Question about regression analysis

[nynysf]

[delete]

[startz]

Hi, guys, I'm new to econometrics and statistics, and would like someone to please answer a simple question about regression analysis for my research project. Please excuse any lack of proper statistical/econometric terminology, I'm learning :)

Here is the set up:
Two groups (pre-economic crisis and post-economic crisis) of data about family expenditures and educational attainment. I'm asking how educational attainment was affected by the economic crisis.

How useful would it be for me to run a t-test?

Also, what about running a simple regression? I ran a regression y = bx + e where y = xeduc and x = total (expenditure)
The level of significance is 99.99% and found that educational attainment went down after the crisis. Is this even an interesting result?

Thank you very much for all your help. Even if you just respond by saying this is wrong, it will help me. Thank you again.

The usual approach would be to define a "dummy variable," call it D, which equals zero before the crisis and 1 afterwards. You might then run a regression something like

y = a0 + b0*x + a1*D +b1*d*x + e and use a t-test to test whether b1<0.

[nynysf]

Thank you so much! I already wrote this in the last post, but the data set I am using is the Indonesian Family Life Survey from RAND. It's so complicated... and not everything can be put into real values, it's impossible to find a way to put educational expenditures into real values. I already consulted a Ph.D. in statistics and he spend a long time trying and said he couldn't do it.... :( I guess if inflation is low it won't matter?
Thank you so much. This was very helpful.

[nynysf]

Hi, I ran the t-test you suggested.
From the points I get so far, the difference is definitely significant as in the t-test for the dummy not equal to 0 is significant. t-test for the dummy > 0 is also completely significant. T-test for dummy < 0 is NOT significant.

What does this mean?

Thank you.