The instrument, by design, skipped the screens EMPLOC (whether the employer operated in more than one location), EMPALL (number of employees of this employer at all locations) and EMPSIZE (number of employees of this employer at this location), when any of the following conditions existed:
EMPLOC, EMPALL, and EMPSIZE were intended to be asked every other wave. They were asked predominantly in waves 1, 3, 5, 7, and 9.
In the edit, the variables, EEMPLOC, EEMPALL and EEMPSIZE, are filled with the previous wave data when the unedited data reflects that the respondent has the same job, and the value for the employer having more than one location is not reported. During waves 2, 4, 6, and 8, we have these criteria satisfied more often than not. Contingent workers is a notable exception.
These variables show an inconsistent flagging in the 2001 Panel, though the collection of the data is similar. When using previous wave data as current data for EEMPSIZE and EEMPLOC, an allocation flag of zero was given, (AEMPSIZE=0), (AEMPLOC=0), indicating no imputation; the data was considered reported. In contrast, EEMPALL was given an allocation flag indicating a statistical or logical imputation of previous wave data, (AEMPALL=4). Therefore we show a high imputation rate for EEMPALL compared to EEMPSIZE and EEMPLOC. This higher imputation for EEMPALL is most evident in waves 2, 4, 6, and 8.
Resolution of the Problem:
The imputation flag for EEMPALL will remain a four (AEMPALL=4) and for EEMPLOC and EEMPSIZE will remain zero, (AEMPLOC=0, AEMPSIZE=0). The problem should be addressed in the 2004 Panel.
|
Main |
Introduction to SIPP |
Survey Content |
Technical Information |
Using & Linking Files |
SIPP Publications |
|
Access SIPP Data |
SIPP Users' Guide |
Tutorial |
User Notes/ListServe/News |
SIPP Help |
Page Last Modified: May 12, 2006
