Hello Statalisters,
I am creating an index with several made up of several indicators and some of these indicators have missing observations. If I aggregate 3 variables a, b, and c, even if a and b are totally complete, if c has any missing observations, the aggregate variable seems to completely ignore the data from a and b, and that observation year is only counted as missing. I am losing so much information because of this.
When I was looking up how to deal with missing observations, I saw the command collapse repeatedly mentioned, but I don't think it's relevant here, since I am not looking to only use summary statistics.
So far I am using averaging with arbitary weights. I even just added all the variables together to see if the problem was not the method but the numbers themselves and they were just balancing each other out, but even when I simply add variables with all positive observations or missing observations, the missing variable seems to overpower any other information.
How would you recommend aggregating variables/creating a composite variable when some observations are missing?
Related Posts with Aggregating several variables with some missing observations
Calculations of quarterly stock holding data by shareholderDear Statalists, I am dealing with an unbalanced panel data for calculating the share holdings by a…
How to edit or delete a "signature" that is auto-added to one's post?Many years ago I created a signature (e.g., name & a quotation) that was automatically added to …
Major change in coefficient when adding a squared variable in fixed effects model. Interpretation?When I use the xtreg command with fixed effects to look at the correlation between use of marijuana …
Categorical dependent variable, how to choose the right modelI would like to Regress the likelihood that collateral are pledged on firm variables, loan contract …
Interpreting Panel data coefficient estimates where the variable doesnt change across observationsHi, I am looking at the effect of different variables on the recycling rate in England. I have 311 …
Subscribe to:
Post Comments (Atom)
0 Response to Aggregating several variables with some missing observations
Post a Comment