Maryland referendums: Parsing the polls

The Sun's newest poll shows Maryland voters strongly in favor of allowing same-sex marriage, strongly opposed to an expansion of gambling and split almost evenly on the Dream Act, which would grant illegal immigrants in-state tuition at colleges and universities. Another independent poll released days earlier, by Gonzales Research and Marketing Strategies, found a big lead for the Dream Act, strong support for marriage equality and a statistical tie on the question of gambling expansion. And the backers of the gambling expansion plan released their own survey on Saturday showing overwhelming support for the Dream Act and marriage equality and a 7-point lead for the casino question.

These contradictory results come at the same time that prominent Republicans are calling into question the integrity of polling in the presidential race, alleging that a biased news media is tilting the results in favor of President Barack Obama. That notion is absurd; it implies a level of coordination among polling firms and media outlets that would be impossible to achieve. But conspiracy theories aside, the complaints do highlight the fact that polling involves some art as well as science, and that may help explain why three polls conducted in Maryland at roughly the same time came up with such different results.


Both the Gonzales poll and The Sun poll, which was conducted by Annapolis-based OpinionWorks, asked about the presidential election, and the results were effectively identical. The Sun found President Obama winning Maryland 57-34, and Gonzales reported that he was up 55-36. That suggests the differences in the turnout models between the two polls — the Gonzales survey, for example, assumes that African-Americans will make up a slightly higher percentage of the electorate than The Sun does — are not skewing the results in one way or the other. (The Mellman Group polling memo released by the casino backers does not include detailed information about its sample, and it doesn't include a question about the presidential race.)

The most obvious difference among the polls, and perhaps the one that most likely explains the difference between them, is in the wording of the questions. Both the Gonzales and Mellman Group polls used the exact wording of the questions as they will appear on the ballot. The Sun poll summarized the questions. Given the limited history of polling in Maryland on referenda, it's not clear which approach is more accurate. Do people actually read the whole question presented to them on the ballot, or do they just skim the language to figure out they're being asked to vote on?


The answer may well determine the outcome of these referenda. The ballot language in all three cases was written by the secretary of state, an appointee of Gov. Martin O'Malley, who supports all three measures. Thus, in the case of the Dream Act and same-sex marriage questions, the ballot language includes key caveats that the measures' opponents typically don't mention — for example, that clergy cannot be forced to marry people in opposition of their religious beliefs. The list of official provisos on the Dream Act runs to 125 words and includes mention of other effects of the law, such as an extension of the time honorably discharged veterans can qualify for in-state tuition rates.

Among the questions polled, the one that showed the least difference between The Sun and Gonzales surveys was gay marriage. Given how familiar people are with this issue, that's not surprising. The biggest difference was on the question of gambling expansion, and there, one point of emphasis in the ballot language may be the key. It begins with the phrase, "Do you favor the expansion of commercial gaming in the State of Maryland for the primary purpose of raising revenue for education..."

That's not technically inaccurate; most of the taxes collected from the state's gambling program go into an educational trust fund. It is, however, somewhat misleading in that the advent of gambling has not increased the state's spending on education above what it was legally required to be in the first place, and nothing about this referendum changes that fact. Gambling expansion opponents are hammering that point home in a multi-million dollar ad blitz, and the difference between these polls shows why.

Another crucial point to remember is that a poll is a snapshot of opinion at one point in time. The divergent results in the polling over the gambling expansion suggest that public opinion on that question is not fixed. Much will depend on what impression voters are left with when they head to the polls in November. These polls may not tell us the outcome of the gambling referendum, but they do allow us to make one solid prediction: We're going to be hearing a lot more from the deep-pocketed interests on both sides of this question before Nov. 6.