Scandinavian Journal of Educational Research. Online first.
Forfattere
Marianne Klem, Jan-Eric Gustafsson og Bente Hagtvet.
Abstract
The Norwegian government recommends a systematic language assessment of all four-year-olds as part of the general health surveillance program for the purpose of identifying children at risk of language delay. This study aimed to investigate the construct validity of the recommended language screening tool called LANGUAGE4 [SPRÅK4] by first examining the dimensionality of the underlying construct of the tool, after which the concurrent convergent validity was established by regressing an external language factor, defined by four standardized language tests, on a single higher-order factor. The findings provide support for a higher-order model with one general language factor, suggesting that a large amount of the variance in LANGUAGE4 is attributable to a single common factor at the second-order level. Furthermore, this single factor explained a considerable amount of the variance in the external language factor. Our findings are interpreted as support for satisfactory construct validity of LANGUAGE4.