Need advice on appropriate method to use..

[Isla Herrera]

I have survey data of 20+ countries with 9 waves done in 18 years.. however, the respondents are different every year..
Since my dependent variable takes the value of 1 or 0, I am doing probit and logit on stata..
However, I would like to ask for some advice on how to further improve this since (1) it is not really a "panel" data, only a stacked cross-section data and (2) I am suspecting endogeneity problem with my main independent variable..
Hoping for a reply..

[Carlo Lazzaro]

Isla:
if respondents differ every year, you do not have a panel.
If what you have is a survey, take a look at -svy- entry in Stata .pdf manual.

[Isla Herrera]

Hello, Carlo, thanks for your response.. since I only have a pooled cross-section.. would you advice to break down the dataset and do the regression separately for each year?
Or would it still be possible to do 1 regression for the entire dataset?

Also, since I am doing a binary response regression model, how could I deal with the endogenous variable?

[Carlo Lazzaro]

Isla:
1) perform your regression on the entire dataset including -i.year- in the right-hand side of your regression equation;
2) see https://www.stata.com/meeting/sandie.../sd12_baum.pdf

ETA: why bumping your query (https://www.statalist.org/forums/for...ethod-to-use)?