[Gigi Chen]

Simon:
1) if you actually have panel data, (pooled) OLS should not be your first choice;
2) if you have repeated time values, you can simply -xtset- your panel dataset with your -panelid- only:

xtset panelid

However, this fix comes at a cost of making time-series operators unfeasible;
3) if your regressand is continuos, you can go -xtreg-.

Hello, I am having a similar problem with repeated time values. I have a pooled panel data, that is the interviewees are selected at random each year, but are selected from the same region. I want to look at regional effects, but since the combination of region and year are not unique I failed to "xtset region year, yearly", so I was wondering if I can simply "xtset region", but then how would I control for the time variable?

Thank you, Gigi

[Carlo Lazzaro]

Gigi:
welcome to this forum.
You can -xtset- your dataset with the -panelid- only and add -i.year- as a predictor in the right-hand side of your regression equation.

[Gigi Chen]

Hi, I tried doing that but my independent variable became negative, but when I ran the regression earlier without the i.year (v007=year in my data) it was fine. What I expected is that the probability of stunting is higher in the "mita" districts.

variabels explained:
"stunting" is a binary variable =1 if the child is stunted, =0 otherwise. is my dependent variable
"mita" is the binary variable =1 if the district historically belonged to an extractive institution called 'mita' =0 other wise.
"b4" is the gender of the child
"v218" is the number of living children in the household
"v007" is the variable for year of interview


I want to run a logit regression on the probability of stunting and 'mita' districts with fixed effect on 'region' and 'gender' but I am really struggling with the commands. Please suggest what I can do. Thank you ver much!

Sorry for the screen shot I don't know how to import Stata results into the forum.

Gigi

without i.v007



with i.v07 added on

[Carlo Lazzaro]

Gigi:
1) please use CODE delimiters to share what you typed and what Stata gave you back (just clincing the #.shaped toggle you will get . Just put the stuff in between);
2) -mita- is never different from zero, no matter your -xtlogit- code.