The predictive validity and stability of standardized assessment in early childhood education

There has been a lot of controversy surrounding the use of standardized achievement tests in preschool. Several researchers claim that the performance of young children is too fickle to be reliably and validly tested. The goal of this study was to examine the predictive validity for future performance and the score-stability of two widely administered Dutch preschool tests. Language and arithmetic scores of 431 children were collected retrospectively over a four-year period. First, percentile scores of low scoring children were plotted to assess the stability of scores over time. In addition, predictive validity of arithmetic and language scores was assessed by means of a multilevel model. Both the language and arithmetic tests were poor identifiers for low scoring children in first- and second grade. The majority of low scoring first- and second graders achieved above average in preschool, or fluctuated between top- and bottom range scores. A small group did not show large fluctuations in scores. Low correlations (r= .09 to .30) between the preschool tests and subsequent tests indicated that both tests are weakly to moderately associated with first and second grade performance. The results are discussed in light of practical applications of these tests.