The purpose of this work is to look for possible causes for the patterns of between/within wave transitions reported by Burkhead and Coder (1985). They found that much larger proportions of transitions between receipt and nonreceipt of several income sources and food stamps were reported as occurring between waves than within waves. In this investigation we found no major influences from demographic characteristics, interview status, imputation procedures or longitudinal editing on this reported pattern. However, some smaller scale effects were noted.

1.   For food stamps and social security, larger proportions of receipt of sources were reported by self-respondents than by proxies. There is usually a higher proportion of transitions between waves when at least one of two consecutive months has a proxy response than when both of the months are self-reported. (See Appendix 1)

2.   There is a larger proportion of between wave transitions when at least one of two consecutive months is imputed than when both of the months are reported. For unemployment compensation the same is true for within wave transitions, where almost all imputes are needed when both of the months require imputation. (See Appendix 2)

3.   For state and supplemental unemployment compensation a small but noticeable increase in the ratio of between to within wave transitions results from reimputation by the longitudinal edit. (See Figure 3.8)

There are certainly other relationships of small magnitude that will be found by further examination of the data, but nothing on the order of magnitude of the between/within wave differences. Further work should concentrate on improving the quality of responses.