With the battle between Hillary Clinton and Bernie Sanders turning to the question of electability in November, it’s often suggested that nominating Sanders could risk a blowout on the order of Barry Goldwater’s epic 1964 loss to Lyndon Johnson or George McGovern’s epic 1972 defeat at the hands of Richard Nixon. That’s because the political science tells us that perceptions that a candidate is outside the ideological mainstream can really matter in general elections, and Sanders would seem vulnerable to being painted as such.

One political scientist who doesn’t think Sanders would necessarily face such a fate, however, is Alan Abramowitz of Emory University. That’s because he’s done a lot of work studying the rise in “negative partisanship,” i.e., the increasing tendency of voters to shape their loyalties around opposition to the other party. As Jonathan Chait noted recently, this means the electorate is more polarized and less fluid than it has previously been, and also that we have very few presidential elections to go on to judge what this trend might mean for 2016.

In an email, Abramowitz explains why he thinks Sanders might not be blown out in a general election or could even possibly win:

Although I generally agree that Sanders would be a risky candidate, I don’t think it’s obvious that he couldn’t win or that the election would end up as a blowout for a couple of reasons:
1) The big reason is polarization and negative partisanship This is also why a Trump or a Cruz nomination would not lead to a blowout loss for the GOP against Clinton. In both parties, voters are much more ideologically cohesive and dislike the opposing party much more than in the past. On the Democratic side, there are far fewer conservative voters who would prefer a Republican to even a very liberal Democrat like Sanders.
On the Republican side, there are far fewer moderate to liberal voters who would prefer a Democrat to even a very conservative Republican like Cruz (not sure how to classify Trump here). Therefore defections would likely be far smaller than in past elections like 1964 or 1972 when one party nominated a relatively extreme candidate.
2) The second reason is that the GOP may well nominate an extreme or high-risk candidate of their own in Trump or Cruz. Even Rubio is no moderate. In fact, his views are more extreme in some ways than those of any recent Republican candidate.

Other political scientists disagree with this, however. Vox talked to a number of them and found a range of views, with some fearing a Sanders nomination could cost the Democratic Party a loss, relative to a more conventional nominee, of anywhere from 2-3 to 6-10 percentage points.

If we are going to debate this, however, one hopes we’ll adhere to some general standards. For instance: polling of general election head-to-head matchups between Clinton or Sanders and the GOP candidates is borderline meaningless, since the general election voters just aren’t paying serious attention now. These numbers shouldn’t be pressed into service for either candidate’s electability case.

For another, it might not matter that Sanders is widely identified in press reports as a Democratic socialist — and thus would not be vulnerable to being portrayed as extreme — because, again, voters aren’t paying attention, and there hasn’t been any sustained national effort to frame Sanders and his policies in the eyes of the general election audience. We have no idea how they will be perceived once these things change. And it’s not enough to argue that Sanders’s general ideas are popular: we should take into account the possibility that a billion-dollar campaign to distort his policies might actually make a difference.

In other words, even if Sanders makes a good argument that middle class tax hikes to pay for single payer would be offset by savings in premiums, it can hardly be assumed that making this case to voters will be easy or that it will be a piece of cake to rebut Republican attacks on single payer as a full government takeover of a huge chunk of the economy. (Yes, Republicans said the same about Obamacare, but a transition to single payer, whatever its virtues, might actually be a good deal more disruptive.)

That’s the Clinton campaign’s apparent assumption, beyond its real substantive differences with Sanders: That there are some policy positions (middle class tax hikes; transitioning to single payer; universally free public college) that actually could become serious liabilities in a general election, even if we’d like to believe that arguing boldly for the most progressive agenda possible will necessarily prove persuasive if we only do so with enough fearlessness and conviction. We can be overly constrained by being risk-averse, of course, but that fact doesn’t mean that there aren’t actual limits on what sort of agenda can prevail in a general election or that we can wish away the possibility that the Republican attack machine might have a much better chance at succeeding if given certain fodder to work with.

On the flip side, however, the fact that Sanders excites young voters, and Clinton appears not to, is a serious and legitimate factor when weighing electability. So are Clinton’s dismal personal numbers and the possibility that the email story could take a very bad turn. And as political scientist Abramowitz notes above, facile comparisons to 1964 and 1972 aren’t sufficiently mindful of how much our politics has changed since then. Also, those who believe conventional Democratic politicians are no longer speaking to the depth of the country’s real challenges or think the only way to address them is with truly transformative bottom-up change (even if the chances of it working are deeply uncertain) might see nominating Sanders as worth the risk, particularly since (as Abramowitz also notes) the GOP could nominate an extreme candidate, which could actually result in a Sanders victory.

On electability, by all means consider the evidence and reach your own conclusions. But do consider the evidence — don’t decide what you want to believe and then make up justifications. The stakes are too high for that, and history will not forgive you.

The unfortunate truth is that we’re all flying a bit blind here. But that’s what the primary process is for — to try to sort this kind of thing out.