Interaction terms

[Julianna Grato]

I am looking at how returns to education vary based on major selection. I am also interested in looking at how returns to education vary by ability within the same major. I am using SAT scores as a proxy for ability. Could I interact SAT score ranges (i.e. 500-600 math SAT scores) with college major selection estimate the importance of math ability is within the same major?

                                                       [IMG]file:///C:\Users\Owner\AppData\Local\Temp\msohtmlclip1\01\clip_image002.png[/IMG]

Your help is greatly appreciated.

[Carlo Lazzaro]

Julianna:
your code is unreadable.
Please post what you typed and what Stata gave you back via CODE delimiters without screenshots and/or images. Thanks.

[Julianna Grato]

That was actually my model I typed in word. I haven't gotten to the STATA yet.
My simple model would be:
income= B1 major selection+ B2 student characteristics (i.e. married, age ect.)+ B3 GPA+ B4 SAT+ B5 SAT*Major selection
Before I start working with my data I wanted to know if STATA's output would show how much ability within the same major explains income.. My code in STATA will be like this but I haven't put it in STATA yet:

generate interaction= major selection* SAT

[Phil Bromiley]

First, you need to read the users guide and the manual coverage on regress. You probably need to take a course that covers regression. It sure sounds like you don't understand the basics. As for the interaction, it is easiest to use factor variable notation for interactions rather than calculating them beforehand. Then, margins will work nicely with the interaction. Look these things up in the pdf documentation subject index.

[Carlo Lazzaro]

Julianna:
what Stata shows is obviously conditional on the code we enter.
That said, you may want to consider -fvvarlist- notation instead of creating categorical variables and/or interaction yourself:

regress income b1 b2 b3 b4 i.Major_selection##c.SAT

Ability is exactly what can bias this kind of model due toendogeneity, as individual ability is correlated to income (the regressand) because, on average, being smarter in negotiating can give back higher income coeteris paribus (cp) and smarter people can achieve, on average, better SAT, cp.

[Julianna Grato]

Thank you Carlo. I appreciate it.