Increasing Turnout for Down Ballot Races: Findings from the 2024 Montana Public Service Commission Election

GoodPower and One Minus Beta tested a digital and mail intervention in the Montana Public Service Commission (PSC) race. The goal was to increase voter turnout and reduce “rolloff,” when people vote for president but leave local races blank. Most voters focus on top-of-ticket races and leave down ballot races blank. In fact, this rolloff effect is disproportionately larger for Democratic candidates. PSCs are important governing bodies that determine energy rates for constituents, yet they’re often unknown to voters.

To test the intervention, One Minus Beta first identified 406 precincts across PSC Districts 2, 3, and 4, covering about 475,000 registered voters. The team then paired precincts with similar characteristics and randomly assigned one precinct in each pair to treatment or control. Treatment precincts received three pieces of mail and layered digital ads about the PSC race. The materials explained the PSC’s role in setting electricity rates and shaping Montana’s future energy mix. They also highlighted the election date and early voting dates. We did not include information on candidates.

GoodPower deployed digital ads about the PSC race between October 24 and November 5, 2024. Additionally, three mail pieces went out on October 22, 25, and 28, 2024. Each mail piece included social normative language: “The last time your commission seat was up for reelection, in 2020, 97.3% of Montanans voted their full ballot.” The mail also highlighted the election date and early voting information.

Rolloff was 4.5% in control precincts and 4.3% in treatment precincts, a statistically significant difference (p<.10, one-tailed, consistent with the study’s directional hypothesis). The effect was stronger where the program reached more households: in precincts where at least 55% of households had a targeted voter, rolloff dropped by 0.3 percentage points (p<.05), a more conservative threshold than the topline result. In precincts with lower saturation, the program had no meaningful effect.

Increasing participation and shifting the vote margin are different questions, and this study answered them differently. The Democratic vote margin was -16.6 percentage points in control precincts and -17.0 in treatment precincts (p=.40), not a statistically significant difference. The program got more people to vote in the PSC race. It didn’t change who they voted for.

A few design choices likely contributed to the results. Saturation mattered: in the average treatment precinct, 67% of households received the program, and the effect was strongest where saturation exceeded 55%, high enough to detect smaller effects than earlier rolloff studies could. Layering mattered too. Most prior rolloff research tested a single mail piece; this program combined three mail pieces with digital ads. And due to legal restrictions, the program excluded any information about individual candidates, focusing solely on the PSC race itself, a constraint that may have actually helped, since it kept the message simple and race-specific. Given how hard prior research has found it to move down ballot rolloff at all, these results are promising for scaling similar work in 2026, particularly for other local, low-salience races like Public Service Commissions.