Most Senate races don’t really have a big pivotal moment. There might be a memorable ad or a bungled debate somewhere along the way that people remember, but for the most part, they’re just characterized by two long straight lines where voters’ opinions of the candidates don’t change, or maybe two lines that very slowly converge and then cross as one candidate becomes better known and the other one wears out her welcome. (Colorado may be a good example of the first type—people seem to have decided long ago they want Cory Gardner gone—while Iowa is more an example of the second type, where it took a while for Iowans to become familiar with Theresa Greenfield or to figure out how Joni Ernst is underwhelming.)
There are a few counterexamples that people like to cite though, probably most commonly the 2006 Senate race in Virginia. Republican incumbent George Allen seemed to be well on his way to reelection, but at a mid-August public appearance he called an Indian-American tracker by a little-used racial slur, at which point perceptions of his campaign dramatically changed in the media and he wound up losing narrowly to Jim Webb. We may have had a similar thing happen in Georgia this year, where Republican incumbent David Perdue repeatedly and seemingly intentionally mangled Kamala Harris’s first name at a rally last Friday.
An analyst who tends toward the “campaign effects don’t really matter” side of political science would probably point out that, for starters, Allen’s screwup didn’t have a big impact on the race right away; Allen continued to lead in the polls all the way through mid-October, when Webb started to pull even. Webb did start to regularly come within single digits of Allen in late August, but it’s possible this was one of those races with long lines that slowly converge, and they only happened to converge right before Election Day, which might have happened anyway without any gaffes, just based on the sharp downturn in Republican prospects nationwide in October of 2006.
Whether that’s true or not, though, media coverage of Allen changed dramatically afterward, switching from a confident incumbent to one who was flailing and maybe out of step with his demographically changing state. That change in media depictions, you might say, created a permission structure where more voters in Virginia were left suddenly open to breaking with tradition and considering a Democrat like Webb.
Which brings us to Georgia in 2020, another southern state with a growing number of well-educated suburban voters that appears to be on the same precipice that Virginia was a decade ago. Perdue’s racist gaffe seems, on the face, that it would have the same effect; it’s already somewhat altered the way Perdue appears in the media, and is likely to lead to some “is this really who we want being the face of our state?” among swing voters. However … there’s one big difference here, and that’s that this incident comes in mid-October, not mid-August, so there’s much less time left for those kinds of internal conversations to take place.
But—and this goes back to my original point—it may not, in the end, matter whether this incident in Georgia happened or not, at all, in terms of the final result. Perdue’s line and the line of his Democratic challenger, Jon Ossoff, converged months ago, pretty much as soon as the Democratic primary was over. This was already an extremely close race and, given how little time remains, will continue to be an extremely close race, where the real story may simply be that Georgia’s reaching a tipping point where people of color plus college-educated white suburbanites outnumber the rest of the state.
Another factor is that there’s actually much more negative partisan polarization than there was in 2006; in other words, “breaking news” stories like this one are simply less likely to move the needle as much as they did in the ‘00s (and certainly less than they would in the 1980s or ‘90s, when many more people were willing to split tickets), because more people are firmly in the camp of one party or the other, and not persuadable by the failings of one or the other candidate because they’re already paying attention mostly to which party the candidates belong to. A big case in point is the only other “holy crap!!” story that’s popped up in a Senate race recently, and that’s the leak of adulterous-sounding texts by North Carolina challenger Cal Cunningham.
Enough time has elapsed in North Carolina that a narrative that may have seemed like a bombshell if you just woke up from a decades-long coma turned out instead to be largely a nothingburger; enough time has gone by for that story to be fully reflected in the polls, and Cunningham’s lead over Republican incumbent Thom Tillis currently six points, almost entirely unchanged from before the story. Voters already decided months ago how they feel about Tillis, and new information that speaks to one candidate’s character doesn’t really change the equation. Just as with Georgia, the real story in North Carolina is that it’s at a similar demographic tipping point where the old Republican coalition doesn’t add up to a majority any more (partly because of suburbanites moving to the Democrats; partly just because of who’s moving into the state, who’s entering the electorate, and who’s dying).
At any rate, this is all a long-winded way of saying “we don’t know yet if the polls in Georgia will have changed; you’ll need to wait a week or two to really know.” Perdue’s screwup was only a few days ago, so it’ll probably be most of a week before we see any polls where all the field dates fall after last Friday. In fact, if you look at our averages in Georgia without thinking deeper about the field dates, you might think that this incident actually benefited Perdue, in that he’s pulled narrowly back into the lead in Georgia! (When, actually, every poll in our average was completed before Friday’s incident.)
Perdue’s currently up by two points in our averages, compared with Ossoff by one at this point a week ago. Of the seven polls that came out this week, Ossoff led in three while Perdue led in four, with Perdue’s biggest lead (up 8 in a University of Georgia poll) slightly bigger than Ossoff’s biggest lead among them (up 6 in a Quinnipiac poll), so that explains the change in position. That’s all within-the-margin-of-error stuff, and if you prefer to frame it simply as a “dead heat” or a “near-tie” or whatever other media euphemism you like, please do. (Just don’t call it a “statistical tie.” That’s not a thing.)
Let’s see how the rest of the “totem pole” is doing this week:
STATE |
DEMOCRAT |
D AVG. |
REPUBLICAN |
R AVG. |
DIFF. |
FLIP? |
NEW HAMPSHIRE |
Shaheen (i) |
54 |
Messner |
39 |
+14 |
|
COLORADO |
Hickenlooper |
49 |
Gardner (i) |
41 |
+8 |
D FLIP |
MICHIGAN |
Peters (i) |
49 |
James |
41 |
+8 |
|
MINNESOTA |
Smith (i) |
46 |
Lewis |
38 |
+8 |
|
ARIZONA |
Kelly |
49 |
McSally (i) |
42 |
+7 |
D FLIP |
NORTH CAROLINA |
Cunningham |
48 |
Tillis (i) |
42 |
+6 |
D FLIP |
IOWA |
Greenfield |
48 |
Ernst (i) |
44 |
+4 |
D FLIP |
MAINE |
Gideon |
46 |
Collins (i) |
42 |
+4 |
D FLIP |
|
|
|
|
|
|
|
KANSAS |
Bollier |
43 |
Marshall |
43 |
0 |
|
SOUTH CAROLINA |
Harrison |
46 |
Graham (i) |
46 |
0 |
|
GEORGIA |
Ossoff |
44 |
Perdue |
46 |
-2 |
|
MISSISSIPPI |
Espy |
42 |
Hyde-Smith (i) |
44 |
-2 |
|
MONTANA |
Bullock |
46 |
Daines (i) |
48 |
-2 |
|
ALASKA |
Gross |
43 |
Sullivan (i) |
46 |
-3 |
|
TEXAS |
Hegar |
42 |
Cornyn (i) |
46 |
-4 |
|
ALABAMA |
Jones (i) |
40 |
Tuberville |
52 |
-12 |
R FLIP |
KENTUCKY |
McGrath |
39 |
McConnell (i) |
51 |
-12 |
|
One other race that was giving Democrats heartburn in the last couple weeks is Michigan, where Gary Peters is probably the most endangered Democratic incumbent (well, except for Doug Jones in Alabama, who’s more on the “possibly extinct” tier). This week, Peters is up to an eight-point lead on average over his Republican opponent, John James. And that’s despite a poll from Siena for the New York Times that put Peters’ lead at only one point; that was counteracted by a variety of polls from competent but less widely touted pollsters like Ipsos and Morning Consult that found the race in the high single digits.
Don’t get too comfortable, though, because another Trafalgar poll that showed up on Sunday will soon be in our database and that’ll push his average back down again somewhat. Again, that helps reinforce my frequent nagging that if there is a shift in polling in a race, often it’s not because of something big happening in the race, but rather is just about which pollsters showed up that week, and what their particular models and house biases look like.