Panel Paper: Designed Redundancies: How Respondents Answer Similar Questions in the SIPP

Thursday, November 6, 2014 : 2:45 PM
Isleta (Convention Center)

*Names in bold indicate Presenter

Elizabeth Crowe, American University; Congressional Research Service
The Survey of Income and Program Participation (SIPP) contains inherent redundancies.  In the SIPP, respondents are asked in multiple questions whether or not they received Food Stamps or food assistance.  This paper compares responses to these questions.  It asks whether respondents answer questions about receipt of food assistance from the government similarly across multiple questions.  It examines whether differences in how questions are asked and whom questions are asked of change responses to the questions.   

This paper uses the 2008 Panel of the SIPP to examine these redundancies.  The SIPP is a nationally representative, multi-year, longitudinal survey conducted by the US Census Bureau.  This paper focuses on questions about receipt of food stamps [er27], total household and family food stamps received [thfdstp, tffdstp], receipt of food assistance [er61], and if the source of food assistance source is a government agency [efoodsc1].   The first three questions are stand-alone questions, not prompted by any other question.  The final question [efoodsc1] is asked of all respondents who indicated that they received food assistance [epaothr3] and responded that the type of food assistance received was money and/or vouchers [efoodtp1]. 

Responses to these questions should be identical.  All respondents who received Food Stamps or food assistance from government programs should have the same responses to all questions.  This ensures consistency in the measurements across the data.  However, responses are not identical across the SIPP.  This is especially true for respondents who indicated that they received food stamps and respondents indicating that the source of their money and/or voucher assistance was from government programs.  Preliminary results suggest that how questions are asked, who responds to the questions, and imputation processes used to fill in missing information are important determinants differences in respondent answers. 

This paper provides important insight into both the variables available in the SIPP to measure receipt of food assistance as well as how respondents answer questions.  The SIPP is one of the best available sources of information about low-income individuals and families.  This paper provides an understanding of how accurate the information is.  The findings of this study can be used by researchers and SIPP administrators to better understand the measures in the SIPP.  It is important for researchers to understand how variables differ within the panel and how responses vary across the panel. Findings can help researchers choose the best measure for their research and help SIPP administers improve the accuracy in the data.