Photo finish?

For months the narrative of Nashville’s mayoral race has revolved around a gaggle of hopefuls with a legitimate shot at making the two-candidate runoff, but little consensus among journalists, pundits, and assorted political obsessives on which two it will be. Now, with just a week until election day, the picture is coming into focus. Sort of.

A few key factors make this open race especially challenging to handicap. One is the presence of a sizable field of mostly experienced individuals who have raised sufficient money to be taken seriously. Another is a dearth of viewpoint diversity across them coupled with the usual reluctance to risk causing offense by drawing attention to differences in a runoff-bound election. A third is a shortage of good public polling offering data-based snapshots of the race. And, when there are polls, the rather hefty share of undecided voters blurs their meaning.

Poll shows just over one week from Nashville mayoral race, O’Connell leads pack

It is tempting to over-interpret a recent poll that emerged earlier this week, commissioned by the Tennessee Laborers PAC and conducted by respected DC firm GBAO, because it looks to be fairly consistent with a few not-publicly-reported polls that insiders have been chattering about the last few weeks. Also, from the Department of Unwarranted Superstition, comes the factoid that in Nashville’s last open mayor’s race in 2015, this same PAC hired this same firm to do a poll at the same time, with results that nailed the entire actual order of finish and came pretty close on spreads. 

This new poll has Freddie O’Connell leading at 21% with Alice Rolli, Jeff Yarbro, and Matt Wiltshire (in that order) clustered in a low-teens scrum for second. But if you creatively factor in margin-of-error fantasies (plus or minus 4.4% on any given number in this poll) then almost everyone conceptually stays in the hunt, or at least persuades themselves that they do. 

Although the last few election cycles have made a lot of people cynical about polls, the basic laws of sampling and probability in social science have not been repealed. And so polls remain informative in races with lots of them we can aggregate. Unfortunately this isn’t that sort of race, but it is worth looking at the sequence and trajectory of the few public polls (meaning polls made public by their sponsors) we have had for some clues as to how this race has unfolded and where it seems headed down the home stretch. 

I’ve compiled the polls since March in this chart, with the candidates ordered by result of the mid-July GBAO poll. The overall sequence tells a few key stories about the race.

First, it shows pretty clearly how early advantages in name recognition eventually give way to the reality of the campaigns themselves. The first two polls essentially proxied name ID, with the state senators Jeff Yarbro and Heidi Campbell out front, other electeds (Freddie O’Connell, Sharon Hurt, and Vivian Wilhoite) clustered behind them, and the newcomers (Matt Wiltshire, Alice Rolli, and Jim Gingrich) playing catch up. Then, as spring veered toward summer, the name-ID edges softened, revealing in June the excruciating tightness that has defined much of the race. By July we see some separation opening up as the proportion of undecideds finally dwindles from massive to merely substantial. 

Second, we can see that two of the newcomers, Wiltshire and Rolli, elevated themselves from longshots into contenders, though they did so in different ways. Wiltshire’s approach has involved the brute force of massive fundraising and spending, especially on TV and radio, plus lots of digital and field. Rolli has raised far less and spent far less, but as the lone Republican in the field she is targeting and reaching voters in that lane, and she also benefits from a PAC buying lots of TV on her behalf. The fact that

Wiltshire has been able to spend his way into contention whereas Jim Gingrich could not (as Linda Rebrovick could not eight years ago) is a reminder that for Nashville voters relevant public service experience matters; a fancy pedigree and resume (and bank account) are not enough.

Third, the numbers offer some hints about how all those undecideds will break. In the rightmost column of the chart I show change from the PPP poll in early June to the GBAO poll last week. It is statistically hazardous to infer much in the way of shifts based on difference scores from just two polls, but let’s do it anyway.

Barring late-race developments or surprises (“but her emails!”) one would normally expect undecideds to break in a way that more or less parallels where decided voters are. That seems to be the case, here, with O’Connell and Rolli showing more gain in the last six weeks than the rest. It may not be a coincidence that they are arguably the two with the clearest ideological lanes to themselves, offering undecided voters a pair of relatively distinct alternatives.

And fourth, it is intriguing to ponder the fate of the two state senators Yarbro and Campbell here in the final drive to the finish. Their initial advantages in name ID were not enough to sustain them for the long haul, and while they raised similar amounts of money, at the end we now find them waging two very different sorts of campaigns. 

Yarbro’s team has produced a slick set of upbeat TV ads. With a fairly austere approach to campaign staffing and spending all along, they have saved plenty of scratch for a final advertising push.

For the last two weeks the Yarbro campaign is flooding local broadcast airwaves with ad spending that comes close to matching Wiltshire’s endgame blitz, though Wiltshire is also dropping copious numbers of dimes on cable and radio. If they buy into gathering conventional wisdom that O’Connell will snag one of the runoff spots, then Yarbro and Wiltshire find themselves battling mano a mano with each other and with Rolli for that precious second spot. All three are all over TV here at the end (though for Rolli it’s mostly a PAC created by a right-wing hedge fund guy spending on her behalf). 

The Campbell campaign, in contrast, is doing no TV. I don’t mean very little TV. I mean literally zero. She did release an odd sort of ad in early July that featured her as a member of a band, but her campaign spent a paltry $9,800 to air it minimally, and that was only during the week ending July 13. Since then, nothing on the air. I asked campaign manager Cyrus Shick if a last minute TV buy might be in the works, but he confirmed they are all in with field and digital: “Our campaign is focused on communicating directly with voters in ways that are more efficient and trackable than broadcast and cable.” His theory is given a large field in a low-turnout race, “it makes the most sense for us to be focused on serving our ads directly to the specific voters we feel need to see them most.” 

Certainly the way political campaigns do media has evolved in big ways over the last decade, and the way many humans consume television has changed dramatically. But chronic voters who show up for low-turnout races may not ‘evolve’ quite so rapidly. In campaigns, TV does still dominate and does still work, especially in local races where voters don’t know much about candidates and in close races where voters are persuadable. I’m not saying that Campbell can’t possibly finish in the top two in a big city mayoral race with an almost exclusive focus on digital and field, but if she does pull this off in what looks like a come-from-behind situation it will be one they talk about in campaign consultant school. 

Lastly, with less money raised and less to spend, the polling data suggests that Sharon Hurt has struggled to expand her base much beyond where it stood at the start of the race. Hurt has ramped up spending on TV for the last weeks of the campaign and has been doing a lot of urban radio since early June. Hurt’s campaign has been energetic and honorable throughout, but it feels like perhaps too little too late. 

I’m not one for making predictions, and I’m bad at it anyway. So I’ll go with the one we can make with confidence: When the polls close next Thursday undecideds will no longer be in charge.