Time to ride my hobbyhorse for a while. Like Harris, which found a 36-30 Democratic lead over Republicans, Gallup has now released their information on partisan identification for 2005. Oddly, without leaners, Gallup's numbes are incompatible with those from Harris: Harris also had a large sample size and a very low MoE, The strange thing is that 33-33 and 36-30 are so far apart. Even with rounding and both MoEs, the numbers remain incompatible. I have no idea why this is the case, but it does seem to strongly indicate that there is something structural within the Gallup methodology that causes them to find more Republicans in their public surveys than other outfits. Whether this is because of the way they ask the question, the way they structure their random phone samples, or something else, I have no idea. Whatever the cause, this must be the primary reason why Gallup's polls are somewhat more pro-Bush than those of most other outfits.

Of equal interest, in this survey Gallup discusses the phenomenon of "Democratic-leaning independents" and "Republican-leaning independents." In the past, I have argued that these two groups are the primary cause for rapidly changing partisan self-identification nationwide. (My argument is that people don't actually move from identifying as Democrats to Republicans or vice versa, but that "independents" are an unstable category within partisan identification because so many of them actually have strong ties to either the Democratic or Republican party, and shift rapidly between independent identification and party identification). Gallup has an interesting chart that shows how this phenomenon can change the structure of party identification over the course of a year:

	  Bush Apprvoal    Dems     Reps
Jan-Mar       51.3	   46.1     45.9
Apr-Jun       47.5	   47.4     43.2
Jul-Sep       44.8	   48.7     41.8
Oct-Dec       40.6	   48.5     42.0

(How to read this chart: "Dems" are the average number of people who self-identifying as Democrats, plus the average number of independents who, when pushed, self-identify as Democrats. The same goes for Republicans.)

I think that this is a particularly interesting chart. Over the long term, it is probably of more use to election watchers than a straight-up Dem-Rep-Ind chart that does not push independents. After all, independents tend to not vote for independent candidates, and determining which party has the advantage in partisan self-identification when independents are included gives a better sense of the overall balance of favor within the electorate between the two parties. On that front, according to Gallup this is the largest Democratic lead since 1999, which is a very good portent for Democrats. Of course, this data also shows that the mindset of independents can change rapidly in nine months, and that Republicans actually made a very slight gain in the final three months of 2005. We clearly still have a lot of work to do before November, and there can be no letting up now.