When Numbers Lie: Looking Beyond the Current Polling Data

The presidential primary race isn’t about receiving the most media attention or donor dollars, or even about winning early primaries and caucuses that are so often declared to be crucial by the news media. This is especially true in light of the large fields that have contested Republican primaries in recent election cycles. For all the attention over swings in the polls and debate performances, it’s important to remember that polling numbers now, still early in the election cycle, are not indicative of eventual primary success. Case in point, the polling averages in the 2012 Republican primary showed eight different lead changes between five different candidates, and eventual winner Mitt Romney did not even solidify his lead in the polls until March, well after the primaries had already started.

For as much media attention as the first states receive, they simply aren’t that indicative of who will eventually win the national primary race. Mitt Romney didn’t win the Iowa caucuses in 2012. McCain got 4th in Iowa in 2008. Going back all the way to 1980, only two winners of the Republican Iowa caucuses in contested primaries have actually gone on to win the primary: Bob Dole in 1996 and George Bush in 2000.

Early wins are important to demonstrate the legitimacy of one’s campaign, as is early media attention, but organization, money and party support mean a lot more. Following that logic, even though many polls and the shifting leads they show are highlighted in the news at this time of the election season, it’s important to look beyond the fluctuating polling data and instead focus on longer-term fundamentals when making election predictions.

One way to compare current polls with long-term expectations is by examining the prediction markets. Polls show how people would vote if the race were held today, but online prediction markets allow people to bet on who they expect to ultimately triumph in the primaries. PredictWise takes market data from PredictIt, Betfair and bookie data, aggregates it and derives a likelihood of winning from that data. While not a perfect predictor of a candidate’s probability of winning, a side by side look at current polling data versus the PredictWise probabilities demonstrates the effects of long-term thinking in primary analysis.

The difference between the current polling data and prediction markets is obvious. Outsider candidates like Trump and Carson may enjoy high polling support now, but their long term path to electoral success is rockier than more established candidates, like Bush and Rubio, who see large increases in their prediction market percentages compared to their current polling data.

Another key thing to point out is that the “leads” that candidates like Carson and Trump have had in recent polling are not even close to majorities. These candidates currently appeal to larger percentages of the electorate than anyone else, but that’s also because the many establishment candidates still in the race are splitting the vote. Hypothetical head-to-head polls show that even though more establishment candidates are behind Carson and Trump in the polls right now, they will benefit once the field starts narrowing. For example, even though Trump is still well ahead of candidates like Rubio and Fiorina when all candidates are included in the polls, he loses to them in head-to-head matchups.

Carson is high in the polls and making a great deal of money through fundraising, but his use of campaign funds so far has called into question his campaign’s ability to handle money. Carson has achieved his high fundraising totals largely because he has been spending large sums of his current campaign cache to acquire more campaign funds – a whopping 54% of funds he has raised have been poured right back into fundraising efforts. The result of this oddly cyclical spending is that the Carson campaign has spent a great deal of its war chest to get to where he currently is.

This might not be a bad strategy, if Carson had unlimited funds, but the most recent filing of PAC and Super PAC funds shows that Carson lacks the critical support needed from outside groups to raise large amounts of campaign dollars. Compared to other major candidates (ignoring Trump, who has refused support from Super PACs but also has the personal wealth to finance his own campaign), Carson’s PAC support barely even registers.

So yes, Carson has been able to raise a lot of money for his campaign, but his high spending rate and lack of PAC support could make funds tight down the road as the competition heats up. Carson’s campaign would do well to focus on the long term, and perhaps be willing to accept a few less percentage points of support now in exchange for having plenty of campaign money when delegates actually start being awarded a few months from now.

A recent CBS/NYT poll garnered a great deal for showing Carson having passed Trump in national polling. What didn’t make the headlines was that 71% of those sampled in the poll may have a preference for a candidate, but also said they had not yet made up their mind for sure. The message is clear: polls at this stage of the race simply don’t serve as viable predictions of ultimate electoral success.  Even us quants here at Electoral Statistics realize that sometimes you have to toss the polling data aside and look at the fundamentals of the race.  The candidates and news media would be wise to do the same.