Thursday, February 26, 2015

General Social Survey 2005 Episode File


I’m trying to clean up weight variable declarations in Equinox, and have encountered a problem. There are two weight variables released for the episode file for GSS cycle 19 – WGHT_EPI and WGHT_PER. However, the documentation for the episode file indicates that WGHT_EPI is the appropriate weight variable for each variable in the episode file. There also isn’t any note in section 7 (Estimation) specifically about the episode file. What is WGHT_PER (on the episode file) to be used for? (Should it be listed as a weight variable for all variables for which WGHT_EPI is indicated?)


Response from subject matter:

"If the client is looking at each episode, then WGHT_EPI is the correct weight. If they are looking at each respondent, then WGHT_PER is used.

The episode is one-to-many. Therefore, each respondent has many episodes, each episode gets a record. The main file, each respondent has a record.

So if the client is looking at the number of episodes, then use WGHT_EPI, otherwise, use WGHT_PER."