A couple of polls suggest shifts in the tide in House racesshifts in both directions.
First, in Connecticut's Fifth District, the Hartford Courant/University of Connecticut poll has Democratic challenger Chris Murphy leading 24-year Republican incumbent Nancy Johnson 46 to 42 percent. This is a reversal of previous polls, except for those of a Democratic firm, Grove Insights, whose numbers always seem to show Democrats doing very well. The poll shows Murphy winning 77 percent of Democrats and Johnson winning 83 percent of Republicans; it's unclear from the Courant's write-up whether these party descriptions are self-identifications (the case in most polls) or from party registration. The Courant says that 45 percent of respondents were independents and that they favored Murphy 45 to 36 percent. It also says that in the Farmington Valley, the affluent suburbs west of the mountain that is the western edge of West Hartford, Murphy leads 52 to 36 percent.
That last is a striking number. In 2004, when she won districtwide 60 to 38 percent, Johnson carried the five towns that I would classify as Farmington Valley (Simsbury, Farmington, Avon, Canton, and Burlington) by 65 to 34 percent (you can check my arithmetic by consulting the Connecticut secretary of state website, and here's where you can get a district map).
If you include in the Farmington Valley area industrial New Britain (Johnson's hometown) and next-door Plainville, then Johnson's margin in the area was 58 to 40 percent. That latter, larger area accounts for 25 percent of the district's votes. The Courant says the margin of error for the survey was plus or minus 4 percentage points, which suggests a sample size of about 600. That means about 125 respondents for Farmington Valley if it includes New Britain and about 100 if it doesn't. The margin of error for such sample sizes gets up to the vicinity of plus or minus 7 percent. Yet even so, this still means a big loss of percentage for Johnson in what should be a strong area for her.
My experience with polls is that results for geographical subgroups should be taken with a big grain of salt. In most polls, a region that comes in surprisingly Republican is usually compensated for by another region that comes in surprisingly Democratic. Errors tend to cancel each other out. This could have happened in this poll. Or this could have been the only geographic region that was out of whack. Or the poll could be right, and an area that stayed Republican through the Gingrich years (when Johnson nearly lost in 1996) and the Bush years has now turned decidedly Democratic. In my ratings of House races last week, I rated Johnson's seat as sure Republican. This poll would cause me to move it to lean Republican, but not to lean Democratic. For that, I'd like to see some confirmation of the Farmington Valley numbers.
On the other hand, there's another new poll that would lead me to change a seat I rated as sure Democratic to lean Democratic and maybe think about putting it in the lean Republican column on the basis of later information (as could happen in the reverse direction in Connecticut's Fifth). That poll was in Texas's 22nd District, conducted by John Zogby for the Houston Chronicle and Channel 11. This is Tom DeLay's seat, where his name has been taken off the ballot and no Republican name appears for the November contest. However, the name of Republican nominee Shelley Sekula-Gibbs appears on the ballot for the special election contest to fill DeLay's unexpired term, and Republicans are campaigning for her as a write-in candidate for the full term.
I am told that Texas has a rather liberal law governing the counting of write-ins that allows for different spellings, initials, etc., to be counted for a registered write-in candidate. The poll showed that 35 percent would vote for Sekula-Gibbs as a write-in and 36 percent for Democrat Nick Lampson, who used to represent Texas's Ninth and was defeated in 2004 as a result of the redistricting plan DeLay pushed through the Texas Legislature in 2003. When people were asked preferences without a mention of the write-in, the result was Sekula-Gibbs over Lampson 52 to 35 percent. It is interesting that the poll also had a DeLay-Lampson pairing, with DeLay ahead by 48 to 40 percentan underwhelming score in a district that voted 64 percent for George W. Bush in 2004.
In my blog last week I counted Texas's 22nd as a sure Democratic seat, as have most handicappers. But write-in candidates don't always lose. Strom Thurmond won a Senate race in 1954 in South Carolina. More recently, two write-in candidates were elected to the House in the 1980s, Joe Skeen in New Mexico's Second in 1980 and Ron Packard in California's 43rd in 1982.
Here's my New Mexico's Second account in The Almanac of American Politics 1982:
The incumbent, Harold Runnels, a conservative Democrat from Little Texas . . . was renominated with ease against a weak opponent in 1980 but died in the summer following the June primary. The Democratic state committee met to name a new nominee and decided on David King, then state finance secretary and also nephew of Gov. Bruce King. The Republicans nominated Joe Skeen, their nearly successful gubernatorial candidate in 1974 and 1978. But Skeen was ruled off the ballot, on the plausible grounds that since the Republicans never nominated a candidate in the first place there was no vacancy to fill. Meanwhile, King's nomination was greeted with scathing criticism, and the young Democrat sought to withdraw; but state election officials ruled that it was past the withdrawal deadline. Both rulings are entirely defensible, but to many it looked as though state officials were trying to elect King without opposition. Skeen decided to run as a write-in candidate, and so did Dorothy Runnels, widow of the congressman, who had lost the Democratic nomination to King.
In a district that Gerald Ford carried 52 to 47 percent four years before and that Ronald Reagan carried 60 to 35 percent that year, Skeen won with 38 percent of the vote, to 34 percent for King and 28 percent for Dorothy Runnels. Skeen won 61,564 votes and Runnels 45,343106,907 write-ins against 55,085 for King, the only candidate listed on the ballot.
Here's California's 43rd from The Almanac of American Politics 1984:
The 43rd District was a new creation, in effect half of the old 43rd, represented by Clair Burgener, who was retiring; even a Democratic redistricter as creative as Phillip Burton could not avoid drawing a new Republican district in the San Diego area, given its recent growth. The new congressman, it was confidently assumed, would be determined in the Republican primary that attracted no less than 18 entrants (which made it theoretically possible for a candidate to win with 6 percent). But it turned out to be more complicated than that.
The reason was that the winner of the Republican primary, Johnnie Crean, struck most of the other contendersand many othersas a despicable candidate. Crean spent some $500,000 of his own money [that was a lot in those days] on television advertising that implied he was the choice of President Reagan and on direct mail that charged one of his opponents, spuriously, with vote fraud. He won the primary by only 92 votes out of 83,000 cast, and the second-place finisher, Carlsbad Mayor and sometime dentist Ron Packard, ran as a write-in candidate. This was not as quixotic as it sounds: This is a highly literate district, and a lot of voters are over 65 and regularly vote on absentee ballots, on which write-ins are easy. However, it was also risky: Democrat Roy Archer tried to launch a serious campaign and must have dreamed of becoming, though presumably for only one term, the Democratic congressman from what is arguably the most Republican district in the country [it voted 71 percent to 20 percent for Ronald Reagan in 1980]. Republican officials seemed to think it was worth a gamble. Theoretically they supported Crean, but with enough winks and nudges that it was apparent that they really preferred Packard.
Packard won the election with 37 percent of the vote66,444 write-ins. Democratic nominee Archer got 32 percent and Republican nominee Crean 31 percent. This despite the fact that Packard spent only $366,000 to Crean's $1.1 million.
So the lesson seems to be that in an unusual and highly controversial situation, a write-in candidate can winat least a Republican candidate in a heavily Republican district. Perhaps I should have given more thought to that when I was rating Texas's 22nd. I don't think it's anything like a sure hold for Republicans; when you choose a write-in candidate, do you really want to choose one with a hyphenated name? But voters in the district are certainly familiar with the controversies over DeLay, and Zogby reports that two thirds of them are aware of the write-in candidacy.
The situation may be similar, though not identical, in Florida's 16th, Mark Foley's seat, where Foley's name is still on the ballot. The Republicans' actual candidate, nominated by a party committee, is state Rep. Joe Negron; votes for Foley will be counted as votes for Negron. And Republicans have come up with a clever slogan; Byron York tells the story in National Review Online.
Getting voters to understand what they need to do in the voting booth is the top priority for Negron's campaign. But it's not at all clear that the clunky "Vote the placeholder name to vote for Joe Negron" instruction will get the job done.
"That's on the website?" asks Todd Harris, the Washington-based Republican consultant who has come to Florida to help Negron. With a laugh, Harris explains that the campaign has better ways to get its message across. "The one I use is 'Punch Foley for Joe,'" Harris says. "We'll have signs made up that say that."
Harris, whom campaign trail reporters will remember as a press aide to John McCain in 2000 and Arnold Schwarzenegger in 2003, evidently has followed through, and "Punch Foley for Joe" is the theme of the campaign. In addition, Florida officials will allow signs to be placed near voting places that will state that a vote for Foley is a vote for Negron as long as they also include a statement that a vote for Democrat Tim Mahoney is a vote for Mahoney.
Polls since Foley's resignation have shown Mahoney ahead by 50 to 43 percent, 49 to 46 percent, and 48 to 41 percent. This is not as heavily a Republican seat as Texas's 22nd (or as California's 43rd was in 1982 or New Mexico's Second in 1980); it voted 54 to 46 percent for George W. Bush in 2004. But a victory for Foley/Negron does not seem utterly impossible either.
The bottom line: There's still room for movement in lots of House races. People like me can predict results in each district, but if we're playing fair and not putting lots of races in a tossup category, we're going to be changing our minds every day on which party is ahead in more than a few races.