Use of Likely Voter Model Does Not Explain Rasmussen "House Effect"

Both critics and defenders of Rasmussen Reports' polling have frequently cited Rasmussen's use of a likely voter model to explain why their polls have tended to show substantially more favorable results for Republican candidates than the average of other surveys. I have often mentioned this myself, for that matter.

The argument goes like this: those people who vote most reliably in midterm elections tend to be older, whiter, and to have higher social status -- which are also characteristics of voters that generally lean toward the Republican candidate. When coupled with what also appears to be a Republican enthusiasm advantage this cycle, it is quite reasonable to believe that a poll of likely voters (like Rasmussen's) should show more favorable results for the Republicans than one of registered voters or adults (like most others).

This argument is completely true, insofar as it goes. But it is not sufficient to explain the bulk of the Rasmussen house effect, particularly given that Rasmussen uses a "fairly loose screening process" to select likely voters.

In fact, this is quite readily apparent. Although Rasmussen rarely reveals results for its entire adult sample, rather than that of likely voters, there is one notable exception: its monthly tracking of partisan identification, for which it publishes its results among all adults. Since Labor Day, Rasmussen polls have shown Democrats with a 3.7-point identification advantage among all adults, on average. This is the smallest margin for the Democrats among any of 16 pollsters who have published results on this question, who instead show a Democratic advantage ranging from 5.2 to 13.0 points, with an average of 9.6.



To be clear, the partisan identification advantage among registered or likely voters is much smaller. A 3- or 4- point gap would be quite normal there. When making an apples-to-apples comparison to other polls of all adults, however, it is something of an outlier and would reflect a house effect of about 6 points when measuring the net difference between Democratic and Republican preferences.

Meanwhile, an increasing number of pollsters have begun to publish results among likely voters in their take on the Congressional generic ballot. Six pollsters apart from Rasmussen, in fact (these are GWU, Bloomberg, NPR, Democray Corps, OnMessage and McLaughlin) have done so since December. They show the Republicans leading the generic ballot by an average of 2.8 points among likely voters, on average (if explicitly partisan-affiliated polls are included, the margin is similar at R +3.3). This is a potentially excellent result for them -- one which might imply a massive, 50+ seat swing in the House, but is less than the 9-point advantage that Rasmussen now shows, and has shown consistently throughout this period.



Note that the house effect here, again, is about 6 points (the difference between the R+9 that Rasmussen shows and the R+3 that the other likely voter polls do). This is of the same magnitude of the 6-point house effect that was introduced in their construction of the all-adult sample, as described above. In other words, Rasmussen does not appear to be applying an especially stringent likely voter model. Instead, the house effect is endemic to their overall sample construction and is "passed through" to their likely voter sample.

Why might these differences emerge? Raw polling data is pretty dirty. If you just call people up and see who answers the phone, you will tend to get too many women, too many old people, and too many white people. This is especially the case if you rely on a landline sample without a supplement of cellphone voters.

Pollsters try to correct for these deficiencies in a variety of ways. They may use household selection procedures (for instance, asking to speak with the person who has the next birthday). They may leave their poll in the field for several days, calling back when they do not contact their desired respondent. An increasing number may call cellphones in addition to landlines.

Rasmussen does not appear to do any of these things. Their polls are in the field for only one night, leaving little or no time for callbacks. They do not call cellphones. They do not appear to use within-household selection procedures. In addition, their polls use an automated script rather than a live interviewer, which tends to be associated with a lower response rate and which might exacerbate these problems. So Rasmussen's raw data is likely dirtier than most.

But pollsters then have a second line of defense: they can massage their data by weighting it to known demographics, such as age, race, gender, or geographic location. This can work pretty well, but it is not foolproof; it requires some finesse. Moreover, some differences in response rates may not intersect neatly with these broad demographic categories. Pew has found, for instance, that those people who rely primarily or exclusively on cellphones tend to be somewhat more liberal, even after other demographic considerations are accounted for.

The bottom line is this: the sample included in Rasmussen's polling is increasingly out of balance with that observed by almost all other pollsters. This appears to create a substantial house effect, irrespective of whether Rasmussen subsequently applies a likely voter screen.

It also appears to be a relatively new facet of their polling. If one looks at the partisan identification among all adults in polls conducted in September-November 2008, Rasmussen gave the Democrats at 6.5-point edge, versus an average of 8.7 points for the other pollsters; their house effect was marginal if there was one at all.



As I've speculated before, I suspect this has to do with shifts in enthusiasm among different types of voters: that it's now become somewhat easier to get Republicans on the phone because they've relatively more excited about their political prospects. Techniques like weighting can correct for some of this response bias, but it can be an imperfect defense, particularly for pollsters like Rasmussen who have very low response rates (because of their "flash" one-night samples and their use of IVR technology).

If, on the other hand, this is a feature rather than a bug, it requires a more robust explanation from Rasmussen. It is not sufficient, after all, to believe that Rasmussen is getting it right: you also have to believe that almost everyone else is getting it wrong.

Their use of a likely voter model alone is not sufficient to explain the differences. Citing Rasmussen's success in calling past election outcomes, which is formidable, is also somewhat non-responsive, since their house effect was not so substantial in past election cycles. Moreover, most objective attempts to rate pollsters, including ours, rely on an evaluation of the accuracy of polls in the week or two immediately preceding an election (when pollsters have strong incentives to "behave" themselves). They may reveal little or nothing about the accuracy of polls months ahead of one.

There's More...

Contract Post

You Asked For It, You Got It

Dr. Christopher Parker is still in the process of reporting and analyzing data from his recent poll, but in response to some of 538's reader requests he was kind enough to provide me some data he had not yet planned to release. This post piggybacks on two earlier posts I wrote on the subject of tea partier attitudes, and also comes a day after a new New York Times survey and article about tea partier attitudes. That survey, which Nate wrote about earlier this morning, did not poll racial or racist attitudes directly, but did tap into a strong sentiment among tea party supporters that the Obama administration "favors blacks" and that "too much has been made of the problems facing black people." Though I have my own very strong views about the meaning and proper label for such sentiments, I'll leave it to 538 readers to conclude for themselves whether or not such views constitute "racist" or "racially-intolerant" attitudes.

Below is a table I created from all the data Dr. Parker provided me. Dr. Parker informs me that he has been doing deeper analyses and will be publishing his findings on the WISER website by early next week.

The data report averages--cell entries are percents--for four subsets and the full universe of white poll respondents to nine survey questions related to race, immigration and gay rights. The data include two subtotals we already knew about—whites who “strongly approve” or “strongly disapprove” of the tea party movement—as well as subtotals for whites who “somewhat approve/disapprove” and those who have no opinion and/or couldn’t identify the tea party, plus the total for all whites. I sorted the columns in a more or less left-center-right fashion, with the strong disapprovers on the left, strong disapprovers on the right, and the other three categories in the middle sorted based on the results.

What we can see, not surprisingly, is that in every case but two the means for those for what I’ll call the three non-polar middle columns—i.e. those who somewhat approve/disapprove, those who didn’t know, and all whites—fall in between those for the two polar subsets. The two exceptions are the “don’t know” subset on Question 5, who have greater worries about immigration taking jobs away than strongly-approving whites; and the somewhat approving/disapproving subset on Question 7, who have the same opinion as strongly-approving whites on the issue of gay workplace discrimination. The remaining 25 of the 27 cell entries in the middle three rows fall, not unexpectedly, in between the poles.

As to the question of how far from the mean response for all whites the means for the strong approver and disapprover subgroups fell—let’s just steer away from the Gallup-suggested term “mainstream” I regret using in my first post because it’s simply too subjective and normative as a label—in most cases the distance is about the same, or within a few percentage points of the same distance. On Question 2 (13 points/10 points); Question 3 (12/10); Question 4 (12/12); Question 6 (5/5); and Question 8 (16/14), the differences are small enough to be effectively the same, given margin of error. However, I would caution that the totals for all whites of course include the responses of both the strongly approving and strongly disapproving subsets, and because the former account for almost twice as many respondents (117 to 66), their weighting artifactually pulls that average closer to their subgroup response averages. (#If this statistical point is not immediately obvious to some, see the footnote below.)

Put another way, white strong approvers are about equally far from the all-whites mean as white strong disapprovers on issues related to black disadvantages caused by slavery, whether blacks are deserving of their situation, black work effort, immigration rates, and military service for gays. On questions 5, 7 and 9 the strong approvers are closer to the mean, and on question 1 the strong disapprovers are much closer to the mean. So if the point some 538 commenters were making is that the strong approvers are outliers but no more outlier than the strong disapprovers, that’s certainly true.

But I maintain that this is, at best, a minor and perhaps semantic point, one that grew out of this whole business of what constitutes "mainstream." The bottom line is that—and here I’ll try to use a better term—both strong approvers and strong disapprovers have abnormal views about black, immigrants and gays--abnormal in the literal sense of deviating from the mean, not deviant. (The poll, like many polls, also cannot account for social desirability biases that may in fact result in lower levels of expressed racism, xenophobia and homophobia--although, of course, those effects would not necessarily be limited only to whites who say they strongly approve of the tea party movement.)

More specifically, for tea party supporters, those abnormal views in every case but one or maybe two reflect higher expressed levels of racism, xenophobia, and homophobia. Some commenters wondered if TP approval is merely a proxy term for "white conservatives," and vice versa for disapprovers/"white liberals." However, in detailed multivariate results that control for ideology which Dr. Parker assured me by email he will soon publish in detail on the WISER website, tea party approval or disapproval is not simply a proxy for these labels, and does in fact have independent explanatory power of the level of intolerance expressed by whites. In any case, the idea that the recent Gallup demographic study "proves" that the views of tea partier supporters are "mainstream" is, it's fair to say, a fiction.


(#N.B.: Because the strong approvers are about twice the share of strong disapprovers, and because they are included in the all whites universe, the larger share of approvers moves the average for all whites closer to them than it would be, say, if we took just the average of all whites excluding both strong approvers and disapprovers. A simple mathematical example can demonstrate this. Imagine there are 10 people with an average of 6 jelly beans in their pockets, or 60 total. There are four Ronald Reagan-lovers in the group, each of whom have 10 jelly beans each, for a subset average of 10; and there are two (half as many) Ronald Reagan-haters, each of whom has zero jelly beans each. That means the remaining four people in neither of these two polar groups have a total of 20 jelly beans (60 overall total minus 40 held by the Reagan-lovers), and thus their subgroup average is 5 jelly beans. And 5 is obviously closer than 6 is to zero, the Haters’ average, and farther than 6 from 10, the Lovers’ average. In other words, if Dr. Parker had provided (or if he later provides me or reports publicly) the average for all whites except strong approvers and strong disapprovers, those averages will move closer to those of the strong disapprovers. None of this, of course, alters in any way the fact that strong approvers still outnumber strong disapprovers by two to one.)

There's More...

Contract Post

Tea Party Bears Beck's Imprint

Wednesday's NYT/CBS poll on the 18 percent of Americans who identify themselves with the tea-party movement was myth-busting in certain ways. Tea-partiers are wealthier and more educated than most Americans, the poll found. They are largely disinterested in the existence of a third party. And they don't particularly care about social issues (three-fifths favor some form of legal recognition for same-sex couples).

Another stereotype, however, rings true: tea-partiers are disproportionately attached to, and perhaps influenced by, FOX News. And they are particularly enamored of Glenn Beck. Nationally, just 18 percent of people have a favorable opinion of Beck (the majority have no opinion whatsoever about him). But most tea-partiers do. Do the math, and you'll find that 59 percent of those who do think highly of Beck consider themselves a part of the tea-party. This is, in fact, the single biggest differentiator of any of the items that the NYT asked about: not ideology, not any particular political belief, but whom they watch on television:



Relatedly, about half of those who get most of their news from FOX consider themselves part of the tea-party. This compares with 37 percent of those who identify as Republican.

This does not necessarily mean that the tea parties have become an offshoot of FOX News -- an allegation that liberals throw around a bit too loosely. But FOX, and particularly Beck's program, have become intimately intertwined with the movement. Indeed, once one begins to think of the tea parties through this paradigm, everything else starts to fit together. The tea-partiers skew older and college-educated: that's basically the cable news demographic.

Their viewpoints, too, reflect the general tenor of FOX News, whose hosts frequently assert that Obama is socialist or extremely liberal, but are often circumspect in providing evidence for those claims. The tea-partiers, likewise, are deeply distrustful and in fact quite angry at government -- but have more trouble at putting their finger on exactly why. They aren't especially likely to want Roe versus Wade overturned, for instance, or to favor restricted rights for gays, or to be preoccupied with illegal immigration, or to think their taxes are unfair, or to want Medicare and Social Security undone. It's mostly, rather, in the way that all these events fit into their meta-narrative about American society, a society which they see as leaving them increasingly victimized and disempowered.

This meta-narrative -- especially when articulated by Beck -- may have conspiratorial undertones. But the tea-partiers aren't especially likely to believe in outright, specific instances of conspiracy. "Only" 30 percent believe Obama was born in a foreign country, for instance, and only a small minority think violent action against the government is justified. (A possible exception: if one regards disbelief in global warming as concomitant with a belief in a global warming 'conspiracy', that is one of the more characteristic beliefs of the movement.)

Academic studies have found that FOX News in fact has quite a bit of ability to persuade voters and reformulate public opinion. That Beck (whose program debuted one day before Barack Obama's inauguration) has become one of its prominent voices may help to explain why the tea-parties have become so influential, and why public opinion appears to have shifted so dramatically within the past 18 months. FOX may, or may not, have a role in directly promoting the tea-parties. But its opinion programming has become the water-cooler around which the tea-partiers gather.

There's More...

Contract Post

Romney, Not Paul, Fares Best in '12 Matchups

The most electable Republican in America is ... um ... Ron Paul?

That's what you might think if you looked at today's Rasmussen poll, which shows Obama leading the libertarian by just 1 point, 42-41. The 'shock poll' is getting its share of attention, particularly since few other polls have shown any Republicans running that competitively against Obama.

But as regular readers of this website will know, the person conducting the poll can have a profound impact on its results. Rasmussen, in particular, has had a substantial Republican-leaning house effect thus far this year. Perhaps they will turn out to be right (although their idea of trying to apply a "likely voter" model 2.5 years in advance of an election is dubious). But it would be wrong to take a Rasmussen poll (or any other) at face value without taking into account this context.

Instead, I've compiled all polls conducted since July of last year(**) that test Obama against a Republican opponent. For this exercise, I use the Real Clear Politics rule of only using one poll from each firm for each matchup (the most recent). I then adjust the polls for house effects using a simplified variation of our standard method; both Presidential approval polls and the numbers from these 2012 matchups are used to calibrate the house effect. The results for each matchup that have been surveyed by at least two different pollsters are compiled below:



After adjustment for house effects, Obama's lead over Paul is not 1 point but more like 10. This result is closer to that obtained by a PPP poll in November, which had Obama ahead 46-38 against Paul (PPP's 2012 polls have also had a very slight Republican-leaning house effect.)

Paul's 9.9-point deficit is not awful -- it's better than of Newt Gingrich (-12.2), Jeb Bush (-13.4) or Sarah Palin (-14.4) do -- but lags behind the performance of Mitt Romney, who is just 5.6 points behind, or Mike Huckabee, who is down 6.6. It also lags behind the performance of a so-called generic Republican, who is actually slightly ahead of Obama. However, it is a problem for Republicans that no actual Republican can approach the performance of the generic candidate, probably because the generic candidate is Rorschach blot that allows each respondent to create what amounts to their fantasy candidate.

Paul does hold Obama to his lowest share of the vote -- 45.7 percent -- among the six actual candidates, although his share in the adjusted average (35.8) is also the lowest. This is very likely a reflection of his comparatively low name recognition.


(**) Although I do go back to last July to compile these figures, the only polls we use that are older than January are from Rasmussen (their Romney and Palin matchups) and PPP (their Gingirch and Paul matchups). Both these pollsters have shown very flat trends in Presidential approval since the time those particular polls were conducted, so the comparison should be reasonably kosher. Moreover, the differences if they were excluded would be trivial.

There's More...

Contract Post

Specter Remains Vulnerable in Primary

Can you trust that new Rasmussen poll that shows Joe Sestak closing to within just two points of Arlen Specter in the Democratic senatorial primary in Pennsylvania?

I don't know. Rasmussen polling released just in advance of election day has been quite accurate in the past. But their take on the race is inconsistent with what other polls have had to say -- including other polls of 'likely voters' -- a phenomenon that has become increasingly common with Rasmussen's polling this year. The poll also has a small sample size, with 435 likely voters.

But suppose you ignore the Rasmussen poll and look at the Quinnipiac poll instead. That one, released last week, shows Arlen Specter ahead 53-32.

If this were a general election, the race would be almost over. Incumbent candidates very rarely lose once they've reached 50 percent in the polling. But instead, this is a primary, and the dynamics of primary elections are completely different. Most notably, in a primary, the ideological gap between the candidates is much smaller, and the ideological range among the voters is much smaller. In other words: most everyone's a swing voter, and it doesn't take that much to flip their preferences from one candidate to another.

Quinnipiac's poll actually gets at his a little bit. They ask Specter and Sestak voters whether their mind is made up -- or whether they could still change it. (More pollsters should ask this type of question more often.) The poll shows that only 58 percent of Specter voters, and 67 percent of Sestak voters, are firmly decided. So another way to look at the electorate is like this:



By this reckoning, about half (49 percent) of the votes are still in play: Sestak would need to win about 60 percent of them to get to 50 percent. That hardly seems impossible, nor even particularly difficult, especially given that Sestak has largely conserved his resources in the campaign. (It would be silly to conclude that such a strategy is ill-advised: even if it worked perfectly, we wouldn't know it yet!)

Mind you, I don't think Sestak is a perfect primary candidate. His voting record is not especially liberal. He ran a good campaign in 2006 to defeat incumbent Curt Weldon, but essentially all districts like Sestak's, which has a PVI of D+3, were going to the Democrats that year. Meanwhile, Specter has become a reliable liberal vote. Although liberals (and others) might not trust him, a campaign against Specter might not produce the energy among the base, and especially within the netroots, that one against a candidate like Blanche Lincoln might. But Rasmussen poll or no, this race remains competitive.

There's More...

Contract Post

Pollster Responds to Your Questions

Because the post I wrote Monday about a new tea party-related poll sparked quite of bit of controversy--not to mention insinuations that either I or the pollster are engaged in "hackery" of some form or other--I contacted the pollster, Dr. Christopher Parker# of the University of Washington's WISER institute, so he could introduce himself to 538 readers, explain his methods and findings, and answer some of the respectful questions raised by thoughtful readers.

I wrote that post to provide a partial response to the emergence of a new meme--one deriving from, and citing as "evidence," a recent Gallup demographic study--which asserts that, because tea party identifiers are similar in some (but not all) demographic characteristics to non-identifiers, their political views can therefore be taken as similar or "mainstream" or "not fringe" or "non-racist," to borrow terms utilized by others in their interpretations of the Gallup results. Although people who are similar demographically often do have overlapping political beliefs, it's fallacious to presume two groups with certain shared demographic characteristics automatically share the same attitudes. The way to determine that is to actually survey their attitudes, which Dr. Parker did.

In any case, here are the questions and answers from my email interview with Dr. Parker, with any emphases in the text belonging to him:

1. First, tell us briefly about your training and background and polling expertise.

I received my undergraduate degree in political science, cum laude, from UCLA. Under the direction of Michael Dawson (chair), along with John Mark Hansen and John Brehm, I earned my PhD in political science from the University of Chicago in 2001. I've since been on the faculty at University of California, Santa Barbara, and held a Robert Wood Johnson post doctoral fellowship at the University of California, Berkeley, where Jack Citrin was my principal advisor. On the principal investigator front, I conducted the California Patriotism Pilot Study (2002), from which I published a paper in Political Research Quarterly.

Since arriving at the University of Washington in 2007--where, though it doesn't officially take effect until this Fall, I've been promoted with tenure to associate professor, and appointed the Stuart A. Scheingold Professor of Social Justice and Political Science--I have collaborated with my colleague, Matt Barreto, on the Washington Poll for the last three years (2007-2010), where he's been the lead investigator. I have also published from this data in the Du Bois Review. We also collaborated on the current study, the Multi-State Survey of Race and Politics. This time, however, I assumed the role of lead investigator.


2. Next, tell us how you constructed the instrument and conducted the poll. Specifically, some readers want to know how questions were phrased and why you picked the states you did. There is also some confusion about how the feeling thermometer questions—about the degree to which respondents believe that African Americans or Latinos are “hardworking” or “trustworthy” or “intelligent”—were constructed. As I understand it, there is a precedent for this in the UMich/ICPSR American National Election Survey, but perhaps you could take a moment to explain to us how your questions conform or differ from that, if at all.

Much of the instrumentation for the survey was extracted from existing polls, word-for-word. For instance, the items assessing the extent to which blacks and Latinos are hardworking, etc., are taken directly from the General Social Survey (GSS), one of the finest survey instruments in the social sciences. Likewise, the gay rights questions on the survey were borrowed from the American National Election Survey (ANES), another social scientific jewel. In fact, I'd say that approximately 70% of the items on the Multi-State Survey of Race and Politics were recently on the ANES or GSS. Beyond the theoretical import of the items, I drew so heavily upon these surveys because I need proven items, ones that had been thoroughly vetted. I also wished to have items that facilitated comparisons to other, more national comparisons. Obviously, the tea party question is new.

This leads to the question of why I chose the states that are in the survey. Because Georgia, Michigan, Missouri, Nevada, North Carolina, and Ohio were battleground states going into the 2008 election cycle, Matt Barreto suggested that we should examine racial and political dynamics in those states. California, the seventh state, was chosen because we thought it wise to include a state in which the election was never in doubt. Thus, it provides a basis for comparison.

A final note on the methodology: The survey is drawn from a probability sample of 1006 cases, stratified by state. On average, it took 45 minutes to complete the survey; the survey had a 51% cooperation rate (COOP4). The study, conducted by the Center for Survey Research at the University of Washington, has a margin of error of plus or minus 3.1 percent, and was in the field February 8-March 15, 2010.


3. Some readers were puzzled that you reported data for those who either strongly approve or strongly disapprove of the tea party movement. Could you say what share of respondents fall into that category and how each differs from the remaining subset of white respondents who neither strongly approve nor disapprove?

I understand why some readers are curious about support for the tea party among whites who are not on either end of the distribution. Here's what I have: Based upon 354 valid cases for this item (30% say they never heard of the tea party or have no opinion), 19% (N = 66) strongly disapprove of the tea party; 17% (N = 59) somewhat disapprove of it; 32% (N = 112) somewhat approve of the tea party; and 33% (N = 117) strongly approve of it. (Of course, when those that have never heard of the tea party (30%; N= 157) are included, increasing the number of observations to 511, the cell sizes change: 13% strongly disapprove; 12% somewhat disapprove of it; 22% somewhat approve; and 23% strongly approve of it.)

More details on how these categories differ on a range of relevant issues will follow in the coming days. The results will be available on the WISER website at the University of Washington.


4. You computed predicted probabilities as well, correct? Can you provide us with those results, and your interpretation and explanation of them?

I understand the skepticism associated with the simple cross-tabs that have been presented thus far. I don't wish anyone to have the impression that I, or anyone else, would draw firm conclusions based on the bivariate relationship between support for the tea party and the items listed in the table. Hence, I've estimated a few models in which I control for, among other things, the effects of partisanship and ideology. The relationships hold.

I'll draw upon three for illustrative purposes. For the first two models, the dependent variables are ordinal, so I report predicted probabilities. The dependent measure for the third model is an index, and is therefore continuous. For this, I estimate a simple regression model. Controlling for political ideology and party identification, support for the tea party (as it goes from its minimum to maximum value) results in a 23% increase in the likelihood that whites believe that "recent immigration levels will take jobs away from people already here." Moreover, support for the tea party decreases support, by 22%, for gay or lesbian adoption. Support for the tea party also promotes racism. In this example, I draw on Kinder and Sanders' (1996) work on racial resentment. I use the following four items to represent racial resentment: "Irish, Italians, and many other minorities overcame prejudice and worked their way up. Blacks should do the same without any special favors"; "Generations of slavery and discrimination have created conditions that make it difficult for blacks to work their way out of the lower class"; "Over the past few years blacks have gotten less than they deserve"; and "It's really a matter of not trying hard enough; if blacks would only try harder, they could be just as well off as whites." (alpha = .75) I use this instead of the stereotype items because it better captures the contours of more modern racism, one in which whites perceive blacks in violation of traditional American values.

In any case, racial resentment increases by approximately 25% as support for the tea party increases from its minimum to its maximum value. Again, each model controlled for possible confounds associated with partisanship and ideology. In sum, based upon this analysis, the data suggest that increasing support for the tea party is likely associated with xenophobia, homophobia, and racism.


5. Pulling it all together, what can we safely and confidently conclude about those who identify with the tea party movement and those who do not? Are their attitudes fundamentally different from other whites, from the American population as a whole, and if so, how so?

One way in which to view these preliminary results is that we should remain cautious, and not jump to firm conclusions. I say this, first, because the sampling frame I use differs from, say, recent polls conducted by Pew, Qunnipiac, the Washington Post, and USA Today/Gallup. Indeed, my results are relevant only to the states in which the survey was conducted, four of which (NV, MO, GA, and NC) voted for the Republican presidential candidate in at least seven of the last ten election cycles. Perhaps this is why my results appear at variance with national polls.

Another reason to proceed with caution is that I don't have an item that directly measures tea party membership. Indeed, support for the tea party isn't the same as accounting for group membership much less group identification, both if which tend to powerfully predict attitudes and behavior. With that in mind, it's entirely possible that I've underestimated the effect of the tea party on political attitudes, and will likely do so in future analysis on its effect on political behavior.

Moreover, I make no claim that the data is representative of the country. Rather, they are representative of the states that were sampled. Appropriate weights, based upon the American Community Survey, have been constructed.


6. Are you aware of other polling out there that either confirms or disconfirms your analyses or any portions thereof?

At first glance, there are a few polls that may corroborate my findings. Recent polls conducted by Pew, Quinnipiac, the Washington Post, and USA/Gallup all ask questions relevant to the Tea Party movement. However, these surveys are absent questions that tap racism. Pew and Qunnipiac ask questions relevant to gay rights, but immigration isn't present. So, I believed it possible for these to answer questions concerning relationship between gay rights and support for the tea party, but not cannot address the association between the tea party and racism or immigration.

Turning to more academic surveys, it's possible to enlist the 2008 ANES, for instance, as a means of corroboration. But, for any items that bear upon race, immigration, or gay rights, it's likely the case that too much has happened between now and then to make reliable predictions. I suspect, however, that when data is collected for the next ANES or GSS, it will then be possible to check my results.


(#As a matter of full disclosure, I had never communicated with nor even heard of Dr. Parker until I wrote about his poll. I do know his sometime-collaborator Matt Barreto; we have attended some political science conferences and/or sat on panels together. Other than my professional/personal acquaintance with Dr. Barreto--and he and I do not research or co-author together--I have no affiliation with either the University of Washington, WISER, Dr. Parker, or anyone else involved in the poll.)

There's More...

Contract Post

Consolidate the Base, Then Move to the Center

New post up on Guardian comment is free today on Cameron's inability to "unite the right."

Successful candidates reassure their base, then appeal to crossover voters to fill out a winning majority. In a US context, this is nicely staged in the primary process, with Democratic candidates angling leftward and Republicans to the right in order to secure their respective nominations, before turning back to the center for the general.

In the UK, the timing is not quite as clear, which may be what has tripped up David Cameron. He's waited to late to shore up support in the right-wing, and will now try to do it with loads of skittish moderates watching. Many of these centrists previously were not so interested in the campaign and would likely have missed the more hard-core overtures if they had been done last year or early in 2010.

---
Renard Sexton is FiveThirtyEight's international affairs columnist and is based in Geneva, Switzerland. He can be contacted at sexton538@gmail.com

There's More...

Contract Post

Why Turnout Matters

With voters unconvinced by each of the major UK parties and their leaders and with third and fourth parties only seriously competitive in a handful of seats beyond those they already carry, political observers in the UK have been questioning the level of turnout for 6 May's voting.

The highest national turnout in recent years in the UK was the 1992 election, which concluded with a resounding win for the Conservatives. With 77.7 percent turning out to vote, the Tories pulled 42.3 percent of the national vote, earning 336 of Parliament's 651 seats, while Labour won just 271 constituencies.

Since then, national turnout has gone from bad to worse. In Labour's 1997 landslide, just 71.4 percent showed up to vote, the lowest since 1935. But this was small compared to the major drop in participation that occurred during the 2001 election, when less than 60 percent of voters cast their ballots.



While seats held by all parties experience a drop in turnout from 1992 to 1997 and then again from 1997 to 2001, Labour seats throughout had the lowest overall turnout -- with the gap growing in 2001 and 2005.

In Conservative-won seats in the 2001 election, when the big turnout drop occurred, turnout was relatively consistent throughout, with quite heterogeneous majorities.

Besides the main zero slope cluster, there was also a group of seats on the upper edge of conservative support where increased turnout resulted in higher majorities.

For the Lib Dems, it was a similar pattern. In short, the party's majorities were independent from turnout.


However, in constituencies won by Labour, a different pattern emerged.

In Labour constituencies in 2001 -- the same seats they won in the Blair landslide of 1997 save 5 extreme marginals lost -- the relationship between turnout and victory margin is stunning. Basically, in safe Labour seats, fewer people showed up to vote -- knowing that Labour would win nonetheless. In particular, you would guess, Conservatives, who were demoralized from the 97 loss.

However, when you look at the change in turnout against the seat swing (Con to Lab swing) in Labour won seats, this trend does not come through.


In short, it appears that in Labour seats from 1997, both Conservative and Labour voters became disenchanted -- a significant number of Labour voters who were caught up in the Tony Blair furor sat out the 2001 election, while Conservative voters who were certain that they could not win in safe or relatively safe Labour constituencies also stayed home.

Why does this matter to us in 2010 context?

There are 4 reasons that this is important to us this election season:

1. Demoralized voters from 2001 and 2005 may choose this election to make a comeback. Polling numbers are showing 10-20 percent undecideds, and significant number of these voters did not vote either in 2001 or 2005. These voters could either stay on the sidelines or decided to come back in the mix. However, given that both Labour and Tory voters left the electorate in 2001 and 2005, it will be a matter of which party can motivate more of their former base to return, something that should favor the Conservatives.

2. Labour middle marginals -- the ones the Tories need to win in order -- have relatively low turnout. In the middle marginal Labour seats, where low turnout has dominated the last several elections, a new influx of voters could push the overall totals 3-5 points in either direction, which could be enough to win. Particularly the seats that used to be quite safe for Labour going into 2001 and 2005.

3. Polling numbers tend to favor the most committed voters. The polls that we have seen so far have had relatively stringent likely voter screens (generally, only 10 out of 10 certain to vote respondants have been included), meaning that we are not really sure of how returning voters may react.

4. Labour seats with 30+ leads will still see quite low turnout. For the reasons said above, the safest of Labour's seats will undoubtedly still get quite low turnout, and big margins of victory. This will have similiar implications for national popular vote numbers, and associated polling that precede them.

---
Renard Sexton is FiveThirtyEight's international affairs columnist and is based in Geneva, Switzerland. He can be contacted at sexton538@gmail.com

There's More...

Contract Post

Additional Information on New York Neighborhood Ratings

So far, my project with New York magazine to rate the city's neighborhoods seems to have stirred up a roughly equal amount of love and hate -- which is better than I expected. A bit of additional context on the ratings and exactly what's included in them is provided below.

-- This is a model of reality. It's a thoughtfully designed model, based on relatively high-quality and objective data sources. But like all models: it's a simplification. I think it passes the test of providing some worthwhile information which can serve as a complement to one's subjective take on the relative quality of different neighborhoods. That is, it points you more in the right direction than the wrong one. But it's not designed to be, nor could it possibly be, definitive.

-- The choice of neighborhoods, and the geographic boundaries assigned to them, were determined by New York magazine staff. I thought they did a very comprehensive job, on balance. It's not trivial to include additional neighborhoods because a lot of this involves counting things -- whether landuromats, toxic waste dumps, or murders -- by hand. The 60 neighborhoods within our scope are not necessarily the 60 best neighborhoods. Yes, we'll get Forest Hills included if we do this next year.

-- A bigger problem, frankly, is that the neighborhoods vary a lot in size (both in terms of geography and population). To be on the right "side" of a large or varied neighborhood may entail significantly better amenities, but also significantly more cost. If we do this again, we will probably break some of the relatively larger neighborhoods (Upper East Side, Upper West Side, Flushing, Astoria, Williamsburg, Ditmas Park/Kensington) into two or more parts.

-- While we can get something approaching an objective answer regarding the relative importance of the non-cost categories (e.g. through survey data), it is inherently very hard to know where to set the cost parameter since this will vary so much according to individual needs. We went with 25 percent simply because it reflects the midpoint between the cost-is-no-object (0 percent) and best-bang-for-the-buck (50 percent) scenarios. However, this is best handled by establishing your own priorities via the livability calculator.

-- The apartment prices listed at various points throughout the article reflect the monthly cost of 800 per square feet of commercially marketable space. We refer to this off-handedly as a "two-bedroom" on various occasions, which is probably about right for Manhattan, but is somewhat stingy for something in the outer boroughs. To reiterate, the calculations are ultimately made on a price-per-square foot basis -- which is the only real way to do it (IMO).

-- The list is targeted somewhat explicitly toward someone who commutes frequently into lower and/or middle Manhattan. If this isn't you, you should reduce the influence of the 'transit/proximity' variable in your calculations. While it would be nice to customize the ratings to individual commuting needs (e.g. you work in Hell's Kitchen, the wife works in Connecticut, you fly out of JFK twice a week, you have Rangers' season tickets, and you have a bunch of family in Bay Ridge) that would be a considerably more massive project.

-- I don't even like Park Slope, particularly.

-- A fairly comprehensive list of the metrics included within each category proceeds below. All component indicators are not necessarily weighted equally.

Housing Cost
An index of housing prices aggregated from six sources: StreetEasy, Zillow.com, PropertyShark, Trulia, Jonathan Miller and the Census Bureau. The index reflects the cost of both rental properties and purchases. It was compiled on a cost-per-square-foot basis. StreetEasy and Zillow.com data were determined to be the most reliable and were given (by far) the most weight, although they may not be available in some outlying areas.

Housing Quality
- Rates of housing code violations and housing complaints
- Average monthly utility bills (a proxy for energy efficiency)
- The number of cockroaches per rental unit (per housing survey)
- Percentage of units with dishwashers and washing machines (per Streeteasy listings)
- Self-assessment of area housing quality (per housing survey)
- Percentage of residences within Historic Districts.

Schools
Quality of public K-8 schools in or near the neighborhood, as measured by test scores (both raw scores and relative to the demographics of the student body), parent satisfaction ratings, and attendance. Private schools are not factored into the calculation.

Transit and Proximity
The largest single component in this category was commute times to four busy stations in Lower Manhattan and Midtown (42nd Street, Fulton Street, Union Square, and Columbus Circle), as measured by MTA’s Trip Planner. We also looked at the number of trains, train lines, and train stations that pass through each neighborhood, survey data on satisfaction with bus service, and a separate calculation on commute times from the Census Bureau.

Crime
- Rate of murders in last 5 years (counted by hand assigned to individual neighborhoods)
- Rate of other violent and non-violent crimes (as extrapolated from police precinct data).
- Perceived safety in walking alone at night, per NYC’s 2008 citywide survey.

Shopping and Services
The density (relative to both population and square mileage) of 15 types of businesses within each neighborhood: supermarkets, delis, laundromats, day care facilities, banks, libraries, hardware stores, gyms, bookstores, clothing stores, coffee shops, liquor stores, doctors’ offices, dentists’ offices, and nail salons. Special attention was paid to the supermarkets category, and especially the presence of a highly-rated produce mart (including but not limited to Whole Foods, Trader Joe’s, Fairway, etc.)

Food and Restaurants
The density of restaurants per square mile in the neighborhood, as well measures of quality as determined from Yelp.com reviews and New York magazine critics’ picks. Quality restaurants far from the city center, which were less likely to be reviewed, were given a bonus multiplier.

Nightlife
Similar to the food category: we looked at the density of bars per square mile, as well as their quality as measured by Yelp and New York magazine critics' picks. We also took a count of dance clubs, and the number of restaurants in each neighborhood that are open past 3 AM.

Creative Capital
- The number of art galleries, theaters (movies and live performance), and concert venues in each neighborhood.
- The percentage of a neighborhood’s workforce that is engaged in the arts (measured both relative to the population and to the number of bankers), per Census Bureau data.
- The number of out-of-state students enrolled at four-year colleges and universities within the neighborhood.

Diversity
Racial and income diversity, as measured on a neighborhood-by-neighborhood and (via census tract data) block-by-block basis. The statistics are race-blind: neighborhoods did not automatically receive points simply for having a large number of nonwhites, with the highest scores going instead to those with a roughly even mix of races. We also looked at the number of distinct ethnic populations within each neighborhood, the percentage of foreign-born U.S. citizens, and the number of same-sex partner households.

Greenspace
- The amount of park acreage: large parks like Central Park were divvied up between several different neighborhoods as appropriate.
- The percentage of residences within each neighborhood that are within walking distance of a park
- Street tree density as measured by New York City’s Tree Census
- Survey data on satisfaction with local parks
- The number of points of beach and waterfront access within each neighborhood.

Health and Wellness
- Air quality complaints
- Asbestos complaints
- Noise complaints
- Asthma rates (a good proxy for pollution)
- Lead poisoning rates
- Infant mortality rates
- Street cleanliness (per NYC government data)
- Counts of rodents, potholes and graffiti
- Frequency of pedestrian and bicycle accidents
- The number of environmentally hazardous facilities within each neighborhood
- Satisfaction on access to emergency services, and to hospitals (per NYC 2008 survey)
- Satisfaction with garbage pickup.

There's More...

Contract Post

New Data on Tea Party Sympathizers

Last week, the ongoing debate over what we know about the tea partiers took a new turn, with scores of conservative commentators like The LA Times' Andrew Malcolm and Glenn "Instapundit" Reynolds assuring us that a new Gallup poll proves the TPers are not a "fringe" or "racist" group.


But the Gallup results only confirm that tea partiers are "mainstream" in their demographics, when what really matters are their attitudes. Results released Friday of a new multi-state poll of white voters conducted by the University of Washington's Christopher Parker paint a more complicated picture. The survey asked white respondents about their attitudes toward the tea party movement--and their attitudes toward non-whites, immigrants and homosexuals.


The charts contained herein show the disparity between whites who strongly approve and disapprove of the tea party movement. In a few cases -- attitudes toward Latinos, for instance -- the differences were small. But only in a few cases: tea party sympathizers believe blacks are less intelligent, hardworking and trustworthy. They appear to be particularly wary of immigrants. And they don't much care for gays, either. (Although note that two-thirds of them support gays in the military, an issue on which policy has long lagged public sentiment.)


Again, this is a comparison of white attitudes, not differences between whites and non-whites. Which means that avid white tea party sympathizers do not even hold mainstream attitudes among whites. If we included the attitudes of non-whites, the views of white tea party sympathizers would be even more aberrant.


Some conservatives believe there is a systematic attempt to demonize the tea party movement and their supporters. There is probably some truth to that. But in the same vein, aberrant opinions espoused by tea partiers or their sympathizers should neither be ignored nor papered-over.

Parker's study shows much higher levels of intolerance among whites who sympathize with the tea party movement. To be clear, the splits compare those who strongly disapprove or approve of the tea partiers, so the differences reflect those whites with polarized attitudes toward the movement. And the study might be more instructive if it compared compared the tea-partiers views toward those of white conservatives or white Republicans, or differentiated between those who are merely empathetic toward the tea-party movement and those who have actively participated in it. But it's safe to say that those who are sympathetic to the tea party movement do not hold mainstream views on issues related to race and identity politics.

There's More...

Contract Post

Newt Gingrich: Dominated Strategy?

A couple of people whose opinions I respect (including one of my colleagues) are starting to jump on the Newt Gingrich in 2012 bandwagon. I'm not there yet -- and I don't think the numbers are there either.

Mind you, I can somewhat see the logic. Mitt Romney has Romneycare problems in addition to all the other problems that made him Mr. Silver Medal in 2008. Mike Huckabee has performed surprisingly poorly in straw polls and there is reason to doubt whether he's going to run in the first place. Sarah Palin has electability problems that even the Republican base is starting to perceive. And, while there are a number of other candidates -- from Tim Pawlenty to Mitch Daniels -- who could presumably leapfrog the putative front-runners, none of them start with the name recognition that Gingrich does.

But Gingrich, I think, has just as many problems. For starters, he has marginally the worst favorables with GOP voters among the four major brand-name competitors. (The numbers below are simple averages of all polls of Republican voters taken over the past year, with a maximum of one entry per polling firm.)



Gingrich also ranks third of the four candidates -- ahead of Palin, of course -- in favorability among the broader electorate:



Moreover, he has baggage. The Republican brand is doing marginally better of late, but as recently as 2-3 years ago, when it wasn't, voters associated Gingrich with some of its less desirable qualities and his favorablity ratings were squarely in the red:



Republicans do not view him as any more electable than Huckabee or Palin (Romney dominates this category) -- this data is from the Clarus poll that I teased earlier:



Basically, I think Gingrich is fairly close to being a dominated strategy. If Republicans just want to pick the candidate they like the best personally -- electability be damned -- they'll go with Sarah Palin. If they want someobdy who they think can win, they'll hold their nose and vote for Romney. And Huckabee, not Gingrich, seems to be their second choice in both those departments. Nor do I think that Gingrich can credibly be re-branded as a fresh face if that's the direction in which Republicans want to go; voters (both Republicans and the broader electorate) remember him pretty well, even though his peak of political activity came around 15 years ago. Finally, he's easily the least electorally accomplished of the four brand name candidates, having never been elected to an office higher than the U.S. House, and from an easy district at that.

As I've been saying for some time now, I continue to think that the Republican nomination will be contested principally by Mitt Romney, Sarah Palin and X, where X is relatively unlikely to be Mike Huckabee (who I think is somewhat unlikely to run), Tim Pawlenty (who I don't think has the charisma to pull himself out of his name recognition deficit -- but we'll see) or Gingrich. If Palin indicates that she will not run, that certainly opens up a slot for a base candidate and Gingrich might then become more viable. But I think it's going to be hard for him to run when Palin is running as well. He's playing for somewhat the same half of the Republican electorate that she is, but he generates less enthisiasm -- and his electability argument is no more than marginally better, and could backfire if he presses the point too much and appears insulting to Palin in the process.

There's More...

Contract Post

Do Manhattanites Get Their Money's Worth?

The results of my project with New York magazine to rate New York City's neighborhoods along a number of objective statistical dimensions are now live. This was a fairly grueling and exhaustive project; people are rightly proud of their neighborhoods, and I don't think you dare to do something like this without being pretty careful about it. With that said, there's a lot of reasonably high-quality data available on New York and its neighborhoods, so I hope we were at least able to get a little closer toward the truth.

If you want to look at a static set of rankings, Brooklyn's Park Slope ranked first, followed by Manhattan's Lower East Side and Queens's Sunnyside. But that's not really where the utility of this project is. The coolest part, rather, is an interactive applet that allows you to determine for yourself the relative importance you attach to 12 different categories of data (housing cost, housing quality, transit/proximity, crime, schools, green space, food/restaurants, health and wellness, shopping and services, diversity, "creative capital" and nightlife). If you play around with the applet for long enough, you'll find that it's fairly easy to slot any of 15 or so neighborhoods into the top position, and any of 40 or so of the 60 that we evaluated into the top ten.

So in some sense, every neighborhood is a potential winner, especially given that 12 dimensions isn't really enough to capture all the eccentricities that go into someone's choice of neighborhood. Maybe you don't just want good food nearby, but good Italian food; maybe you don't just want to be near any old park, but near Central Park. We evaluate commute times based mostly on travel times to a group of destinations in lower and central Manhattan (as well as things like the density of train coverage), but most households will have one or two particular locations that they have to commute to for work (or recreation). If you've got Mets season tickets, fly out of LaGuradia 50 times a year, and work near Grand Central, something on the 7-train like Woodside, Queens will be more convenient for you than something in Brooklyn's brownstone belt.

With that said, there were some general patterns that emerged:

* I strongly suspect that there is something of a premium associated with living in Manhattan itself. For example, if you compare Brooklyn Heights with the Upper West Side, they are similar in most respects -- both are leafy, quiet, not-very-diverse, family-oriented neighborhoods with average (relative to NYC) food and nightlife, aesthetically pleasing and well-maintained homes, and similar transit times. But 800 square feet of apartment space costs a ridiculous $2,575 per month in Brooklyn Heights and a much-more-ridiculous $3,650 per month in the Upper West Side. There are a lot of examples like this, where a Manhattan neighborhood isn't obviously superior to a Brooklyn or Queens neighborhood but is considerably pricier; a comparison between Harlem (Manhattan) and Fort Greene (Brooklyn) follows this pattern, for instance.

* Housing costs in New York appear to increase somewhat exponentially with increasing neighborhood desirability, whereas in the other city I've lived in as an adult (Chicago), they're fairly linear. In New York, the five most expensive neighborhoods are about 5x as expensive as the median neighborhood on a price-per-square-foot basis; in Chicago, the ratio is just 2:1. There might be any number of reasons for this, but I suspect that a lot of it has to do with the fact that in cities like Chicago that are even slightly less dense, if you want to pour more money into your home you can usually do so by getting a bigger home. In New York, the constrains on space are such that to a large extent, the only real variable you can permute is where you live, and the most desirable neighborhoods may be priced in at a super-premium. Neighborhoods like the West Village, Tribeca and Soho are really nice places to live (they'd rank 1, 2 and 3 on the list if cost were not considered), but they're really, really, really expensive.



* There seem to be times when the idea of a neighborhood that's on its way back gets a little ahead of itself. Two neighborhoods, for instance, that are much nicer places to live than they were 10 years ago (Chinatown and Harlem) nevertheless came out as probably being a bit overpriced. The schools and crime rates, in particular, still lag behind in Harlem, as does the housing stock and cleanliness of Chinatown. Perhaps people are speculating about which neighborhoods will be cool five years hence and prices (especially in the for-purchase market) reflect that. But who knows if they're guessing right: the new new thing might have stolen their thunder by that point.

* Not to encourage gentrification, but if I were buying property as an investment right now, I'd look toward places that are cheap relative to those qualities that take the longest time to upgrade or repair -- particularly transit lines, and to a lesser extent greenspace and the quality of the housing stock. That might mean places like Long Island City and Sunnyside (Queens), Prospect Heights and DUMBO (Brooklyn), Washington Heights (Manhattan) and perhaps even some portions of the South Bronx near the train.

There's More...

Contract Post

Southern Republicans Prefer Romney?

Let me condition everything to follow by noting that the results of Saturday's Southern Republican Leadership Conference straw poll are, well, the results of a straw poll. It's not necessarily representative of mass Republican preferences, or even those of southern Republicans. That said, there are some interesting results, especially because the poll asked respondents for their secondarypreferences.

First, the results:



By a single vote over Ron Paul, Mitt Romney won among first-choice preferences, 439 to 438. Sarah Palin and Newt Gingrich were just a few votes apart from each other at 330 and 321, good for 18 percent each.

Some reports from New Orleans indicate heavy, coordinated lobbying on behalf of the Paul and Romney campaigns. I don't doubt it; we saw the same thing at the recent CPAC meetings on behalf of Paul, who brought a lot of motivated young people out to support him. So it's interesting to look at the secondary preferences because, presumably, strategic supporters of Romney and Paul would know better than to cast their second-choice votes for the other guy.

And sure enough, neither Romney nor Paul led among second-choice preferences: Southerner Gingrich was first with 339 votes, followed by Palin (332), with Romney (242) in third. But the glaring result here is that Paul, with just 98 votes, finished ninth and last among second-choice preferences--even behind libertarian former New Mexico Gov. Gary Johnson. In other words, Republicans either like Ron Paul or they don't like him. He's almost nobody's fall-back choice. (Btw, I must confess I didn't even know Johnson was running for president until I saw this straw poll. What's going on here? Is he a potential slate ticket running-mate for Paul, himself a libertarian non-interventionist?)

With two choices revealed, we can compute an abbreviated Borda Count by assigning two points for every first-place vote and one for a second-place vote. The Borda Count, displayed in the final column, re-sorts the results in interesting ways. Romney is still in first, but Palin and Gingrich and Paul all move into a virtual tie for second, with just a few points separating them. Palin and Gingrich obviously have much deeper appeal than Paul.

One final observation, in the form of a question: Why is Huckabee struggling among southern Republicans? Obviously, he and Gingrich eat from the same, southern bowl. And I gather that many social conservatives in the region are in still in love with Palin, even if she hails from a state as far away from the South as possible, which also draws support away from the former Arkansas governor. But Huckabee was not just fifth but a very, very distant fifth. I emphasize this point to reiterate my gut feeling that Gingrich is the GOP's 2012 sleeper candidate.

There's More...

Contract Post