One of the figures in this morning’s WBUR poll suggest ed the tantalizing prospect of stronger support for the Olympics if the games were spread out across the state rather than clustered in Boston as originally proposed. Support in the poll (39 percent) started out trailing opposition (49 percent). But when poll respondents were told that “some have proposed spreading out the Olympic venues across Massachusetts,” support rose to 51 percent, 14 points higher than opposition. This is the first potentially promising polling news for the Olympics in quite a while.

A less centralized Games appears to be the vision for the so-called “Bid 2.0,” due out later this month. It was foreshadowed by the announcement that New Bedford would be the sailing venue. So does this poll show that Bid 2.0 will pull higher levels of support since voters say they are more likely to support a statewide Games? Possibly. But there are realities of polling to keep in mind before placing any bets.

First, the most reliable figure in any poll is the base support/oppose question, done at the beginning of a poll. It does not include any assumptions, and does not elevate any specific information in voters’ minds. Questions that follow the initial support/oppose question, while valid, are imperfect substitutes for future polling. The responses to these questions may not be predictive.

To see why this is so, let’s say the statewide venue plan in the new proposal turns out to appeal to voters. Overall support may move, or it may not. Voters have manystrong opinions about Olympic issues beyond where the events will be held. Asking about a specific issue such as the prospect of statewide venues elevates that issue in voters’ minds at that moment as a way of measuring its specific appeal. But all of the other factors are still there in voters’ minds and will play a role in how they answer future polls.

So while the idea of a statewide Games is appealing, it may or may not be a very important piece of information going forward. After Bid 2.0 is released, the initial support/oppose question will be posed to voters again-voters who may prioritize any number of factors other than the venue location.

There is a precedent that illustrates the hazard of over-interpreting hypothetical questions in Olympic polling. The initial polling for Boston 2024 was conducted by Kiley & Co. in April of 2014. A poll memo and slide deck from this poll were obtained via open records requests from Jonathan Cohn and kindly sent to us by Kyle Clauss of Boston Magazine. The poll memo was previously published by Boston2024, but the slide deck was not.

That poll started out with 48 percent support, but after hearing “brief background info,” support rose to 56 percent. As the poll progressed, “support increases steadily – up to 66% – as residents heard more information about a potential Boson-area Olympics,” said the memo. It turns out the information the poll used was not very predictive of how support would evolve. This is not a fault of the poll itself, just an illustration of the dangers of counting on knowing how the information environment will evolve.

Relying on anything other than the initial level of support can be a hazardous way of making predictions about support levels. Support may rise as a result of the statewide venue plan, but this is far from certain.

– Steve Koczela



2016 update

Fox News and CNN have said they plan to use each candidate’s standing in the polls to winnow the field for the GOP debates. CNN has made it clear how this averaging will work, Fox has not. FiveThirtyEight’s Harry Enten explains why this leaves the bottom of the top 10 uncertain.

Nationally, polls of GOP primary candidates show a tightly clustered group with no clear leader. The Gravis Marketing/Howie Carr poll finds Jeb Bush ahead with 21 percent of the vote in New Hampshire primary.

Hillary Clinton’s support of automatic voter registration appears to have polarized a formerly bipartisan issue.

It seems Darth Vadar and the Terminator have better favorability ratings than most presidential candidates, writes WaPo’s Chris Ingraham. He warns us not to “take any of this too seriously.” We won’t, until it comes time for the Voldemort/Darth Vadar debate. That sounds serious.

On the issues

A majority of Americans support same-sex marriage, according to a new Pew poll, with support continuing to increase across demographic groups. Amazing fact: in 1982, the first national poll on this issue found 11 percent support for same sex marriage, 82 percent against.

Support for Obamacare continues to slide, per a new ABC News poll. But it’s complicated, and some parts of the law are very popular.

Polling rules of the road

Many poll calls are made with machines that then connect a live interviewer when a potential respondent answers the phone. Proposed FCC rules may change this,writes Politico’s Steve Shepherd, which could fundamentally alter the way telephone polling is conducted.

Nate Silver fires back at the FCC, writing in part, “And the FCC probably ought to go back to policing ‘wardrobe malfunctions’ and not making pollsters’ jobs any harder.” The whole piece is worth reading for his defense of the importance of survey research.


– – – – – – – – – – – – – – – – – – – – – – – #NerdAlert Tearline – – – – – – – – – – – – – – – – – – – – – – –


There was another event on poll data fabrication yesterday in Washington, DC. This was the third in the series, illustrating what appears to be growing recognition of the need to deal with the problem systematically.