Therefore.

The Anti-Marriage Ballot Initiative Fairy Tale

Once upon a time there was a man named George W. Bush who happened to be President of the United States. George wasn't a very good President, but wanted to run for reelection. In order to win reelection, he would have to somehow beat the evil Democrats. He had his bestest friend in the whole world, Karl Rove, help him so he could win.

Karl thought he could help his friend George win reelection by making sure that lots more right-wing religious people than usual came to vote on Election Day. He thought that he could get a whole bunch of states to put initiatives on the state ballots to ban marriage for gay people. Religious people don't like gay people, especially gay people who want to get married. And they also don't like evil Democrats. If he could get lots of religious people to come to vote against marriage, they would also probably decide to vote for George at the same time.

And so Election Day came and, in eleven states, lots more right-wing religious people than usual came to the polls just so they could vote against marriage. And, just like Karl planned, while they were there they also voted for George for President. So many more right-wing religious people showed up that George won. And the evil Democrats blamed the gays for wanting to get married and causing so many more religious people than usual to come to the polls to vote for George. And so George and Karl lived happily ever after.

THE END.

So that's how the fairy tale goes. And like most fairy tales, there's not a lick of truth to it.

Eleven states had anti-marriage initiatives on the ballot on Tuesday. All eleven, of course, passed. The theory goes, of course, that the presence of these initiatives on the ballot in key states like Ohio and Michigan motivated religious conservatives to come to the polls in record numbers to vote against marriage and cast a vote for Bush at no extra charge.

If there's any truth to this theory, then, in the eleven states in which anti-marriage initiatives were on the ballot, Bush should have done significantly better in 2004 than he did in 200o. And sure enough, in those eleven states, he improved his vote share over how he did in 2000.

Arkansas +3.0%
Georgia +3.3%
Kentucky +3.1%
Michigan +1.8%
Mississippi +2.2%
Montana +0.7%
North Dakota +2.2%
Ohio +1.0%
Oklahoma +5.3%
Oregon +0.8%
Utah +4.2%

But in the national popular vote, Bush improved on his vote share by 3.2% (47.9% in 2000 to 51.1% in 2004). So, in the eleven states in which anti-marriage initiatives were on the ballot, only in three (GA, OK and UT) did he outperform his overall national vote share improvement. Indeed, the presence of the anti-marriage initiative correlates with a below-average vote share gain by Bush.

Perhaps comparing each state to the national average is unfair, and a more accurate comparison would be to compare each state in which an anti-marriage initiative took place with a neighboring, politically-similar state in which there was no anti-marriage initiative on the ballot. For example, Arkansas can be compared with Missouri, Georgia with Florida, etc.

Missouri +2.9 (AR +3.0)
Florida +3.4 (GA +3.3)
Tennessee +5.7 (KY +3.1)
Wisconsin +1.5 (MI +1.8)
Alabama +6.0 (MS +2.2)
Idaho +1.2 (MT +0.7)
South Dakota -0.4 (ND +2.2)
Pennsylvania +2.0 (OH +1.0)
Texas +1.8 (OK +5.3)
Washington +1.2 (OR +0.8)
Wyoming +1.2 (UT +4.2)

In Georgia, one of three states in which Bush improved his share of the popular vote at a rate greater than he did nationally, his improvement was not out of line with Florida. And out of all eleven comparisons above, there are only five examples where Bush improved more in a state with an anti-marriage initiative on the ballot than he did in a neighboring state which did not. In two of those examples (AR-MO and MI-WI), the difference is neglible. Of the three remaining examples, the ND-SD comparison may be skewed due to the hotly contested Senate battle in South Dakota. A comparison with Nebraska (+3.0) shows that the vote for Bush in North Dakota may not have been affected significantly by the presence of the anti-marriage initiative after all.

Conversely, the KY-TN comparison might have been affected by an artificially strong performance by Gore in Tennessee in 2000. A comparison with West Virginia (+4.1) confirms that Bush's improvement in Kentucky is indeed below that which he realized in neighboring states.

That leaves two states, Oklahoma and Utah, in which the presence of an anti-marriage initiative on the ballot correlated with an improvement in Bush's vote share in that state which is greater both relative to the nation and relative to the neighboring and politically similar state selected. And just because a correlation exists, does not mean there is any causation.

It is certainly possible that the fact that the Bush administration raised the issue to the level to which did led to increased turnout among religious conservatives nationwide, which then resulted in Bush's overall improved vote share over his 2000 performance. However, one would also expect that this vote share improvement would have been particularly high in states in which the marriage issue was particularly relevant. On the contrary, there is no evidence that suggests that the strategy of putting the anti-marriage initiatives on the ballot in several states did anything to improve Bush's performance in those states.