831)alsonotethat '#& censusinterviewerswereallowedtouseproxyrespondentsoutsidethehousehold,were (#' inexperienced,andusedpaperandpencil,asopposedtocomputeraidedinstruments.#X\XX}X\#B  ԍ(47page h+&* 2831,R1,R2)d}X\XXX\ԍC  dTheyalso )$( notethatC2SSinterviewerswhodidnotworkonCensus2000weremorelikelytoprobewhen |*%)  Hispanicwasgiveninanswertoraceandthatmanyoftheseinterviewersworkonother h+&* surveysthatdonotallow Someotherrace#X\XX}X\ #(_Raglin_ԀandLeslie,2002:2830).B  ԍ(47page2830,R1)d}X\XXX\ԍC  d#X\XX}X\ #}X\XXX\ T,'+  @-(, &  AmongnonHispanics,_Raglin_ԀandLeslie(2002:2831)alsonotedgoodconsistencyinreporting  whenboththeCensus2000andC2SSdatawerecollectedviamailforWhite,Black,andAsian'/   respondents.TheyfoundonlymoderateconsistencyforAmericanIndianandAlaskaNative,  Someotherrace,andTwoormoreracesrespondents.#X\XX}X\ #Ԁ}X\XXX\Ԁ#X\XX}X\#B  ԍ(47page2831,R3)C  }X\XXX\_Raglin_ԀandLeslie(2002:2831)conclude: t 0  Thereisoftenconcernabouttheconsistencyofracereporting,butthesedataL (#(#   indicatethatforalargeshareofthepopulation!nonHispanicswhoarewilling 8    tofilloutthemailforms!racereportingisconsistentwiththeexceptionofpeople $ t   reportingTwoormoreraces#X\XX}X\~#}X\XXX\.#X\XX}X\#B  ԍ(47page2831,R4)d}X\XXX\ԍC  d#X\XX}X\#}X\XXX\  ` #X\XX}X\l#}X\XXX\#X\XX}X\#}X\XXX\Accordingto_Raglin_ԀandLeslie(2002:2830),therewasanotabledifferencebetween  8  Census2000andC2SS#X\XX}X\#}X\XXX\ԀracedataforHispanicscollectedby#X\XX}X\#}X\XXX\interviewers#X\XX}X\L#}X\XXX\.Thissuggested $  thatinterviewersprobablyaffectedthereportingofracebyHispanics.Theauthorssuggest   thatthereasonforthiswasthatmanyCensus2000enumeratorsweretemporaryemployees   withlittleinterviewingexperience,whileC2SSenumeratorswerepermanentCensusBureau   employeeswithmoreexperience.B  #X\XX}X\#ԍ(47page2830,R1)d}X\XXX\C  d   #X\XX}X\#Thus,itseemslikelythatenumeratorsandinterviewersmayhavecauseddifferencesinthe \ reportingof Someotherracealoneorincombinationwithotherraces.Totheextentthat H C2SSinterviewershadexperiencewithotherdatacollectionthatdoesnothavea Someother 4 racecategory,itislikelythattheywerelesswillingtoaccept Someotherraceresponses.  p Asdiscussedpreviously,anobservationstudy #  24      ךreportedbyHoughand_Borsa_Ԁ(2003:42)showed  \ thatsomecensusenumeratorshaddifficultyaskingaboutrace.Somedidnotshowthe_flashcard_, H readthequestionasworded,orreadalloftheracecategories.B  ԍdd(new=HoughandBorsa6Bpage  42,R1)(original=Borsa6page55,R1)dC  d 4        h ProcessingDifferences.Adifferenceintheprocessingofenumeratorforms(whichhadonly   onewriteinareaforrace)comparedtomailforms(whichhadthreewriteinareasforrace),  ledtoanoverstatementofSomeotherraceby6percent,B  dd(55page3,R1)C  ddandTwoormoreracesresponsesby  about15percentB  dd(55page3,R2)C  dd(see_Cresce_,2003foramoredetaileddiscussion).   jDiscussionofDifferences.Perhapsthequestionweshouldbeaskingiswhytherearent more X  differences betweenCensus2000andC2SSracedistributions,notwhythereareanydifferences D! (toparaphrasesociologistKingsleyDavis).Evenifwetooktwoindependentdecennialcensuses 0"  atthesametime,itwouldbereasonabletoexpectdifferencesduetononsamplingerror. #l! IncomparingCensus2000andC2SS,weknowtherearesubstantialwordingandmethodological $X" differences,andsomeprocessingdifferences,asdiscussedabove.However,theinvolvementof $D # interviewersprobablyhadalargeeffectonracereporting,particularlythatofHispanic %0!$ respondents.Selfselectionbythe difficulttoenumeratethroughnotrespondingviathemail &"% questionnairemayjustcomplicatethetaskofenumerators.ButasnotedbyLeslie,_Raglin_,and '#& _Schwede_Ԁ(2002:2065): (#'  j )$( dd8  Probingisonepartofthequestionandanswerprocessthatcannotbecompletely  standardizedandthus,thereisanopportunityforinterviewerstobeinconsistent  acrossrespondentsandacrossinterviews.Thatistheonesituationinwhich  interviewerrelatederrorcanoccur(_Mangione_,Fowler,andLouis,1992).B  ԍ(49  L  page2065,R8)d}X\XXX\ԍC  d#X\XX}X\s##t   Weshouldnotethattherearemanyothersituationswhereinterviewerrelatederrorcanoccur,B  ԍ(per   BetsyMartin7.23.03e.mail)C  Ԁ L  butitisclearthatresponsestendtobemostconsistentwhencollectedviamail}X\XXX\Ԁ(_Raglin_Ԁand 8  Leslie,2002:28302831)#X\XX}X\$#.Thus,itseemsimperativethattheCensusBureaustudywaysto $ t maximizemailresponse,andtoensurethatinterviewershaveastandardizedapproachto  ` collectingraceinallitssurveysinordertominimizeinterviewereffectsondatacollection.   L  ;XX\   7r5Pr P   6.PuertoRicoCensus2000RaceandEthnicityQuestions on?' #X\X;&#  }X\XXX\Census2000wasthefirsttimethatresidentsofPuertoRicowereaskedtocompleteandreturn  theirquestionnairesbymail(_Berkowitz_,2001:1).#X\XX}X\'#Italsomarksthefirsttimequestionsonrace  andHispanicoriginwereaskedofindividualsinPuertoRico,althoughracewascollectedby r enumeratorsthroughobservationinthe1950census.Thedecisiontoincludetheraceand ^  Hispanicoriginquestions occurredbecausethegovernmentofPuertoRicorequestedthesame J  questionnairecontentasstatesideinordertospeedtheprocessingandreleaseofPuertoRico 6  censusdataandsothatPuertoRicocouldbeincludedinstatewidestatistics(_Christenson_, " r 2003:1).  ^  Accordingto_Berkowitz_Ԁ(2001:iv): 6  0  A}X\XXX\lmosteveryonehadheardsomethingaboutCensus2000fromtelevisionandradioads,   newspapers,schools,orfrominformalsourcessuchasrelatives,neighbors,and    brothersor sistersintheirchurches.Mosthadalsodiscussedsomeaspectofthe   processwithsomeoneelse.Manyparticipantsindicatedtheyhadconsultedwithfamily  membersorneighborswhiletryingtocompletetheirquestionnaires,sometimesinan n efforttoreachaconsensusastowhatwasbeingaskedorhowtheyshouldanswer.#X\XX}X\+#Z(#(# eBecauseofthenewnessofthequestions,itisprobablynotsurprisingthat_Berkowitz_Ԁ(2001:2122) 2 foundthat}X\XXX\there#X\XX}X\y/#}X\XXX\Ԁwere concernsthatsomequestionsweretooprivate,andthat therace/ethnicity n questionsinspiredthemoststrenuousnegativereactionsofanyquestionsontheCensus2000  Z questionnaireinmoreurbancoastalcommunities#X\XX}X\/#ofPuertoRico.Itisalsopossibleddthattherole F ofinterviewersinPuertoRicomightbedifferentfromthatrolestateside.}X\XXX\About53percentof 2 PuertoRicoshouseholdsreturnedtheirCensus2000questionnairesbymail,comparedwith65  percentstateside(_Berkowitz_Ԁ2001:1)#X\XX}X\1#.  #  25      }X\XXX\Ԁ#X\XX}X\13#Ԁ_Berkowitz_Ԁ(2001:16)found }X\XXX\astrongpreferenceforthe   morepersonal,doortodoorapproachtakeninthe1990census.Theyfoundtheideaofdropping  offthequestionnaireatthegatetooimpersonalandbureaucraticfortheirtaste.  e.  6.1.#X\XX}X\3#Studydesignandlimitations on4 j  Thebasicmethodfollowedby_Christenson_Ԁ(2003:23)wastocompareraceandHispanicorigin B"  distributionsbasedonCensus2000100percentdatacollectedstatesideandinPuertoRico. .#~! ThemainlimitationoftheevaluationofPuertoRicosraceandHispanicorigindataisthe $j"  lackofanypreviousquantitativemeasuresforcomparison.Thelackofcognitivestudies %V # preventsdrawing definitiveconclusionsaboutwhatledrespondentstoanswerthewaythey %B!$ did.Finally,wemaynotknow theextenttowhichtheresponsesofPuertoRicanswereshaped &."% bytheirunderstandingoftheirracialidentityasopposedtothewaytheyinterpretedandreacted '#& tothequestionitself.d ($' }X\XXX\   *%) 6.2.#X\XX}X\8#_Nonresponse_ԀtoraceandHispanicorigin ss9  Table6.1showsthatthe_nonresponse_ԀtoraceishigherinPuertoRicothanintheUnitedStates  (5.0and4.1percent,respectively),butjusttheoppositeistrueoftheresponsetoHispanicorigin t (3.4and4.8percent,respectively). `  &  tablesTable6.1 _Nonresponse_ԀtoRaceandHispanicOriginintheUnitedStatesandPuertoRico tables9; 8  *Xdd 44 44 X(#(#,( dd ,( dd ,( dd +  ( x( x  Question -#!  ff -  UnitedStates -#!    ff -  PuertoRico 6,&      ff 46Race -#! , X -  4.1% -#! ,  X -  5.0% 7-' ,   40 X 47HispanicOrigin -#!,| f -  4.8% -#!,|  f -  3.4%7-+,|   40  f 7Source:TabulationofCensus2000HundredPercentDataFile(HDF).B  ԍ(52,page2,6,R1)C   H  '8 *;Table6.2showsthatHispanicsoverallandPuertoRicansintheU.S.aremuchmorelikely  p  nottoanswertheracequestionthantheircounterpartsinPuertoRico._Nonresponse_Ԁtoraceby  \  Hispanics  #  26      ׀intheUnitedStateswasover14percent,andover17percentbyPuertoRicansin H theUnitedStates,comparedwithunder4percenteachontheislandofPuertoRico.| #  27       4  &  tablesTable6.2 _Nonresponse_ԀtoRacebyHispanicsandPuertoRicansintheUnitedStates   andPuertoRico tablesuB  *Xdd( dd ( ( dd ( ( dd ( X(#(#,( dd ,( dd ,( dd +    Hispanicgroup -#!L ff -  UnitedStates -#!L  ff -  PuertoRico 6,&L    ff 46AllHispanics -#!LX -  14.3% -#!L X -  3.4% 7-'L  40 X 47PuertoRican -#! f -  17.5% -#!! f -  3.4%7-+"  40  f 7Source:TabulationofCensus2000HundredPercentDataFile(HDF).B  ԍ(52,page3,4,7,8,R1)C  d " 'eB}X\XXX\&   6.3.#X\XX}X\aG#Hispanicoriginreporting ncG $ Itisnotsurprisingthat98.8percentofPuertoRicosresidentswereidentifiedasHispanic h& orthat'!GԀ95.1percentwereidentifiedasPuertoRican.Another1.5percentwereofDominican T ' origin;1.4percentwereidentifiedas OtherHispanicorLatino;Cubanswereabout @!( 0.5percent,andMexicans0.3percent(_Christenson_,2003:4).Lessthan4percentofthe ,"|) HispanicoriginresponsesinPuertoRicowerewriteinentries,and37.6percentofthose #h* reflectedthecheckboxresponses(Mexican,PuertoRican,etc.).Another52.8percentofthe $T+ writeinresponsesweredetailedHispanicresponses;6.5percentweremultipleresponses; $@ , and3.1percentwereotherresponses.AmongthespecificHispanicgroupswrittenin, %,!- 71.4percentwere Dominican;11.6percentwereSouthAmericanentries;5.7percentwere &". Spaniard;4.8percentwereCentralAmerican;andonly6.4percentweregeneraldescriptors(e.g., '#/ Hispanic,Latino,etc.)(_Christenson_,2003:5). (#0 5PB 5  5  }X\XXX\6.3.1#X\XX}X\JM#ReportingofHispanicoriginbyenumerators  Ingeneral,thedistributionofHispanicoriginresponsesthatwereenumeratorfilleddoesnot t varymuchfromthosethatwererespondentfilled.Table6.3showsnodifferencesinthe ` proportionHispanicandnonHispanicinPuertoRicobymodeofcollection,buttherearesome L  differencesinthespecificcategories.Comparedwithrespondentfilledreturns,enumerator 8  filledreturnsshowedproportionatelyfewerPuertoRicans(0.7percent),Cubans(33.0percent), $ t andallotherHispanicgroups(33.0percent),butmoreDominicans(110percent)andMexicans  ` (150percent) .  L   & P tablesTable6.3 DistributionofHispanicOrigininPuertoRicobyModeofDataCollection tablesP $  *Xdd( dd ( ( dd ( ( dd ( X(#(#,dd%,dd%,dd%,dK+      ЈHispanicOrigin -#!d U ff -  Respondentfilled (x  (1) -#!d % ff -  Enumeratorfilled (x  (2) d      6,!P  ff 6Percent (x  Difference d  3=(21)/1 ?5&P "   ff 4?Hispanic -#!PX -  98.8% -#!P%X -  98.8% 6,!P%X 6  0.0% @6'P2 40 X 4@U  ӀPuertoRican -#!X -U95.5% -#!%X -  94.8% 6,!%X 6  ӄ0.7% @6'2 40 X 4@SV  ӀDominican -#!@X -VӀ1.0% -#!@ %X -  Ӏ2.1% 6,!@!%X 6  110.0% @6'@"2 40 X 4@W  ӀCuban -#!@#X -WӀ0.6% -#!@$%X -  Ӏ0.4% 6,!@%%X 6  ӄ33.0% @6'@&2 40 X 4@X  ӀMexican -#!'X -6YӀ0.2% -#!(%X -  Ӏ0.5% 6,!)%X 6  150.0% @6'*2 40 X 4@Z  ӀOtherHispanics -#!0+X -qZӀ1.5% -#!0,%X -  Ӏ1.0% 6,!0-%X 6   @F 33.0% @6'0.2 40 X 4@F[  NonHispanic -#!0/f -[1.2% -#!00%f -  Ӏ1.2% 6,!01%f 6Ѐ0.0%7-+02" 40  f 7Source:Adaptedfrom_Christenson_Ԁ(2003:14,Table9). L2 'PPTable6.4showsmorestrikingdifferencesintheproportionofHispanicsandnonHispanics $t4 intheUnitedStatesbymodeofcollection.EnumeratorfilledreturnsintheUnitedStates `5 showedproportionatelyfewernonHispanic(6.5percent)andmoreHispanic(52.7percent) L6 responsesthanrespondentfilledreturns.EnumeratorfilledreturnsintheUnitedStates  87 showedproportionatelyfewerCubans(20.0percent),butmorePuertoRicans(45.5percent), !$8 Dominicans(50.0percent),Mexicans(80.3percent),andallotherHispanicgroups(12.9percent) "9  thanrespondentfilledreturns. #:  tablesTable6.4 D& P  istributionofHispanicOriginintheUnitedStatesbyModeofDataCollection tablesK`  *Xdddd%dd%dd%dKX(#(#,dd ,dd ,dd ,dd +    HispanicOrigin -#!@ ff -  Respondentfilled T (1) -#!@  ff -  Enumeratorfilled T (2) -#!@  ff -  Percent T Difference @  3=(21)/1 6,&,     ff 46Hispanic -#!, | X -  11.0% -#!, |  X -  16.8% -#!, |  X -  Ӏ52.7% 7-', |  40 X 47ЀPuertoRican -#!| X -  Ӏ1.1% -#!|  X -  Ӏ1.6% -#!|  X -  Ӏ45.5% 7-'|   40 X 47ЀDominican -#! X -  Ӏ0.2% -#!  X -  Ӏ0.3% -#!  X - F 50.0% 7-'  40 X 47ЀCuban -#! lX -  Ӏ0.5% -#! l X -  Ӏ0.4% -#! l X -  ӄ20.0% 7-' l  40 X 47ЀMexican -#!l X -  Ӏ6.1% -#!l  X -  11.0% -#!l  X -  Ӏ80.3% 7-'l   40 X 47ЀOtherHispanics -#! X -  Ӏ3.1% -#!  X -  Ӏ3.5% -#! ! X - F 12.9% 7-' " 40 X 47NonHispanic -#! \ #f -  89.0% -#! \ $ f -  Ӏ83.2% -#! \ % f -  ӄ6.5%7-+ \ &  40  f 7Source:Adaptedfrom_Christenson_Ԁ(2003:14,Table9). (x & 'Pt`}X\XXX\6.3.2.#X\XX}X\Pk#DiscussionofHispanicoriginreporting P( AlthoughthereisnobenchmarktoevaluatethereportingofHispanicorigininPuertoRico, (* theresultsfromCensus2000appeartobereasonableprimafacie,andthereappearstobeno + particularbiasincomparisonsofrespondentfilledreturnsandenumeratorfilledreturnsin , PuertoRico.V  ԀWFmIncontrast,therearesignificantdifferencesinthedistributionsstateside:V   - WmenumeratorfilledreturnsshowedproportionatelymoreHispanics(withtheexceptionof . Cubansamongthegroupsexamined).IntermsofreportingofdetailedHispanicgroups, t/ theredidnotappeartobeexcessivereportingofgeneralHispanictermsprobablybecausethe `0 overwhelminglydominantgroupontheIsland(PuertoRican)appearsasareportingcategory L1 inalldatacollections._Cresce_ԀandRamirez(2003:11)suggestthatthePuertoRicancategory, 82 alongwiththeCubancategory,weretheleastaffectedbychangesintheHispanicorigin $t3 questionusedinCensus2000(seeChapter2foramoredetaileddiscussion). `4 }X\XXX\  6.4.#X\XX}X\q#Racereporting ׀"q  86 DespitethenewnessofracereportinginPuertoRico,reportingwasverycomplete,asseen "8 inthesectionabove,andquitedifferentthanmighthavebeenexpected.Table6.5showsthe #9 distributionofraceinPuertoRicoforallpeople.Abouteightineverytenpeople(80.5percent) $: werereportedas Whitealone,and84.0percentreported Whitealoneorincombinationwith % ; oneormoreotherraces.Nearlyoneintwelve(8.0percent)reportedas BlackorAfrican p&!< Americanalone,but10.9percentreported Blackaloneorincombinationwithoneormore \'"= otherraces.About6.8percentreportedas Someotherracealone,and8.3percentdidsoin H(#> combinationwithotherraces.Onein25(4.2percent)reportedbeingofmorethanonerace. 4)$?    *p%@  PD7rD   7  7  &   tablesTable6.5 RaceDistributioninPuertoRico tablesu  *Xdddd dd dd dd X(#(#, dd ,dd , dd +    Selectedracecategories 6,!T ff 6RacealoneV   @i  @D Ww 6,!T" ff 6Racealoneorincombination T withotherraces 6,&@"   ff 46WhiteV   l  A   Wx -#!@X -  80.5% -#!@ X -  84.0% 7-'@  40 X 47BlackorAfricanAmerican -#!  X -  Ӏ8.0% -#!   X -  10.9% 7-'    40 X 47AmericanIndianandV  ԀWzAlaskaNative -#! 0 X -  Ӏ0.4% -#! 0  X -  Ӏ0.7% 7-' 0  40 X 47Asian -#!0 X -  Ӏ0.2% -#!0  X -  Ӏ0.5% 7-'0   40 X 47Someotherrace -#! X -  6.8% -#!  X -  Ӏ8.3% 7-'   40 X 47Twoormoreraces  -#! f -  Ӏ4.2% -#!  f -  !7-+   40  f 7Source:SummaryFile1,TableP3andTableP9. <  ' uBecausemostresidentsofPuertoRicoareHispanic,itisimportanttocomparetheirrace   distributiontothatofstatesideHispanicsandPuertoRicans.Table6.6showstherace  distributionsofHispanicsinPuertoRicoandtheUnitedStates.ComparedwiththeUnited  States,HispanicsinPuertoRicoaremuch morelikely toreport Whitealone(68percent)and   Blackalone(295percent).Ontheotherhand,HispanicsinPuertoRicoaremuch lesslikely  t toreport AmericanIndianandAlaskaNativealone(75percent), Someotherracealone ` (84percent),and Twoormoreraces(35percent)._Christenson_Ԁ(2003:8)reportsasimilar L patternwhenlookingatthe racealoneorincombinationdistributionofrace. 8         h      p      x & ` tablesTable6.6 RaceDistributionofHispanicsinPuertoRicoandtheUnitedStates tables| `" *Xdd dd  dd  dd X(#(#V  , dd ,rdd ,rdd ,dd W+  P#P  Selectedracecategories -#!d$ ff -  PuertoRico d% (1) -#!P&  ff -  UnitedStates d' (2) -#!P(  ff -  PercentDifference d) 3=(12)/2 6,&P*    ff 46WhitealoneBBB +  -#!,X -  80.7% -#!- X -  47.9% -#!. X -  Ӏ68% 7-'/  40 X 47Blackor ,0 AfricanAmericanalone -#! 1X -  Ӏ7.9% -#!,2 X -  Ӏ2.0% -#!,3 X -  295% 7-',4  40 X 47AmericanIndianand "h5 AlaskaNativealone -#!#T6X -  Ӏ0.3% -#!"h7 X -  Ӏ1.2% -#!"h8 X -  ӄ75% 7-'"h9  40 X 47SomeotherracealoneV   T$: W_ -#!@% ;X -  Ӏ6.9% -#!T$< X -  42.2% -#!T$= X -  ӄ84% 7-'T$>  40 X 47TwoormoreracesV   &!? W -#!|'"@f -  Ӏ4.1% -#!&!A f -  Ӏ6.3% -#!&!B f -  ӄ35%7-+&!C  40  f 7Source:Adaptedfrom_Christenson_Ԁ(2003:6)Table3. (#C '`p  p*%E HowdifferentaretheresponsesofPuertoRicansonV  ԀW{theIslandfromthosestateside?Table6.7  showssimilarresultsasforHispanicsoverall.ComparedtotheUnitedStates,PuertoRicansin  PuertoRicoarealsomuch morelikely toreport Whitealone(72percent)butonlysomewhat  morelikelytoreport Blackalone(17percent).Ontheotherhand,PuertoRicansontheIsland t arealsomuch lesslikely toreport AmericanIndianandAlaskaNativealone(50percent), `  Someotherracealone(82percent),and Twoormoreraces(46percent)thanarePuerto L  Ricansstateside. 8   tablesTable6.7& ` RaceDistributionofPuertoRicansinPuertoRicoandtheUnitedStates tableso  ` *Xdd dd rdd rrdd rdd X(#(#V  , dd ,dd ,rdd ,dd Wː+   P  P  Selectedracecategories -#!d   ff -  PuertoRico d   (1) -#!P   ff -  UnitedStates d   (2) -#!P   ff -  PercentDifference d  3=(12)/2 6,&P     ff 46Whitealone    -#! X -  81.4% -#!  X -  47.4% -#!  X -  72% 7-'   40 X 47BlackorAfricanAmerican ,  alone -#!X -  Ӏ7.6% -#!,  X -  Ӏ6.5% -#!,  X -  17% 7-',   40 X 47AmericanIndianand h AlaskaNativealone -#!TX -  Ӏ0.3% -#!h X -  Ӏ0.6% -#!h X -  ӄ50% 7-'h  40 X 47SomeotherracealoneV   t  T  W_ -#!@!X -  Ӏ6.6% -#!T" X -  37.3% -#!T# X -  ӄ82% 7-'T$  40 X 47TwoormoreracesV     % W -#!|&f -  Ӏ4.0% -#!' f -  Ӏ7.4% -#!( f -  ӄ46%7-+)  40  f 7Source:Adaptedfrom_Christenson_Ԁ(2003:910)Table5aandTable5b. ) '` _Christenson_Ԁ(2003:11)reportsthat9.2percentoftheresponsestoraceinPuertoRicowere p+ writeins.Ofthose,82.8wereclassifiedas Someotherrace.Ofthe Someotherrace \, responses,63.8percentinvolvedaHispanicoriginanswer(e.g., Hispanic, PuertoRicans, H- etc.),31.9percentwereacolorresponse(e.g. Moreno, Brown,etc.),1.8percentan 4. undefinedmixedraceresponse(e.g., Mixed, Mulatto, Multiracial,etc.),andtherest  p/ wereotherresponses.  !\0 }X\XXX\6.4.1.#X\XX}X\#Reportingofracebyenumerators "42 eUnlikeHispanicorigin,racedatacollectedbyenumeratorsshowreportingthatismoderately $ 4 distinctfromthatinrespondentfilledreturns.AsshowninTable6.8,enumeratorfilledreturns % 5 forHispanicsinPuertoRicoareproportionately lesslikely tobe Whitealone(8percent), &!6  Blackalone(17percent),and AmericanIndianandAlaskaNativealone(400percent), '"7 but morelikely tobe Someotherracealone(54percent)and Twoormoreraces(35percent). l(#8  e X)$9 rF     & L  tablesTable6.8 RaceDistributionofHispanicsinPuertoRicobyModeofDataCollection tables  *Xdd dd dd rdd rdd X(#(#,A dd ,dd ,dd ,'dd +    Selectedracecategories t  T  l  A -#!@ ff -  Respondentfilled T (1) -#!@  ff -  Enumeratorfilled T (2) -#!@  ff -  Percent T Difference @  3=(21)/1 6,&,     ff 46Whitealone , |   l -#! h X -  83.0% -#!, |  X -  77.1% -#!, | X -  Ӏ8% 7-', |  40 X 47Blackor h  AfricanAmericanalone -#!T X -  Ӏ8.3% -#!h  X -  Ӏ7.1% -#!h  X -  Ӏ17% 7-'h   40 X 47AmericanIndianand   AlaskaNativealone -#! X -  Ӏ0.5% -#!  X -  Ӏ0.1% -#!  X -Ѐ400% 7-'  40 X 47SomeotherracealoneV   0  Wߦ -#! X -  Ӏ4.7% -#!0  X -  10.3% -#!0  X -  Ӏ54% 7-'0   40 X 47TwoormoreracesV   l  W -#!X f -  Ӏ3.4% -#!l ! f -  Ӏ5.2% -#!l " f -  Ӏ35%7-+l #  40  f 7Source:Adaptedfrom_Christenson_Ԁ(2003:15,Table10). $t# 'LEnumeratorfilledreturnsforHispanicsintheUnitedStates(asshowninTable6.9)are L% proportionately lesslikely tobe Whitealone(8percent), Twoormoreraces(43percent), 8& and AmericanIndianandAlaskaNativealone(63percent),but morelikely tobe Black $' alone(5percent)and Someotherracealone(13percent).UnlikeHispanicorigin,racedata ( collectedbyenumeratorsshowedmoderatelydistinctreporting. )  tablesTable6.9 RaceDistributionofHispanicsintheUnitedStatesbyModeofDataCollection tables +  *XddA dd A dd dd 'dd 'X(#(#, dd ,dd ,dd ,dd +  t,t  Selectedracecategories -#!(- ff -  Respondentfilled (. (1) -#!/  ff -  Enumeratorfilled (0 (2) -#!1  ff -  Percent (2 Difference 3 3=(21)/1 6,&4    ff 46WhitealoneV    P5 W -#! <6X -  49.2% -#! P7 X -  45.7% -#! P8 X -  Ӏ8% 7-' P9  40 X 47Blackor <": AfricanAmericanalone -#!(#x;X -  Ӏ2.0% -#!<"< X -  2.1% -#!<"= X -  Ӏ5% 7-'<">  40 X 47AmericanIndianandV   x$? WAlaskaNativealone -#!d% @X -  Ӏ1.3% -#!x$A X -  0.8% -#!x$B X -Ѐ63% 7-'x$C 40 X 47SomeotherracealoneV   &"D W-Ԁ -#!'"EX -  40.0% -#!&"F X -  46.1% -#!&"G X -  Ӏ13% 7-'&"H  40 X 47TwoormoreracesV   (@$I W` -#!),%Jf -  7.0% -#!(@$K f -  Ӏ4.9% -#!(@$L f -  Ӏ43%7-+(@$M  40  f 7Source:Adaptedfrom_Christenson_Ԁ(2003:15,Table10).X  ԍ *H&M ЎY p      x  *H&M }X\XXX\  , (O 6.4.2.#X\XX}X\R#Discussionofracereporting  AtleastinsomeareasofPuertoRico(urbancoastalareas),theracequestione #  28      ׀ elicitedthe  strongestnegativereactionsfromparticipantsinfourfocusgroups._Berkowitz_Ԁ(2201:17) t notesthatseveralparticipantsreportedthatthey stoppedfillingouttheirquestionnaireupon ` reachingtheracequestion.Someparticipantsfeltthequestionswerediscriminatory,divisive, L  andnotappropriatefor theCreoleormixedrealitiesofPuertoRico.Forexample, 8  _Berkowitz_Ԁ(2201:1718)reportssomeparticipantsreactions[ emphasisadded ]: $ t }X\XXX\0  Ihavereceivedtrainingonequalemployment.Iunderstandthatabouttheraces.  L  WhenIsawthecensusformandreadtheracequestionIthought IamnotWhiteor  8  BlackoranythingelsebecauseIamHispanic andsoIwasupsetanddecidednot $  tofillitout.#X\XX}X\C# (#(# }X\XXX\0   Ididnotfindanalternativeanswerformyrace becauseweareneitherAfricanBlacks   norAmericanIndians.Thecensusdidnothavetheoptionalanswerof#X\XX}X\#}X\XXX\PuertoRican,#X\XX}X\"#}X\XXX\   ourrace.ThequestionupsetmebecauseIthoughtwhydowehavetobedividedasa p race,ifwehaveallkindsofraceslivinghere:Chinese,Arabs,Dominicans,Cubans. \ ItoccurredtomethatthisquestionwassomewhatracistandIdidnotwanttofillout H theformandsoIdidnot.#X\XX}X\u#4(#(# }X\XXX\0  Therewas nooptionforLatino,orPuertoRican,orHispanic .Thisbadlydesigned  \ questiondemonstratedthatourculturedoesnotexist.IfeltoffendedandsaidIwouldH(#(# 0  notfillitout.MywifetoldmeIhadtofillitout,accordingtolaw.Isaidletthemcome 4 andgetmeandhavethemputmeinjail! (#(# #X\XX}X\R#Despitethestrongreactionstotheracequestioncitedabove,itV  ԀWisclearthattheracereporting  inPuertoRicoisgoodintermsofcompleteness(about5percentdidnotrespond).Unlike  statesideHispanicsandPuertoRicans,Islandresidentsweremuchmorelikelytorespondtorace  (seeTable6.2).ResidentsofPuertoRicoweremuchmorelikelytoreportas Whiteandmuch l Ѐlesslikelytoreportas Someotherracethantheirstatesidecounterparts.Itispossiblethat X  thehigherproportionof Someotherracestatesideispartlyafunctionofthemuchlargerrace D! _nonresponse_ԀamongHispanicstateside,andtheconsequentimputation.Hispanicresidentsin 0"  PuertoRicowerealsomorelikelytoselfreportas Blackonrespondentfilledreturnsthantheir #l! statesidecounterparts(8percentand2percent,respectively;seeTable6.8),butPuertoRicans $X" intheUnitedStateswereonlyslightlylesslikelytoreport BlackthantheirIslandcounterparts $D # Ѐ(6.5and7.6percent,respectively;seeTable6.7). %0!$ BecauseofthelargerolethatenumeratorsplayedinPuertoRico,therewassomeconcernthat '#& enumeratorsmayhaveaffectedracereporting.Wereenumeratorssomehowresponsibleforthe (#' largeproportionreporting WhiteinPuertoRico?Thatdoesnotappeartobethecase. )$( AsseeninTable6.8,enumeratorfilledreturnsshowslightlylessreportingof White,and  morereportingof Someotherrace.Theyalsoshowslightlylessreportingof Blackthan  respondentfilledreturns.Interestingly,enumeratorfilledreturnsintheUnitedStatesshowedthe  opposite:slightlyhigherproportionsof Blackthanrespondentfilledreturns(seeTable6.9).In t anycase,itishardtoconcludethatenumeratorssomehowsignificantlybiasedordistortedthe ` racedataofHispanics.Theracereportingpatternofrespondentfilledreturnsissimilar, L  althoughcertainlynotidentical,forHispanicrespondentsbothintheUnitedStatesandinPuerto 8  Rico. $ t eOntheotherhand,theracereportingpatternisverydifferentamongHispanicandPuertoRican  L  respondentsinPuertoRicocomparedwiththeirstatesidecounterparts(seeTable6.6and  8  Table6.7).Itisalsoclearthatthesedifferencesarenotexplainedbyenumeratorbehavior. $  ThedifferencesintheracereportingpatternofPuertoRicansV  ԀWontheIslandandintheUnited   Statessuggestthat,despitethecontroversy, raceisconceptualizedandunderstooddifferently   ontheIslandthanintheUnitedStates.    eY ;XX\P r [MP  7.Conclusion #X\X;#      ThemajorobjectiveofthisTopicReportistosynthesizeresultsfromtheCensus2000Testing,  Experiment,andEvaluationsProgramresearchrelevanttoraceandHispanicoriginand,  ifpossible,toanswersomeoralloftheresearchquestionsthatguidedthereport. r   7.1.Effectsofquestionnairechanges Q J  WhatwastheoveralleffectonreportingofraceandHispanicoriginengenderedbythechanges " r inquestionsequencing,wording,questionnairelayout,anddroppingexamplesthatwere  ^  includedin1990?Wascompletenessofreportingadverselyaffected?  J  Thelessonwelearned,onceagain,fromthe AlternativeQuestionnaireExperiment(_AQE_) n #  29       "  (seechapter2)isthatchangesinthequestionnaire(inthiscasethemailoutform)have   unintendedconsequences.V  ԀW]ԀSomeofthechangeshadaperverseeffectanddidnotfully   resolvetheissuestheyweredesignedtoaddress,asexplainedbelow.   7.1.1.SequencingandinstructionsV   r WTInCensus2000,Hispanicoriginwassequencedaheadofraceandaninstructionwasaddedto J answerbothquestions.Thesechangeshadtwomainobjectives:V  ԀW5a)decrease_nonresponse_Ԁto 6 Hispanicorigin;andb)increasereportinginstandardracecategoriesbyHispanics.Whilethe "r _AQE_Ԁcouldnotdifferentiateexactlywhateffectswereproducedbyaspecificchange,thereis ^ evidencethatallthechangeshadaneffect.First,_nonresponse_ԀtotheHispanicoriginquestion J droppedquitedramaticallyinthe2000styleformcomparedtothe1990styleform.Second, 6 the2000styleformelicitedbetterracereportingbyHispanicrespondents,although_nonresponse_ " toraceisstillmuchtoohigh.ProportionatelyfewerHispanicrespondentsreportedas Some  otherraceinthe2000styleform,butthischangedidnotevencomeclosetoeliminatingthe  problem.Also,moreHispanicrespondentsresponded Whiteinthe2000styleformthan  inthe1990styleform.    7.1.2.Twoormoreraces sV   Z! WOneofMartins(2002a:iv)interestingconclusionsisthat, contrarytowhatmighthavebeen 2#! expected,thereislittleevidencethatallowingrespondentstoreportmorethanoneracereduced $n" thesingleracereportinginthe5majorracecategories.However,oB  ԍ(1pageiv,R5)dC  dnereasonisthat,evenwith  %Z # instructionstoreportonerace,somerespondentstothe1990styleformreportedtwoormore %F!$ racesanyway.B  ԍ(1page '#& 12,R1)dC  dInaddition,almostonethirdofHispanicrespondentsdidnotreportarace.B  ԍ(1page  11,R1)dC  d &2"% So,theactualimpactonpublishedracedatadependsonhowtheseresponsesareimputed. '#& PJ     Itisimportanttorememberthatthesefindingsare_generalizable_Ԁonlytothemailoutuniverse.B  ԍ(1page5,R2)dC  d  Itisalsopossiblethat,withamuchlargersample,wemighthavereacheddifferentconclusions  becausewehadfewcasesinthesmallercategories.Smallsamplesarearecurringproblemwith  allresearchonthesetypesofquestions. t 7.1.3.QuestionwordingandexamplesV   L  W*Oneunintendedeffectofredesigningthemailformtobemoreuserfriendlywastochange $ t thereportingofspecificHispanicgroups.Fortunately,Martin(2002a:v)reportsnoevidenceof  ` anydifferenceintheproportionofpeoplereportingasHispanic,butthisconclusioncould  L  changeif_nonresponses_Ԁareimputed.Nonetheless,therewasprobablymorecompletereporting  8  bynonHispanicsinthe2000styleform.Theproblemisthatthe2000styleformelicitedfewer $  reportsofspecificHispanicsubgroups,andmorereportsofgeneralHispanicidentity(Martin   2002:v).DatausersweredisturbedbythereduceddetailfortheHispanicpopulationinCensus   2000(seeGAO2003a,Logan2002:3,and_Suro_Ԁ2002:8).   Manyofourcriticsblametheproblemonthedroppingofexamplesandthechangein p questionwording,butitisnotclearthatthisistotallytrue.First,Martin(2002a)showedthat \ the Mexicancategorywasaffected,butlogicallythiscategoryshouldnothavebeenaffected H sinceitappearedasacheckboxinbothforms.AsMartinpointsout,someofthesedifferences 4 mayhaveresultedfromotherchangestotheform.B  dd(perBetsyMartins7.23.03e.mail)dC  dSecond,_Cresce_ԀandRamirezs(2003)work  p suggeststhat PuertoRicanand Cubangroupsmayhavebeenaffectedeventhoughboth  \ appearedascheckboxesinbothforms.Third,Martin(2002b:2)showedtheoppositeeffect H amongAsianandPacificIslandercategories:the2000styleformhadhigherproportionsamong 4 theexamplegroupsthanthe1990styleform.However,itisimportanttoconsiderthatthese   resultsmaybeanartifactoftherelativelysmallsamplesizeforthesmallerracecategories.   Hadthesamplesizebeenmuchlarger,wemighthavereacheddifferentconclusions._Cresce_Ԁ  andRamirez(2003)didnotundertakeasimilaranalysisforAsianandPacificIslandergroups,  butitshouldbedoneforcompletenesssake.         h & ` DroppingexamplesinthequestiononHispanicoriginmayhavegivensomerespondentsthe X  impressionthatwewereattemptingtogetthemtoselectamongtheterms Spanish, Hispanic D! or Latino.The printgroupinstructionmayhavereinforcedthatnotion,resultinginfewer 0"  specificandmoregeneralresponses.Itmayhavealsocreatedinconsistentreporting,asexplored #l! below.Inanycase,the2003NationalCensusTestdatashouldbeabletoshedadditionallighton $X" theeffectofexamplesandrevisedinstructionsontheresponsestotheHispanicoriginandrace $D # questions. %0!$ '`X   '#& &    7.2.Consistencyinreporting en  The ContentReinterviewSurvey(_CRS_)  #  30      ׀(seechapter3)allowsustoassesstheconsistencyof'  reportingrace,Hispanicorigin,placeofbirthandancestry,amongotheritems.The_CRS_Ԁreport x consideredHispanicoriginandplaceofbirthreportingtobeofgoodconsistency,andraceand d ancestryreportingtobeofmoderateconsistency.T #  31      ׀V  ԀWOver95percentofrespondentsanswered P  boththeraceandHispanicoriginquestioninCensus2000and_CRS_.B  ԍdd(2page51,R1)dC  d <  Hispanicoriginreporting.AccordingtoSingerand_Ennis_Ԁ(2002:xxiixxiii),FX\XXX\theeditedHispanic  d origindatawereofgoodconsistency,#X\XXFX\@#ԀB  ԍ(2pagexxii,R2)dFX\XXX\ԍC  dbutthelackofclearinstructionsonthequestionmayhave  P  causedsomerespondentstoreport#X\XXFX\#FX\XXX\multiplecategorieswhenthequestionwasintendedtoelicit  <  #X\XXFX\#FX\XXX\one.#X\XXFX\N#ԀB  ԍ(2pagexxiii,R1)dC  dV  ԀWBasedonunediteddata,therewasgoodconsistencyforthe NotHispanicandthe (   Mexicancheckboxcategories,butmoderateconsistencyforthe PuertoRican, Cuban,   and OtherHispaniccheckboxcategorieswhenconsideredseparately.Examiningeight   categoriesalsoshowedgoodconsistency!about3.3percentofrespondentschangedtheir   answers.However,asSingerand_Ennis_Ԁ(2002:5354)remindus,allbutthe NotHispanic   and Mexicancategorieswere rare,whichcancausemeasurementerrorintheindexes. t Singerand_Ennis_Ԁ(2002)alsonotedthatsomerespondentschangedanswersbetweenCensus ` 2000and_CRS_,butwhatisclearisthatmostoftheinconsistencyarisesinthe OtherHispanic L categoryandthemultiplereports. 8 ExaminingthedifferencesinthequestionsusedinCensus2000and_CRS_,arespondentmight ` concludethatthe_mailback_ԀHispanicoriginquestionisaskingifapersonis Spanishor L  Hispanicor Latino,whereastheenumeratorand_CRS_Ԁquestionsareaskingaboutspecific 8 groups(e.g., Mexican, PuertoRican, Cuban,orofanotherHispanicorLatinogroup). $ The printgroupinstructiononthe_mailback_Ԁformmayhavereinforcedthenotionthatwe  wereaskingrespondentstoselect,orevenreject,thegeneralresponses.Ifso,arespondentin  Census2000mayhavereplied No,notSpanish/Hispanic/Latinoand Yes,Cuban,meaning  aCubanwhodoesnotidentifywiththegeneralterms,butduringreinterviewin_CRS_Ԁthe  respondentsaid, Yes,Cuban,thuscreatinganapparentlyinconsistentresponse.Similarly,a p respondentmighthaveidentifiedas LatinoinCensus2000,andthenidentifiedas Yes,Puerto \  Ricanin_CRS_,alsocreatinganapparentinconsistency. H! eRacereporting.ByexaminingreportingofHispanicrespondentsseparately,Singerand_Ennis_  #p! (2002:59)concludedthattheHispanicpopulationcontributegreatlytotheracedatavariability.B  ԍdd(2  $\" page59,R1)C  ddFX\XXX\Ԁ  $\" Thisfindingreconfirms#X\XXFX\5#thatHispanicrespondentshavemoredifficultyansweringtherace $H # rL     questionthandononHispanics.However,amongnonHispanics,onlyBlacks,Asians,and  Whitesshowgoodconsistency,whileAmericanIndiansandPacificIslandersshowonlymoderate  reportingconsistency. Someotherraceand Twoormoreracesshowedpoorreporting  consistency.Thereissomeevidencefromobservationsthatenumeratorsdidnotalwaysread t thequestionaswordedandmayhavefailedtoshow_flashcards_Ԁ(Houghand_Borsa_Ԁ2003:42).B  dd(new=HoughandBorsa6Bpage39,R1)(original=Borsa6page48,R1;page55,R1)dC  ԀAs ` withHispanicorigin,thereweredifferencesinquestionnaires.Butonereoccurringdifficultyisa L  samplesizethatisinsufficienttoproperlymeasuredifferencesinreporting,especiallyforthe 8  smallerorraregroups. $ t eAncestryreporting.OneoftheinterestingfindingsreportedbySingerand_Ennis_Ԁ(2002:27)was  L  thatresponsescollectedbymailshowedmoreconsistencythanthosecollectedbyenumerators,  8  althoughbothwereinthemoderaterange.B  d(2,page27,R2)dC  dInexaminingthedataforspecificHispanicorigins $  ofsufficientsize,wenotedmoreconsistency.Therewasmoreinconsistencyinreporting    Hispanicand Spanish,andsomeoftheinconsistencycamefrommovingbetweengeneral   HispanicandspecificHispanicresponses.   Placeofbirthreporting.Generallyspeaking,theconsistencyofplaceofbirthreportingas p identifiedbythewriteinresponsewasquitegood.ButSingerand_Ennis_Ԁ(2002:32)warnof \ FX\XXX\evidencethatthemodelassumptionswerenotmetforsomecategories.Evenso,subgroups H showedgoodconsistency.#X\XXFX\J #B  ԍ(2,page32,R1)dC  dWhenweexaminedplaceofbirthreportingfromCentraland 4 SouthAmerica,theseresponsesappearedtobereportedconsistently.Theseresultssuggest,  p forHispanicgroupsatleast,thatplaceofbirthandancestrymaybeconsideredreliable  \ supplementsforHispanicorigin.Theiruseforsupplementingraceresponses,however, H needstobefurtherexplored. 4        h   &   7.3.Sequencingand_nonresponse_ nx    DidsequencingofHispanicoriginaheadofracehavethedesiredeffectofreducing'l   _nonresponse_ԀtoHispanicorigin?DidthesequencingofHispanicoriginaheadofrace  resultinproportionatelyfewer SomeotherraceresponsesinraceanddidHispanics l havemorecompletereportingofrace? X  ThereisveryclearevidencethatsequencingofHispanicoriginaheadofracedidreduce 0"  _nonresponse_ԀtoHispanicorigin.Thereissomeevidencebasedonthe_AQE_Ԁthatsequencing #l! ofHispanicoriginaheadofraceresultedinproportionatelyfewer Someotherraceresponses. $X" Nonetheless,itisstillthethirdlargestracecategoryafter BlackorAfricanAmerican,and $D # showsnoindicationofdisappearing.The_AQE_ԀalsoofferssomeevidencethatHispanicsV   %0!$ WEreportedracemorecompletelyin2000styleforms,butverylargeproportions(about21percent) &"% ofHispanicsstilldidnotanswertheracequestion.InCensus2000,about17percentof '#& raceresponsesforHispanicswereimputed.B  ԍ(57page2,R1)dC  d (#'        h    |*%)   7.4.ComparingCensus2000tootherdatasources n2  Howdothedecennialdataonracecomparewiththosecollectedinothersources?  SeveralrecentstudiescompareCensus2000dataonraceandethnicitytodatafromother ` sources.BasedontheworkofBennettandGriffin(2002);Leslie,_Raglin_,and_Schwede_Ԁ(2002); L  _Raglin_ԀandLeslie(2002);and_Schwede_,Leslie,andGriffin(2002),B  dd(originally=5page1,R1)ddC  Ԁweexaminedhowraceand 8  HispanicorigindifferinCensus2000andtheCensus2000SupplementarySurvey(C2SS). $ t OneofthemainobjectivesoftheACSistoserveasareplacementforthelongformforthe  ` 2010Census.Therefore,itisveryimportanttounderstandhowCensus2000andC2SSdiffer  L  forraceandHispanicorigin,andwhatrevisionstoproceduresandthequestionnairescanreduce  8  thesedifferences. $  Hispanicoriginreporting.BennettandGriffin(2002:210)foundnodiscernibledifferencesin   theproportionofHispanicoriginresponses,butfoundsignificantdifferencesinthedetailed   Hispanicoriginresponses.B  dd(45page210,R1)C  ddB  dd(Originally=5page51,R1)C  ddSpecifically,theyfoundthatcomparedwithCensus2000,C2SS   producedproportionatelymoreMexicans.Ontheotherhand,the OtherHispaniccategory p declinedbyabout17percent.Presumablythisreflectsproportionatelylowerreportingofgeneral \ Hispanicresponses,asshowninotherresearchby_Cresce_ԀandRamirez,2003;Logan2002;and H _Suro_Ԁ2002. 4 BennettandGriffin(2002:210)speculatethattheobserveddifferencesareduetotheuseof  \ examplesintheC2SSduringtelephoneandpersonalvisitinterviewing.Theseaidswerenot H providedduringCensus2000operations,althoughonecouldarguethatthepresenceofthe 4 Hispanicorigincheckboxgroupsactasexamples.B  ԍ(perBetsyMartins7.22.03e.mailcomment)C  ԀThisreasoningdoesnotexplainwhythe   MexicanpercentageisalsolowerinCensus2000.   dRacereporting.BennettandGriffin(2002:208210) foundsignificantdifferences between  C2SSandCensus2000 distributionsforboththeracealoneandracealoneorincombination  categories.Theauthorsfoundanumberofdifferencesintheracedistr_ibutions_,butthe l dpercentageof Whitealone and Someotherracealoneshowedthegreatestdifference. X  TheC2SSshowedproportionatelymore Whitealoneresponsesandfewer Twoormore H! racesresponses.Census2000showedproportionatelymore Someotherracealone 4"  responses,asexplainedinmoredetailbelow.Smallbutsignificantdifferencesalsoexistedfor  #p!  BlackorAfricanAmericanalone, AmericanIndianorAlaskaNativealone,and Asian  $\" Ѐalone(BennettandGriffin2002:208).Theauthorsalsoexamined}X\XXX\theracedistributionsfor $H # HispanicsandnonHispanicsbecausereportingpatternstendtobequitedifferentforHispanic %4!$ respondents.#X\XX}X\o"#     h }X\XXX\ & "% #X\XX}X\##}X\XXX\RacereportingbyNonHispanics.#X\XX}X\##}X\XXX\ԀComparedwithCensus2000,C2SShadslightlymorereports (#' of Whitealoneand Asianalone,andfewerreportsof AmericanIndianandAlaskaNative )$( aloneand Twoormoreraces.#X\XX}X\R$#}X\XXX\ThelargestdifferencebetweenCensus2000andC2SSwas *%) forthe Someotherracealonepopulation.#X\XX}X\%#}X\XXX\When Twoormoreraces#X\XX}X\k&#}X\XXX\isbrokeninto l+&*  TworaceswhichincludeSomeotherraceand Allotherracecombinations,Census2000 X,'+ hadproportionatelymoreracecombinationsthatincluded Someotherrace#X\XX}X\&#}X\XXX\Ԁasoneoftheraces D-(, PN     thandidC2SS.Ontheotherhand,C2SShadproportionatelymorereportsofallotherrace  combinationsthandidCensus2000#X\XX}X\(#}X\XXX\.#X\XX}X\Q)#  RacereportingbyHispanics.ComparedwithCensus2000,C2SShasabout31percentmore t reportsof}X\XXX\ Whitealone,about30percentfewer Someotherrace#X\XX}X\>*#}X\XXX\Ԁreports,andabout24percent ` fewerreportsof Twoormoreraces#X\XX}X\*#}X\XXX\amongHispanics.Lookingat Twoormoreraces#X\XX}X\r+#}X\XXX\broken L  into TworaceswhichincludeSomeotherraceand Allotherracecombinations,Census2000 8  hadproportionatelymoretworacecombinationsthatincluded Someotherrace#X\XX}X\+#Ԁasoneofthe $ t racesthandidC2SS,andproportionatelyfewerthatincluded Allotherracecombinations.  ` )}X\XXX\Basedontheapparentreportingdifferences,theCensusBureauconductedtwostudies.Thefirst,  8  asemistructured#X\XX}X\-#}X\XXX\studyofdebriefingdata,suggestedsomeC2SSinterviewersusedactiveprobes $  thatmayhaveinfluencedreportingofspecificraceresponses#X\XX}X\.#}X\XXX\Ԁ(Leslie,_Raglin_,and_Schwede_,   2002:2068)#X\XX}X\/#}X\XXX\.#X\XX}X\S0#}X\XXX\B  ԍ(49page2068,R1)#X\XX}X\0#}X\XXX\ԍC  Inanotherstudyofthedebriefingdata,_Schwede_,Leslie,andGriffin(2002)found   thatfeweryearsofexperience,regionofthecountry,andinterviewerinterpretationofwhatthe   racequestionwasaskingmayhaveaffectedtheresponses.#X\XX}X\0#Thesestudiesalsofounddifferences   ininterviewersinterpretationsof whattheracequestionisasking,andnotedthatB  ԍdd(48page3136,R2)ddC  interviewers p pressedresearcherstoexplain justwhatitisheadquarterswantstocollectwiththerace \ question. H        h      p      x e}X\XXX\_Raglin_ԀandLeslie(2002:2827)matchedrespondentsinterviewedintheC2SStotheirCensus2000  p recordsandcomparedresponsestotheracequestion.#X\XX}X\4#}X\XXX\B  dd(47page2827,R1)dd#X\XX}X\5#}X\XXX\ԍC  #X\XX}X\*6#}X\XXX\ԀTheauthorsfoundmuchmoreconsistent  \ raceresponsesamongrespondentswhoansweredbothCensus2000andC2SSviamailthan H amongthosewhowereinterviewedinboth.ThiswastrueforbothHispanicsandnonHispanics. 4 #X\XX}X\y6#Ԛ}X\XXX\Inexplainingthefinding,_Raglin_ԀandLeslie(2002:2831)notethathouseholdswerenotassigned   randomlytomailvs.personalinterviews,butratherwereinterviewedbecausetheydidnot   respondtothemailquestionnaireandmayrepresentthehardtoenumeratepopulation.B  ԍ#X\XX}X\7#}X\XXX\dd(47page2831,R1)#X\XX}X\}9#}X\XXX\ԍC  dd  #X\XX}X\9#}X\XXX\AmongnonHispanics,_Raglin_ԀandLeslie(2002:2831)notedgoodconsistencyinreporting  whenboththeCensus2000andC2SSdatawerecollectedviamailforWhite,Black,andAsian l respondents.TheyfoundonlymoderateconsistencyforAmericanIndianandAlaskaNative, X  Someotherrace,andTwoormoreracesrespondentsB  ԍ.#X\XX}X\O:#ԍ}X\XXX\ԍ#X\XX}X\?<#}X\XXX\dd(47page2831,R3)dd#X\XX}X\<#}X\XXX\ԍC  .#X\XX}X\<#}X\XXX\_Raglin_ԀandLeslie(2002:2831)concluded D! that,for nonHispanicswhoarewillingtofilloutthemailforms!racereportingisconsistent 0"  withtheexceptionofpeoplereportingtwoormoreraces.#X\XX}X\D=#}X\XXX\B  ԍ(47page2831,R4)#X\XX}X\>#}X\XXX\ԍC  #X\XX}X\>#}X\XXX\ #l! AccordingtoGriffinetal.,(2002:63)thesestudiessuggestthatdifferencesininterviewing $D # techniquesusedinCensus2000andC2SSmayhaveledtomorereportingof Whiteinthe %0!$ C2SSandmorereportingof SomeotherraceinCensus2000forHispanics.Thisresearchdid &"% notexplainthedifferencesseenfornonHispanics.Thesefindingsledresearcherstoinvestigate '#& processingdifferencesbetweenCensus2000andACS.#X\XX}X\L?#Oneprocessingdifferenceintherace (#' editsforenumeratorformsinCensus2000mayhaveledtoanoverstatementofthenumberof )$( respondentsinthe Someotherraceand Twoormoreracescategories.B  ԍdd(5page5,R2)ddC  ԀOtherresearch |*%) suggeststhismaynotbetheentireexplanation,althoughitmayaccountforsomeofthe h+&* differencesindistributions.ThisprocessingdifferencemayhaveexaggeratedtheTwoormore T,'+ racescategorybyabout15percent(see_Cresce_,2003). p  @-(, &   7.5.ComparingCensus2000andthe1990census XD  GiventhechangesintheraceandHispanicoriginquestionin2000,howcanthesedatabe'D  comparedtodatafrom1990?Whatarethelimitationsofsuchcomparisons?Whatlessons t havewelearnedaboutbridgingtheCensus2000racedatasothattheyaremorecomparable ` todatacollectedpreviouslyandtodatainotherdatacollectionsthatdonotallowformorethan L  oneraceresponse? 8  Hispanicorigin.AlthoughtherewerewhatturnedouttobesignificantdifferencesintheCensus  ` 2000andthe1990censusHispanicoriginquestions,theoveralltotalHispanicpopulationdata  L  arereasonablycomparable.Forexample,Logan(2002:3,4)concludedthatCensus2000hada  8  goodcountofHispanics,butdidnotdowellinidentifyingtheirspecificorigin.B  dd(35page3,R1)dC  dSeveralstudies $  indicatethattheobservedchangesinthedistributionofdetailedHispanicgroupsinCensus2000   werenotdueentirelytoashiftinhowpeopleofHispanicorigindefinethemselves.Rather,   thismayhavebeenaffectedbysomechangeinthewayweaskedtheHispanicoriginquestion.   Weareleftwiththequestionofwhethertheeliminationofexampleswastheprobablecauseof   thereportingdifferencesindetailedHispanicgroups.TheGAOreport(2003a)highlightedthe p discontentwiththereportingofspecificHispanicsubgroupsinCensus2000.Thisreportmarks \ animportantturningpointinfeedbackgiventotheCensusBureau.Publicconcernisnow H focusedonaverycompletecountofspecificsubgroupswithinminoritycategories,ratherthan 4 theconcerninpreviouscensuses(e.g.,_Choldin_,1986)withthedifferentialundercountofminority  p groups.  \ Race.ThefundamentalchangestotheracequestioninCensus2000whichallowedrespondents 4 toreportmorethanoneracehavecomplicatedcomparisonswithpastcollectionsthatallowed   onlyonerace.TheCensusBureauconductedtheCensusQualitySurvey(_CQS_)toassistdata   usersincomparingracedataobtainedunderthenewschemawiththatcoll_ected_Ԁundertheformer  format.  The_CQS_Ԁisveryimpressivebecauseofitslargesamplesize,highresponserates,representative l sample,andthehighmatchingratewithCensus2000records.Butdespiteanenviablesurvey X  execution,the_CQS_Ԁhasseverallimitations:toofewcasesreportingmorethanonerace,whichare D! furtherdiminishedbyinconsistentracereporting,reluctancetoselectonerace,andthesplitpanel 0"  design.Thecomplexmethodologyandmultiplemodesofdatacollectionwillmakeitdifficultfor #l! userstodecidehowbestto bridgemultipleracedatafromCensus2000toothersingleracedata $X" collections.Butbeforewedismissthe_CQS_,weneedtoconductadditionalresearchandanalysis, $D # andweneedtoexplorehowtopoolthepaneldata.B  dd(perBetsyMartin7.23.03e.mail)C  dd %0!$ Inretrospect,itseemsthatthe_CQS_Ԁmethodologymayhaveintroducedmanymoresourcesof '#& bias,suchastimelag,movergainsandlosses,interviewereffects,modedifferences,proxy (#' reporting,andpossiblymatchingproblems(allofwhichmaygiverisetoinconsistentreporting) )$( withoutentirelyeliminatingconditioningeffectsorensuringtheindependenceofobservations. |*%) Itisquiteclearthatmuchmoreanalysisisrequiredtofullyexplorethe_CQS_Ԁdataandunderstand h+&* itsimplicationsforracereportingandbridging. T,'+  @-(, MPP M  M   7.6.PuertoRico XV     GiventhattheCensus2000ofPuertoRicowasthefirsttoaskraceinadecennialcensusinmany  decades,whatweretheissuesincollectingthosedata?Whatwerethegeneralattitudesand t problemsexpressedbythePuertoRicanpublicintermsoftheracequestion?Howdotherace ` andethnicitydatacollectedinPuertoRicocomparewithdatacollectedstatesideforthetotal L  population,Hispanics,andPuertoRicansintheUnitedStates? 8         h      p     Hispanicorigin.AlthoughthereisnobenchmarkforHispanicorigininPuertoRico,theresults  ` fromCensus2000appeartobereasonableprimafacie,andthereappearstobenoparticularbias  L  incomparisonsofrespondentfilledandenumeratorfilledreturnsinPuertoRico.V  ԀW7[Incontrast,  8  therearesignificantdifferencesinthedistributionsstateside:enumeratorreturnsshow $  proportionatelyfewerHispanics(withtheexceptionofCubansamongthegroupsexamined).   IntermsofreportingofdetailedHispanicgroups,theredidnotappeartobeexcessivereporting   ofgeneralHispanicterms,probablybecausetheoverwhelminglydominantgroupontheIsland   (PuertoRicans)appearsasareportingcategoryinalldatacollections._Cresce_ԀandRamirez   (2003:11)suggestthatthePuertoRicancategory,alongwiththeCubancategory,weretheleast p affectedbychangesintheHispanicoriginquestionusedinCensus2000. \ Race._Berkowitz_Ԁ(2201:17)reportedthatatleastinsomeurbancoastalareasofPuertoRicothe 4 racequestionelicitedstrongnegativereactionsfromparticipantsinfocusgroups.Shereports  p thatseveralparticipantsreportedthatthey stoppedfillingouttheirquestionnaireuponreaching  \ theracequestion.Someparticipantsfeltthequestionswerediscriminatory,divisive,andnot H appropriatefor theCreoleormixedrealitiesofPuertoRico.Thereisnoevidenceonhow, 4 orevenif,thesenegativereactionsaffectedresponserates.Whatwedoknowisthat,inspite   ofthesereactions,racereportinginPuertoRicowasquitegoodintermsofcompleteness.   UnlikestatesideHispanicsandPuertoRicans,Islandresidentsweremuchmorelikelytorespond  torace.Islandresidentswerealsomuchmorelikelytoreportas Whitealoneandmuchless  likelytoreportas Someotherracealone.Thehigherproportionof Someotherracealone l responsesstatesidemaybeafunctionofthemuchlargerrace_nonresponse_ԀamongHispanicsinthe X  UnitedStates,andconsequentlytheneedtoimputeracedatafornonrespondents.Hispanic D! residentsofPuertoRicowerealsomorelikelytoselfreportas Blackalonethantheirstateside 0"  counterparts,butPuertoRicansintheUnitedStateswereonlyslightlylesslikelytoselfreport #l!  BlackalonethantheirIslandcounterparts. $X" eTherewasconcernthatenumeratorsmayhaveaffectedracereportingbecauseofthelargerole %0!$ enumeratorsplayedinPuertoRico.Enumeratorreturnsshowslightlylessreportingof White, &"% andmorereportingof Someotherrace.Theyalsoshowslightlylessreportingof Black '#& thanselfreportedreturns.Interestingly,enumeratorreturnsintheUnitedStatesshowedthe (#' opposite:slightlyhigherproportionof Blackthaninmailreturns.Inthiscase,itishardto )$( concludethatenumeratorssomehowsignificantlybiasedordistortedtheracedataofHispanics. |*%) Theracereportingpatternofmailreturnsissimilar,althoughcertainlynotidentical,forHispanic h+&* respondentsbothintheUnitedStatesandinPuertoRico.Thedifferencesinracereportingsuggest T,'+ thattheunderstandingandconceptualizationofraceisdifferentforPuertoRicansontheIsland @-(, [Q    thanintheUnitedStates.Despitethecontroversythataskingraceengendered,ahigherpercentage  ofPuertoRicansontheIslandreportedaracethandidPuertoRicansintheUnitedStates.  ee &   7.7.Futureresearch be[k    h      p  t Whatresearchandtestingshouldbeconductedbeforethe2010Censusinordertoimprove L  theCensus2000questionsonraceandHispanicorigin? 8  'tOkThesuggestionsarisingfromthisreviewareconsistentwiththosealreadyunderwaywiththe  ` 2003NationalCensusTest(e.g.,examiningtheroleofexamples,changingquestionwording  L  andinstructions,dropping Someotherrace,changingresponsecategories,andexamining  8  newapproachestocollectingdataonraceandethnicity). $  Examples.Weneedtotesttheeffectofexamplesingettingbetterinformationaboutdetailed   Hispanicoriginandracegroups.Thedetailwillhelpnotonlytoensureacompletecountbutalso   togetthedetailedtabulationsthatdatausersareexpectingustobeabletogenerate.   Questionwordingandinstructions.Weneedtotesttheeffectofrestoring origininthe \ Hispanicoriginquestion,improvingtheinstructionforthe OtherHispaniccategory,and H clarifyingtheinstructionstorespondentstoanswerbothquestionsandtonotgiveanethnicity 4 responseinrace.  p  Someotherrace(_SOR_).Weneedtotestthefeasibilityofeliminatingthe_SOR_Ԁcategory H becauseitisnotveryconsistentlyreportedoruseful,exceptasacollectioncategory,andbecause 4 wehavetoeliminateitforotherpurposes,suchassurveycontrolsandpopulationestimates.   However,wemustalsounderstandwhatwillhappenifrespondents,especiallyHispanics,   continuetoreportanidentitywhichisnotoneoftheOMBraces.  ЀFewerResponseCategories.Weneedtocontinuetotestapproachestoreducethe national  origincategoriesinbothraceandHispanicorigin.Someoftheseissuesarebeingexploredin l the2003NationalCensusTest,butadditionalresearchneedstobeconducted.The_CRS_Ԁfindings X  suggestthatdetailedcategoriestendtobelessconsistentlyreported.Partofthismaybedueto D! theconfusionassociatedwiththepresenceofsomeAsianandPacificIslandernationalorigin 0"  groupsintheracequestion,andHispanicnationalorigingroupsintheHispanicquestion. #l! Thiscreatesconfusionforsomerespondentsaboutthepurposeofbothquestions.Asnotedby $X" _Schwede_,Leslie,andGriffin(2002:3136),ourresponsecategoriesare astrangepasticheofskin $D # color(whiteandblack),internalindigenousethnicgroups(e.g.,AmericanIndian/AlaskaNative), %0!$ U.S.IslandAreas(e.g.,Samoa),nationality(e.g.,Japanese),andgeographicalregionformany &"% countries(otherAsian).B  ԍ(48page3136,R1)dC  dTheauthorsalsonotethatevenourCensusBureaufieldrepresentatives '#& expressedsomeconfusionaboutexactlywhatheadquartersintendedtocollectwiththese (#' questions.B  ԍ(ChangesmadeperTerryAngueirascommentsatExecStaff T,'+ hearing)C  Previousexperiencesuggeststhatremovingsomeofthecategorieswouldbedifficult )$( becausemanyconstituentsexpectthatexistinggroupswillberetainedontheform,andinfact |*%) wehavebeenunderpressuretoexpandthenumberofcategoriesshown.But,aswithother h+&* questions,beforechangescanbemade,extensiveresearchandtestingneedstobedone. T,'+  @-(,  rR       ;XX\  8.Recommendations bey|#X\X;Z|#   Basedonthestudiesreviewedinthisreport,wemakethefollowingrecommendations(please  notethatwedonotattachanyparticularimportancetotheorderinwhichtheyappear):  3r XlA  X30 r  8.1.0rrs#rs# Pretestandevaluateallquestionnairechanges,reduceuncontrolledvariationinthe J  questionsthatareasked,andconductmoreresearchonmodeandmethodological :  influencesonthedata. B  dd(perBetsyMartin7.23.03e.mail)C  dd& vrs#rs# Itisimportantthatwepretestandevaluateallquestionnairechangespriortoimplementation.  N  Weneedtoreduceuncontrolledvariationinthequestionsthatareaskedandweneedto :  understandhowmodeandothermethodologicaldifferencesaffectthedatathatwecollect. &  Census2000had54differenttypesofforms,andmanyformshaddifferentraceandHispanic   originquestionsthanthe standardmailform.The_AQE_Ԁshowedthatevenwhatappeartobe   minorchangesontheHispanicoriginquestionproducednoticeabledifferencesintheresponses   wecollected.ThestudiescontrastingCensus2000andC2SSdatasuggestthatwedonot  understandhowmodeandothermethodologicaldifferencesaffectedtheresponsesineachdata r collection.(Seechapter2formorediscussion.) ^  r     0 r  8.2.0rrs#rs# Uselargersamplesizesfortests."rrs#rs#  Asthe_AQE_,_CRS_,and_CQS_Ԁstudiesshowed,therearemanyinstanceswherelargersamplesizes N wouldhaveimprovedourabilitytoevaluateeffectsonnumericallysmallgroups.Whilesmaller : samplesizesmaysavemoneyintheshortrun,theymayendupcostingmoreinthelongrunifthe & testsmustberepeatedtoyielddefinitiveresults.Ontheotherhand,largedatasets,suchasthe  matchedCensus2000andC2SSdata,willoftenproducetoomanystatisticallysignificant  differencestoyielddefinitiveresults.However,onbalance,itisbettertohavealotofdata  ratherthantoolittle,particularlywhenweseektounderstandhowproposedchangeswillaffect  numericallysmallgroups. r  8.3.0 r Avoidoverlycomplextestdesigns!thesimplerthebetter.6#!rs#rs#  Itisimportantthatweavoidoverlycomplextestdesigns.Forexample,thecomplexdesignof %b # the_CQS_Ԁmadeitdifficulttointerprettheresultsandanswerthequestionsthetestwasdesigned %N!$ toexplore.Havingtwopanelsinthe_CQS_Ԁeffectivelyreducedthesamplesizeavailableforusto &:"% analyze.Weneedtodoalotmoreanalysisofthe_CQS_Ԁdata,andweneedtodeterminewhether '&#& wecaneffectivelypoolthedatainordertoobtainlargersamplesizesforanalysis. ($'   +&* 8.4.0 r Explorewaystoimprovemailresponse!notonlyisitlessexpensivebutwemayalso  getmoreconsistentlyreportedracedata.rs#rs#  A}X\XXX\ԀstudyofmatchedCensus2000#X\XX}X\ً#}X\XXX\andC2SSrecordsfoundmuchmoreconsistentraceresponses x amongrespondents#X\XX}X\9#}X\XXX\whoansweredbothCensus2000andC2SSviamail.Thiswastruebothfor d HispanicsandnonHispanics#X\XX}X\܌#}X\XXX\(_Raglin_ԀandLeslie,2002:2831).However,#X\XX}X\#}X\XXX\weknowthathouseholds P  inCensus2000andC2SSwerenotassignedrandomlytomailorinterviewmodes.Infact, <  householdswhowereintervieweddidnotrespondtothemailquestionnaireand,therefore,may ( x representaparticularsegmentofthepopulationforwhomitishardcollectdata.#X\XX}X\"#}X\XXX\B  ԍ(47page2831,  d R1)d#X\XX}X\Ï#}X\XXX\C  d#X\XX}X\=#Thecombined  d benefitsoflowercostandpotentiallymoreconsistentraceresponsesmakethemaildata  P  collectionmodeevenmoredesirable.  <  0 r  8.5.0rrs#rs# Explorewaystoimprovetrainingandmonitoringofenumeratorandinterviewer   behavior. rs#rs#  }X\XXX\Nomatterhowmuchweimprovemailresponse,therewillstillbeaneedforenumeratorsand x interviewerstoconductnonresponsefollowupandotherdatacollections.Therefore,interviewer d behaviorwillalwaysbeanimportantissuefordatacollection.Basedonasemistructured P debriefingstudyofACSinterviewers,Leslie,_Raglin_,and_Schwede_Ԁ(2002)#X\XX}X\#}X\XXX\speculatethat < interviewerbehaviorcausedbydifferences#X\XX}X\<#}X\XXX\inexperienceandtraining#X\XX}X\̔#}X\XXX\mayaccountforrace (x reportingdifferencesinCensus2000andC2SS.#X\XX}X\)#}X\XXX\B  ԍ(49page2065,R3)dC  dAlthoughthisstudywasbasedonreported, d notobserved,behavior,itsuggeststhepossibilitythatsomeinterviewersusedactiveprobes P whichmayhaveinfluencedreportingofspecificracesresponses(Leslie,_Raglin_,and_Schwede_, < 2002:2068)#X\XX}X\•#}X\XXX\.#X\XX}X\#}X\XXX\ ( #X\XX}X\# r 1) Improveinterviewerunderstandingofraceandethnicityquestions.   }X\XXX\0 r InordertoensurethatwecollectreliableinformationonraceandHispanicorigin,  interviewersmusthaveagoodunderstandingoftheseconcepts.Inanotherstudyof  interviewerdebriefingdata,_Schwede_,Leslie,andGriffin(2002)foundvariabilityin x interviewerinterpretationsof whattheracequestionwasasking.Theyfoundthat d  thisvariedby#X\XX}X\#}X\XXX\yearsofexperience,typeofexperience,andregionofthecountry. P! Recognizingthis,weneedtoexplorewaysofensuringacommonunderstanding <"  amonginterviewersabouttheraceandHispanicoriginquestions.(#x!rs#rs# Ѐ#X\XX}X\ߚ# $d" 0 r   r 2) Provideastandardizedapproachforcollectingraceandethnicitydata.%P #rs#rs#  }X\XXX\0 r InordertoobtainreliableinformationonraceandHispanicorigin,interviewersmusthave %@!$ trainingandstandardmethodsfordatacollection#X\XX}X\#}X\XXX\.Itisimportanttomaintainconsistency &,"% acrossdatacollectionswithinmodesothatinterviewershavesimilarexperiencecollecting '#& thesedata#X\XX}X\ޝ#}X\XXX\.($'rs#rs# Ѐ#X\XX}X\ٞ#}X\XXX\ )$( #X\XX}X\<# r 3) Improvemethodstomonitorenumeratorandinterviewerbehavior. *%)  }X\XXX\0 r Inordertoreducetheeffectofinterviewerbehavioronthecollectionofconsistentraceand |+&* Hispanicorigindata,weneedtoexplorewaystomonitorinterviewerbehaviorthrough h,'+ training,feedback,andrewardorpunishmentofbehaviors#X\XX}X\#.B  dd(perBetsyMartin7.23.03e.mail)C  dd  T-(,rs#rs# _ԇ8.6.0 r Explorewaystominimizethedifferencesbetween,ifnotstandardize,raceand  ethnicityquestionsacrossdatacollectionsrs#rs# }X\XXX\ Thestudiesreviewedinthisreportpointoutmanydifferencesinthemethodsandmaterials d usedforraceandethnicityinourdatacollections.Inordertomaximizeourabilitytocollect P  consistentraceandHispanicorigindata,weneedtoconsiderstandardizingthequestionson <  raceandethnicity#X\XX}X\#}X\XXX\acrossourdatacollectionsasmuchaspossible.Werecognizethatmode ( x differencesmayrequirespecificapproaches,butthequestionsshouldbeconsistentwithinmode.  d Thiswillalsoreducevariabilityarisingfromdifferencesinthetypeofexperienceamong  P  interviewers#X\XX}X\_#}X\XXX\.  <  #X\XX}X\#}X\XXX\ r   #X\XX}X\I# 0 r  8.7.0rrs#rs# Withineachdatacollection,minimizeoreliminatevariationinresponsecategories   acrossformstoavoidintroducingdataprocessingdifferences. rs#rs#  }X\XXX\Wehaveonedocumentedinstanceinwhichdifferencesinthenumberof#X\XX}X\#writeinareas}X\XXX\forrace#X\XX}X\>#}X\XXX\ x responses#X\XX}X\#}X\XXX\causeddifferencesintheoutputdata.#X\XX}X\#Adifferenceintheprocessingofenumerator d forms(whichhadonlyonewriteinareaforrace)andtomailforms(whichhadthreewritein P areasforrace)ledtoanoverstatementofSomeotherraceby6percent,B  dd(55page3,R1)C  ddandanoverstatementof < Twoormoreracesresponsesbyabout15percent(seechapter5formorediscussion)B  dd(55page3,R2)C  dd.}X\XXX\Reducing (x thenumberofformsandstandardizinginputfieldswillreducetheprobabilityofspuriouserrors d indataprocessing#X\XX}X\N#. P  r        h 0 r  8.8.0rrs#rs# Considerremoving Other...checkboxesandkeepingthewriteinarea.rs#rs#  Davisetal.(2001:III16)noteaninabilityofmanyrespondentstousetheexistingcategories.  Onesourceofincomparabilityariseswhenrespondentscheckaboxandwriteinanentryinan  inappropriatearea.Forexample,Davisetal.(2001:III18)notedtheinstanceofarespondent x whoreportedherAmericanIndiantribalaffiliationinthe Yes,otherSpanish/Hispanic/Latino d  writeinarea,aftermarkingthe No,notSpanish/Hispanic/Latinobox.Bycheckingbothboxes, P! therespondentcreateda mixedHispanicoriginresponsewhichwasprobablynotintended. <"  Similarly,ifarespondentweretomarkthe OtherAsiancheckboxandwrite Irishinthe $d" writeinarea,wewouldhavetomakethedecisionofwhethertoclassifytherespondentas %P #  Asian and Whiteortoremovethecheckboxandkeepthewriteinresponse.Withoutthe %