Hello, I am not so familiar with the sampling weight. I am using PISA to do the analysis. I take the non-response data out from the original data, and do a student weight adjustment . However, when I was doing CFA, I am not sure should I put a final student weight(after adjusted) or 80 replicate Fay weights.
Is it fine that I only include a final student weight in the CFA analysis?
You should use both the replicate weights and the student weights. You can leave the non-response data in the file, just designate it as missing data.
Ali posted on Monday, November 09, 2015 - 11:52 pm
Thank you! I have another question for non-response data. The student questionnaire has three forms, and there are some items are overlapped, but not every student answers the same item. For example, there are 1000 students answering the questionnaire, but only 500 students answer item 1 due to different forms.So, the missing rate is 50%.
Is it fine that I leave the non-response data as the missing data in the file and don't change the original student weight?Because I am doing across 11 countries, if I leave the non-response as the missing data, it could have a lot of missing data in the file.
Or, could I remove the non-response data out then use the original weight?