Definition
The Partisan Score predicts the probability that a person will vote for Democrats in federal races (e.g., U.S. Senate and House). Scores range from 0-100, where higher values indicate greater likelihood of voting Democratic and lower values indicate greater likelihood of voting Republican. A score of 50 means the model assigns equal probability to voting for either party.
Note: This score measures predicted voting behavior, not ideology. A score of 50 may reflect uncertainty in the prediction or moderate/independent political views.
Technical Details
The model was built from the following survey question:
Separate from the president, when thinking about elections for US Congress and US Senate between Democrats and Republicans, do you almost always vote for Democrats, almost always vote for Republicans, or do you often vote for candidates of both parties in elections for US Congress and US Senate?
[If both, followup]: Even though you vote for candidates of both parties, do you vote for Democrats slightly more often than Republicans, Republicans slightly more often than Democrats, or do you vote for Democrats and Republicans equally often for US Congress and US Senate?
- Democrats - always
- Democrats - mostly
- Republicans - always
- Republicans - mostly
- Both - Democrats more
- Both - Republicans more
- Both - roughly equal
- Neither / Third Party
- Don’t know/rather not say
Respondents who answered “Democrats - always”, “Democrats - mostly”, or “both - Democrats more” are labeled as Democrats, while anyone who responded “Republicans” are labeled as Republicans. Those who responded “Both - roughly equal”, “Neither / Third Party”, or “Don’t know/rather not say” are dropped.
The Partisan Score was trained on polling data from the Summer 2025 Murmuration poll. We collected 7,721 responses online from registered voters in July and August 2025. The survey used stratified random sampling to ensure national representativeness across age, race, gender, partisanship, education, and geography.
The model was 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.
Scores range from 0-100, where higher scores indicate greater likelihood of voting for Democrats and lower scores indicate greater likelihood of voting for Republicans. The score represents the model's predicted probability (scaled to 0-100) that an individual consistently votes for Democrats in Congressional races.
We validated the model's accuracy using a held-out set of 1,524 polling respondents (20% of the original survey sample) whose data was not used during model development. Among individuals with scores in the top 20% of the Partisan Score, 97% report voting for Democrats (in federal races), making them 75% more likely to report voting for Democrats than the average voter. At the other end of the distribution, individuals in the bottom 20% of the Partisan Score are 88% less likely than average to report voting for Democrats.
The model achieved a test set AUC of 0.914, demonstrating strong discriminative ability in identifying partisan voting patterns. In held-out test data, the model correctly distinguished between Democratic-leaning and Republican-leaning voters at rates significantly above baseline.
Use Cases
- Persuasion efforts in partisan primaries: In states where party-registration is not available and/or primaries are open, you can use the Partisan Score in conjunction with a turnout score to find likely Democratic (80-100 Partisan Scores) or likely Republican (0-20 Partisan Scores) voters. You would then layer on top of this a turnout score (e.g. Statewide Primary Turnout Score) and filter for high-propensity voters (e.g. turnout scores of 70+).
- Base Mobilization/GOTV: Partners working to mobilize Democratic voters can target high Partisan Score voters (e.g., 70+) who may need encouragement to turn out but are highly likely to vote Democratic when they do. In this case, the effort should use a turnout score, and target voters in the middle range (40-70), who are most likely to be responsive to GOTV outreach.
- Primary Election Targeting: In heavily Democratic or Republican geographies where the primary winner often becomes the de facto general election winner, this score helps identify voters aligned with moderate candidates. For example, in a Democratic primary with a progressive candidate and a moderate candidate, partners supporting the moderate candidate can combine high Ideology Moderate Scores (scores of 55 and above) with high Partisan Scores (scores of 80 and above) to target voters most likely to support their candidate. This is particularly valuable when issue scores don't differentiate candidates within a target universe (e.g., when the vast majority of Democrats within a location support reproductive rights and gun control).
- Finding Competitive Districts for Resource Allocation: Partners working on multi-district campaigns (state legislative races, congressional campaigns, ballot measures) can use aggregate Partisan Scores to identify the most competitive districts where resources will have the greatest impact. By calculating the average Partisan Score across voters in each district, partners can identify districts with scores close to 50, indicating genuine competitiveness. This is particularly valuable for state-level organizations deciding where to invest in field programs, advertising, or candidate recruitment.
Targeting Table
The table below shows the score values associated with each decile to help you more easily target using the Partisan Score nationally. Note: these score cutoffs may be different in your local districts.
| To target the top... | Set the minimum score value as... |
| 10% | 95 |
| 20% | 90 |
| 30% | 80 |
| 40% | 67 |
| 50% | 54 |
| 60% | 42 |
| 70% | 30 |
| 80% | 11 |
| 90% | 5 |
| 100% | 0 |