How to interpret command xtcsd pesaran and friedman ?

[Clemence Sauve]

Dear every stata users,

I want to know if my cross-sectional units are independent. N=24 and T=3.
My research is about the evolution of the employment rate (TxEmploi) performance between the 24 provinces of Argentina, between 2019 and 2021.

My xtreg regression only use the independent variable because I just want to know the trend of dependence between the provinces for this only specific variable.

I did :
- xtreg TxEmploi
- xtcsd, pesaran abs show
- xtcsd, friedman abs show

The results are :

Can you help me interpret the results and tell me if this is ok to do the xtreg and the xtcsd tests like that ?

Also, I think it is strange that the Friedman and Pesaran Pr results are different, don't you think ?

Thank you very much,

Regards,

Clémence Sauvé

[Joseph Coveney]

I don't know about xtcsd, but your xtreg model doesn't include the time variable (year) as a predictor. Wouldn't omitting that variable imply that you consider observations of employment rate within each province to be exchangeable between the years 2019 and 2021? Didn't something happen during this interval?

In lieu of xtreg, which is for large sample sizes, with only twenty-four provinces, you might want to look into something like icc and seeing whether the intraclass correlation coefficient is larger than zero. You can do something similar with mixed,which has adjustments for small-samples.