Sunday, November 4, 2012

Romney’s Reason to Play for Pennsylvania

The Saturday before the election produced a predictably large volume of polling in battleground states — but also some predictable-seeming results, with most of the polls coming close to the average of other polls.

Because President Obama leads in the polling average in most of the swing states, this means that most of the polls there on Saturday showed him ahead as well. Among the 21 polls in battleground states on Saturday, 16 had Mr. Obama ahead as compared with just two leads for Mr. Romney; three other battleground state polls had the race tied.

Some of the consistency in these results may reflect a tendency of polls to converge or “herd” around the polling average close to Election Day. This may occur because some polling firms alter their turnout models or other aspects of the polls so as not to produce outliers — a dubious practice if the goal is to provide an objective take on the race.

At the same time, Mr. Obama’s state polls continue to show more strength than they did just after the Denver debate. As we wrote on Saturday, we are at the point where the polls would have to be biased against Mr. Romney (in a statistical sense) in order for him to win the Electoral College.

It is worth emphasizing the point once more that, for all the distractions caused by individual polls, the polling averages have been very reliable in the era of rich state polling.

In the table below, I have listed every instance in the FiveThirtyEight database since 1988 in which at least three polling firms issued likely voter polls in the final two weeks before an election. I have taken a simple average of the last poll that each firm released, similar to the Real Clear Politics method for averaging polls.

Of the 77 states with at least three late polls, the winner was called correctly in 74 cases. (I exclude Missouri in 2000, where the polling average showed an exact tie.) There has been little tendency for the state polling averages to overrate either Democrats or Republicans, or either incumbents or challengers. The state polls also performed fairly well in two years, 1996 and 2000, when the national polls were somewhat off the mark.

The chances of a miss are higher, of course, when the polls show a closer race. Even among the 33 cases where the final polling margin in a state was within five percentage points, however, the polling average identified the winner correctly in 30 cases.

The FiveThirtyEight forecast model uses a more sophisticated method than this to estimate the chance of a polling miss. For instance, it looks at how much the polls have historically missed the final margin in the election, rather than simply whether they called the winner correctly. In Ohio in 2000, for example, George W. Bush was projected to win by eight points in the polling average, but carried the state by about three points instead. That is a reasonably large error, even if it did not reverse the outcome.

Furthermore, the FiveThirtyEight estimates of the uncertainty in the polling averages are based partly on earlier years, like in 1980, in which there were few state polls but when the national polls performed poorly.
There are other characteristics of this year’s polling that make upsets less likely. The polls tend to be more accurate when there are fewer undecideds in the race, and fewer voters who say they will vote for third-party candidates. The number of undecided voters is now very small in most states, and third-party candidates are not a major factor in the race (as they are in volatile polling years like 1980 and 1992).

Instead, Mr. Obama is at about 50 percent of the vote in the polling average in Pennsylvania, Wisconsin, Nevada and Michigan; at close to 49 percent in Ohio; and at about 48 percent in Iowa, New Hampshire, Virginia and Colorado.

There are not really any recent precedents in which a candidate has led by something like 49 percent to 46 percent in the final polling average, as Mr. Obama does now in Ohio, and has wound up losing the state. That does not mean such misses cannot or will not occur: there have only been a few elections when we have had as much state polling data as we do now, which is why the model allows for the possibility of a 1980-type error based on how the national polls performed that year.

But the reasonably high level of confidence that the model expresses in Mr. Obama’s chances of winning Ohio and other states reflects the historical reality that the polling average normally does pretty well.
That brings us to Pennsylvania — where the forecast model puts Mr. Obama’s chances at better than 95 percent.

One poll of Pennsylvania on Saturday, from Susquehanna Polling and Research, showed a different result, with the two candidates tied at 47 percent. But in context, this is not such a great poll for Mr. Romney.

The polling firm has had a very strong Republican lean this cycle — about five percentage points relative to the consensus, a much larger lean than firms like Rasmussen Reports and Public Policy Polling that are often criticized for having partisan results. Susquehanna is the only pollster to have shown Mr. Romney ahead in Pennsylvania at any point in the race, as they did on one occasion in February and another in October (Mr. Romney led by four points in their previous poll of the state). Perhaps they will be proven right, but it is usually a bad bet to bank on the one poll rather than the many.

Still, Mr. Romney’s campaign is making a late play for Pennsylvania with advertising dollars and a visit there on Sunday.

That is probably a reasonable strategy, even though Mr. Romney’s chances of pulling out a victory in Pennsylvania are slim. What makes it reasonable is that Mr. Romney’s alternative paths to an Electoral College victory are not looking all that much stronger.  FULL POST

No comments: