Yesterday morning, political pundits were abuzz with the news that Biden's polling lead had finally vanished. Biden, Warren, and Sanders were now pretty much tied to win the nomination, each possessing nearly 20% of the vote. On its face, this was great news. Monmouth's not some fringe pollster, they have an A+ rating from FiveThirtyEight. Don't fall for the confirmation bias. One poll like this is more likely to be just a single outlying data point rather than a signal of a growing trend. This is especially true when you look at the Emerson Poll released this morning, showing Biden right back up and Warren/Sanders in the positions they were in all summer. So what was wrong? Wasn't Monmouth one of the DNCs, FiveThirtyEight, and Real Clear Politics' approved pollsters?

Why can't we rely on really good pollsters to be accurate? Just because a pollster has experience and has performed well in the past isn't enough. Various factors involved in the actual methodology of polling impacts the reliability of a poll. One of these factors is sample size. Like Rachel Bitecofer of the Wason Center in Virginia emphasized on Twitter this morning, there were just 298 people referenced in the poll. For context, other polls taken around this time involve thousands of people. Even the best pollster can't get a truly representative sample with only a couple hundred answerers, and Monmouth continued to break down the results into subcategories like age group with little more than 10 to 12 people responding. The margin-of-error of nearly 5.7% was also troubling. That's a big difference. For reference, even among the distorted representation polled by Monmouth, that could mean that Warren and Sanders only got 14% of the vote (closer to where they stand in poll aggregates) and Biden got 25% (closer to where he stands in poll aggregates). That's an extreme example, but it gives cause for warning.

Ok, so I can't rely on professional pollsters to give me a good view of who's winning. Are there any other ways to tell? The short answer is that polling really sucks this far out and we really shouldn't be trying to analyze the horse race. The media tends to treat these primaries like football drafts, which is unfortunate because there's a variety of actually meaningful stats that impact drafts while there's almost no established factors that have a high degree of impacting the primaries. History shows that polls taken further than two months out before the election aren't accurate. The annoying, but real answer is that the only thing these polls are picking up on is name recognition. The media reports that Pete Buttigeig and Elizabeth Warren aren't doing well with black voters, but that's actually just because black voters have no idea who these people are. That's really the cause of the 'so-called' black focus on electability. Black voters are diverse and don't share the same political beliefs. What the mainstream media's trying to push is the narrative that Black voters always vote for the safest choice, using Obama not catching fire until he proved he could win Iowa as a example. But this is the wrong way of thinking about this phenomenon that ties into how we're so focused on polling at this early stage. Obama got the support of black voters, AND white voters in South Carolina after he won Iowa because they had heard of him! That's just one example of media distortion based on polling, but the whole environment is a confusing mess and shouldn't be taken seriously at this stage.

Fine. We can't rely on polls. Is there anything I can use to evaluate who has the best chance of winning? Absolutely! A lot of FiveThirtyEight's staff has recently tried to establish a couple factors to be able to tell who's doing the best in the race so far. Before I go into them, it's important to decide why you want to look at the candidates this way in the first place. Are you trying to use this to decide who to vote for? Because voting for a primary candidate based on their perceived electability is a recipe for disaster. Just vote for the candidate that shares your positions! It's the primary, that's what they're there for! Don't worry about what everybody else wants to vote for. That's how we end up with unrepresentative candidates running for President and it's the reason everyone's uncomfortable with the idea of 'invisible primaries' where the Democratic and Republican elites just select the candidate they want. However, if your desire to predict who might win is from a scholarly perspective, there are a couple of indicators you can use to get a sense of who's got enough staying power to remain in the race for as long as possible. After all, a candidate actually being in the race is the strongest correlate to winning votes.

I briefly go through an overview of the candidates' campaign organizations here, but in this post I'd like to go over some extraneous factors that FiveThirtyEight and other data-driven resources are looking at when they evaluate the staying power of the candidates. The first is favorability ratings. These are useful because, as FiveThirtyEight's Nathaniel Rakich has found, it captures the reality that primary voters are comparing and contrasting several options for president, not just one. At the moment, Warren, Sanders, and Biden are roughly tied in their favorability ratings. The next feature is the number of activists that are excited about their candidacy. FiveThirtyEight's Seth Masket has been running a project where he interviews Democratic activists in early primary states. As of his most recent installment, it seems as though the most energy is concentrated around Harris, Buttigeig, Biden, and Warren. A third factor involved is the number of mentions that each candidate has in the news. FiveThirtyEight's Dhrumil Mehta's been examining these statistics, and in his most recent report, he found that while Biden's dominated the cable news mentions through the summer, Sanders and Warren have forced him to share the spotlight. The final external element to consider is who is getting the most endorsements from people within the party. This falls back into the 'invisible Primary' trope that always surrounds Presidential Primaries, which is ironic when you consider that almost every other nation on the planet is governed exclusively by people with ironclad establishment ties. This is the hardest metric to use, because the field is so large. None of the biggest players (we don't see you Obama) have ventured out in support of a single candidate yet. Most Democratic leaders, like their Republican counterparts in 2016's crowded race, are adopting a wait-and-see approach. However, Democrats have proven to be much more willing to endorse someone else from their home state. Harris, Warren, Booker, and Klobuchar have all gotten big legs up from establishment players from Senators to former Vice Presidents who come from their home states. However, Biden's really the only candidate to garner widespread establishment appeal from outside his geographic base (which should surprise absolutely nobody).

None of these is the final word on anything. As Nate Silver at FiveThirtyEight has aptly stated, all of these candidates, including Biden, can be considered an underdog. None of them has Hillary Clinton's obscene advantage in 2016, and all of them are just as equally likely to lose as they are to win. Will Biden win because he simply has the most name recognition and largest number of establishment Democrats interested in his candidacy? Will Warren win because she has the best organization on the ground and most activists interested in her candidacy? Will Sanders prevail because he has the largest donor base and most widespread appeal throughout the country?

What the Monmouth poll showed was that this race has begun to narrow in scope. That the three main competitors shown in the poll have the highest name recognition out of any of the other candidates should be your main hint to take it as just another data point at this stage in the race, and not a sign of a greater trend. There are three frontrunners, but it's anyone else's race to lose. We've got two debates coming up in September and October and a hectic media push that'll happen over Christmas and January. All of the top ten candidates that qualified for the September debates will have the staying power to remain in the race until Nevada. As a result, all of them have a pretty decent chance of winning - at least in comparison to the others.