 # correlation

If the correlation between age of an auto and money spent for repairs is +.90 A. 81% of the variation in the money spent for repairs is explained by the age of the auto B. 81% of money spent for repairs is unexplained by the age of the auto C. 90% of the money spent for repairs is explained by the age of the auto D. none of the above

I like C, but I’m doubting myself.

i like D

C, not sure

i don’t like the Q

A

a

Well, that goes without saying. Not sure where the 81% values are coming from. Thinking that it’s a trick to get people to say, oh, i’ll square 90% because it might be some stddev or variance issue… which it’s not.

I like D…think 90% is in there to confuse us because we have seen it in portfolio management with regard to returns on asset classes being 90% due to asset class selection rather than security selection. without any statistical significance on the correlation, i do not think you could make any reasonable conclusion…still not sure though

C. 90% of the money spent for repairs is explained by the age of the auto reminds me of a regression coefficient

But it’s correlation is +.90. So 90 percent of the movement in x is demonstrated in the movement in y.

can we get a ruling herre?

(correlation coefficient)^2 .9^2=.81, so R^2=.81, that tells us that 81% of variation in the dependent variable “money spent for repairs” is explained by independent variable "age of the auto

indeed, it/R^2 is called coefficient of determination.

R^2 is regression material is this actually in the 2008 curriculum?

saurya_s Wrote: ------------------------------------------------------- > indeed, it/R^2 is called coefficient of > determination. disclosure: I’m drilling a 2007 Qbank … never saw this before so I might have been off here :S

no no no no no

supersharpshooter Wrote: ------------------------------------------------------- > R^2 is regression material > > is this actually in the 2008 curriculum? exactly, this method of R^2 is used when finding the linear beta/coefficient after using OLS in a regression model (not the simple correlation coefficient)…but thanks for the refresher, in the case that we do see it on the test

just google it coefficient of determination http://www.stat.tamu.edu/stat30x/notes/node47.html I assume this to be L2 material

sorry mates, my mistake