U.S. flag

An official website of the United States government

Skip Header


Reliability and stability of the 6-question disability measure in the Survey of Income and Program Participation

Written by:

Introduction

Researchers  have  long  acknowledged  that  disability  is  a  dynamic  characteristic  (Adler   1992,  Verbrugge,  Reoma  and  Gruber-Baldini   1994,  Wolf  and  Gill  2008).  Nonetheless,   the  concept  is  often  treated  as  static  over   short  periods  in  longitudinal  studies.  The   disability  status  of  a  respondent  is  asked   during  one  interview  and  assumed  to  remain   constant  over  several  interviews  or  for  the   life  of  the  panel.  I  explore  this  assumption   using  reoccurring  data  on  disability  status   from  the  Survey  of  Income  and  Program   Participation  (SIPP).  In  the  2008  panel,  the  six-question  set  of  disability  questions  from   the  ACS  were  added  to  a  reoccurring  topical  module.  I  employ  structural  models  from  Heise  (1969)  and  Wiley  and  Wiley  (1970)  to   separate  reporting  error  from  real  change   under  two  assumptions  about  the  measures’   reliability.  Both  methods  assume  that   disability status follows a first-order Markov process. With these methods, I find that the  disability measures in the SIPP had relatively mode rate to low reliability with coefficients  between  0.414  and  0.638.  Conversely,  an  individual’s  true  disability  status  is  strongly   correlated  with  the  person’s  status  one  y ear  later  (r=0.937).  Thus,  the  supposition  that disability remains relatively consistent ov er short periods has some validity.

Related Information


Page Last Revised - October 8, 2021
Is this page helpful?
Thumbs Up Image Yes Thumbs Down Image No
NO THANKS
255 characters maximum 255 characters maximum reached
Thank you for your feedback.
Comments or suggestions?

Top

Back to Header