In general I agree with you that this is unrepresentative for a nation as a whole - see my other comment for that.

However one advantage of a sample like this is that significant independent variables like the ones you mentioned are widely held constant, which means that the evaluated variables will have a stronger power. So it is informative relative to others.

It would be best to take this study into context with other studies which compare educational achievement and poverty level.