@Irishk,
Interesting article, Irishk, by the lad from Dartmouth. I did find his logic a bit hard to follow. That was probably do a combination of his smarts and my stupidity about regression analysis in statistics.
The gist of a "generic" poll is to collect responses to this question: "If there was an election today for Congress between a Republican and a Democrat, which would you vote for?"
The results, he says, favor the Republicans by about 2% if you exclude polls that (like our friend, Scott Rasmussen) are perceived to have a bias.
In fact, Rasmussen, as of a week ago, had the Republicans at 44% vs Democrats at 35%. Different polls have different metrics, so I am not sure we can compare Rasmussen to the composite poll. Interestingly, the % of folks who said "neither" in the Rasmussen poll is rising.
Election day is 8 months away and this survey is based on polling data from a relatively small sample and completely ignores the fact that elections are held in districts where the winner takes the seat.
Nonetheless the student's prediction is that the stats point to the Republicans taking back the House. I think that, if things progress as they are, he could be right.
Polling on individual House races - 435 of them - is pretty sparse. I should find some updated Senate stuff this week.