Free Republic
Browse · Search
News/Activism
Topics · Post Article

Skip to comments.

What to Expect in 2010: Predicting House Seats from Gallup’s Final Likely Voter Poll
Sabato's Crystal Ball ^ | 09/16/10 | Alan I. Abramowitz

Posted on 09/26/2010 9:11:17 PM PDT by freespirited

 

The Gallup Poll’s final likely voter poll has proven to be a highly accurate predictor of the results of midterm elections for the House of Representatives. In 13 midterm elections between 1950 and 2006 for which Gallup estimated the preferences of likely voters, the correlation between the Republican margin among likely voters in the final pre-election poll and the actual Republican margin in the national popular vote is an impressive .94. Moreover, the average difference between the predicted Republican vote margin and the actual Republican vote margin in these 13 elections is only 2 percentage points.

Given the accuracy of the Gallup’s final likely voter poll in predicting the outcome of the national popular vote for the House of Representatives, it is not surprising that this poll also does a very good job of predicting the division of seats between the parties. The correlation between the Republican margin among likely voters in the final pre-election poll and the number of House seats won by Republicans in the election is .91 which means that the final likely voter poll explains 82 percent of the variation in Republican House seats.

Despite the strong relationship between the results of Gallup’s final likely voter poll and the division of seats in the House, we can substantially improve the accuracy of our seat prediction by adding two additional variables to the regression equation: a dummy variable specifying the party of the incumbent president and the number of seats held by Republicans prior to the election. Table 1 displays the results of a regression analysis predicting the total number of Republican House seats based on these three independent variables.

Table 1. Results of Regression Analysis of Republican House Seats in Midterm Elections



Source: Gallup Poll and data compiled by author

All three variables have large and statistically significant effects. The evidence in this table shows that even after controlling for the results of the final likely voter poll, Republicans tend to win more House seats in midterms under Democratic presidents than in midterms under Republican presidents. In addition, the more seats Republicans hold before an election, the more seats they tend to end up with after the election. While more seats generally lead to larger seat losses or smaller seat gains, the end result is still generally more seats in the next Congress. With these two additional independent variables, the accuracy of the prediction is much greater—the regression equation now explains an astonishing 97 percent of the variation in Republican House seats.

Based on the results of the regression equation in Table 1, we can estimate the number of seats Republicans would be expected to win in the 2010 midterm election depending on the results of Gallup’s final likely voter poll. Those estimates are displayed in Table 2 for a range of possible final likely voter results.

Table 2. Predicted Republican Seats in 112th House



Source: Gallup Poll and data compiled by author

The results displayed in Table 2 indicate that in order to win a majority of seats in the House, Republicans will need a lead of 2-3 percentage points in Gallup’s final likely voter poll. A tie among likely voters would leave Democrats with a narrow majority. Gallup will not begin providing polling results for likely voters until October. However, their recent polls have shown an average Republican lead of about 5 percentage points among registered voters which, given the normal Republican turnout advantage in midterm elections and the unusually high level of enthusiasm of Republican voters in this year’s election, should result in a somewhat larger GOP lead among likely voters.

Even a 5 point Republican lead among likely voters would be expected to yield about 226 Republican seats, 8 more than the 218 needed for a majority. A 10 point Republican lead among likely voters, which would not be surprising based on a 5 point Republican lead among registered voters, would be expected to yield about 241 Republican seats. That would be an increase of 62 Republican seats. It would also be the largest number of Republicans elected to the House since 1946 when the GOP won 246 seats.




TOPICS: Politics/Elections
KEYWORDS: 2010; gallup; genericballot; midterms
Looking good.
1 posted on 09/26/2010 9:11:20 PM PDT by freespirited
[ Post Reply | Private Reply | View Replies]

To: InterceptPoint

Statistics junkie ping.


2 posted on 09/26/2010 9:13:44 PM PDT by freespirited (This tagline dedicated to the memory of John Armor, a/k/a Congressman Billybob.)
[ Post Reply | Private Reply | To 1 | View Replies]

To: freespirited

Yeah, but what about the dead-people voters? Bet they haven’t been polled.


3 posted on 09/26/2010 9:16:16 PM PDT by DefeatCorruption
[ Post Reply | Private Reply | To 1 | View Replies]

To: freespirited

The statisticians are wrong again... The results are going to be much more dramatic but what do they know...

This is the beginning of the year of the people - actually, the next 12 years (or as long as necessary) is the year of the people. It will take that long to get rid of the professional politicians...


4 posted on 09/26/2010 9:19:44 PM PDT by Deagle
[ Post Reply | Private Reply | To 1 | View Replies]

To: freespirited

Gallup doesn’t take into account the felon, dead, pet and double voters. Those “likely voter” populations have increased about 200% in the last 4 years.


5 posted on 09/26/2010 9:28:33 PM PDT by Recovering_Democrat
[ Post Reply | Private Reply | To 1 | View Replies]

To: freespirited

Looks good... But there is still far too much time between now and Election Day.


6 posted on 09/26/2010 9:31:35 PM PDT by Thunder90 (Fighting for truth and the American way... http://citizensfortruthandtheamericanway.blogspot.com/)
[ Post Reply | Private Reply | To 1 | View Replies]

To: freespirited

It’ll look much better if we’re really past the tipping point.


7 posted on 09/26/2010 9:34:37 PM PDT by Mobties (Let the markets work! Reduce the government footprint!)
[ Post Reply | Private Reply | To 1 | View Replies]

To: Recovering_Democrat

Ye who have little faith!

I dont like Gallup much, but I suspect its data will have predictive value again this year.


8 posted on 09/26/2010 9:38:00 PM PDT by freespirited (This tagline dedicated to the memory of John Armor, a/k/a Congressman Billybob.)
[ Post Reply | Private Reply | To 5 | View Replies]

To: freespirited

Interesting. Looks good in the House.


9 posted on 09/26/2010 9:42:32 PM PDT by truthfreedom
[ Post Reply | Private Reply | To 1 | View Replies]

To: DefeatCorruption

“Yeah, but what about the dead-people voters? Bet they haven’t been polled.”

And once the Military vote is thrown out and the convict vote is in ... the Dhimmi chances are looking better.


10 posted on 09/26/2010 10:05:31 PM PDT by jessduntno ("If anybody believes they can increase taxes today, they're out of their mind." -- Mayor Daley)
[ Post Reply | Private Reply | To 3 | View Replies]

To: DefeatCorruption
"Yeah, but what about the dead-people voters? Bet they haven’t been polled."

Dead people cannot help in House seats; only for statewide elections.

Meaning, the 105% turnouts in 100% black districts in St Louis, Miami, and Philadelphia, would have no effect this year because the vote fraud only increases the win margin in districts that are already guaranteed to the Democrats.

Also, vote fraud only works when the margin of error is less than about 0.5%. With a GOP margin of 8-10% among likely voters, there is not enough fraud possible for the Dems to steal seats.

11 posted on 09/26/2010 10:56:17 PM PDT by tom h
[ Post Reply | Private Reply | To 3 | View Replies]

To: freespirited
One thing yet to be accounted for is Sept unemployment. I guess it will come out on Oct. 1 or on the 8th. The first may be too soon to have the figures but it usually comes out the first Friday of the new month. We already know that 9 states estimated their employment figures last month. We also know that 27 states had falling employment numbers last week. It looks pretty bad to me.

If the unemployment rate goes from 9.6 to 9.7 or above, look out below in the markets. Each day they keep feeding us that there will NOT be a double dip but I'm not so sure. A bad GDP figure could also just about do the markets in. Bad unemployment, bad GDP, and a bad market, right before Nov. 2nd? Holy cow!

I think the Repubs will do OK even if the unemployment rate goes to 9.5, but to go the other way would be devastating a month out, IMHO.

12 posted on 09/27/2010 12:35:39 AM PDT by chuckles
[ Post Reply | Private Reply | To 1 | View Replies]

Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.

Free Republic
Browse · Search
News/Activism
Topics · Post Article

FreeRepublic, LLC, PO BOX 9771, FRESNO, CA 93794
FreeRepublic.com is powered by software copyright 2000-2008 John Robinson