The problem of respondent attrition: Survey methodology is key

Monthly Labor Review

Volumn 128, Issue 2, 2005, Pages 63-70

  Olsen, Randall J ^a  

^a OHIO STATE UNIVERSITY   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

EID: 28244449085     PISSN: 00981818     EISSN: 00981818     Source Type: Journal    
DOI: None     Document Type: Review

References (11)

1. This is the rule determining to which respondents interviewers would return in the event the respondent did not complete one or more interviews.
2. Census interviewed the young men and young women in person for several consecutive years and then reverted to the interview pattern for the two older cohorts - personal, skip a year, telephone, skip a year, telephone, personal. In the late 1980s the women's interviews were done in person, although many cases were done over the phone when circumstances dictated.
3. See Frank Mott, "Looking Backward: Post Hoc Reflections on Longitudinal Surveys," in Erin Phelps, Frank Furstenberg, and Anne Colby, eds., Looking at Lives: American Longitudinal Studies of the Twentieth Century (New York, Russel Sage, 2002).
4. The two original men's cohorts - NLSY79 and NLSY97 were cancelled in the early 1980s.
5. With the rise of nonmarital unions, event histories on marriage and cohabitation have become more complex. The approach of asking the retrospective question with explicit reference to the respondent's previous answers is referred to as "bounded interviewing."
6. The discrepancy is handled at Ohio State. This style of data collection generates fewer "seam problems" than histories that are not collected using bounded interviewing. Our current practice is to accept the date given initially and place the "seam" on the day after the date of interview.
7. The low-income, non-black, non-Hispanic oversample was nearly eliminated in 1993 and most of the military oversamples were dropped earlier.
8. Currently the normal incentive is $40.
9. In the past 15 years, we have also experimented with monetary incentives to interviewers. Unfortunately, those experiments failed. At best, they simply shifted the timing of when cases were completed.
10. Our CAPI system is based entirely on Web browsers. Some interviews done over the phone utilize the Web to present the interview to the interviewer, so offering a Web interview to respondents is a minor step.
11. The other side of repeated contact is that the respondent may be more likely to give normative responses to an interviewer once they have established rapport.