Dear Statalist users,

I hope you are well.

I am a PhD candidate at RMIT University, Melbourne. I am using Australian household dataset (HILDA). I am using seven waves (wave 11 to 17) of the HILDA survey. My independent variables are financial wellbeing and risk-taking attitude. My dependent variable is debt-taking behavior that is a binary response variable. I filtered my dataset to make it balanced. I have few questions:

(1) In the original dataset, 25% respondents were those who take on debt. However, with the filtered dataset, this percentage reduced to 10%. My question is - can I still generalize the results? I am afraid someone can simply say that your filtered dataset is not a representative of original dataset, so your results are not acceptable.
I am using Stata. While filtering, I included only those respondents who responded to my variables of interest across the seven years. That means I want to see the life-cycle effect. In original dataset, the respondents were aged 15 to 99. However, when I filtered the dataset, the respondents are aged 20 to 88 in the first wave, while in the seventh wave, they are 26 to 94 because I tracked same respondents across the seven waves.

(2) As my dependent variable is a binary response variable, I am using panel binary logistic regression -xtlogit-. I opt to use random-effects model. Except of coefficients and odds ratio, what else can I generate in Stata? I heard about marginal effects. Could someone please guide on this?

(3) I also want to see the effect of independent variable of t-1 time on the dependent variable of t time. I know I can use lag of independent variable. My question is - Can I include both the current and lagged independent variables in one model?

(4) I also suspect that if a person takes on debt in one time, (s)he is most likely to take on debt in next period. Can I include lagged dependent variable in the model? I tried in Stata, but the value of 'rho' became zero when I did this. That means there is an issue in the model when I include lagged dependent variable. Could someone please help me on this?

I am sorry if anything confuses you. I can explain more if this happens.

Thanks for the help.

Regards,
Muhammad Tahir