wage rate model

[Tom Schmidt]

Hello everyone,

I am asking for support for the following tasks. I received a dataset that includes the variables yrs of education, yrs of experience, yrs of experience squared, amount of hours worked and several dummy variables (black, hispanic, union, married).
Firstly, I need to choose which variables should be included in my regression of logwage (wage rate per hour). Do I just think logical here? E.g. more school --> higher wage rate --> include school variable? Or should I just plot it?
Secondly, we need to investigate correlations between them. What do these correlations tell me in order to decide which variables to include?
Which ones could be good interaction terms and how do i interpret them?
Finally, we have to create different models and have to compare them with our basic model. Please give me hints of what could be changed in these models?

Thank you very much,
TS

[Carlo Lazzaro]

Tom:
I assume that this query is the prequel of your last one.
Some comments about it:
- skim through the literature in your research field and see how others rendered the data generating process when presented with the same research task;
- beware of endogeneity: if you do not have a proxy for individual ability, that implicit predictor lurks in the residuals, being at the same time correlated with your dependent variable (the wage rate: other things being equal, individual ability can well play a role in negotiating one's wage) and one of your independent variable (education: other things being equal, personal ability can well play a role in achieving outstanding marks and/or excellent educational levels).