ACS Labor Force Questions: Results from Cognitive Interview Testing
Jennifer M. Rothgeb
KEY WORDS: cognitive testing, American Community Survey, ACS, labor force questions, interviewer-administered survey, working, looking for work, weeks worked
ABSTRACT
During mid-2004, the Housing and Household Economic Statistics Division (HHES) determined that employment and unemployment results obtained through the Census 2000 data revealed substantial differences from those obtained through the Current Population Survey (CPS.) The American Community Survey (ACS) is utilizing the same questions as those in Census 2000. Examination of 2003 ACS estimates of employment and unemployment revealed that estimates of employed persons were about 4 percent below CPS and the unemployed estimate in ACS was approximately 25 percent higher than CPS. (CPS is the official source of labor force estimates in the U.S.) The objective of the research in this study is to develop and cognitively test alternative question wording that can be used in the split-panel ACS 2006 Content Test. Content Test results will be compared to CPS to determine if revisions to question wording brought the ACS estimates closer to CPS. A multi-method approach was used to serve as input to the development of alternative labor force questions. Data was obtained through behavior coding research conducted on fifty one interviews with the current ACS labor force series, debriefing of 15 ACS interviewers and expert reviews by staff from the Bureau of Labor Statistics, the Bureau of Transportation Statistics, and the Population Division of the U.S. Census Bureau. Cognitive testing of various question wording was conducted with 40 laboratory participants. Cognitive testing was done using the three interviewing modes used in ACS: self-administered questionnaires, computer-assisted telephone interviewing, and computer-assisted personal interviewing. The recommendations resulting from cognitive testing were included in the ACS 2006 Content Test and results of that test were favorable towards the recommended test items when compared to CPS estimates.
CITATION:
