DJ Drummond takes a close look at the fluctuation in presidential polling over the last few weeks, and digs down into the partisan identification of respondents. He discovers something curious: multiple polls show John McCain holding steady or gaining among every partisan subgroup, yet losing ground overall. The only explanation for this is that the polls have incorporated a higher percentage of Democratic voters — either accidentally or intentionally. That begs a question: what would the polls show if the partisan breakdown were tweaked to reflect the historical norm.
The partisan breakdown of polling samples is a pretty charged topic. Rasmussen has recently revised its methodology for weighting its poll, to more readily show shifts in party identification. Drummond’s point however, is that regardless of the week-to-week changes in party ID, the affiliations of those who show up at the polls doesn’t seem to change that much — at least over the last decade. In 2000, when Republicans were enthused about their nominee and eager to end the Clinton era, Republicans comprised 35% of the voters and Democrats 39%. In 2004, when both parties were highly motivated and overall turnout was the highest in recent history, the split was 37%-37%. In 2006, when Democrats were extremely motivated and Republicans were dispirited, 38% of voters were Democrats and 36% Republican. Prior to the selection of Sarah Palin, there was clearly a significant enthusiasm gap between the two parties. As things currently stand however, it’s hard to imagine that one party will have a historicallly unusual advantage on election day. It seems more likely that this contest will be decided by the independent voters. Right now John McCain seems to have an edge with this group, but election day is a long way off.