Measure outcome at three time points

[Sara Hansen]

Hi,

I have a dataset with outcome = heart rate, measured at baseline and time 1 and 2. Further, I have some potential confounders (age, sex, country).
I would like to test if there is a difference in heart rate from baseline to time 1 and 2.

Which analysis would be best (and simplest)? Perhaps a linear regression? Or a Linear mixed-effects model?

Thanks for your time,

[Carlo Lazzaro]

Sara:
I would go -xtreg,fe- as you have 2 (or possibly 3, if you include baseline heart rate, as I surmise from the title of your post).
The right-hand side of your regression equation can safely host age, sex, country.
As far as .age- is concerned, you may want to search for non-linearities via:

c.age##c.age

.

Usually time-invariant prdictors, such as -i-sex- and -i.country- cannot be estimated by the -fe- estimator, because demeaning a constant gives back zero.