Monday :: Jan 7, 2008

Postscript on the Iowa Gold Standard

by eriposte

Shortly after the stats on Sen. Obama's victory in Iowa were available, I wrote this comment in response to a post by Turkana on the accuracy of the final Des Moines Register poll, which is considered the "gold standard" in Iowa polling (emphasis mine):

Funnily enough, the "gold standard" poll just got lucky if you look at the exit polls. DMR correctly projected that Obama will get about 40% of the independent vote. However, in their model, they assumed 40% of Dem primary voters would be independents. Guess what, they were *completely wrong* on this by a whopping factor of 2 (the actual % of independents in the Dem caucus was 20%). So, they dramatically overestimated the % of independents, which was my #1 concern with the DMR poll. So, how did Obama win? Well, unlike the projection of the DMR poll that he trailed Clinton among registered Democrats by 6%, he ended up even or slightly ahead of Clinton among registered Democrats (32-31%). This was a big reason why he won yesterday.

So, the DMR got close to the final numbers, but it was largely a fluke - not because their model was right! I doubt through that this will be made clear and we will go on to the next election with [everyone] thinking DMR is a genius.

Now, ARG, which was mocked for their last pre-caucus poll, has this note from Dick Bennett posted on their webpage (emphasis mine):

John Edwards should be mad that the Des Moines Register Poll was wrong

Conventional wisdom on the Internet is that the final Des Moines Register poll accurately predicted the official outcome of the Democratic caucus. When the results of the Des Moines Register poll are compared to the results of the Democratic entrance poll, however, it becomes clear that the party composition of the Register's Democratic sample was deeply flawed.

From the Register poll:

The support from non-Democrats is significant because a whopping 40 percent of those planning to attend described themselves as independent and another 5 percent as Republican.

The entrance poll shows, however, that only 24%, not the 45% in the Register's sample, of Democratic caucus participants were independents or Republicans. That is a "whopping" polling mistake.

Contrary to what the results of the Register's poll suggest, Barack Obama did not exclusively energize and turn out a wave of new caucus participants.

If the results from the Register's poll are adjusted based on party composition of the entrance poll, the Register's final poll results would be Obama 30%, Clinton 29%, and Edwards 24%. Using the final caucus results that include only the viable candidates, Obama picked up 8 percentage points of second choices (thanks in large part to the Richardson campaign), Clinton picked up about 1 percentage point, and Edwards picked up 6 percentage points. [Eriposte note: The only reason Edwards narrowly managed to get to second place is because of the second choice votes that went to him].

Obama's margin of victory in the final delegate tally had more to do with the energy generated by the non-viable candidates than it did with his own campaign.

As the campaign moves to New Hampshire and its 373,000 undeclared (independent) voters, the conventional wisdom is that only Obama will vie for the undeclared voters with John McCain. Undeclared voters in New Hampshire, representing 44% of all registered voters in the state, are not a monolithic block. It is hard to imagine, for example, anti-war undeclared voters even considering McCain.

The Obama myth fueled by the flawed Register poll sample damages John Edwards because Edwards gains significant strength from middle-aged independents. Of all the Democrats, Edwards is more apt to compete with McCain for undeclared voters and those undeclared voters could make Edwards very competitive with Obama and Clinton in New Hampshire.

This is not to say ARG should somehow be given a free pass for getting it so wrong. Rather, it is important to keep in mind that DMR's prediction, to a significant extent, was a fluke.

eriposte :: 5:19 AM :: Comments (2) :: Digg It!