In 2024, Civitech partnered with the Idaho Democrats to bring thousands of new Democratic voters into the electoral process. While Idaho is not typically considered a battleground state, the Idaho Democrats recognized the potential to engage underrepresented communities and turned to Civitech to make it happen. With data-driven strategies and precise execution, Civitech transformed a bold idea into measurable results, proving that voter registration efforts can thrive even in the toughest environments.
Background
Civitech ran three distinct voter registration programs focused on adding new Democratic voters to the Idaho voter file over the course of the 2024 election cycle. While Idaho is certainly not typically considered to be a competitive state overall, Civitech is committed to helping our partners reach and turn out folks typically left out of the electoral process.
Civitech sources unregistered voters using two main methods:
Pinpointing people who had been registered to vote, moved addresses, and did not update their registrations
Collecting public information records from universities, licensing bureaus, state agencies, etc.; standardizing the data; and matching against the voter file
We then apply our proprietary, third-party-validated partisan models to narrow down our outreach universes.
Methodology
RCT Experiment 1: June VR mail + three SMS chase attempts (where possible)
19,508 unregistered likely Democrats in Idaho Legislative Districts 6, 15, 16, 17, 18, 19, 26, and 29 received a pre-filled registration form in the mail along with an insert explaining the program and a pre-stamped, pre-addressed return envelope to send it back to the SOS. Those with known phone numbers received three rounds of SMS chase following delivery. These districts were chosen by our partner because they appeared likely bluer than the state as a whole.
A control group of 977 (approximately 5% of total eligible targets) was held out at the time of universe creation.
RCT Experiment 2: October VR postcard + three SMS chase attempts (where possible)
10,674 unregistered likely Democrats in the same Idaho Legislative Districts (6, 15, 16, 17, 18, 19, 26, and 29) received postcards with voter registration instructions with a QR code for the Idaho online voter registration (OVR) portal. They were also informed that same-day registration is available during Early Vote. Those with known phone numbers received three rounds of SMS chase following postcard delivery.
A control group of 535 (approximately 5% of total eligible targets) was held out at the time of universe creation.
Observational Experiment 3: October VR SMS only
9,531 unregistered likely Democrats in all the other untargeted, typically redder, districts with known phone numbers received three rounds of SMS outreach including a link to the Idaho OVR portal and information educating them about same-day registration options during Early Vote.
A control group was not held out for this test, as our partner wanted to leave no stone unturned given the perceived state of the race. As an imperfect proxy, we compared results to observed voter registration behavior for the untargeted known unregistered folks with known phone numbers in these districts, all of whom modeled as likely Republicans (N = 36,270).
Results
Experiment 1:
Using a November Idaho voter file (exact vintage: 11-18-2024), we observe 3,906 successful registrations in our treatment group - a success rate of 20.02%. We observe 171 successful registrations in our control group - a success rate of 17.5%. This amounts to a raw observed 2.52 point lift, or a 14.4% relative lift.
Observed partisan breakdown of treated registrants is as follows:
Partisan Breakdown of Registrants
We intend to follow up with folks who registered without a party affiliation in an attempt to further validate candidate choice.
Experiment 2:
Using a November Idaho voter file (exact vintage: 11-18-2024), we observe 1,940 successful registrations in our treatment group - a success rate of 18.18%. We observe 91 successful registrations in our control group - a success rate of 17.01%. This amounts to a raw observed 1.17 point lift, or a 6.9% relative lift.
Observed partisan breakdown of treated registrants is as follows:
Partisan Breakdown of Registrants
We intend to follow up with folks who registered without a party affiliation in an attempt to further validate candidate choice.
Experiment 3:
Using a November Idaho voter file (exact vintage: 11-18-2024), we observe 1,274 successful registrations in our treatment group - a success rate of 13.37%. We observe 3,520 successful registrations in our untreated cohort - a success rate of 9.70%. This amounts to a raw observed 3.66 point lift, or a 37.8% relative lift.
Observed partisan breakdown of treated registrants is as follows:
Partisan Breakdown of Registrants
We intend to follow up with folks who registered without a party affiliation in an attempt to further validate candidate choice.
Confidence Intervals
June Reg Form + Chase
Observed Lift: 2.52%
CI Lower: 0.07%
CI Upper: 4.97%
October Postcard + Chase
Observed Lift: 1.17%
CI Lower: -2.10%
CI Upper: 4.43%
October VR Text
Observed Lift: 3.66%
CI Lower: 2.91%
CI Upper: 4.41%
Probability Distributions
Conclusions
All programs were successful in increasing the likelihood of voter registration in our treated universes, with all confidence intervals pointing firmly in the positive direction. Interestingly, the intervention with the lowest raw registration rate - October SMS only to likely Dems living in red districts - appears to have the greatest lift and relative impact.
This should all be caveated that comparisons between an observational cohort result and true RCTs should not be taken as gospel truth, but it does lead us to a new hypothesis for future experimentation in 2025 and beyond: inexpensive, low-effort outreach to typically ignored districts is an efficient way to add net new Democratic voters to the electorate.
