Reference no: EM132368778
Instructions:
A problem with the OLS model. You need to code non-linear variables like education, HHclass etc. with separate categories. They are not linear variables. Given the time pressure, Not sure how much more you can get done before you need to submit, so you think it is fine to estimate models separately for boys and girls, by wave, but you need a coherent model that makes sense.
So, use categories for all non-linear variables. You would recommend estimating models with the following set of independent variables:
Partner, Gender of respondent, Income, Social Class (full set of indicators for each category), Education (Full set of indicators for each category), Difficulty making ends meet (full set of indicators for each category). You can always run t tests to see if the coefficients of two similar indicators are the same, and if so combine them.
So, for instance if having moderate difficulty and have severe difficulty have statistically identical coefficients, you can combine them into one group. It is not clear to me why you estimated three different models with different sets of control variables -- You would estimate one core set given that you are short on time, but make sure it makes sense.
Attachment:- Data Dictionary.rar