Choose Y and X variables to model on the Household and the Environment Survey 2006. Using Ox software to write a program to do estimation, and then write a report based on the analysis. The report would be a Word. Doc., include the Ox. Program file and other supplementary files as appendix to the report.
a. Go through the variable lists and other documentation for the Household and the Environment Survey 2006 (HES survey)
b. Choose a Y variable to model, which can be either binary or (modeled as) continuous.
P.s:(You may want to transform the survey variables to construct Y . One way to make good choices is to search for and find research papers that analyze the same data or another data set containing similar variables.)
c. Choose a set of X variables to include on the right hand side of your model(s) in the X matrix. As with Y the choice of X is not best made without exploring the data.
P.s: (As you can see from articles, people often run a set of models on one Y variable, starting with a basic set of X's and then adding columns to the model to see how this affects the estimates of the other coefficients. You can do this, starting with a basic set of variables for which you think the path of causality to Y is clearest. Then add to X other variables that are more interesting but possibly not exogenous to Y or plausibly correlated with unobserved factors that cause Y. In this case you report your models in one table: each column is one estimated model (specification). Each row is a X variable, with blanks appearing when X was excluded.)
d. The survey asks sequences of related questions, such as ways household recycle...You can choose to model a set of related questions as different Y variables. If you model several related Y 's the X variables should be the same for each endogenous variable (do not show different specifications). And report all your estimates in one table: each column is a different model / Y variable, and each row is a different X variable.
e. Use the analysis in some way to illuminate what the data say about the topic.
(Examples include: test a hypothesis about the estimates that answers an interesting question, test for differences in coefficients across specifications, compare expected outcomes across groups, predict outcomes for some type of observation not in the data..)
f. After the process of choosing X's and Y 's, write a program to do the estimation in your paper.
(That is, don't show in this program all the experimenting you did to get to your analysis, but show every step that contributes directly to the report. The submitted program is to load the raw data set, clean it up by transforming and recoding variables, send the clean data to a model, estimate the probit or regression models, and produce any extra output needed for your use of the estimates.)