# Dummy Variables

Only 1 dummy variable cud be taken from the 3. As including more than 1 will cause MULTICOLLINEARITY! Coz they are mutually INDEPENDENT! Perfect Negative correlation! Inflation less than 3% Inflation from 3-6% Inflation Higher than 6% Any comments?

Nope, you need n-1 # of dummy variables, where n = # of qualitative statistics you are trying to model.

You have to say: 1 when for the case you care about and 0 when it is not the case.

You can use multiple dummy variables. When you do its number of variables - 1. So if you want 10 qualitative variables, you use 9 dummys

Oh SHIT! Now i remember! I had solved a question on Quarter 1, Quarter 2, Quarter 3, Quarter 4 They were independent, but still correctly specified! DAMN got it wrong!

wasnt the question asking about how many choices u can have for each dummy variable, which is two–either on or off

No it was just a really round-a-bout way of asking you how many dummy variables are needed. 2 of the choices were 3, the other one was 2.

so which one was right–i remember A and C said 3 variables and B said 2 variables

should be 2 … the choice was B. i spent some 5 minutes on it and eventualy figured out at the last moment. the fist option was no asnwer where it said use 1 for expansioanry, use 2 for rstrictive. 2nd one said use 1 for expansionary otherwise neutral and 1 for constractionary otherwise neutral. So basically all you have to do was to find whether it was contractionary or expansionary, the intercept should represent neutrality. 3rd one used three dummy… which is a not accurate. The right choice is B (2 dummy)

dont call me a dummy fool. you dont even know me.

there were 3 independent variables…n-1, so the answer was 2, and I think B was the correct option

I dont think those examples are in context. Dummy variables are clearly stated as being “on” or “off”. The 3 states you highlighted above arent independent variables. The independent variables were different I recall. That would mean you need a dummy for each of those states of the world, wich I suppose would mean C. But I do now recall the n-1 thing, so this might change it.

n-1, huh, what…i think this is being overthough. dummy variables are on or off which means theyre 0 or 1. A) and C) had 3 possible states of the economy and B) had two so asnwer was B.

I’m not too sure about this one, but I definitely answered C because of the reason that dummy variables are either on or off, and because we were given three states for the independents, that would lead me to include 3 dummy variables in the regression. I answered this purely out of my memory of taking econometrics 5 years ago, so I may be wrong, but the whole n-1 argument, does not seem to make all that sense to me given the context of the question. I could be wrong on this one though.

As far as I remember with choice B (2 var) we would not have an influence of policy - none of the options (exp., neutr. o r cons) would be present in regression equation - that’s why I chose C. Was it a mistake?

I did Dummy Variables thoroughly but 3 month before the exams. Since i was dealing with Heteroskedasticity/Multicollinearity in the final week, i got skewed towards Multicollineartity! DAMN IT! Did it wrong even thou i knew the concept!

just confirmed with the company statistician… if you have three values for a variable and you use dummy you use three-1 dummyvariables… FOr example for months in a year… each month has a 1 or 0 value, You would actually choose 11 variables…

yes because you can deduce the nth term using n-1 terms… one of the terms need to be true, if all terms were false, then ya u know what I mean =)

king_kong Wrote: ------------------------------------------------------- > Nope, you need n-1 # of dummy variables, where n = > # of qualitative statistics you are trying to > model. i spent a loooooooong time reading the answer choices back and forth before realizing that i needed to use n-1.

right tsetse/financial_novice …