# Regression Analysis

Perform a regression with one dependent and two independent variables.  For example:  Y = b0 + b1*X1 + b2*X2

Or more specifically from the lecture:  Temp = b0 + b1*CO2 + b2*Year

I would like 50observations on each variable.  You have to find the data yourself.

Mostly, I would like you to explain what and why.  For example from the class regression in the slides you might say:

I was curious whether global temperatures rise with global CO2 and time.  If they rise with CO2, then it global warming is probably human caused, if it rises with time then maybe it is just whether fluctuations.  To test this I ran the following regression: Temp = b0 + b1*CO2 + b2*Year.  I found that b1, the coefficient on CO2 was XXX, and more importantly that this was significantly different than zero, so it seems like a real result.  For every part per million of CO2 that goes into the atmosphere global temperature rises by XXX.   So on…

I suspect that with tables and explanation this will be around 4 pages.  Use as much space as you need though.

Remember that a low p-value indicates a low probability that the coefficient in question came from a distribution with a mean of zero and hence would lead us to reject the null hypothesis.  Low p-value = Significance, High p-value = non-significant.

For the report I would like you to collect some data and perform a regression analysis to tell me what the data suggests.  You can collect any data that you want to examine any question you want but there are some rules.

1) You must explain what question you are trying to address with your regression.2) You must explain why you picked the particular variables that you picked.3) You must have at least three variables:  one dependent and at least two independent variables.4) You must have at least 60 observations on each variable.5) You must report the full regression equation (that is both independent variables in a single equation)along with the standard errors and p-values of the estimates.6) You must explain what the estimates mean in plain English.7) You must provide an F-test of the full regression and explain what it means.8) You must provide the R2 of the regression and explain what it means.9) You must explain what actions the regression suggests you should take.10) You must explain what problems your regression might have.This should take between 4 and 6 double spaced pages including tables.