Individual fixed effects

[Adrienne Wold]

I am estimating the results of a survey experiment where each respondent received two different questions, is it appropriate to use individual fixed effects when there are just two observations per individual?

[Clyde Schechter]

Yes, it is. You have paired observations nested within respondents. It is perfectly fine to use the respondent id as the fixed effect and regress the response against the question. However, you might consider a simpler alternative: do a paired t-test. The two procedures are completely equivalent, and give the same exact results.