Census Bureau

Appendix B:

Supplement on Survey Design and Methodology

Line Divider

B.1 Modifications to the Race and Hispanic Origin Questions

The 1996 RAETT question on race had a number of substantial changes from the 1990 Census version. The order of the race and Hispanic origin questions was changed in the RAETT, so that the Hispanic origin question came immediately before the race question (except for Panel D where the race question preceded the Hispanic origin question). A number of wording and formatting changes were made to test reporting in general and for certain racial response categories in particular. For example, the "NOTE: Please answer both questions 5 and 6," was added in the RAETT, and no examples for the "Other Asian or Pacific Islander" response category were provided. African American was not included in the Black response category in the 1990 Census. The wording for the "Other race" category was changed to "Some other race" in the RAETT. The spanner Asian or Pacific Islander (API) that was included for the 1990 Census was deleted.

The Hispanic origin question underwent some format and wording changes for the RAETT. The instruction "Mark [X] the 'No' box if not Spanish/Hispanic/Latino" was added to the RAETT, and "Latino" was added to the instruction. The examples given in the 1990 Census for "Other Hispanic" were deleted. The RAETT bolded "No" in the response category "No, not Spanish/Hispanic/Latino." See Figure B-1 on the following page for a copy of the 1990 Census questions on race and Hispanic origin; a copy of the RAETT questions on race and Hispanic origin is in Appendix A.

B.2 Sample Selection

Each sampling frame in the RAETT had a unique set of criteria for determining which tracts should be included on the frame. Column two of Table B-1 details these criteria, while column three indicates the number of states and tracts on each frame. The last column is the number of housing units in each targeted sample. For example, the frame for the White ethnic targeted sample consists of those tracts in which more than 66.4 percent of the households are of Scandinavian origin, and those tracts where more than 31.7 percent of the households are of French origin, etc. Also, 29 states and 133 tracts are on the sampling frame for the White ethnic targeted sample, and the sample consists of 17,550 housing units.

It is apparent from Table B-1 that not all states are represented in each of the targeted samples. Those states that did not have high concentrations of some of the population groups were excluded from the sampling frame. Table B-2 identifies the states that contained tracts which were included on the sampling frame for each targeted sample.

Figure B-1. Race and Hispanic Origin Questions from the 1990 Census


Table B-1. Final Frame Results

Targeted sample Population density criteria1 States/tracts (numbers) Sample housing units
White ethnic Scandinavians>66.4 or French>31.7 or
British>61.7 or Polish>58.5 or
Russian>29.1 or Baltic>37.6 or Eastern
Europe>49.9 or German>83.6 or Italian>69.2
or US/Amer>43.4 or White Other2>59.3
29 / 133 17,550
Black Black>50, and rural Black>5;
CA, IL, NY have higher Black percentage
and no rural
35 / 1266 26,550
Hispanic Mexican>75 or Puerto Rican>29 or
Cuban>30 or Other Hispanic>50; TX,
CA, NM, AZ have higher percentages
15 / 861 26,550
American Indian Off reservation>20.
On reservation - 10 selected tracts in CA,
ME, MN, NM, NY, NC; specific American
Indian reservations with more than
20 mailable addresses
18 / 68 15,870
Asian or Pacific Islander Japanese>35 or Filipino>35 or
Hawaiian>25 or Samoan>17 or
other API>60; HI has higher
Japanese and Filipino percentages;
CA and IL have 15 selected tracts
8 / 78 23,700
Alaska Native Specific areas/villages selected
by Census Bureau
AK Only;
20 villages
1,950
1 The population density criteria are expressed in terms of percentage of households per censsu tract.
2 This group consists of other European and Canadian ancestries.


Table B-2. States Included in Sampling Framess

States Targeted population groups
White
ethnic
Black Hispanic American
Indian
Asian or
Pacific
Islander
Alaska
Native
Alabama X X   X    
Alaska       X   X
Arizona     X X    
Arkansas   X        
California X X X X X  
Colorado X X X      
Connecticut   X X      
Delaware   X X      
District of Columbia   X     X  
Florida X X X      
Georgia X X        
Hawaii         X  
Idaho     X      
Illinois X X X   X  
Indiana X X        
Iowa X X        
Kansas            
Kentucky   X        
Louisiana X X        
Maine X     X    
Maryland X X        
Massachusetts X X X   X  
Michigan X X   X    
Minnesota X X   X X  
Mississippi   X        
Missouri X X        
Montana       X    
Nebraska X X        
Nevada   X   X    
New Hampshire            
New Jersey X X X      
New Mexico     X X    
New York X X X X X  
North Carolina X X   X    
North Dakota X     X    
Ohio X X X      
Oklahoma   X   X    
Oregon   X        
Pennsylvania X X X   X  
Rhode Island X          
South Carolina   X        
South Dakota       X    
Tennessee X X        
Texas X X X      
Utah X     X    
Vermont            
Virginia X X        
Washington       X    
West Virginia X X        
Wisconsin X X   X    
Wyoming            

B.3 Edits and Recodes

The mail return data on race were recoded based on the response categories checked by the respondent and additional data provided by the respondent in the entry spaces associated with American Indian or Alaska Native, Other Asian or Pacific Islander, Other race, and reporting more than one race. For Panel A and the panels with an explicit multiracial category (Panels B, D, E, and G), race data were recoded as "Other Asian or Pacific Islander" if two or more boxes were checked in the 10 Asian or Pacific Islander detailed categories, or if no response was provided and two or more written entries were entered in these 10 categories. If two or more boxes were checked and one was outside of these 10 Asian or Pacific Islander categories, or if no response was provided but one of two or more written responses were outside of the 10 categories, the response was considered "multiple race" and placed in a separate category within the race distribution. If the respondent checked the multiracial category or provided any valid response in the associated multiracial write-in space, the response was considered multiracial regardless of other checked boxes or write-in responses.

For the panels that allowed respondents to mark more than one race category (Panels C, F, and H), race data were recoded as "Other Asian or Pacific Islander" if two or more boxes were checked in the 10 Asian or Pacific Islander detailed categories, or if no response was provided and two or more written entries were entered in those 10 categories. If two or more boxes were checked and one was outside of these 10 Asian or Pacific Islander categories, or if no response was provided but one of two or more written responses were outside of the 10 categories, the response was treated as multiple race.

For the panels with separate race and Hispanic origin questions (Panels A, B, C, D, G, and H), the written race responses were recoded in another major race group only if the respondent had marked one of the check boxes associated with the write-in spaces or no box was checked. For the Hispanic origin question, written responses to "Other Spanish/Hispanic/Latino" were used to recode the responses.

For the panels with a combined race and Hispanic Origin question (Panels E and F), the written responses to the ancestry question were used to recode if the "Other race" box were marked. If the American Indian or Alaska Native box were marked, the write-in responses were used to recode into another major race group if necessary. If no response box were marked, both the American Indian or Alaska Native written response and the ancestry written responses were used to code the response to the combined question.

Source: U.S. Census Bureau, Population Division and
Decennial Statistical Studies Division

Questions? / 1-866-758-1060
Last Revised: October 31, 2011 at 10:03:15 PM