Anyone studied regression?

Mad old tory · 19 May 2009 at 19:25

You have my sympathies. I did an economics degree but luckily was able to drop econometrics after 2nd year as I frankly didn't have a scooby what was going on most of the time.

killari · 19 May 2009 at 19:53

Something else that is pretty important is residual analysis. If you are given the residuals you can run a plethora of tests to see if the model is a good fit.

killari · 19 May 2009 at 19:59

monkentage said:
you said n-2 degrees of freedom. if you wanted to know what df for the regression as a whole i think it would be n - k - 1 where k = number of parameters

Shouldn't it be n - 1 - k where k is the number of ESTIMATED parameters?

Barks · 19 May 2009 at 20:28

killari said:
Shouldn't it be n - 1 - k where k is the number of ESTIMATED parameters?

It's confusing because one textbook uses one method and one another.

In general, N - k (N - sample size, k = number of variables)

k is always the number of unrestricted variables, when testing for joint significance a third variable (q - number of restricted variables) is introduced into an alternate F equation.

muon · 19 May 2009 at 20:38

What module are you doing at which university? I'm curious to see how it compares to the LSE module.

Barks · 19 May 2009 at 20:42

ghost101 said:
What module are you doing at which university? I'm curious to see how it compares to the LSE module.

Statistical Methods for Economics 1st year.

It doesn't go any further than regression, but it's all new to me having only done S1 A Level maths.

The LSE slides are slightly more advanced but they're bloody good at explaining things. For example, when you posted the link to that slide, the stuff after page 25 is not covered in our module. The question I posted does not provide an alternate (restricted) value of R2 for comparison.

muon · 19 May 2009 at 20:51

Oh okay. In which case I have no idea what they want you to do. The mandatory LSE second year metrics course is relatively easy compared to some other top places. However they have an optional (highly recommended if doing a masters) intermediate course

http://econ.lse.ac.uk/courses/ec221/

which you take in the second year. And its harder than anything offered by a university in the UK in the second year. My maths isnt as good as some so I did the easier one, but I still did relatively poorly

. I wish I had put in the effort, because I wouldnt have to work my socks off this year.

Barks · 19 May 2009 at 20:55

We definitely have an econometrics module next year and from what I've heard it's a nasty piece of work. I should really have paid attention in lectures though it has taken me two hole days to master the module and I'm still probably not going to do that well in it! Can't thank you enough for the LSE slides though, ours are useless by comparison.

Forgot to say, Newcastle uni

muon · 19 May 2009 at 20:59

scott212 said:
We definitely have an econometrics module next year and from what I've heard it's a nasty piece of work. I should really have paid attention in lectures though it has taken me two hole days to master the module and I'm still probably not going to do that well in it! Can't thank you enough for the LSE slides though, ours are useless by comparison.

Forgot to say, Newcastle uni

You're welcome. Although the credit should go to Christpher Dougherty. Brilliant lecturer.

Barks · 19 May 2009 at 21:00

ghost101 said:
You're welcome. Although the credit should go to Christpher Dougherty. Brilliant lecturer.

I wouldn't expect anything less at somewhere as prestigious as LSE though

Barks · 20 May 2009 at 15:52

Guys sat the exam this afternoon and I was able to answer everything on the paper relatively easily. Thanks again for the help!