Main Menu

Partial poll - 2/28 PDF Print E-mail
Written by Administrator   
Friday, 29 February 2008 10:41
Turnout in Texas is extremely high and includes a high percentage of first time primary voters. For this reason, I decided to begin with a sample population from all registered voters and screen responses to determine likely voters. Initial questions dealt with general election matchups and presidential choice in 2004. The resuts of the 2004 question matched election results fairly closely, so I believe it to be a fairly representative sample. In the general election matchups, Republicans win all matchups easily though not at Bush-like margins. Obama does slightly better than Clinton against Huckabee while Clinton does much better than Obama against McCain. Clinton had almost the same percentage against each Republican, while Obama's percentages varied.

Unfortunately, the screening questions produced too few primary voters to be as precise as I would like for the tight Democratic primary race. The five point margin of error is higher than the difference between the two candidates in many recent polls, and I will need to increase the sample size before I am comfortable releasing detailed percentages. That said, after weighting to expected demographics, Obama received the support of one more respondent than Clinton did.

Obama increased his numbers among African-Americans and Latinos. In previous polls, women over 60 was the only African-American segment where Clinton was competitive. Obama is now dominating that group as well. Younger Latinos are now going for Obama as heavily as younger voters overall.

Clinton improved among white voters and Bush voters. She was stable among white Kerry voters, but there has been an increase in younger white Bush voters who will vote GOP in November, but vote for Clinton in the primary. Obama still has a significant number of older white Bush voters crossing over temporarily, but the effect is mostly offset now.

Kerry voters are much more likely to vote in the primary than Bush voters, and almost none will vote in the GOP primary. One in four Bush voter who votes in a primary will do so in the Democratic primary. A little less than half of the Bush crossovers would vote Democratic in November. Among these longer term crossovers, slightly more would stay for Obama, but many would stay for Clinton. About one in four Democratic primary voter voted for Bush.

Among likely Democratic primary voters, half of Kerry voters have voted early. Only forty percent of Bush crossovers have. If lines are long on election day, this may reduce the crossover effect. Early voting and intended primary convention attendance are evenly split between Clinton and Obama, with Clinton receiving one more early vote than Obama while the same number intend to support each candidate at the convention. A disproportionate amount of Obama's support in these areas is in Harris and Travis Counties (Houston and Austin) so Clinton tends to lead in other parts of the state.

McCain beats Huckabee by about 20 points, and Paul's numbers have faded.

1100 registered voters surveyed, about half were likely primary voters. Expected demographics calculated by taking historical primary turnout from the voter file, and calculating percentages for each combination of gender, ethnicity and age group. This was modified by increasing each variable by a factor to reflect increased turnout. The two variables given the highest factor were voters under 40 and African-Americans. Latinos were given a smaller factor than whites and African-Americans as south Texas Latinos have turned out at much higher rates than the rest of the state in past Democratic primaries and therefore don't have as much room to grow.
Last Updated on Friday, 29 February 2008 10:42
TX 2/20 - Pre-debate PDF Print E-mail
Written by Administrator   
Thursday, 21 February 2008 11:17
Update - Analysis of the data has revealed a surprisingly large number of Democratic primary voters that would vote against the same candidate in the general that they voted for in the primary. Almost all are older white Bush voters, and they break heavily for Obama. Those that break for Clinton are mostly female, while the Obama voters are equally split on gender.

This group of voters is outside of the expected model, and will require revisions to be made. Backing them out of the Democratic part of the model will tilt the numbers towards Clinton, but then adding them on top of the Democratic model should boost Obama's numbers, possibly giving him the lead.

I will not revise these numbers, but the next poll will use the revised model.

These 'Democrats for a Day' seem to be more likely to support Huckabee than McCain, though both candidates have support in the general. This may correlate with Huckabee's drop as some of his supporters may have decided his miracle was increasingly unlikely and crossed over to affect the other race.

This group of crossovers is much less likely to say they will attend the primary convention, so this may help Clinton in the battle for convention delegates.



Ran a poll Feb 20 using a new turnout model. Anyone who claims to know what Democratic turnout will be is lying, but analysis of earlier contests indicates many new primary voters resulting in higher proportions of women, African-American and younger voters. My previous polling has relied on historical turnout models and a sample of voters with past primary voting history. That worked fine when it was assumed that Texas would vote long after the nominations were effectively decided. For this round of polling, I have made several modifications. First, I have added general election voters who haven't voted in past primaries to the sample. I have added an oversample of young voters, not filtered by voting history. I have also added an oversample for certain areas with a high percentage of African-Americans, but historically low primary turnout. Finally, I have increased weightings for young voters and African-American voters. Those increased weightings also slightly boosted female turnout, so I did not specifically modify that weighting.

To filter the results for likely voters, I asked a few non-primary questions and then asked about intent to vote in the primary. Past primary voters were significantly more likely to say they intended to vote than those that had not previously voted in primaries.

Much more below

Last Updated on Sunday, 24 February 2008 12:34
TX Pres Primary 1/30-31 PDF Print E-mail
Written by Administrator   
Friday, 01 February 2008 11:21
With Edwards and Giuliani dropping out, I decided to poll the race in Texas before Super Tuesday next week. Among Republicans, Mitt Romney has gained the most from the narrowing field. While he has stayed in the low to mid teens since summer, conservative Thompson and Giuliani voters have picked Romney over John McCain, raising his support to 30%, a hair ahead of the Arizona Senator's 29%. Mike Huckabee follows with 20%, down for the second poll in a row. Ron Paul has his best showing at 8%, up from 4% last month. Alan Keyes, who has decided to base his campaign in Texas until March, follows with 3%. 10% are undecided.

Looking into the numbers, McCain and Romney have opposite gender gaps, with men choosing Romney by a nine point margin and women choosing McCain by a nine point margin. Huckabee had no gender gap. Huckabee And Romney tied for the lead among voters under 40, followed by Ron Paul with McCain well back. In both the 40-59 and 60+ age groups, McCain led, followed closely by Romney. Huckabee was just below his overall average in these groups, while Paul had very little support.

Voters were asked to rate different issues of importance to the next president. Compared to earlier ratings, it is clear that Romney's new supporters are much more conservative than McCain's. Romney supporters were 30% more likely to give illegal immigration the top rating. Half gave global warming the lowest rating, compared to only 20% for McCain supporters. Ratings for both healthcare and the economy were much lower. Only a quarter of Romney's supporters would support even a gradual reduction in troop levels in Iraq, while this idea had around 40% approval from supporters of McCain and Huckabee, as well as undecided Republicans. In earlier polls, Romney supporters were much closer to the average. In contrast, McCain's supporters were more moderate this time when compared to the last poll. This suggests conservative supporters of former candidates are moving to Romney while the moderates are moving to McCain.

The issue profile of undecided Republicans most closely matches Romney on illegal immigration and the economy, McCain on troop levels and Huckabee on healthcare.

  • Romney 30% (11%)
  • McCain 29% (24%)
  • Huckabee 20% (26%)
  • Paul 8% (4%)
  • Keyes 3% (0%)
  • Undecided 10% (8%)

November and December polls both gave Hillary Clinton a hefty 51% to 17% lead over Barack Obama, but that has changed considerably over the last two polls. Current results give her just a ten point lead over her main rival, 48% to 38%. Mike Gravel received 3% while 12% were undecided. Much of Obama's increase has come from his increase in support among Latinos. In December, Clinton had a 70 to 7 lead in this group. The January 10 poll was 63-18. In this poll, the margin was down to 60-29. That is still a wide margin, but 2-1 is a lot different from 10-1. Edwards was at 5% among Latinos in the January 10 poll, so Obama's gain cannot be completely explained by his departure. There is a significant gender gap as Latino men under 60 were more likely to support Obama while Latinas under 60 gave Clinton large margins. Clinton easily led both sides of the 60+ group.

Obama's support among African-Americans actually dipped slightly as the number of undecided women increased. Obama still has a large lead 55-21, with 24% undecided.

Both candidates gained among white voters with Edwards departure. Obama gained eight points while Clinton gained five. I can't distinguish between former Edwards voters and prior movement, but my best guess would be that it was a combination of the two. Undecided was also up two points.

Clinton's support was even between men and women, but Obama had seven points higher support among men. Women were seven points more likely to be undecided.

Clinton supporters gave all issues higher importance ratings than Obama supporters did. Maybe I asked about the wrong issues. There was usually high correlation between the rankings of undecided Democrats and Clinton supporters. The exceptions were global warming and troop levels. Undecided Demoocrats were midway between the candidates on global warming, and closer to Obama on troop levels with more opting for removing troops gradually as opposed to bringing them home quickly, but the differences on troop levels was much smaller than other issues.

  • Clinton 48% (46%)
  • Obama 38% (28%)
  • Gravel 3% (0%)
  • Undecided 10% (8%)

1573 past primary voters polled Jan 30-31, 2008
734 Republicans and 534 Democrats considered likely voters
R MoE 3.5% - D MoE 4.3%
This model is consistent with the model I have used on these tracking polls since May. If the Texas primary is meaningful at the presidential level, I will modify the model to reflect the higher expected turnout in future polls.
Last Updated on Friday, 01 February 2008 11:30
GOP Post Romney PDF Print E-mail
Written by Administrator   
Thursday, 07 February 2008 22:00
Did a quick poll tonight to see how Mitt Romney's exit from the GOP primary would change things in Texas.

Last week, Romney was just ahead of John McCain, 30 to 29, with Mike Huckabee following at 20. Ron Paul and Alan Keyes were in single digits. Huckabee did best among voters under 40, while McCain did best with voters over 60. Romney's support was fairly even through the different ages.

This week, Huckabee has a small lead among the <40 and 40-59 age groups, while McCain has a large lead in the 60+ age group. McCain has fairly even support between men and women in each age group, but Huckabee does better among women in the younger groups. If Huckabee could attract the portion of the <60 male vote that is going to Paul and Keyes, he would be competitive with McCain. I have no reason to expect that they will shift, just noting the gender gap in the <60 'not McCain' vote.

  • McCain 43% (29%)
  • Huckabee 33% (20%)
  • Paul 9% (8%)
  • Keyes 3% (3%)
  • Undecided 13% (10%)

510 polled 2/7/08 Margin of error 4.3%
Last Updated on Thursday, 07 February 2008 22:24
TX Jan 10 - Obama gaining, McCain challenging PDF Print E-mail
Written by Administrator   
Friday, 11 January 2008 11:58
I polled Texas primary voters January 10 and again found much more movement on the GOP side, John McCain, who was in single digits last month surged into a statistical tie with Mike Huckabee. Huckabee, who still leads McCain 26% to 24% lost some of the moderates that discovered him last month, but added a few more conservative supporters. Huckabee is a 24 point favorite among voters under 40 and a 6 point favorite among voters 40-59. McCain has a 16 point lead among Republicans 60+.

Fred Thompson and Ron Paul lost two points, Mitt Romney lost three and Rudy Giuliani lost six points from last month. Thompson held steady among men, but lost support among women. Romney lost all his younger supporters to Huckabee and Giuliani lost support from every sector. Hunter ticked up one and Keyes held steady at about a half percent. Eight percent were undecided, though it would appear that a lot of the others aren't all that decided either.

725 polled 1/10/08, Margin of Error 3.6%

  • Huckabee 26% (29%)
  • McCain 24% (7%)
  • Thompson 12% (14%)
  • Romney 11% (14%)
  • Giuliani 10% (16%)
  • Paul 4% (6%)
  • Hunter 3% (2%)
  • Keyes 0% (0%)
  • Undecided 8%

Among Texas Democrats, Barack Obama had his best showing yet, but Hillary Clinton still has a solid lead. Clinton maintains her strength among Whites over 60 and among Latino women, while Obama maintains his strength among African-Americans, but there is significant movement among Latino men. While Clinton has had a very strong lead among all Latinos, Obama is now almost tied among Latino men. There was no other demographic segment that showed significant movement. Bill Richardson had received most of his Latino support from men, so his supporters may have moved to Obama when he dropped out of the race. That is not enough to account for all of Obama's gain, but it may help explain that one noticable shift.

John Edwards did particularly well among middle aged white men, but received almost no support from African-Americans and did much worse among women than men in most age and ethnic groupings. Dennis Kucinich held steady at 1%, Mike Gravel surged to 1% and 10% were undecided.

564 polled 1/10/08, Margin of Error 4.1%

  • Clinton 46% (51%)
  • Obama 28% (17%)
  • Edwards 14% (15%)
  • Kucinich 1% (1%)
  • Gravel 1% (0%)
  • Undecided 10%
Last Updated on Friday, 11 January 2008 12:04
<< Start < Prev 1 2 3 4 5 6 7 Next > End >>

Page 3 of 7
Joomla Templates by Joomlashack