Aggregating three items, on the casino referendum, Martha Coakley and Charlie Baker:
Casino Referendum Chances Fading As Margin Widens
Earlier this year, it looked distinctly possible that a referendum to repeal the casino law would pass. The margin had narrowed from its pre-2014 levels of 25 or 30 points in favor of casinos, down to single digits, and even flipped to net opposition in one poll.
No longer. Now, multiple polls show the margin widening once again, meaning the referendum to overturn the casino gaming law is looking like more of a long shot. No recent poll has shown a margin of less than 9 points, and the two new polls from UMass Lowell and WBUR show 23- and 15-point margins against repealing the law, respectively. The Boston Globe poll has held steady with a 9- or 10-point margin against repeal for some time now.
In the case of the WBUR poll, we are now using the actual language that will appear on the ballot. This makes it more likely that the margin is indicative of how people will vote on Election Day. By using the actual ballot language, we capture any confusion or other mental processes that go into deciding on a yes or a no vote, rather than just whether voters support or oppose the idea of casinos. In the case of ballot questions, pollsters don’t want to clear up confusion about wording for voters or the meaning of a yes versus a no vote. That confusion is part of what voters will face when they step into the booth, which means there is value to capturing it in a poll.
Partly due to confusion, and partly due to lower information about the ballot question’s impact, many voters default to choosing “no” when they are not fully informed, since doing so makes no change to the status quo. This is another tendency we can capture by using the yes/no choice and the actual ballot wording, and a possible reason why the margin appears to have widened in the new WBUR poll. Jon Krosnick of Stanford University will be publishing research on this issue soon, and we will highlight it when it is available publicly.
12, 21, 24, 32? How Big Is Coakley’s Primary Lead?
The last 10 days have featured Democratic primary poll results that show Martha Coakley leading Steve Grossman by 12, 21, 24 or 32. So what gives?
Primary polling is typically less accurate in terms of predicting the margin than polling for the general election. So some volatility should be expected, even if all of the pollsters in question were using similar methods. And the margins in all of the polling for the last few weeks, except for the Suffolk/Herald poll (12 points), are within a 21- to 32-point range in favor of Coakley. This is a reasonably narrow range, given the higher average error in primary polling.
Further, there is a possible explanation for why the Suffolk poll showed a narrower result. Looking at the methods behind the Suffolk poll may offer a reason why their result differed. Since the beginning of this primary season, the contest has been largely about Coakley’s name recognition against Grossman’s insider support. Grossman is better known among insiders, garnering a slew of endorsements from elected officials, and handily winning the party's endorsement at the June state convention. This dynamic extends to voters, where the better informed, more frequent voters favor Grossman, while those with less information know and support Coakley. The Suffolk poll focused on very frequent voters, screening for voters who have participated in three of the last four off-year Democratic primaries going back to 2002. By doing this, they may have reached a group of voters who are much more likely to know Grossman and support him.
Looking at the data from the WBUR poll, we see that voters who have voted in the most Democratic primaries are the most likely to support Grossman. Put another way, among the same group of voters Suffolk targeted, we show a 13-point margin for Coakley, just 1 point different from the 12-point margin they found.
So who’s margin is right? Sad to say, we won’t know until we see the results next week. It is true that in past primaries, there are a lot of voters who haven’t consistently voted in previous primaries. And these less frequent voters go for Coakley. But if turnout is exceptionally low, and Grossman’s last minute campaigning succeeds in reaching and persuading newly tuned-in voters, the final electorate may end up being a more informed group. If not, the less informed voters who have nonetheless heard of Coakley will keep the margin wide in her favor.
Baker’s Strong Performance In Globe Poll Appears Likely To Last Another Week
The primary polls aren’t the only place with a considerable range of results over the last week or so. After months of consistent leads for Coakley in the polls, the Globe last week released the first poll showing her trailing Republican Charlie Baker in the general election. Then this week, the UMass Lowell/7News poll and the WBUR poll both showed Coakley with a 9-point lead over Baker.
Until this week, the Globe’s poll was more in line with the UMass Lowell and WBUR poll, making it more likely (though far from certain) that the Globe's result from this past week will turn out to be the outlier. The HuffPollster chart below shows that this approximate margin has been consistent at least since June.
Even if the Globe turns out to be the outlier, we may see another release this week from the Globe showing a steady or even more favorable margin for Baker. The reason for this lies in their use of a two week average. Each week's result is actually the average of two weeks worth of data. In last week's release, their margin went from Coakley up 7 to Baker up 1.
To show a shift this large suggests the most recent interviews last week showed Baker with a considerable lead. Globe pollster John Della Volpe confirmed this. Because their method calls for averaging two weeks of data, Thursday night's release will include new interviews as well as the pro-Baker interviews from last week. So, unless this week is extremely pro-Coakley (say, by 20 points or so), we will not see the Globe result bounce back to Coakley +9 overall.
None of this is meant as a criticism of the Globe’s method. Instead, it is meant to show why their method could produce an apparently stable result this week that differs from the two other polls released in the last week.
Steve Koczela is the lead writer for Poll Vault and president of The MassINC Polling Group.