After the slayings at the Pulse nightclub in Orlando, much speculation surfaced as to whether this massacre — or terrorist attacks in general — will help or hurt Republican presidential candidate Donald Trump. As political scientists Jennifer Merolla and Elizabeth Zechmeister noted earlier, the research is ambiguous. On the one hand:

Generally speaking, we find that terrorist threat advantages Republicans more than Democrats, in part because Republicans are traditionally perceived as better able to handle issues related to national defense.

On the other hand:

Although the public may evaluate Democratic women less favorably when terrorist threat is salient, as of 2015, Hillary Clinton is likely to be bolstered by the foreign policy experience she gained as secretary of state and by her tendency to take stands that are more hawkish than those of Bernie Sanders.

We can get modest purchase on this by examining the Clinton-Trump polls back in the fall of 2015, when the Paris and San Bernardino, Calif., attacks occurred. Below is a graph of the polling averages for each candidate, where I’ve allowed the averages to be very sensitive to possible changes in the polls — even at the risk of mistaking random bumps and wiggles for real changes:

There’s no clear story here. After the Paris attacks, Clinton’s numbers increased by a couple of points. After the San Bernardino attacks, both candidates’ poll numbers slipped a bit. Trump’s increased a few points late in December, but that was about three weeks after the attacks. It’s tough to know if there was any connection.

And this is contingent on treating these bumps and wiggles as meaningful. If we make the polling averages even a little less sensitive, there wasn’t much change at all around the two attacks:

Of course, it is too soon after the Orlando attacks to know if they will have any immediate effect on the head-to-head polls. The scale of that attack is greater, and its impact could be magnified by the fact that it happened during the general-election campaign.

But this evidence suggests that previous attacks did not have an obvious impact on voters’ support for Clinton or Trump.