Most pollsters will use random sampling to produce polls; smart analysts will look at moving poll averages and compare these to historic patterns. We can consider the first factor objective because it is quantitative. The problem is that the second factor--assessment of the data is usually too subjective.

I myself employ three variables: 1) Moving averages; (2) Comparison with other analyses, especially Sabato, Fitzgerald, Zogby and the Leip Community; (3) History and Demography. This last merits explanation. Very often, pundits will reflexively call state A "blue" or state b "red" based on historic voting patterns. These can be useful, but things do change. For example, who foresaw the reddening of West Virginia in 2000? (Karl Rove, that's who). As recently as 1988, New Hampshire was a Republican landslide state and New Jersey reliably red. Things do change. Some factors to look at include age, immigration rates and--surprise-- the two parties actual political strategies.

Let us take the three aforementioned states as examples: I have wondered: "who the hell who voted for Dukakis in 1988 voted for Bush in 2000?" Well, one answer is that there may have been older new dealers, not around in 2000. In the second instance, people do MOVE. New Hampshire has an influx of people born in other parts of New England. This can also explain the "new" swing state status of Florida, Colorado and Nevada--a chunk of progressive voters are transplants; conversely, some mountain states may attract ex-Californians who don't want to live in a multi-cultural state (a phenomenon pejoratively called "Fuhrmanization," after the ex-detective).

What do I mean by political strategies? Simple: what attracts one group may be what repels another. THIS IS ESSENTIALLY WHAT ELECTORAL POLITICS IS ABOUT. In the case of West Virginia, in 2000 the Republicans realized that there are enough cultural conservatives receptive to their message (you know, "God, guns and gays"). In WV, a big part of this was, alas, environmentalism; Gore was prtrayed as a threat to jobs. This is the strategy that took WV, Kentucky, Missouri, ArkansasAt the same time, pundits refer to Connecticut and New Jersey as blue states, and they are. But why?
I would suggest that successful people in the northeast are not immune to anti-tax allergies; when given the chance, they will often vote for "moderate" Republicans. Here comes the flip side of Rove's strategy: what works in MO or WV repels people here, who are mostly pro-choice, tolerant of gays, and not hostile to science, eg.,evolution, stem cell research; A good test of this hypothesis will be the electoral reaction to the Michael J. Fox ads.

I write this to point something out to the geniuses in the Democratic Party: ROVE already knows this. That is why he is ahead of the curve. Where from here? I am beginning to agree with the hypothesis that Democratic future gains lie in the belt from Colorado to Oregon (CO, NM, AZ, NV, OR) and NOT in the upper South. (As such, for example, we should definitely be targeting Allard in Co and Smith in OR in 2008.) As this Tuesday may show, it is time to start talking to people who may actually listen us.

Update (11/6):

11/6/06 FINAL:

SENATE:
DEM 51 REP 49

HOUSE:
DEM 231 REP 204

So Who Is Right--Sabato or Todd?

After all the turmoil, the Democrats will take both Houses on Tuesday. In the House, Dems have 16 leads and 10 more leans: I give them 2 of the remaining 7 tossups for 231. Prof Larry Sabato (www.centerforpolitics.org) gives Dems 232; he suggests that Republican turnout will leave the remaining Dem challengers "stuck on third base" (i.e, with 48-49%). Chuck Todd of the Hotline (www.hotline.com) demurs: he suggests that the Dem majority is held under 220, or they win 235+--no in-between. He reasons that if the Democrats win the aggregate popular vote by 8-10 pts, they win virtually ALL the contested races; if they win 51%-49%, they are stopped. In other words, the Dems win all three Rep seats in Conn, or none. Myself, I just do not see Dems actually winning seats in ID, KS, NE or WY; this is where the Republican turnout effort will hold (and besides, such seats would be sitting ducks in 08). For what it's worth, I see Dems winning 2 seats in CT (Farrell and Murphy), 2 in NY, 3 in PA (no pickups in NJ). But it is an interesting debate...

In the Senate, the Democrats will hold MD and NJ, and squeak by in RI, MT and yes, Virginia (but not TN). Jim Webb will win VA (or, to put it more precisely, George Allen will lose). The cliffhanger tonight willl come in--Missouri. Neither side has generated momentum, but I believe the stem cell debate will flip this to McCaskill. If there is one Republican sleeper, it may be RI: will some Democrats inside the voting booth feel a need to "excuse" Chafee? I would like to think not.

Finally, I highly recommend Ed Fitzgerald's 11/6 essay on unfutz.blogspot.com on Republican strategy. Through the ads, robocalls, voter suppression stunts et al, let us hope that the fix ain't in--that we are not fighting a Sisyphean battle against a stacked deck. That is all.