Comparing Burnout in Sport-Specializing Versus Sport-Sampling Adolescent Athletes: A Systematic Review and Meta-analysis

1,429 athletes (1,371 specializers / 58 samplers; 1,422 completed the ABQ); 8 studies, of which only TWO contained any samplers at all

Adolescent athletes, mean age 15.59 y (range 12.5-17.2); 825 male / 528 female in the 7 studies reporting sex

8 studies pooled — 7 cross-sectional (level-of-evidence 4) plus 1 longitudinal prospective survey (LOE 3) — all self-report on Athlete Burnout Questionnaire (ABQ) subscales, scored 1-5. THE DESIGN FLAW IS WORSE THAN A SAMPLING IMBALANCE, and this is the thing full-text reading turns up: SIX of the eight studies contain ZERO samplers. All 58 samplers come from just two studies — Russell 2018 (24) and Strachan 2009 (34) — while the 1,371-strong specializer mean is pooled across all eight. The headline contrast is therefore a BETWEEN-STUDY comparison dressed as a within-study one, and the two arms are drawn from different populations: five of the eight specializer samples were recruited from elite academies of excellence, national talent programmes or Olympic sport classes (Appleton, Gustafsson, Isoard-Gautheur, Hill, Gerber), whereas both sampler-containing studies are small non-elite community samples. The authors concede it — "only 2 studies appropriately identified a sport sampler" — and also that "athlete specialization status had to be determined by our coauthors" post hoc, since the source studies were mostly built to study perfectionism, motivation or fear of failure rather than specialization. Reported as raw differences of ABQ means via z test, not standardized effect sizes. There is NO publication-bias assessment, NO heterogeneity statistic, and no formal quality appraisal beyond level-of-evidence labels. No adjustment for training volume, competitive level, sex, sport or age; causal direction untestable, and the authors state "no research is available showing whether burnout or reduced sport performance occurs first". Of the samplers, 24 are female and 34 have no sex reported.

The pooled result says specializers score higher on all three ABQ burnout subscales, largest on reduced accomplishment (delta 0.87 on a 1-5 scale). The two studies that actually compared specializers with samplers INSIDE one sample say something close to the opposite, and this is the finding that governs how much the pooled number is worth. Russell & Molina 2018 (24 samplers vs 53 specializers) "found no significant differences between sport specializers and sport samplers across all 3 dimensions" — and its sport-devaluation means run the wrong way (samplers 1.48 vs specializers 1.40). Strachan 2009 (34 vs 40) found specializers higher on exhaustion only (p<0.05), explicitly "not in reduced sense of accomplishment or sport devaluation". So across the only 6 within-study subscale contrasts available, 5 are null, and the one significant one is exhaustion — while the pooled headline effect is on reduced accomplishment, the subscale that was null in BOTH within-study tests. The arithmetic shows why: reduced-accomplishment means in the elite-academy specializer samples run 2.31-3.07, whereas in the two studies that also measured samplers the specializers score 1.73 and 1.82, essentially level with their own samplers (1.62, 1.65). The pooled delta of 0.87 is therefore mostly an elite-academy-versus-community-club contrast, not a specialization contrast. It cannot distinguish "specialization causes burnout" from "athletes already disengaging report both fewer sports and more burnout" (reverse causation), from a pure training-load effect, or from a between-sample selection artifact. Suggestive at best, and the archive should not cite the 0.87 without the within-study nulls beside it.

Trait negative affectivity/neuroticism and conscientiousness — substantially heritable — predict both burnout scores and sport persistence, so the specializer-sampler contrast is partly a personality contrast. Nothing in the corpus is genetically informative, and five of the eight studies were designed to study heritable personality traits (perfectionism, fear of failure, motivation) rather than specialization.

Coded `mixed` on the strength of the paper's own contents: the only two studies that directly compared specializers with samplers within a single sample disagree with each other and with the pooled result — Russell 2018 null on all three subscales, Strachan 2009 positive on exhaustion only. Later reviews and a 6-year 7th-12th grade longitudinal cohort (Valenzuela-Moss 2024) are reported to echo the headline, but that claim is not verifiable from this paper, and no prospective study has randomized or cleanly de-confounded specialization from volume, competitive level and dropout.