Question about longitudinal weights

CPS
1

I have created a longitudinally matched dataset for month in sample 1 to month in sample 4, February to May and March to June 2020. I matched respondents based on cpsidp, sex, age, race, education, and Hispanic status. However, about ten percent of respondents have a lnkfwmis14wt value 0. Why would that be the case?

2

LNKFWMIS14WT is zero in these circumstances:

  1. Any cases that do not have MISH = 1 (see the Universe page for LNKFWMIS14WT).
  2. Any individual who does not appear in all of the first four months in sample. They may have a missing survey in one or more of those months, or the sample may not be available.

I suspect these are why you are seeing so many zeros, but please follow up if you think that’s not the case.

3

I see. So the weight would equal zero if someone answers the survey in MISH1, skips MISH2 and/or MISH3, then returns in MISH4?

4

That’s right.