Gallup posts new data this morning on the congressional generic ballot that deserves a look. If anything, I'm a little surprised the numbers aren't worse for the Republicans, given shifts in party identification in the past several years toward Democrats and President Bush's approval level.
A few points of perspective: First, this spread will likely close as the election approaches--particularly if the presidential contest remains highly competitive. Gallup notes:
Voter turnout typically helps the Republicans narrow any Democratic advantage seen in pre-election polls based on all registered voters. That was the case in 2004, when the Republicans trailed the Democrats by four percentage points among registered voters in Gallup's final pre-election survey, 45% to 49%, but the Republicans went on to win 47% of the national popular vote, and a 30-seat majority in Congress.
Second, the generic ballot is a good, but not perfect, predictor of House seats gained/lost. The question is highly correlated with the average total number of votes each party wins in the aggregate. But due to other variables--like the way congressional district lines are drawn--a party's average share of the national vote doesn't always neatly translate to seats gained/lost at the district level. Third, research shows the generic question is more highly correlated with House seats gained/lost in midterm elections than in presidential years.
Still, any party would rather lead than lag when it comes to this generic question. And since 1950, when Gallup first started reporting this data, Republicans have only held an advantage twice (1994 and 2002)--both years saw the GOP win more votes nationally and more House seats than did Democrats.
