First off, if you go to the Harris Interactive web page, you will note that one of the LV models Harris employs shows Bush up 51 to 43 (the one that has been reported shows Bush up 48 to 46) (more on the difference between the two models below).
This caused my heart to skip a beat - until I realized the poll was taken from October 14-17 (last Thursday through Sunday). In other words, it was taken at the height of "Marygate," when Kerry had two of his worst polling days during the entire campaign season - a fact that has been stated by Kerry campaign insiders as well as Charlie Cook. For whatever reason, Harris decided not to release this poll until today (20th).
But why the discrepancy, even amongst LV models? Well the first model, which places Kerry at 46, Bush at 48, is predicated on respondants who say they are "absolutely certain to vote" in 2004, while the second LV model includes those who stated this AND who also voted in 2000. Perhaps this explains a lot about Gallup (and Newsweek). hmmm.
Another interesting tidbit. Amongst the 16 "swing state" sub-sample, Kerry leads amongst the first category of LVs, 51 to 44. Amongst the second category, it is tied. Again, this emphasizes how important it is to vote Kerry even if you do not live in a "competitive" state. The odds Kerry is going to win the electoral vote look pretty good at this point. The popular vote, however, seems to be up in the air. For all kinds of reasons, Kerry would benefit from winning the popular vote.
In conclusion, I would expect the next Harris poll - at least amongst the "first category" of LVs to show Kerry in the lead, as it is unlikely (I pray to God) that Kerry will have similar polling days like those during "Marygate." Also, I think it is more than a 50% likelihood that Kerry is going to win the electoral vote. The popular vote, I'm not so sure - so if you support Kerry, vote for Kerry, whether you live in Oklahoma, Texas, Massachussetts, New York, Ohio, or Florida.
Ben P