Definition
The Statewide Primary Turnout Score indicates the likelihood an individual will vote in primary elections where either statewide races (Senate, Governor, etc.) or federal races (e.g. Congressional) are top-of-ticket. These models are custom trained in each state, to specifically better calibrate the models to past turnout patterns specific to that state.
Technical Details
This turnout model is calibrated within each state to better match turnout patterns in primary vote history behavior specific to that state. That is, the specific model for individuals registered in Georgia will be different from the specific model for individuals registered in Colorado. This reflects both the different predictive power of model inputs across geographies and the different elections used as training targets. In each location, we identified a recent statewide primary election and used it as the basis for labeling voters: those who voted are labeled as targets, and those who did not are labeled as non-targets.
The models were trained using gradient-boosted decision trees. Model features were drawn from the Atlas by Murmuration dataset, which includes demographic (age, race, gender, etc.), commercial, geographic, and vote history information for all registered voters nationally.
The Statewide Primary Turnout Score ranges from 0–100, where an individual's specific score approximately corresponds to their likelihood of voting in an upcoming statewide primary.
We validated the model’s accuracy using a held-out test set of 20% of the original sample. Among individuals with scores in the top 20% of the Statewide Primary Turnout Score, 72% voted in statewide primary elections, compared to 28% of the population at large. Thus by using this score for targeting, our partners are about 157% more likely to reach voters in statewide primary elections than if they targeted at random.
Use Cases
As with turnout scores more generally, this score can be used in two broad ways:
- Persuasion/Name Recognition: Using the higher end of the Statewide Primary Turnout Score, organizations can focus outreach on individuals already likely to vote in upcoming statewide primaries. This is particularly useful in the early stages of a campaign — especially where name recognition is low — to ensure resources are directed toward voters who are most likely to turn out.
- GOTV: Voters who fall in the middle to low range of the Statewide Primary Turnout Score distribution (say with scores of between 30 and 70) can be targeted for GOTV outreach. When used in this manner, the score should generally target known supporters of candidates or issues to ensure partners are communicating with the voters most likely to support their work.
Targeting Table
The table below shows the score values associated with each decile to help you more easily target using the Statewide Primary Turnout Score nationally. Note: as this score is built in each particular state, these score cutoffs differ based on the turnout patterns specific to your state so you can use these as a starting point, but should adjust the thresholds according to your desired universe size.
| To target the top... | Set the minimum score value as... |
| 10% | 72 |
| 20% | 47 |
| 30% | 28 |
| 40% | 17 |
| 50% | 10 |
| 60% | 6 |
| 70% | 3 |
| 80% | 1 |
| 90% | 0 |
| 100% | 0 |