Validation of a screening instrument for young children with teacher assessment of school performance.

Abstract

Teachers work in a context that encourages them to differentiate high performing from low performing students, or rather easy-to-teach from hard and very hard-to-teach children. Once they understand children, teachers' professional judgments are the most reliable source of information about children's success or lack thereof in school. The intent of this study was to identify a screening instrument that would agree with teacher judgments in order to increase the relevance of educational programs for young children early in their educational experience. A group of children was observed and tested, and the results were compared to teacher judgments from kindergarten through third grade. Results were validated with a second group of kindergarten children. An instrument was identified that can be used by the teacher to gather behavior samples in a matter of minutes before s/he has had sufficient exposure to the child to make more informed decisions. Results were predictive of subsequent school performance in kindergarten through third grade.Educators of prekindergarten and kindergarten children find themselves in a dilemma when attempting to identify children who are at risk for learning difficulties. The screening of children entering kindergarten is mandated by state and federal laws (Public Law 99-457) in order to help find those who need further evaluation. The laws, however, do not provide guidelines for appropriate screening methods (Stavrou, Glassman, Kaiser, & McNeil, 1990). A primary consideration in choosing a screening instrument is the degree to which screening information relates to social and educational consequences for young children (Macmann,

Use of tests of statistical significance and other analytic choices in a school psychology journal: Review of practices and suggested alternatives.

“…In only two studies did authors conduct an actual "external" replication with an independent sample of new subjects (Jorgenson, Jorgenson, Gillis & McCall, 1993;Vickers & Minke, 1995 , 1992Jorgenson et al, 1993). The pattern is regrettable, because methodologists are critical of stepwise methods, since these methods yield distorted and non-replicable findings (see Huberty (1989), Snyder (1991), and especially Thompson (1995b)).…”

“…Third, authors of only 2 of the 35 articles invoked an “internal” replicability analysis, such as cross-validation, the jackknife, or the bootstrap (Elias & Allen, 1991; Keith & Cool, 1992). In only two studies did authors conduct an actual “external” replication, with an independent sample of new participants (Jorgenson, Jorgenson, Gillis, & McCall, 1993; Vickers & Minke, 1995). Again, authors who think that statistical significance evaluates result replicability will erroneously find such replicability analyses less necessary, with all the attendant negative consequences for the business of accurately accumulating evidence across studies.…”

“…Three other patterns incidental to the primary focus of our work also must be noted. First, many of the authors who used regression methods elected to use stepwise methods (e.g., Huebner, 1991Huebner, , 1992Jorgenson et al, 1993). The pattern is regrettable, because methodologists are critical of stepwise methods, since these methods yield distorted and nonreplicable findings [see Huberty (1989), Snyder (1991), and especially Thompson (1995)].…”