Pca

[Serena Gallo]

Hi dears,
I would like to have help with the PCA.
I have a dataset with ten variables, most of all from 2015 to 2019 and I want to construct an index.
However, I have a variable only for 2015, 2017, and 2019 because I have gathered it from a survey.
Is it a problem for the implementation of the PCA? How can I fix it?

[Nick Cox]

I guess what you mean is that you have all years from 2015 to 2019 for some variables but only some years for one variable.

It's a problem for anything if you have some missing values. If I understand this correctly, you can use the data you have or interpolate the data you don't have and then use them all.

I've never felt comfortable with PCA for panel data.

[Serena Gallo]

Yes, you have understood the issue.
How Can I interpolate the data I don't have?
Can you suggest me anything?

[Nick Cox]

Did you try

search interpolate

?

I'd say watch out. If 2/5 of your data on one variable are interpolated, many reviewers are likely to be queasy about that. Depending on your status, consult your teachers or colleagues for reactions.