Cross-sectional weights
To study the effect of an experiment or calculate an estimate in one wave, a cross-sectional weight should be used. The specific cross-sectional weight would depend on the origin of the variables used in the analysis:
| Origin of variables used in analysis | Weight for cross-sectional analysis |
|---|---|
| All variables used are from the household enumeration grid (indall file) | w_psnenip_xw |
| All variables used are from the household enumeration grid (indall file) and/or the household questionnaire (hhresp file) and your analysis is at an individual (not household) level | w_psnenip_xw |
| All variables used are from the household questionnaire (hhresp file) and your analysis is at a household level | w_hhdenip_xw |
| All variables used are from the proxy questionnaire (indresp file) and possibly from the household enumeration grid (indall file) and household questionnaire (hhresp file) | w_indpxip_xw |
| At least one of your variables is from an individual adult questionnaire (indresp and associated files) and possibly also from the household enumeration grid (indall file) and household questionnaire (hhresp file) | w_indinip_xw |
| At least one variables is from an individual self-completion questionnaire (indresp file) and possibly also from the household enumeration grid (indall file) and the household questionnaire (hhresp file) | w_indscip_xw |
| Variables from the youth self-completion questionnaire (youth file) | w_ythscip_xw |
Longitudinal weights
If your analysis includes multiple time points or waves of Innovation Panel interviews you should use a longitudinal weight. From Wave 14 onwards we do not provide specific longitudinal weights. Instead, we provide starting points for longitudinal weights, called issue weights (w_psneni#.li). The issue weights are provided at an enumeration level and are created for each refreshment (including those before Wave 14). It takes into account selection probabilities and corrects for nonresponse up until and including refreshment year, and joins all the samples at the time of refreshment. It is expected that a user creates their own longitudinal weight starting with our issue weight and corrects for attrition themselves tailoring it to their own analysis model.
