A pilot self-coach alignment audit in NCAA collegiate rowers: field feasibility, null findings, and...

Testing relational mechanisms like self-verification (SV) in competitive collegiate sport is constrained by field-feasible measurement and the small-N problem. We introduce a brief alignment audit and evaluate its field feasibility and statistical behavior in a single-team pilot.

NCAA Division II women rowers (N = 12) completed four in-season waves aligned with 2,000-m ergometer assessments, reporting self-rated mental toughness (MT) via the Mental Toughness Index (MTI), meta-perceptions of head and assistant coach appraisals, and their preferred coach for feedback (self-reported, not observed). Because coach preference was perfectly invariant within every athlete across all four waves, the design was analytically reduced to a cross-sectional athlete-level comparison (N = 12); GEE and GLMM models were retained only as sensitivity diagnostics, since quasi-complete separation precluded their use for primary inference. The perceived-proximity proxy, ΔAlignment, was the difference between absolute self-coach discrepancies, a coarse proxy that cannot distinguish self-verification from self-enhancement or structural coach-role preference.

Head-coach preference was stable at 33.3% across all 48 observations. Athlete-level analyses converged on a null: logistic regression returned OR = 1.01 [95% CI (0.71, 1.56), p = .950]; Welch t(3.68) = 0.04, p = .968; Hedges' g = 0.03 [95% CI (-1.08, 1.14)]; Fisher exact p = .236 for alignment directionality. The 95% CI width of 2.22 Hedges' g units is the primary index of design insensitivity. Temporal stability of ΔAlignment was moderate to good [ICC(2,1) = .73, 95% CI [.48,.90]]. GEE and GLMM point estimates were uniformly null (all ORs = 1.00). Tiered simulation-based power analysis at N = 12 was flat across OR = 1.5-3.0 (13.5-13.9%), confirming the design's insensitivity to effect size rather than absence of effect.

Perceived self-coach alignment was not detectably associated with self-reported coach preference. This reflects the inability of a single-team, N = 12 design to provide precise information, not evidence against SV theory. The primary contribution is methodological: a reproducible, low-burden audit template with open-science materials (data, code, RSA-ready component scores at OSF) and a design-sensitivity framework to guide prospectively powered multi-team replication.