PUAD630 | Regression Memo | Complete Solution

Instructions for Regression Memo

For this memo, you are to run a lot of regression analyses (maybe 10 to 12) with different combinations of variables from the spreadsheet provided on Sakai. Begin with using the "Female" variable only to predict "Salary," but then move on to using two explanatory variables (multiple regression) and then 5, 6, 7, or 8 variables. Notice how the size of the coefficient for the Female variable changes as you add additional explanatory variables. Of the dozen or so analyses that you run, pick around 5 to report (be sure to include some analyses that make use of a lot of explanatory variables).

Now, your task is to "make sense" of the relationships between the explanatory variables and the outcome variable, and to do so by reinterpreting the variables to fit some context that you are familiar with, such as various factors that are related to "Sales" for a company, or perhaps "Profit" or "Expenses."

You will get the most credit if you use, and correctly explain, analyses with many explanatory variables (8 should be good), including dummy variables for things like the 5 levels of "Education." For full credit, you also need to make this memo look professional and make use of visual aids, such as Graphs and Tables, that would help busy administrators understand what you are reporting.

To see an example of a memo that follows the above instructions (and was rewarded with a good grade), please see the Sample Memo that is here also on Sakai.

Finally, you can earn up to 20 points of extra credit (depending on how well it's done) by including also interaction variables (recall, this involves multiplying two columns together) and explaining, just like it's done in our text (see Figure 10.30), how the influence of one variable is influenced by the level of the other variable (we will discuss in class how the influence of Education on Salary was much higher for men than for women--interesting).