Here's Another Election Day Prediction

John Fund of the Wall Street Journal has this entry in today's Opinion Journal Political Diary:

Most election predictions from journalists aren't worth much, but for years I've been consulting a Republican demographer whose track record has typically been very good, especially regarding Congressional races. So for what it's worth, here's how he saw things late Friday:

White House: A Bush win that includes Florida, Ohio and New Mexico, but not Michigan, Pennsylvania or New Hampshire. He says Catholics, ethnic Scandinavians (especially Norwegians) and the rural vote should help Bush win Iowa and Wisconsin, and get him close in Minnesota, where the Twin Cities vote turnout for Kerry will be crucial. Overall, he says, this year's voting trends look like 2000 only more so, with cities and the older suburbs going more strongly Democratic and the smaller towns and exurbs more emphatically for George Bush. Catholics are moving toward Bush.

Senate: He sees a GOP sweep of the two Carolinas, Georgia and Florida, plus a John Thune win over Tom Daschle in South Dakota. Louisiana goes to a December runoff. Those five GOP pickups will be offset by losses in Illinois, Colorado, and probably Alaska, where Governor Frank Murkowski's nepotism in appointing his daughter Lisa may lose what should be a safe GOP seat. Tom Coburn should hold Oklahoma, thanks to Mr. Bush's big margins. That's a net GOP gain of two seats.

House: A gain of four or more in Texas, thanks to redistricting, should yield a net GOP gain nationwide of two or three seats. "I'd like to get to 230," he says, up from the current 227 to 205 split (with two GOP-leaning seats unfilled). Other than Texas, he says, the GOP should pick up a seat in northern Kentucky and maybe one each in Louisiana and Oregon. But the party is likely to lose one in Connecticut (Rob Simmons), one or two of the open seats in Pennsylvania, an open seat in Colorado (McInnis), as well as Phil Crane in Illinois. You can certainly see why Democrats hate Tom DeLay, the architect of the GOP's Texas gerrymander. Without him, they'd probably be gaining seats in the House this year.

The caveat in all of these estimates is that they don't anticipate a late-breaking trend among undecided voters for the Democrats, of the kind that happened in 2000 and surprised everybody, including my source.