The View From Here . . .

As attention turns more closely to this tightly contested presidential election, a number of seemingly minor issues loom large in an attempt to get a handle on the race.

The first mini-issue concerns partisan allocation of respondents in opinion polls. Basically, pollsters and their analysts debate whether polls should be automatically adjusted to reflect a fixed number of respondents from each party.

It is well known that people who identify themselves to pollsters with the Republican or Democratic party heavily (in the 80-90 percent range) favor the candidates of that party. Thus, oversampling of partisans of either party has the strong potential to skew poll results. Those pollsters, such as John Zogby and Scott Rasmussen, who normalize by parties seem to use models based on the 2000 and 1996 presidential election exit polls, which each showed approximately 39% self-identified Democrats and 35% self-identified Republicans. On the other hand, pollsters who reject this approach claim that changes in party affiliation in effect float over time and should be taken into account in each poll. These pollsters normalize their sample for more definitive demographic characteristics such as age, sex, race, and geographical region but argue that imposing a fixed party ratio does not adequately capture currents of public opinion.

Another mini-issue is the relative importance of the electoral and popular vote in a presidential race.

Some analysts tend to look heavily to the electoral college, pointing out quite correctly that under the Constitution, the election is normally decided by the 538 electors from each state (plus Washington, DC) who cast their votes in December.

On the other hand, in any recent election decided by a decent margin (say, 3 points or more), the electoral college almost always magnifies, rather than contradicts, the popular vote. Take two fairly close elections in recent history - George H.W. Bush's victory in 1988 and Bill Clinton's triumph in 1992. In both instances, the candidates, who won the popular vote by 8 and 5 percent respectively, won decisive electoral vote victories (426-111 and 370-168). In only one election since 1888 -- the last one in 2000-- has a popular vote loser won the electoral college

However, since there may in fact be a very close popular vote, electoral vote strategists are hard at work. Unless the race suddenly blows open at the end, it seems unlikely that that many states will actually change from their vote in 2000. Indeed, with Mr. Bush looking increasingly strong in Florida (and Mr. Kerry ahead in Pennsylvania), the election may come down to one big battleground state -- Ohio, with 20 electoral votes. If the President can hold this state, which basically is a tossup in local polls, it is difficult to come up with a plausible scenario for a Kerry win. If the senator pulls it out in the Buckeye State, however, it likely will be a very, very long night, with great interest in a few smaller battleground states -- Wisconsin, Iowa, New Hampshire, New Mexico and Nevada.

Finally, there is much interest in the effect of a likely larger turnout this year. Registration figures are basically up across the board, but the great question is how many new voters will actually vote, and for whom. The latest Zogby poll indicates that these new voters are largely split (49-43% for Kerry), which would tend to indicate a pretty negligible impact on the final result. However, if they were to vote in huge numbers or move sharply in one direction, it would provide yet another wild card in this tense election.