Last week, my focus was on residual forces in Iraq. This week, it has been on what I have termed "the inflated poll theory" for Hillary Clinton in national primary preference polls (see
here,
here,
here and
here for the best examples). The theory is simple: Hillary Clinton is
weaker stronger among all self-identified Democrats than she is among the Democrats who will actually make up the primary / caucus electorate. For all of the ink that has been spilled, and all of the papers that have been chased trying to prove or disprove this theory, the fact is that this theory can only be demonstrated with one type of test that is not currently available. In order to test this theory, in the same survey of voters, multiple polling organizations would need to include national primary preference results both for all registered voters who are self-identifying / leaning Democrats, and for all likely Democratic primary voters. If there is no noticeable difference between the two samples of voters across these polls, then my theory is wrong. If there is a noticeable difference, but it does not negatively impact Clinton's standing, then my theory is wrong. However, if there is a noticeable difference, and it does negatively impact Clinton's standing, then my theory is correct. And if my theory is correct, then pretty much every standing electoral assumption and media narrative about the current Democratic presidential nomination campaign is wrong. If Clinton's lead is accurately demonstrated to be either very slim or entirely non-existent, then this campaign would change dramatically.
To date, there has only been one poll conducted that serves as a meaningful test of my theory:
the University of Iowa poll taken from March 19-31. One of the reasons I remain confident in my theory is that this poll does in fact demonstrate my central thesis: Clinton does worse among likely Democratic caucus goers in Iowa than she does among all self-identifying Democrats in Iowa. From the poll:
Clinton led among self-identified Democratic registered voters with 29.6 percent, followed by Edwards with 26.3 percent and Obama with 20.9 percent...
Results show that Edwards remains the leader among likely Democratic caucus goers, competing primarily with Clinton for caucus support. Edwards led by a substantial margin with 34.2 percent. Clinton followed with 28.5 percent, and Obama with 19.3 percent. No other candidate reached 2 percent. Twelve and a half percent of Democratic caucus goers were undecided.
In this poll, Clinton leads by 3.3% among all self-identified Democratic registered voters (the type of sample which virtually all national primary preference polls are currently using), but she trails by 5.7% among likely Democratic primary / caucus participants (the sample I am arguing that provides a more accurate view of the campaign for the nomination). Simply put, this first data point demonstrates my thesis: Clinton does 9% better among all self-identifying Democrats than she does among the likely Democratic primary and caucus electorate. Thus, the data to date indicates that national primary polls are inflating her advantage. Considering both a 9% shift, and that Clinton's current national polling advantage is
9.7% according to Real Clear Politics, it should not be hard to see why this is so important.
Now, there are several caveats. First, this is a poll of Iowa, rather than a national poll. It is entirely possible that likely Democratic caucus-goers in Iowa are more engaged in the campaign relative to all self-identified Democrats than they are nationwide. Second, Obama does not perform better in this poll among likely caucus goers than he does among all self-identified Democrats. Third, Edwards has long shown strength in Iowa, and this poll might be picking up on that. Fourth, it is only one data point, and thus not demonstrative of anything definitive.
Right now, this is still just a theory. However, it is an important theory to test, because accurate reports on who is currently ahead in the race for the Democratic nomination, and how much that person is ahead by, are crucial to developing an informed Democratic primary and caucus electorate. Whether or not we like it, and whether or not we think it should, information of this sort has an impact on the nomination campaign. As such, it would be a disservice to the Democratic primary and caucus electorate if we did not work to make certain they had accurate information on who is winning, and by how much that person is winning. I imagine there are quite a few Republicans out there who feel the same way about their party. Preliminary evidence indicates that this is a very legitimate area of inquiry into the accuracy of most national primary polls. It needs to be tested. Hopefully, someone will step up to bat and do just that.
Update: Another professional pollster, one which recently found
Clinton in 3rd place in North Carolina (PDF), chimes in to agree with my theory.
Read here for details.
