Poll Shows Americans Oppose DOMA, Government Defense of it

Domapoll   Domapoll2

New polling released by the Human Rights Campaign in partnership with Greenberg Quinlan Rosner Research reveals that most voters oppose the Defense of Marriage Act as well as Republican efforts to defend it in court.

HRC writes:

"51 percent of voters oppose DOMA while 34 percent favor it. Independent voters, who were instrumental in the Republican House takeover, oppose the law by a 52 percent to 34 percent margin. Additionally, when read statements for and against defending DOMA in court, 54 percent of voters oppose the House Republicans' intervention, while only 32 percent support it. Given a list of issues important in determining their vote for President, voters ranked the economy and jobs (54 percent), Medicare and Social Security (23 percent) and education (19 percent) as most important with only 5 percent of respondents saying marriage was most important to them. Finally, when presented with the actual benefits from which married same-sex couples are excluded, voters overwhelmingly support extending those benefits."

Read the poll results here.


  1. David in Houston says

    800 people ≠ all of America. Show me an online poll on USAToday or CNN with 100,000 voters and I’ll feel better about the results.

  2. Clive says

    You bash NOM when they use push polls but this one doesn’t seem much better, the only difference is that it is telling you what you want to hear.

  3. Mike in Houston says

    David — an online poll, no matter how many people “vote” is not a scientific sampling and therefore not a valid representation of public opinion (especially since they’re easily freeped).

    Clive — read the questionaire, it’s hardly push-polling and certainly not designed like the NOM polls to only give answers that skew in one direction… and it jibes pretty well with the latest Pew and Gallup polling on what Congress should be concentrating on: creating jobs versus ideological flag waving.

  4. Pete n SFO says

    Because only party members vote in Primaries, the party itself will constantly have to pander to the most extreme among them.

    Boner doesn’t care about marriage; he cares about generating funding & staying ahead of the media storm that would come if he didn’t defend it.

    I’m hardly sympathetic to GOP but seriously, the media would have a field-day w/ his hypocrisy would he remain silent.

  5. says

    I want to believe the results of this poll, but I’m reticent. A sample size of 800 doesn’t seem to be a large enough N to allow for generalization of the findings to the population as a whole. However, I might be wrong on that. For the statistics geeks out there, what would be considered a minimal number of respondents for a questionnaire to be considered normative?

  6. CJS says

    The key questions are not valid. For example, in the question, “I am going to read you some benefits married couples enjoy under federal law…”, using the word “enjoy” imposes positive affect (which is roughly the same as emotion) on the responder, so the responder is more likely to want to give those benefits to those who don’t “enjoy” the rights because humans are by nature somewhat altruistic. In the question as to whether Congress should focus on jobs or defend DOMA, it’s not an “either/or” issue. Congress can do both simultaneously. The responses are not an admonishment of Congress for defending DOMA per se, but rather just a ranking of the two issues, where most responders think that job creation is a more important issue than DOMA, rather than disapproval that Congress is defending DOMA. Nate Silver would go crazy if he saw this poll.

  7. Mike says


    The issue of whether the results are generalizable to the population is less an issue of sample size and more an issue of bias. If the sample was obtained in an unbiased way (i.e. a random, representative, sample of all Americans), then any sample estimate you obtain would be an unbiased estimate of the population percentage.

    For example, if you took representative samples of size n=10, n=30, or n=100,000, and obtained sample percentages against DOMA of 49%, 53%, 51%, then those estimates would be unbiased estimates of the true percentage of Americans against DOMA. The estimate from the sample of 100,000 people would clearly be much more precise than the estimate from the sample of 10.

    However, if your three samples were picked in a biased manner (i.e. not representative of the true population), then your estimates will not represent the population percentage of people against DOMA, no matter the sample size. An increase in sample size just makes your estimate more precise– it doesn’t make it correct.

    That said, my own opinion is that there is probably some degree of bias in a poll which samples registered voters and then claims it is representative of all Americans (I might think registered voters tend to be more politically motivated).

  8. Chadd says

    I remember polls that said 73% of Americans supported the repeal of DADT. That did not stop the GOP from defending it until the bitter end. Regardless of whether or not the questionnaire is flawed or whether the sampling is representative of the greater population, the GOP shows little regard for the general consensus when pushing their agenda. It stands to reason that if 73% of the population supported gay marriage, the GOP would still oppose it.

  9. NY2.0 says

    There is over 800 people living on my street, I’m sure they’re a good representation of America! Not. When I see a poll from Gallup or USAToday with these results I’ll probably bat an eye.

  10. Anthony says

    Most national polls use sample sizes of roughly 1000; 800 is not outside the norm of people polled and only carries a margin of error of about +/-3.5%. That’s why people hire polling firms (like HRC did) because you’re able to get statistically accurate information about a large group of people from a very small sample. For example, increasing the number of people polled from 800 to 10,000 would only reduce the margin of error by less than 1%

    As far as CJS’s point about whether Congress should focus on DOMA or Job, he’s right, it’s not an either/or, but if you read the pdf of the whole questionnaire (which I recommend you do before commenting) you’ll see that the person polled could choose to answer “neither,” “both,” or “don’t know,” which a few did.

    NY2.0; Gallup frequently uses sample sizes of approximately 800 people. I don’t know why people keep saying 800 is too few, it isn’t.

    @Mike; you’re correct about explaining the statistics, but if you read the questionnaire and the memo, you’ll see that the poll doesn’t claim to be representative of all Americans but rather all American VOTERS, as the pollsters asked several preliminary questions of those polled to insure they were getting a representative sample of likely voters. Again, read the questionnaire.