Question
I was just checking out the Labour Force Survey on the STC Nesstar site and noticed that it was monthly. But I have a memory of annual LFS files in the past. And this was reinforced when I checked out the
odesi site. My question - which files do we use - or can we use both the annual and the monthly files. eg. LFS 1980. I do remember seeing a note about this in the distant past and went searching for it but was unable to find it.
Answer
I searched the dlilist archives and did find references to "annual" LFS files being received from the author division. According to the dlilist archives, at one time, we received LFS data annually instead of the monthly LFS files we now receive. (http://www.statcan.gc.ca/dli-ild/data-donnees/mhonarc/dli-ild-msg/msg04282.html).
“The LFS "annual file" is NOT a real annual file. That is there has not been any re-manipulation of the weights so that the data reflects an annual estimate. All you get is all the records from each monthly file put into one huge file. The monthly weights associated with the records for that month still has to be used - there is NO annual weight. Trying to use the records in this file to reflect annual estimates means that you must use each of the monthly weights and then divide the result by 12. The result will not be exact and will probably not match any estimates previously released and reported.
My advice would be to use the monthly files only.”
(Full thread available at http://www.statcan.gc.ca/dli-ild/data-donnees/mhonarc/dli-ild-msg/msg08827.html
Showing posts with label Weighting and Estimation. Show all posts
Showing posts with label Weighting and Estimation. Show all posts
Friday, February 26, 2010
Friday, November 21, 2008
Weighting in GSS Cycles
Question
I'm using cycles 1, 9 and 18 of the General Social Survey. Do I have to incorporate the weights for the cases to find correlations and do regression analyses? For example, in cycle 9 the weight variables PERWGHT.
I have a faculty member asking this question regarding GSS cycles and weighting. Your help will be greatly appreciated.
I'm using cycles 1, 9 and 18 of the General Social Survey. Do I have to incorporate the weights for the cases to find correlations and do regression analyses? For example, in cycle 9 the weight variables PERWGHT.
Answer
Yes, you should always use the weighted data to do this type of calculations and you may want to use the program bootvar to calculate the variance.
Hope this helps.
Thursday, January 25, 2007
Bootstrap Weights List
Question
Is there a list anywhere of surveys that have bootstrap weights available or is this something I can find out from the files available
(documentation etc)?
Answer
I have a list of files with bootstrap, jackknife and BRR weights at:
http://www.chass.utoronto.ca/datalib/major/bootstrap.htm
Is there a list anywhere of surveys that have bootstrap weights available or is this something I can find out from the files available
(documentation etc)?
Answer
I have a list of files with bootstrap, jackknife and BRR weights at:
http://www.chass.utoronto.ca/datalib/major/bootstrap.htm
Subscribe to:
Posts (Atom)