A heavily-evangelical electorate might not give Ted Cruz an advantage in Indiana. Voters seem to care more about being inspired by a candidate than the candidate’s competitiveness. And a new report compares the accuracy of 10 different online panel surveys. This is HuffPollster for Monday, May 2, 2016.
INDIANA’S DEMOGRAPHICS FAVOR CRUZ, BUT ECONOMY FAVORS TRUMP - David Wasserman: "Indiana, which bills itself as the Crossroads of America, has long looked like a huge fork in the road to the GOP nomination. Its 57 delegates make it the largest prize remaining besides California, and its high-stakes delegate allocation rules — winner-take-all statewide and by congressional district — assure the winner will claim a commanding majority….In many respects, Indiana should be a terrific state for the Cruz/Carly Fiorina pre-ticket. Indiana has the highest share of evangelical Protestants of any state yet to vote — 31 percent — which is 9 percentage points higher than in Wisconsin, the site of Cruz’s last triumph. However, here’s why Indiana could be an even better Trump state: It boasts the highest share of manufacturing jobs in the country. From steel mills on the shores of Lake Michigan to the medical device hub of Warsaw, to Elkhart, the 'RV capital of the world,' Indiana’s blue-collar workforce — and its blue-collar retirees — are machine-made for Trump….Then there’s the polling: of the seven polls released this month, most have shown Trump with modest leads ranging from 2 to 9 percentage points." 
The demographic/economic interplay has been important for Republican contests - Philip Bump: "Two weeks ago, [Trump's] strongest territory had been the South. Now, it's the Northeast….This week's contest in Indiana is much different territory for Trump, though. He did well around Chicago, but otherwise the areas bordering the state have been as enthusiastic about Cruz and Kasich as they are about him. He leads in the polls, but it's clear that Indiana will be a much tougher contest….A key thing to note from this is that there's no strong correlation between any demographic point and the results of the 2016 election. This isn't like the Democratic contest, where knowing the percentage of the population that's black can give you a good sense of who will win. Figuring out the overlap of geography and economics and ideology at play on the GOP side is a lot more nuanced than that. Which is another reason that Indiana will be interesting." [WashPost]
HUFFPOLLSTER VIA EMAIL! - You can receive this daily update every weekday morning via email! Just click here, enter your email address, and click "sign up." That's all there is to it (and you can unsubscribe anytime).
VOTERS CARE MORE ABOUT CANDIDATES' ABILITY TO INSPIRE THAN THEIR COMPETITIVENESS - Frank Newport and Jim Harter: "We found that, across candidates, the leadership strengths most highly correlated with likelihood to vote for a given candidate were inspiring, cares about individuals, visionary and courageous...For example, the average percentage of those who give each candidate a high rating on competitive and also say they are highly likely to vote for that candidate is 29%...On the other hand, 53% of those who rate a candidate as high on the inspiring dimension also say they are highly likely to vote for that candidate….Overall, the conclusion is that ratings on the 'softer,' not necessarily campaign-related, dimensions are those that really seem to matter to Americans, or at the least are those most differentiating in terms of voting intentions." [Gallup]
HOW ACCURATE ARE ONLINE PANELS? - A new Pew Research report compares a variety of web samples against government data sources like the Census to determine how well the samples represent the population. Pew looked at nine “nonprobability” internet samples -- ones where respondents volunteered to take surveys, rather than being randomly selected -- plus their own online probability-based American Trends Panel, in which respondents are randomly selected.
Estimates of civic engagement and measurements of minorities' opinions are particularly problematic - Pew Research: “All the samples evaluated include more politically and civically engaged individuals than benchmark sources indicate should be present. The biases on measures of volunteering and community problem-solving were very large, while those on political engagement were more modest...Online nonprobability survey vendors want to provide samples that are representative of the diversity of the U.S. population, but one important question is whether the panelists who are members of racial and ethnic minority groups are representative of these groups more broadly. This study suggests they are not….researchers using online nonprobability samples are at risk of drawing erroneous conclusions about the effects associated with race and ethnicity.” [Pew]
So, who was the best? - Pew notes that the surveys they studied aren't "monolithic," with some performing significantly better than others, but the only survey they identify by name is their own. The others are identified only by letters A through I. Survey I consistently showed less bias than any of the other samples on just about every metric, but we don’t know which survey company fielded that one.
What the report doesn’t mean - Pew’s research focuses on comparing online panels to various known government “benchmarks” (the numbers generally considered the most accurate), but not to telephone surveys. It also doesn’t compare telephone surveys to those benchmarks. The biases shown in the report shouldn’t be read as saying these web panels are inferior to telephone surveys -- that’s not addressed.
What the report does mean - Not all web panels are created equal. From the report: “In general, samples with more elaborate sampling and weighting procedures and longer field periods produced more accurate results. The less accurate samples tended to be selected (or ‘balanced’) only with respect to gender, age and region. The best performing samples, by contrast, were balanced not just on those characteristics but also on variables such as education and income….The limitations of this study’s design, however, make these conclusions preliminary at best. Our data come from just nine samples, none of which experimentally manipulated these design features. Consequently, the effects of those features are not well isolated.”
It’s not enough to get the sample to “look” representative - More from the report: “Today numerous online survey vendors condition their samples on nondemographic variables in an effort to make them more representative. When implemented carefully and with full consideration of the survey objectives, this practice may help to improve data quality….The implication is that what matters is that the respondents in each demographic category are reflective of their counterparts in the target population. It does not do much good to get the marginal distribution of Hispanics correct if the surveyed Hispanics are systematically different from Hispanics in the larger population.
THE WEEKEND’S POLLS
-One poll shows Ted Cruz 16 points ahead of Trump in Indiana. [IPFW]
-Hillary Clinton leads Bernie Sanders in Indiana. [ARG]
-Trump extends his lead over Ted Cruz in the national primary race. [IBD/TIPP]
MONDAY'S 'OUTLIERS' - Links to the best of news at the intersection of polling, politics and political data:
-David Rothschild assesses Donald Trump's chances in Indiana by comparing prediction markets to FiveThirtyEight's forecasts. [PredictWise]
-Craig Fehrman explains why Indiana is "weird." 
-Americans who dislike the messy intricacies of democracy are more likely to like Trump. [WashPost]
-Matthew Yglesias thinks Trump will cause long term damage to the Republican Party by turning Latino voters permanently Democratic. [Vox]
-Lynn Vavreck finds that bringing up gender in the election hurts Trump, but doesn't always help Hillary Clinton. [NYT]
-Andrew Gelman looks to see if Google Trends can be predictive of referendum elections. [WashPost]
-SurveyMonkey offers some advice on how to QA a survey. [SurveyMonkey
-Americans are more optimistic about the economy than it may seem. [WashPost]