If you were running for president, and you wanted to make sure that your supporters actually showed up and voted on Nov. 8, should you be warning of a “rigged election”? If you are Donald Trump, the apparent answer is yes.

Already people are dubious about whether this is actually a good strategy:

Now, new research by political scientists Adam Levine and Robin Stiles suggests that, indeed, Trump’s message is not good for mobilizing voters.

Levine and Stiles conducted simple experiments in June 2016 in partnership with the group Vote.org. The experiment went like this.

If you Googled words or phrases suggesting that you were interested in information about voter registration, one of Vote.org’s Google AdWords ads competed at auction with other advertisers. When Vote.org’s ad won the auction, it appeared above or to the right of the information that the Google search returned. We’ve all seen these types of ads many times.

Of course, people who were interested in information about registering to vote are hardly a representative sample of the country. But they are a pretty important group in elections.

People who saw one of Vote.org’s ads were then randomly assigned to see different versions of that ad. The baseline version of the ad — which functioned as the control group in each experiment — said:

Free Voter Registration
Registering is quick, easy, & free
Register to vote now!
vote.org

The experiments then tested that baseline version against an alternative message, in which the second line of the ad was changed to one of the following: “Wealthy Buying Elections,” “The System Is Rigged,” “Your Voice Is Not Yet Being Heard” or “Be Heard This Election.”

Levine and Stiles then looked to see whether respondents actually clicked on the ad. Of course, most people didn’t click the ad, no matter what version they saw. People generally don’t click on Internet ads.

Nevertheless, Levine and Stiles did find that the “system is rigged” message was less effective than the baseline “registering is quick, easy, and free” message. In that experiment, 8 percent clicked the ad when they saw the “quick, easy, and free” message, but only 5 percent did so when they saw “the system is rigged.”

Interestingly, two of the other messages were similarly ineffective, including “wealthy buying elections” and “your voice is not being heard.” Given that these messages have also been prominent in 2016 — for example, in the rhetoric of Sen. Bernie Sanders — the experiment has lessons for more candidates than Trump.

Of course, the experiment isn’t a precise analogue to the present. The ad didn’t suggest that the system was rigged against anyone in particular. And certainly Trump isn’t buying digital ads with this message — or really any ads.

But in Trump’s case, the experiment’s finding suggests that “the system is rigged” is not the message you want to be sending.