Using STATA "mixed" command with autoregressive correlation of residuals for comparing time series of medical Google queries

[Gianfranco Di Gennaro]

Dear all,

thanks in advance.

I need an information.

I have 4 different time series (number of daily google clicks in the last 3 years for 4 different medical keywords).

I want to compare the 4 time series, i.e. estimate whether the variation in these 3 years has been different between the keywords.

I wonder if it makes sense to use a mixed model with autoregressive error correlation.

Something like:

mixed numberofclicks.. indepvars. . . || i.keyword:, residuals(ar 1, t(time))

Thank you very much!