3. Data Analysis

This part of your reports interprets the results of two regressions or two sets of crosstabs or summary tables that test the hypothesis that you set forth in Homework Project #1. It does not matter whether that hypothesis turns out to be correct or not; what matters is whether you test it correctly and how clearly you can explain your findings. You should conduct the analysis in Stata, and your report should include the actual Stata output from your regressions or tables, and touch on all of the following points, one by one:

A. Explain to the reader what your hypothesis is and what the reasoning behind it is. Tell us what your independent variables are and what your dependent variable is. State the type of the variable, the dataset it comes from, and describe any recoding you conducted. If any of the variables are ordinal, will you treat them as categorical (nominal) or quantitative (interval or ratio)?

B. Explain the observed relationship in the data. If you are conducting a cross tabular analysis, review the graph or table that you used in Homework #3, discuss whether or not your key independent variable seems to have a relationship with your dependent variable, and what is the nature or pattern of this relationship. If you are conducting a regression analysis, interpret the coefficients or slopes and the r2 statistic. Conduct appropriate analysis.

C. Test whether the relationship observed is statistically significant. The best way to do this depends on the type of variables that you are using.

1. Explain what type of test you are using and why.

2. State the null and alternative hypotheses.

3. Carry out the hypothesis test of your key IV. What is the significance level?

4. State a conclusion - do you reject or fail to reject the null.

D. Check for a confounding variable or an intervening variable. Is there a significant relationship between your Independent variable and dependent variable, once you control for a third variable? To do this, you will repeat steps B, C2, C3, and C4, controlling for a third variable.

If you are conducting a cross tabular analysis, you will make a two-way table, then examine results with a three-way table, controlling for your intervening or confounding variable. Check to see if the chi-square statistics are still significant in the three-way table, or if the relationship between your IV and DV looks different when controlling for another variable.

If you are conducting a regression analysis, you will run two regressions on your DV: one with only the main IV, and one with both your main IV and your confounding or intervening variable. Check to see if your IV has the same relationship with the DV in both models, and if it is still significant once you control for a third variable.