Shaping opinion by skewing the polls

Cue the Republican primary-voter angst. The mainstream media appear ready to choose the GOP presidential nominee.

Or so it would seem. A new crop of news stories have appeared praising Indiana Gov. Mitch Daniels, who has governed as a conservative but who has said it is time to "call a truce on the so-called social issues" until the country has resolved the pressing economic issues facing the country, particularly the federal deficit and debt. Social conservatives were displeased with the idea of checking their concerns at the door and have said so.

But the Washington Post was enamored and went so far as to say that others already in the race were somehow unserious, or that only if and when Daniels enters the race for the Republican nomination will the contest truly be under way.

This, despite the fact that Daniels is a clear third-tier candidate in the polls, winning an average of just 3 percent support. That number will surely rise if Daniels continues to win fawning MSM coverage.

In another survey on presidential politics, the Associated Press released a poll this week in which Democratic respondents outnumbered Republican respondents by 17 percentage points! Of course, President Obama fared very nicely in that survey.

TRENDING: Trump crushes Dem hope of vote-by-mail: 'A lot of people cheat'

This year, it will be particularly important to watch the polls closely and to look beyond the headlines to see how surveys are constructed, because how a poll is built has a great deal to do with its findings. There is noticeable skepticism on the political right that polling data is used by MSM outlets not just to report public opinion, but to shape it.

By contrast, scientific polling is calibrated, or weighted, to accurately reflect the population it is trying to measure, and it is in the process of weighting surveys that results can be skewed one way or another to obtain a desired result. Polling has been called a process of part science, part art, and it is demographic weighting that is the ultimate mixture of the two. Each polling organization interprets the demographic makeup of its target sample a little differently, and it is these differences that account for much of the variable results in different surveys.

For instance, the New York Times/CBS survey does not weight its surveys for partisan affiliation. Neither does Gallup or the AP. They do properly report the percentage of Democrats, Republicans and independents who participate in their surveys, but sometimes you really have to dig deep into their reports to find that information. What you discover in almost every survey from these sources is that Democrats are over-represented and Republicans are under-represented. Sometimes this ratio is close, but like the AP survey released this week demonstrates, sometimes it is not.

Another factor is the voting habits of those in the sample. Surveys that include all adults might be fine if you are measuring attitudes about flowers or restaurants, but if you are trying to measure anything political, you really need a sample of likely voters. Even a sample of registered voters is not reliable, particularly in this post-ACORN era, as many states have badly swollen voter rolls that include people who registered to vote – or had someone else register them – but who have never actually cast a ballot. Since barely half of registered voters cast ballots in presidential elections, about half of the sample of any registered voter survey includes the opinions of people who don't participate in elections. Who cares what they think about politics?

It is intuitive that a sample of all adults or of registered voters is going to be less informed than a sample of likely voters. Polls using such respondent samples are likely to rely more on popularity and name identification than on issues relevant to the office being sought by the candidates. This is why Donald Trump does better in polls of all adults or registered voters. Among likely voting Republicans, he tends to do significantly less well.

Because public opinion surveys are really the only scoreboard in the world of politics before Election Day, they carry unusually high influence and can sway those who have not yet made up their minds. This is what many Republicans seem to understand – and fear – about what the mainstream political media is about to do with the GOP race for president.

Fritz Wenzel is president of polling firm Wenzel Strategies.