The outcome is measured before ideology is added
Candidate Margin Overperformance is the residual left after subtracting the model's predicted structural legislative-ticket gap from the observed gap. The structural expectation uses symmetric incumbency, era-specific downballot lag, limited demographics, and capped campaign effort. Ideology is excluded from that model and introduced afterward.
Positive candidate-directional values always mean that the named candidate—Democrat or Republican—ran ahead of the structural expectation. Aggregate election results cannot distinguish persuasion from turnout.
Absolute ideology and federal-relative performance
Each point is one candidate-cycle with a nationally comparable Shor–McCarty score. Use the controls to change the electoral baseline or isolate a party; the axes and summary update with the selection.
Moving one national standard deviation rightward is associated with greater federal-relative performance. The relationship survives winner-only, prior-service, finance, incumbency, and leave-one-cycle-out checks.
Point estimates generally favor less conservative Republicans, but most are imprecise. A winner-only CMO result is suggestive; federal-relative and prior-service results do not establish a general moderation advantage.
Coverage is concentrated among successful candidates
The Democratic and Republican distributions have little common support, and Shor–McCarty primarily observes people who reached office. The selection audit shows how different the scored sample is from candidates without an observed score.
Total association and mediator-adjusted association
The first estimate leaves incumbency and fundraising out because electoral strength may help produce both. The second conditions on them. The difference between the bars shows how much of the relationship is absorbed by those possible pathways.
For Democrats, the absolute relationship remains strong for CMO and federal-relative performance after adding incumbency and finance. Presidential-relative performance attenuates in the smaller finance-complete sample.
Issue-level associations are not one left–right scale
The issue chart keeps substantively different positions separate. It does not combine gun access with criminal punishment, racial civil rights with sexual morality, or the level of taxation with who bears the tax.
These are separate historical associations, not additive points in a single score. False-discovery-adjusted values are shown. Tax burden does not say who paid the tax and must not be read as generic fiscal conservatism.
Inspect the candidate-level evidence
Choose an issue, party, and outcome to inspect the underlying candidate-cycles. Hover or focus a point for the candidate, election, position, and measured performance. Unobserved positions are not imputed.
District fit varies by issue
A positive interaction means conservative positioning becomes more favorable as a district's federal vote becomes more Republican. Several issue interactions appear, but uneven coverage and multiple testing make them leads for case research rather than forecast inputs.
The aggregate relationship is durable; recent coverage is thin
The Democratic Shor coefficient remains positive when every cycle is omitted in turn. But only five post-2016 Democratic candidate-cycles have usable Shor scores, so the page does not manufacture a modern-era conclusion from an underpowered cell.
Interpretation and limits
This is historical, observational evidence. Shor scores are career measures and cover successful officeholders disproportionately. Issue evidence is temporally screened but uneven. Associations can reflect recruitment, geography, candidate quality, and coalition composition as well as voter response.