Cross sectional_Panel

[ajay pasi]

Hi statalist community

For cross-section data analysis, I have learned two ways of running regression ( that incorporate fixed effects)

Way 1: logit DepVar IndepVar1 IndepVar2 i.fixed_effect1 i.fixed_effect2, vce (robust)----this gives me Psudo R^2

Way 2: Two steps xtset fixed_effect1 and then xtlogit DepVar IndepVar1 IndepVar2 i.fixed_effect2, fe vce (bootstrap)--- it does not give Psudo R^2


Comparing Way 1 and Way 2, I can not use vce (robust) in Way 2. Moreover, Way 2 does not give Psudo R^2.

Is there any advantage in going for Way 2 instead of going for Way 1? Or is there any advantage in going for Way 1 instead of Way 2?

[Carlo Lazzaro]

Ajay:
do you mean:
1) cross-sectional studies;
2) repeated cross-sectional-studies;
or
3) panel datasets?

[ajay pasi]

Dear Carlo, IHDS is a panel survey led by Sonalde Desai and others at NCAER/UofMd which collected on a wide range of topics from ~42,000 households across India. The survey includes two rounds of data collection. However, I used data for only one wave. There is no time dimension in the data.

[ajay pasi]

Hi Carlo,
Its not panel dataset as has been suggested by you "https://www.statalist.org/forums/forum/general-stata-discussion/general/1691668-fixed-effect_panel" in my previous query about the data. I believe it is repeated cross-sectional data. But I am using data for only one wave.

[Carlo Lazzaro]

Ajay:
go Way 1, then:

logit DepVar IndepVar1 IndepVar2 i.fixed_effect1 i.fixed_effect2, vce (robust)

Some further comments:
1) more details about the fixed effect you're interested in might be helpful;
2) an -if- clause woud help in selecting the wave of data you're interested in;
3) be sure that the option for non-default standard errors is the right one for the issue that you want them to deal with (heteroskedasticity; autocorrelation of the epsilon).

[ajay pasi]

Thanks, Carlo. Appreciate the help.