Here are two examples of "toss up" races:

In New York-26, incumbent Republican Thomas Reynolds is running against Democrat Jack Davis. Reynolds had more than $3 million on hand vs. $63,000 for Davis as of the last FEC filing. That means Reynolds can afford to get out the vote and advertise his unredacted, unedited, and unfiltered message to the voters.

The Western New York district is historically Republican, having voted for George Bush by 12 points over John Kerry. Reynolds beat Davis by 12 points in 2004.

The downsides, according to pundits, are his stewardship of the National Republican Congressional Committee, as if everyday voters care about that. Reynolds admitted he knew about the Foley e-mails, and his top aide resigned. A poll shows Davis up by five points, but that was taken at the height of the coverage of Reynolds' relationship to the scandal. Also, the report doesn't say how "likely" voters were determined, or whether the Republican vs. Democratic breakdown of those polled was reflective of the district. Too often, it's not.

Does knowing about the Foley e-mails erase 12 points and a $3 million fundraising lead in an historically Republcan district? History tells us otherwise.

Another "toss-up" is District 13 in Florida, Katherine Harris' district. Democrat Christine Jennings has a 2-point lead, according to the latest poll. But Republican Vern Buchanan has a $3 million fundraising lead, and a two-point lead among "likely voters." The district voted for George Bush in 2000 and 2004, and of course voted for Harris, who symbolized the Republican side of the contested 2000 election.

Again, the "analysis" here seems to be the latest poll.

I'm looking for a 3- to 5-point win, at least, for the Republican in each of these races.

Here's a "leaning Rep" race: New Hampshire-2 comprises the western 2/3 of the state, but also includes much of the increasingly Democratic area in the south straddling the state line with Massachusetts. Bay State residents have been migrating north to escape taxes and high housing costs, but bring their Democratic votes with them.

However, incumbent Republican Charles Bass beat Democrat Paul Hodes by 21 points in 2004. Hodes is again the Democratic nominee. Bass' lead in a September poll was 25 points.

Republican leaning indeed.

The districts of longtime incumbents Christopher Shays and Nancy Johnson of Connecticut are considered "Republican leaning," though the two have a consistent history of beating challengers. Both won in 1998, a difficult year for northeastern Republicans who had to distance themselves from attacks on Bill Clinton, who's popular there.

Here's a "leaning Dem" race:

In Georgia-12, Democrat John Barrow beat then-incumbent Republican Max Burns by four points in 2004.

Here's the rub: Georgia-12 was redistricted by the legislature in 2005, making it more Republican. It's still a swing district, and slightly Democratic.

Georgia is a place where Democrats outside Atlanta campaign sounding like Republicans. The media likes to point out the problems northern moderate Republicans have -- "a vote for (Republican candidate) is a vote for George Bush's agenda! says the Democratic challenger. In Georgia, it's, "a vote for Barrow is a vote for Nancy Pelosi."

The pundits are mixed on this one. The Rothenberg Report -- which is predicting a Dem sweep -- calls Georgia-12 a real chance for a Republican take-back.

That sounds like a toss up to me.

I could not find an example of a case where the Times analysis seemed to irrationally disfavor a Democrat.

This has been the history of polls and punditry for the last 20 years. Rather than tell it like it is, the pundits tell it as they would like it to be.

When I predicted on my radio program in 1994 that I thought the Republicans would take the House, Michael Dukakis said, "Paul, you're crazy." The pundits thought both Gore and Kerry would win, and predicted the status quo in 2002. Instead, the Republicans took the Senate back.

On election day in 2004, a friend -- a former press secretary for a Republican congressman -- called in panic that Kerry was going to win the election, or so said the polls and pundits that morning.

Polls do tell us some things, and can indicate trends. But they've not had a good track record lately at predicting close elections. In the House, incumbency is powerful, and money is powerful. And all politics is local. That's why watershed elections are rare.

The historic trend has been tilted toward the Republican side for about 20 to 25 years. The historic trend has also been for polls and pundits not to see it.

My prediction, made pre-Foley, does now look a bit optimistic. I'm not so naive as to think the Foley shenanigans won't affect a single race. But I'm still seeing a fourth consecutive election that does not pan out for Democrats as well as they, and the media, had hoped and predicted: A hold for the Republicans -- 220 R and 215 D -- in the House.

Update: Paul's observation that there is an excess of exuberance by prognosticators concerning the lack of possible losses by Democratic incumbents is illustrated by this tidbit in the Chicago Tribune piece on Rahm Emanuel:

[Rahm Emanuel is] even willing to hint that Democrats won't lose a single incumbent, a feat that no party has accomplished in any election cycle, including the 1994 Republican takeover.