Wednesday, April 25, 2007

Variables in National Longitudinal Survey of Children and Youth cycles

Question

A student is asking why, in the primary PUMF files for the NLSCY, the numbers of variables are so different:

cycle 1 - 1415 variables
cycle 2 - 913 variables
cycle 3 - 331 variables

I'm guessing it's because of the increasing need to drop variables to protect confidentiality as the survey continued, but he wants to make sure. Am I right, or were variables dropped on a whim?

Answer

We had a presentation from the folks looking at longitudinal PUMFs at the last EAC meeting. The issue of dropping variables was brought forth and the answer was that they are ensuring that respondent confidentiality does not become an issue when a file is longitudinal. This helps explain when the numbers of variables shrinks from cycle to cycle - the more detail is available, the more risky it becomes.

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.