Virginia gubernatorial nominees Ed Gillespie (R), left, and Ralph Northam (D) at a debate on Sept. 19. (Bill O’Leary/The Washington Post)

It’s the time of year when the statewide candidates start clogging mailboxes with fliers, the airwaves with ads and social media with “gotchas.”

The aim is simple: They want to influence you to vote for them.

None of it may matter.

According to a new paper from David Broockman of Stanford University and Joshua Kalla of the University of California Berkeley, some of the tools campaigns have used to reach, convince and motivate voters probably don’t work.

“Our best estimate of the direct effects of campaign contact on Americans’ candidate choices in general elections is essentially zero,” they write. “Our findings throw cold water on the notion that it is easy, overall, for campaigns to persuade voters.”

Not that it’s impossible to persuade voters. But doing so, Broockman and Kalla argue, requires campaigns to get creative with their outreach and to rigorously test the results. That requires investment and a willingness to take a few chances.

The authors also note that campaigns “should not be fooled by data collected early on in an election that suggests they can persuade meaningful numbers of voters.”

According to their research, Broockman and Kalla write, “by the time Election Day arrives, voters often forget early persuasion efforts. What is more, the tactics that seem to work early on in an election cycle usually stop working as the election draws near.”

There are caveats to their argument. They admit there isn’t enough data to show whether TV ads can really move the numbers and that “experiments have not yet tested or precisely determined the impact of candidates’ qualities, issue positions, or overall messages.”

Those would seem to be very, very big caveats.

But the larger point remains: The ability to get voters’ attention diminishes over time. And voters aren’t too open to changing their votes.

That’s an academic perspective — tested in real campaigns, yes. Yet it’s not airtight and not enough to stop campaigns from using the tools they’ve employed for decades.

What’s a Virginia gubernatorial candidate to do?

There is a special sense of urgency surrounding that question this year. The most recent batch of polling data — from Monmouth University, Christopher Newport University and Roanoke College — show the contest is close.

None of those polls shows a candidate polling at or above 50 percent. They all show a fair number of undecided voters — potentially persuadable to one candidate or the other. Keen observers will also note that none of them show Republican gubernatorial nominee Ed Gillespie ahead.

Maybe part of the answer is in the wildcard: Libertarian nominee Cliff Hyra.

Hyra polls between 3 and 5  percentage points, less than the 6.6 percent of the vote 2013 Libertarian nominee Robert Sarvis won.

Shane Cory, a one-time executive director of the Libertarian Party, once told me that third-party candidates “typically poll two times higher than their end result.”

That was certainly true for Sarvis, who reached nearly 12 percent in the final poll of the 2013 race.

Will it hold true for Hyra this time? If so, his final vote total could range from 1 percent to 3 percent. Where do his voters go? They may split roughly 50-50 for the major-party candidates. Or they may not vote at all. But both the Gillespie and Northam campaigns will try to persuade those voters to choose their candidate.

If the academic data are true, however, those efforts won’t be very effective.

What will be? If the dire tone of this piece from former Democratic staffer Brendan Lilly is true, then at least one of the campaigns — Democratic gubernatorial nominee Ralph Northam’s, to be precise — needs to get the lead out and campaign like the election is tomorrow. If not, Lilly writes, the Northam effort “is a disaster and heading for defeat in November.”

Perhaps Lilly is right. Or, if the Broockman and Kalla study is correct, none of the old methods matters all that much, and the bulk of the Election Day results are already settled.

We will get a real life test of whether they are in a few weeks.