Some numbers crunching

I seem to notice a pattern emerging from poll results for Senate and House races. Republicans are doing better, in comparison with the Bush 2004 performance, in the Northeast and big metropolitan areas, than they are in states and districts with substantial rural and small-town populations. It looks as if the big margins Bush won in rural and small-town counties–margins that were essential to his wins in states like Ohio and Missouri–are not there, at least not yet, for Republican candidates with serious Democratic opponents. But Republicans in the Northeast and perhaps in other big metro areas (as in the two seriously contested House races in the Chicago suburbs) are running even with or ahead of Bush's showing.

Examples:

• New Jersey, where Tom Kean Jr. is currently running ahead of Democratic appointed incumbent Bob Menendez.
• Maryland, where Republican Michael Steele is running about even with Democratic nominee Ben Cardin, who won his primary narrowly over Kweisi Mfume; Steele and Mfume are black.
• Bush lost both New Jersey (46 to 53 percent) and Maryland (43 to 56 percent).
• Ohio, where Republican two-term incumbent Mike DeWine is behind Democratic Rep. Sherrod Brown.
• Missouri, where incumbent Jim Talent is running essentially even with Democrat Claire McCaskill.
• Bush carried both Ohio (51 to 49 percent) and Missouri (53 to 46 percent).
• Then there's Montana, where three-term Republican incumbent Conrad Burns, carrying the baggage of large contributions by Jack Abramoff clients, is trailing Democratic state Sen. Jon Tester; Bush carried Montana (59 to 39 percent).

I think I see a similar pattern in House races. Recent polls have shown Republican incumbents behind in Indiana's Second, Indiana's Eighth, Indiana's Ninth, Kentucky's Fourth, and Pennsylvania's 10th, all districts Bush carried by solid margins. Also, polls show the Republican incumbent only barely ahead in Colorado's Fourth. These are all largely outside large metropolitan areas, and the metro areas included in some (Louisville, Cincinnati) did not trend Democratic as most very large metro areas did in the 1990s. But recent polls have also shown Republican incumbents ahead in Connecticut's Fourth and Fifth and in Pennsylvania's Sixth, Seventh, and Eighth. Bush lost all these districts, though several by only narrow margins. And two polls have shown the Republican only narrowly behind in the race for the at-large seat in Vermont, which gave John Kerry his third-highest percentage of any state.

To be sure, each of these races has its own dynamic. There aren't many publicly available poll results in House races, and many of those are conducted by Democratic or Republican firms and so must be used with caution. But taken together, they raise the question of whether we're seeing a change in the contours of partisan support that largely prevailed from 1996 to 2004. Rural areas seem to be giving smaller margins to Republican candidates (maybe they'd still give large margins to a Republican presidential candidate, but that office isn't up this year). And major metro-area districts in the Northeast seem to be giving larger percentages to Republican candidates. Remember that these metro areas were not always as Democratic as they were in 1996–2004. All these states and districts (the boundary lines have changed, but I think I can interpolate the results here) voted for Ronald Reagan in 1980 and 1984 and for George H.W. Bush in 1988. They moved toward Democrats after the Republicans' big victory in 1994, as Bill Clinton skillfully emphasized his liberal stands on cultural issues and as the threat of Democratic tax increases vanished. Now are they moving in the other direction? An interesting question.

Let's look at some of the races more specifically. Start with the Maryland Senate race, where the interesting question is whether Republican Michael Steele can make inroads among ordinarily 9-to-1 Democratic black voters; I've blogged about this before. The most recent poll, conducted by SurveyUSA, indicates he already has. SurveyUSA shows Steele ahead by 48 to 47 percent–a statistical tie. It shows Steele carrying white voters 54 to 41 percent, almost exactly the same as the 55-to-44 percent lead for George W. Bush in the 2004 exit poll, and Steele winning 33 percent among blacks, compared with 66 percent for Cardin. By way of comparison, blacks voted 11 percent for Bush and 89 percent for Kerry. If blacks had voted 33 percent for Bush, he would have lost to Kerry by a margin of something like 48.2 to 50.6 percent–as close as Michigan or Pennsylvania or Minnesota. And if Maryland had been that close all along, it would have been a target state for the Republicans' get-out-the-vote efforts, which means the Bush percentage would probably have been a little higher and might have been higher than Kerry's.

The regional breakdowns in Survey USA enable us to make the following comparison between Steele's showing in that poll and Bush's performance in the exit poll.

Most voters in Baltimore and in Prince George's County in the Washington suburbs are black; in the September 12 Democratic primary, Kweisi Mfume got 63 percent of the vote in Baltimore and 70 percent in Prince George's County. There are also a fair number of blacks in the Baltimore suburbs and in the other Washington suburban county, Montgomery County. This poll shows that Steele is running well ahead of Bush among blacks but behind among rural and small-town voters. Cardin is surprisingly weak in his home base, the Baltimore suburbs. Of course, this is just one poll, and it may be wrong. But it's interesting.

Let's look now at the New Jersey Senate race. Menendez has a real problem; he's being investigated by the United States attorney. And this is one of the few states where Republicans got the challenger candidate they wanted, Tom Kean Jr., well known as the son and namesake of the only undeniably politically popular Republican governor in the state since the 1940s and one whose record is unbesmirched by scandal. Kean was ahead 48 to 45 percent in a Quinnipiac poll (September 13–18) in which George W. Bush had a 33-to-64 percent negative job approval. From Quinnipiac we can get the following geographic breakdown—into Shore (presumably Atlantic, Cape May, Cumberland, Monmouth, and Ocean counties), Philly land (presumably Burlington, Camden, Gloucester, Mercer, and Salem counties), and New York land (presumably the other 11 counties, which cast 58 percent of the state's votes in 2004)—and compare the results with the 2004 election.

Kean, in this poll anyway, is underperforming in the South Jersey suburbs of Philadelphia and overperforming on the Shore.

Rasmussen (August 28) has it Kean 44 to 39 percent, with interesting economic breakdowns: Under $40,000 voters are heavily for Menendez, voters from $40,000 to $100,000 are heavily for Kean, and voters over $100,000 are split 44 to 44 percent. Protestants are evenly split, Roman Catholics 58 to 28 percent for Kean, and others (Jews, seculars, etc.) 46 to 33 percent for Menendez. Fairleigh Dickinson (August 21–27) has a nearly identical result, Kean 43 to 39 percent. New Jersey is a low-informational state, in which voters tend to know relatively little about candidates and issues; the major local television stations are all based out of state, in New York City and Philadelphia.

Now let's look at Ohio, where four September polls have shown Mike DeWine behind Sherrod Brown: SurveyUSA (52 to 42 percent), Quinnipiac (45 to 44), University of Cincinnati (51 to 47), and Rasmussen (47 to 41). SurveyUSA has DeWine leading by only 50 to 42 in the West, losing 53 to 43 in Central, and getting walloped 59 to 35 in the East. I don't know the boundaries of these regions, but it seems clear he is running far behind Bush's 51 percent showing in 2004 in each of them. The University of Cincinnati gives regional breakdowns, but they look fishy to me, as regional breakdowns sometimes do (the margin of error for any part of the state is significantly larger than for the state as a whole). They show Brown leading 53 to 45 in Northeast Ohio (I would have thought he would lead by more) and 61 to 39 in Northwest Ohio (implausible, since Democratic Toledo is surrounded and outvoted by much more Republican territory). Central Ohio is 53 to 45, and Southwest Ohio is 54 to 43 for DeWine–plausible, but disappointing results for a Republican. Southeast Ohio, which is flagged for having a small sample, is 58 to 34 for Brown. These regional results give some small support for the hypothesis advanced above, that Republicans are running behind Bush 2004 in rural and small-town areas, while running about even with him in big metro areas (the biggest of which in Ohio is Cleveland in Northeast Ohio).

This is confirmed as well by this SurveyUSA poll on the Missouri race between Republican incumbent Jim Talent and Democrat Claire McCaskill. The poll shows McCaskill ahead 47 to 46 in what has been the tightest Senate race for the longest time in this cycle. In urban areas, McCaskill leads 56 to 39; Talent leads 48 to 47 in the suburbs and 52 to 42 in rural areas. That's a weak showing in the suburbs for Talent, who is from suburban St. Louis County, but remember that most blacks in metro St. Louis live not in population-declining St. Louis City but in the separate St. Louis County. Compare that with the 2004 exit poll, while noting that the definitions don't seem to be the same. That shows George W. Bush with 30 percent in urban areas (well below Talent, who has paid a lot of attention to inner-city issues), 52 percent in the suburbs (Talent is a little below that, but not much), and 66 percent in rural areas (far above Talent's 52 percent).

If it's true that Republicans are running behind Bush 2004 in rural areas but not in big metro areas, the question is: Why? I have only tentative answers to what I advance as a hypothesis. One might be disparate impact of the war on terrorism: Rural areas produce a disproportionate number of men and women for the military and may be concentrating on casualties, while people in big metro areas may see them as terrorist targets. Another is economic: People in rural areas aren't necessarily doing as well as most people in major metro areas. Or is it because there's a declining concentration on the two major political figures–Bill Clinton and George W. Bush–who happen to have personal characteristics that people on the other side of the cultural divide (affluent suburbanites; rural and small-town Wal-Mart shoppers) absolutely loathe?

Anyway, I'm not sure there really is a trend–maybe the rural areas will come home to the Republicans in the next seven weeks; maybe metro-area states and districts will vote to oust Democrats–and so I'm even less sure about assigning reasons. But I do think that we may very well see such a trend in 2008, if Rudy Giuliani or John McCain is nominated for president, as I argued in this recent U.S. News column. And perhaps if Mitt Romney is nominated; he's a pretty metropolitan kind of guy, living in metro Boston and having been raised in metro Detroit (where he was three years behind me at Cranbrook School in Bloomfield Hills; I have a hard time thinking of him as someone who's older than 14). Romney has thus lived in the Midwest, East, and West (when he was in college and later when he was cleaning up the Winter Olympics in Utah); McCain has, I suspect, lived in every part of the United States when he was in military service, and he was born in the Canal Zone; only Giuliani is decidedly parochial, having lived in New York City and Long Island all his life, though we did all get to know him on September 11.