Instructions for Assignment:
This is a practical assignment involving regression analysis.
You should estimate the models, print the regression outputs, tables and graphs and prepare notes.
DATA:
The file contains data on wages (wage), IQ scores (IQ) and experience (exper) of 500 individuals.
It also includes a dummy variable (male) which takes the value of 1 when the individual is male and the value of 0 when the individual is female as follows:
DATA ANALYSIS:
a) Study the data and make sure you understand how each variable is measured. How should you go about trying to explain the variation of wages across individuals?
b) Produce a table of “Descriptive Statistics,” containing means, variances, standard deviation, max and min values for all variables.
c) Produce a “Correlation” matrix. Any of the correlations is particularly high?
d) Produce graphs (scatter plots) of the wage against IQ and also wage against experience (exper).
e) Is there anything particularly different about male individuals?
BIVARIATE MODELS:
a) Using these data, estimate all possible bivariate regression models for which wage is the dependent variable, using only one of the continuous explanatory variables IQ and experience (include constant term in each case).
b) Which model is better? Why?
c) Generate 99% and 95% confidence intervals for all coefficients.
d) Analyze the results from the estimation by conducting (step by step) hypothesis tests for the significance of all the coefficients (including the intercept).
e) Transform the variables and estimate the lin-log, log-lin and log-log models.
MULTIVARIATE MODELS 1:
a) Regress wage on experience and IQ including a constant term in your equation.
b) Generate 95% confidence intervals for all coefficients.
c) Analyze the results from the estimation by conducting (step by step) hypothesis tests for the significance of all the estimated coefficients (including the intercept).
d) Transform the variables and estimate the lin-log, log-lin and log-log models.
MULTIVARIATE MODELS 2:
a) Regress wage on experience, IQ and male dummy including a constant term in your equation.
b) Generate 95% confidence intervals for all coefficients.
c) Analyze the results from the estimation by conducting (step by step) hypothesis tests for the significance of all the coefficients.
d) Create interactions terms involving dummy variable and use these interaction terms ((exper*male) and (IQ*male)) to re-estimate all models.