Construct validity of the German general motor fitness and coordination test for children | Die Konstruktvalidität des Allgemeinen Sportmotorischen Tests für Kinder

Büsch, D,Strauss B, Seidel I, Pabst J,Tietjens M ,Müller,L,Kretschmer, J,Wirszing D

Abstract

Whereas numerous studies have confirmed the objectivity and reliability of the "Allgemeine Sportmotorische Test für Kinder" (AST, a general motor fitness and coordination test for children), examinations of its construct validity reveal that: (1) the few available findings fail to confirm the theoretically predicted distinction between the latent variables of condition and coordination, and (2) no studies so far have examined its item homogeneity, even though the test recommends computing a sum score to provide a general estimate of motor fitness. Confirmatory factor analyses show that the two latent variables condition and coordination cannot be extracted from the AST. Models with different latent variables also cannot be confirmed, too. Analyses based on IRT models, in contrast, reveal that one can distinguish between two qualitative classes of persons who are associated with either better running performance combined with lower throwing performance or better throwing performance combined with lower running performance. Nonetheless, it should be noted that class membership does not remain constant over time. In sum, results based on classical and probabilistic test theory negate the ability assumption underlying the AST. For practical assessments of a skill-specific performance, this means that it only seems justifiable to use a sum score for the locomotion items and a sum score for the object manipulation items. © 2009 Springer Medizin Verlag.