Status:Closed    Asked:Jun 24, 2014 - 05:52 PM

Faster way to select variables?

My present task is to create extract(s) from IHIS spanning the early seventies through the present including all variables available from a list of over 500 variables we are looking for. I spent a few hours selecting all necessary variables, then selected all samples. Unfortunately, this reduced my variable list from >500 to only 54, seemingly due to availability restrictions (i.e. if one or more years doesn't have a variable then it is dropped entirely instead of just being left unfilled). This was very dissappointing.

Is there a way to save a list of variables to reuse? Is there a sample year that has ALL variables so I could create one extract on that year and then change the sample in subsequent revisions (perhaps 2010)? I'm trying to find a way to get the data I need without creating separate extracts for each year and having to cross-reference with my list each time. Any help would be greatly appreciated. Thanks.

Do you have the same question? Follow this Question

Staff Answer




My first recommendation would be to select your samples before selecting variables. This will prevent the disappearance of variables once samples are selected. However, the issue you saw, where variables were dropped if they were not present in every selected sample should not be the case. I have just created a few extracts to test this, selecting a number of variables with different availability and then selecting samples. For me, the only variables that are dropped when selecting samples are the ones that were not present in any of the samples I selected, which makes me think that the problem you encountered stemmed from something else. While it hasn't been an issue more recently, previously using the back arrow on your web browser can confuse the data selection system, which may lead to situations such as random deselection of variables.

You should be able to generate the dataset you are attempting, avoiding the need to download a separate file for each year. The only way to save a list of variables is, as you propose, to create an extract with all of the desired variables. Revising that extract and then using the Change tools for samples or variables will automatically select that same list of variables.

I am sorry you experienced such a frustrating error. I hope this helps.


Jun 25, 2014 - 09:55 AM

Report it


Wrong HH weight Tanzania 2012 census?
Assign Individuals to PUMAS, append dataset
Regarding: Does CLASSWKR in ASEC before 1988 include incorporated and non-inc...
Does using the INCLUGH variable over time requires correcting for the VERIFY=...
Does CLASSWKR in ASEC regarding self-employment (code 10) include incorporate...
I am trying to cross age, urban, and race. Which weight am I supposed to use?
Login   |   Register

Recently Active Members

View More »

Share |