As even the casual political bystander has learned during the past two months of the primary season, the influx of political polling in Campaign 2008 has given pollsters more and more opportunities...to miss the mark. New Hampshire and California have been the two biggest black eyes on survey research firms to date. While no poll is a predictor of outcomes per se, the degree to which pollsters in both of those states underestimated the Clinton vote and overestimated the Obama vote (and overestimated the Romney vote in California) does not inspire confidence among interested observers about what will happen on March 4th.
Tuesday may indeed be the last big test for pollsters before the General Election in November, with scores of polls having been released in Ohio and Texas during the past two weeks. The most recent results from these surveys once again demonstrate that pollsters are having a high degree of difficulty in determining who is a 'likely voter'—as the results they have generated vary wildly.
For example, take the Democratic primary in Ohio. One poll released today has Barack Obama up by two points (Reuters/C-SPAN/Zogby), while another has Clinton up by 12 points (Suffolk). Still another survey released a few days ago (Columbus Dispatch) had Clinton up by a shocking 16 points, although that poll was based on registered voters (with a very large sample—2,308). Most polls however, show Clinton with a lead within the margin of error up to 9 points. Since only 1 poll (the Reuters/C-SPAN-Zogby survey) has ever shown Obama ahead in the Buckeye State, the Clinton campaign should be optimistic about their chances there on Tuesday.
In Texas, the Democratic race is tight, but on the Republican side there have been equally curious polling results. American Research Group (ARG) conducted a poll on February 23-24 of 600 likely voters that gave McCain a 45 to 41 percent lead over Huckabee. But, in their next poll conducted just 3 days later, McCain somehow opened up a 62 to 23 percent lead—a 35-point swing! While Huckabee's campaign has not been relevant to the political process for a few weeks now, he did not commit any major campaign blunder to effect such a mammoth change among the Republican electorate. Such a turnaround in such a short time frame is just about unprecedented in political polling.
Polling, to be sure, is a tricky science—even more so during primary season when voter turnout is very unpredictable. We will see whether March 4th becomes the 3rd strike this season for pollsters or whether they 'got it right.'