Wednesday, October 06, 2010
What do the polls tell us about conservative chances?
On this blog during the 2008 presidential primary season, I mentioned that I thought polls tend to be inaccurate during a rapidly changing situation. Specifically, the polls over-estimated the move to Obama in New Hampshire, coming on the heals of his success in Iowa. In fact, the polls over-predicted his vote, and Hillary won New Hampshire. Of course, the move of the Democratic electorate was clearly towards Obama, and later polls got it correct. That move took place over several months.
I've been watching Real Clear Politics fairly closely since mid-July, daily since about September 15. On July 21 RCP had the House of Representatives predicted results as 202 Democrat, 201 Republican, and 32 tossups. On September 15 they had it 205 Republicans, 193 Democrats, 37 tossups. The Democrats were losing ground, Republicans gaining ground. This afternoon, RCP had a number of House races change from polls published this morning. The result: 210 Republicans, 188 Democrats, 37 tossups. Thus the Republicans would have to win only 8 of those tossups to gain control of the House.
As a conservative, I could get giddy from these numbers, assuming all those Republicans were likely to be conservative and all those Democrats more likely to be liberals. But I must temper my excitement with the thought that the polls could be skewing the extent of a moving trend. Clearly the trend is Republican in this election, but how much? If RCP projections are correct, and the tossups split down the middle, the House would end up 228 Republicans, 207 Democrats. But maybe the polls are over-predicting the movement, as they did in 2008. How are we to know?
It seems to me that the difference here is the time factor. Those incorrect 2008 polls were trying to determine a shift of sentiment measured in days. These 2010 polls are measuring a months-long trend. It’s a strong trend, but taking place somewhat slowly. This seems to me to validate the polls more so than the 2008 primary polls. Plus, the polls are for a number of separate races, not a single race. Thus the rapid-trend-over-predicting model doesn't seem to apply.
This makes me feel that the polls are valid and RCP is accurately stating the change in the House: a GOP majority with seats to spare. If in the House, then maybe in the Senate. If in the Senate, then maybe in governor races and State legislatures. Of course, that’s what I want. I hope I’m not just drinking the cool aid.
I've been watching Real Clear Politics fairly closely since mid-July, daily since about September 15. On July 21 RCP had the House of Representatives predicted results as 202 Democrat, 201 Republican, and 32 tossups. On September 15 they had it 205 Republicans, 193 Democrats, 37 tossups. The Democrats were losing ground, Republicans gaining ground. This afternoon, RCP had a number of House races change from polls published this morning. The result: 210 Republicans, 188 Democrats, 37 tossups. Thus the Republicans would have to win only 8 of those tossups to gain control of the House.
As a conservative, I could get giddy from these numbers, assuming all those Republicans were likely to be conservative and all those Democrats more likely to be liberals. But I must temper my excitement with the thought that the polls could be skewing the extent of a moving trend. Clearly the trend is Republican in this election, but how much? If RCP projections are correct, and the tossups split down the middle, the House would end up 228 Republicans, 207 Democrats. But maybe the polls are over-predicting the movement, as they did in 2008. How are we to know?
It seems to me that the difference here is the time factor. Those incorrect 2008 polls were trying to determine a shift of sentiment measured in days. These 2010 polls are measuring a months-long trend. It’s a strong trend, but taking place somewhat slowly. This seems to me to validate the polls more so than the 2008 primary polls. Plus, the polls are for a number of separate races, not a single race. Thus the rapid-trend-over-predicting model doesn't seem to apply.
This makes me feel that the polls are valid and RCP is accurately stating the change in the House: a GOP majority with seats to spare. If in the House, then maybe in the Senate. If in the Senate, then maybe in governor races and State legislatures. Of course, that’s what I want. I hope I’m not just drinking the cool aid.
Labels: 2010 Election
Subscribe to Posts [Atom]