Question

Status:Closed    Asked:Feb 11, 2015 - 01:56 PM

Problem with creating 5 year file

I had asked about a 2013 5 year file. I received a response describing how to create my own 5 year file. The response suggested that I select the file for each year (I am only interested in one city), combine them, and then divide the weights by 5 to create a weighter for the new file. The problem is that once I did that Stata responded that "may not use noninteger frequency weights" as the new weights are no longer whole numbers. Any suggestions? Thank you.

 
Do you have the same question? Follow this Question
 

Staff Answer

avatar

Tim_Moreland

Staff

While frequency weights do not allow for non-integer weights, the other types of weights available in STATA do allow non-integer weights. Depending on the STATA command, you have available as many as four different types of weights: fweights, pweights, aweights, and iweights. In general, it is preferable to use probability weights (pweights) for IPUMS samples.


Also, if you are interested in only one city, you can use the "Select Cases" option in the Data Extraction System to create a data extract that consists of only respondents from your city of interest. This will reduce the size of your data files. See here for more information on this option.

Hope this helps.

 

Feb 11, 2015 - 02:28 PM

0
0
Report it

OTHER QUESTIONS NEEDING ANSWERS

When will "City" be available in the 2013 5 year file?
Login   |   Register

Recently Active Members

View More »

Share |