When two entries disagree

The resolution order, applied in sequence. Stop at the first step that explains the disagreement. (Not from a study; the base's own procedure — see Limitations.)

1. Caliber. Compare the McKay tiers of the two samples (REF-529). In the base's experience (not counted, not from a study), many contradictions are two true findings about two different populations.

Worked example already in the base: in recreationally active women, time to fatigue after 60 min at 60% VO₂max was significantly reduced in the mid-luteal phase in hot-humid air (32 °C / 60% RH) but not in temperate conditions (REF-710, 12 women tested per condition, n=8 analysed in each); in ten well-trained women exercising in the heat, cycle phase did not change work done (257 vs 255 kJ) (REF-711) — both abstract-checked 2026-09-27.

Inference, not tested: the studies may describe different athletes rather than contradict each other — but the protocols also differed (time to fatigue vs self-paced work; for REF-710 a different heat condition, see step 4) and a null in n=10 cannot exclude a small phase effect.

2. Sport and modality. A finding measured in running does not automatically hold in cycling, and vice versa.

The interference literature is the standing example — and also a warning that the example itself is contested: the 2012 pooling found that running, but not cycling, as the endurance mode caused significant decrements in hypertrophy and strength (REF-512, 21 studies), while the 2022 pooling of 43 trials found no significant moderation by cycling vs running (REF-330).

When the two pooled analyses disagree, the newer and larger one does not win by being newer; the honest answer is that modality dependence is unsettled and the robust residual is a small penalty to explosive strength, worse in same-session designs (REF-330).

3. Population. Sex, age and training history can change the answer (the base's practice to check them). For sex this is documented: most high-quality sport-science data come from men and female athletes are under-represented (REF-336, REF-337 — methodology papers; they do not address age or training history).

4. Conditions. Environment can be the whole disagreement. Within REF-711 the same well-trained women did 7% more work in dry heat than in humid heat matched for WBGT (263 vs 248 kJ, environment compared across both phases), while cycle phase produced no difference (257 vs 255 kJ); in REF-710 the luteal cost appeared only in hot-humid air and not at 20 °C. Heat and humidity are not interchangeable, and "matched WBGT" does not mean equally hard.

**5.

If the disagreement survives all four, it is EQUIPOISE — say so.** REF-531 is filed in the base as exactly this, and it shows how equipoise is built: the review itself (C+, narrative) argues that follicular-phase-based resistance training is *possibly* superior for strength and mass and calls for more research, while the meta-analysis of 78 studies finds the pooled effect of cycle phase on performance trivial (early follicular vs all other phases ES −0.06, CrI −0.16 to 0.04; evidence quality low) and recommends a personalised approach (REF-335).

One side proposes a training effect, the other finds no group-level effect on acute performance — different outcomes (chronic strength/mass adaptation vs acute performance), so the base's reading, not either paper's, is that the evidence does not decide.

In equipoise the coach states that the literature is split, gives the athlete the trade-off, and defers to the athlete's own tracked record. Not from a study; the base's own practice: require ≥3 comparable attempts before reading a personal pattern.

What never resolves a conflict: publication year on its own, journal prestige, sample size on its own, or which study the athlete read first.