Compliance Without Control: Selection Bias in Measures of Presidential Party Dominance

As of August 4, 2026, Trump-endorsed candidates had won roughly 98 percent of decided Republican primaries, widely read as evidence of maximal party control. It is not, because the population these rates are computed on is not fixed with respect to the leader. I distinguish three quantities commentary treats as one: compliance, the conformity of surviving officeholders; institutional control, the capacity to achieve desired outcomes; and electoral productivity, the conversion of dominance into general-election seats. Since 2024 these have diverged. The seventeen Republicans who voted to impeach or convict in 2021 show what conventional measures omit: five were defeated in primaries, but fourteen had left Congress or announced departures by August 2026. An analyst reading primary outcomes alone observes 29 percent of the cohort’s disappearance and codes the remainder, implicitly, as loyalty. I propose attrition-adjusted bounds that treat exits as informative. The design is descriptive and identifies no causal effect.