| The purpose of this research was the development of a valid and reliable mathematics curriculum-based measure (MCBM) that can be used by early childhood programs and teachers to assess "number sense," the ongoing development of quantitative reasoning, in preschool children. This is important in light of the current focus on program accountability, as exemplified by the National Reporting System instituted by Head Start, implications of the federal No Child Left Behind Act, and the documentation of progress toward academic content standards required by many states. Previous attempts to develop assessment tools for measuring number sense have focused largely on memorized number facts and have not considered the pre-counting quantification skills described by Piaget and Kamii. Of further concern to educators is the difficulty in accurately assessing preschool children. For this reason, curriculum-based measures, which are aligned to classroom curriculum and can therefore inform planning and instructional decisions, are of particular interest. The hypothesis of the current research was that six measurable variables---quantification level, counting skills, comparison of sets, numeral recognition and understanding, combination of sets, and patterning---would make a significant contribution to the latent construct of number sense. The measurement tool was an interactive game, played between assessor and child, that consisted of a series of cards in a standardized order and teddy bear counters. The game was aligned to the curriculum of a licensed preschool center in Cincinnati, Ohio, from which a sample of 108 children, ages 3 to 5, was drawn during the 2006-07 school year. Confirmatory factor analysis was used to evaluate the hypothesized model. Results indicated a good model fit: chi2 (3, N=108) = 4.246, p=.236; NFI=.990; CFI=.997; and RMSEA=.062. In addition, all standardized path coefficients demonstrated both statistical ( p < .001) and practical significance (beta > .3), with robust standardized regression weights ranging from .71 to .88. The results suggest that the MCBM, which evaluated precounting as well as counting skills, was a valid and reliable measure of number sense for this sample. |
