Some of my best friends are Sanders supporters* and lately a lot of them are touting a Quinnipiac poll that shows Bernie Sanders with a bigger lead over Donald Trump (+13) than Hillary Clinton has over Trump (+7).
The reason why they’re bringing up this poll seems obvious.
Some Democratic voters may be making their choices based, at least in part, on electability. They don’t want to vote for someone who is very unlikely to prevail in November.
But there are five problems with touting this poll.
First, polls far ahead of the general election really don’t mean very much.
As an analysis by Harry Enten shows, sometimes early polls are close to the eventual results but the average error is pretty large.
If you look at polls that tested the eventual Democratic and Republican nominees in the last two months of the year before the election, the average absolute error of the polling average is 10.6 percentage points. [source]
This can be true with an incumbent on the ballot or, like this year’s election, if there is no incumbent.
We don’t know what will happen during the campaign and in the world and nation, so predictions can’t take those into account. That’s why I generally don’t make any election forecasts, certainly not 11 months ahead of time.
Second, how Sanders does in general election polls now doesn’t account for the scrutiny and negative messages he’ll get should he be the nominee.
Thus far, Bernie Sanders has gotten through the nomination fight with not a single negative ad run against him by his opponent.
While the candidates have some policy disagreements, Sanders hasn’t been hit much at all.
His whole national record, work as Mayor as Burlington and political activities and life before his long electoral career hasn’t been picked over the way it would be.
Sanders has spent a lot of time on talk radio and tv and there is likely a wealth of comments that would get a good deal of attention if he ended up with the nomination.
A June 2015 Gallup poll found that fewer than half of Americans said they’d vote for a socialist. That, by itself, could be a problem for Sanders.
And even if you think there’s nothing Sanders could be hit with, there is always something that can be made into an issue — as Michael Dukakis and John Kerry can attest.
In contrast, Hillary Clinton has gotten an enormous amount of scrutiny over the years. It’s hard to imagine that anything at this point can erode her image.
Third, one should never pay too much attention to a single campaign poll.
One key rule of thumb for pre-election polls is that in general, single candidate match-up polls should never be given very much attention.
Sure, there are some especially good polls. The Des Moines Register poll has proven itself in Iowa, as has the Field poll in California and the Pew polls nationally.
However, even the best done poll don’t just give specific results but also a margin of error and an estimate of how likely the actual result in the population is within that margin of error. Fewer people are responding to pollsters and there are new methods, some of which have limited track records. It’s also hard to know who exactly will vote.
Fourth, the polling aggregates show that the differences between the Democratic candidates in general election match-ups are rather small and don’t all show Sanders doing better.
Let’s take the match-ups between Trump and Sanders or Clinton.
Depending on what poll just came out, sometimes Sanders is doing better vs Trump and sometimes it’s Clinton. But whichever Democrat is doing better, the differences between their leads are not big at all.
Plus different groups come up with different results because they include different polls and do different things with those included. Real Clear Politics has a straight average but it’s over an relatively longer period of time, while HuffPost Pollster does a statistical trend analysis which gives more weight to more recent polls.
At the time of this writing, the Real Clear Politics polling average has Sanders ahead of Trump by 2.0 percentage points. Clinton leads Trump in that polling average by 4.8 percentage points. So here, Hillary’s average lead is 2.8 points better than Sanders’.
In the HuffPost Pollster poll aggregate, Sanders leads Trump by 6.7 percentage points, while Clinton leads Trump by 5.5 percentage points. For this one, Sanders’ has a lead over Trump that’s 1.2 points better than Clinton’s.
Fifth, Quinnipiac polls are outliers to some extent.
You can see that by comparing the results with them and without them.
HuffPost Pollster lets readers do that easily, since you can create custom charts based on dropping certain types of polls, pollsters, etc.
As noted above, Sanders comes out 6.7 percentage points over Trump with all polls included in the HuffPost Pollster aggregate.
But look at what happens when the Quinnipiac poll is removed.
Rather than Sanders having a lead over Trump, Trump leads Sanders.
What happens when this pollster is removed from the Clinton-Trump polling average? With it included, Clinton leads Trump by 5.5 percentage points.
When Quinnipiac polls are taken out, Clinton still leads Trump, with just a little bit more of a lead.
For whatever reason, Quinnipiac is an especially good poll for Sanders, at least with this matchup.
So, is one of these Democrats more electable?
These data don’t demonstrate that one way or another.
However, I think my second point, about Sanders not having had any real scrutiny, is a real issue.
But it’s impossible to know for sure which Democrat would do better against whomever the Republicans nominate. We can’t run the election twice and compare results.
Both candidates face the same overall context. There are structural advantages Democrats have in demographic strengths and the electoral map. Moreover, Republicans have moved far to the right. However, it’s not easy for one party to win the presidency three times in a row.
In sum, Sanders’ supporters have a weaker case than they’ve made about his purported greater electoral strength — but it doesn’t make a lot of sense to focus too much on electability, one way or the other.
* Disclosure: I currently support Hillary Clinton for the Democratic nomination. For some, that will be enough to dismiss everything in this post. I encourage such people (and others) to think about and test these claims for themselves.