Multimodal assessment of declarative tactical knowledge in youth academy football players: within-player structure and between-player alignment in U15 and U18

Abstract Background. The gap between what players can verbally report about tactical principles and what they execute in match situations remains poorly understood. This study examined three questions, each at two levels of analysis: (1) whether the association between familiarity judgements and retrieval performance across players, as aggregate coupling, and within players, as item-level resolution differs by age group; (2) whether spatial representation and animated sequence naming converge across players, as the coupling of aggregate scores, and within players, as item-level concordance; and (3) whether distinct multimodal concordance profiles exist and whether their distribution differs between age groups. Method. Forty-four male youth academy players from a Ligue 2 club (U15, n = 23; U18, n = 21) completed three tasks applied to a validated corpus of 32 tactical actions: a prospective familiarity judgement (T1), a spatial representation task (T2), and an animated sequence naming task (T3), permitting item-level pairing of all three measures. The study was cross-sectional and was not preregistered. Results. Within players, judgement-retrieval resolution was substantial and equivalent in the two groups (T1-T2: γ = .75 vs. .76; T1-T3: γ = .41 vs. .35; all p < .001), as was item-level concordance between the two retrieval modalities (mean φ = .28 vs. .24). Across players, aggregate familiarity was strongly associated with both retrieval tasks in U18 (ρ = .70 and .69, both p ≤ .001) but not detectably in U15 players. This between-group difference survived Holm correction for the familiarity. naming pair (p = .024), as did the lower item-level judgement retrieval disagreement in U18 players (23.4% vs. 33.2%, p Holm = .028). The spatial naming coupling (r = .60 vs. .04; uncorrected p = .046) did not survive correction (p = .070) and was sensitive to individual observations. No group difference in raw performance was detected after correction. Two concordance profiles were identified descriptively, with borderline bootstrap stability; their distribution was uninformative with respect to age group (OR = 1.18, 95% CI [0.27, 5.15]). Conclusions. In this single-academy cross-sectional sample, the within-player structure of tactical knowledge resolution and inter-modal concordance was already in place at U15 and equivalent across groups. What differed was the between-player alignment of aggregate levels. Because age group is confounded with football practice, academy tenure, coaching staff, and recruitment cohort, the source of these differences cannot be identified. The findings require replication in a larger, preregistered, multi-site sample before inferences about development or diagnostic practice can be drawn.