Custom Search

Wednesday, March 10, 2010

A Reality Check

From The Marston Chronicles

In our last report, we explained how we use mathematics to project who will control the House of Representatives and by how much. We also explained that we use mathematics to project which House races are in play and will likely flip to the opposite party. The major problem with this approach is that it is applied across the board to all races and does not account for local conditions that could affect the race. For that reason, we need to conduct a reality check to see if what we project seems to be actually happening. That means we have to look for poll results in various House races. Those are relatively rare but there are always some done. Here is what the polls are telling us:

RACE POLL HCV CQ POL COOK ROTHENBERG MARSTON
Alabama 2 D +24 No Toss-up Leans D R toss-up Likely R
California 44 R +14 No Likely R Likely R R favored Safe R
Colorado 3 D +2 Yes Likely D Leans D D favored Leans D
Delaware AL D +10 Open Likely D Leans D D favored Leans D
Indiana 9 R +8 Yes Leans D D toss-up Leans D Leans R
Maryland 1 R +13 No Toss-up D toss-up R toss-up Likely R
Massachusetts 10 R +3 Yes Likely D D toss-up D favored R toss-up
Michigan 7 R +10 Yes Toss-up D toss-up Toss-up Likely R
Minnesota 6 R +16 No Leans R Likely R R favored Safe R
Nevada 3 Tied Yes Leans D D toss-up Leans D Leans R
New Hampshire 1 R +10 Yes Leans D D toss-up Toss-up Leans R
New Hampshire 2 R +7 Open Toss-up D toss-up Toss-up Leans R
New Mexico 1 D +9 Yes Likely D Likely D Safe D R toss-up
New Mexico 2 R +2 No Toss-up D toss-up R toss-up R toss-up
New Mexico 3 D +6 Yes Safe D Safe D Safe D Likely D
New York 1 D +2 Yes Likely D Leans D Leans D R toss-up
North Carolina 8 D +16 No Leans D Likely D D favored R toss-up
North Dakota AL R +6 Yes Leans D Leans D D favored D toss-up
Ohio 1 R +17 Yes Toss-up D toss-up R toss-up Leans R
Oklahoma 2 D +16 No Safe D Safe D Safe D Safe D
South Carolina 5 D +7 No Leans D Leans D Leans D Leans D
South Dakota AL D +7 No Likely D Likely D D favored Safe D
Virginia 5 Tied Yes Toss-up D toss-up R toss-up Likely R

Now we look to see if the letter designating the party which should win the race in the Marston column matches the letter of the party that is ahead in the poll results column. They do match in 16 races but not in the other 7 races. Now we look to see what is going on in those 7 races. In two cases the poll showed a tie which means we will need a newer poll to see which way the political winds are blowing in that race. In four cases we project the race for the Republican when the polls show the Democrat ahead. In one case (ND-AL), the poll shows the Republican ahead when mathematics calls the race for the incumbent Democrat. Those races require further examination.

When we first looked at the results, we thought that the vote on health care might have something to do with these results so we added a column showing that vote. It would seem that a no vote helped in AL-2 and NC-8 to insulate the incumbent Democrat in a red state. Likewise a yes vote in ND-AL put the incumbent Democrat in a red state in trouble. In NY-1, the Democrat has a small lead so perhaps we should move him from R toss-up to D toss-up. In NM-1 we have the Democrat well ahead so we need to move him from R toss-up to leans D.

One other thing that sticks out is the huge lead shown for Bobby Bright in Alabama 2 when we have him as the most vulnerable Democrat in the country. He got 50.23% of the vote in a terrible Republican year while McCain-Palin carried his district with 63% of the votes. In what appears to be a banner year for Republicans, he is ahead by 24 points against his strongest Republican opponent? That does not make any sense. We have never heard of the polling company that conducted the poll but something is clearly off. We do not know what that polling company was smoking but it would appear that it was not legal.

We will make adjustments for a health care vote in cases were this vote made a difference in either direction and fix the two races where local conditions indicate that necessity. We suspect it will not change the final result in the House margin but obviously it will affect which seats make up that margin. We will know more once the changes are made. Now you see how an analyst proceeds. You can evaluate races on a subjective basis race by race and make adjustments as the situation changes like the other analysts do or you can use a mathematical model and adjust your equations or chose the results in a previous year when local conditions warrant it. We prefer the latter approach because it is less subjective.