Individual Submission Summary
Share...

Direct link:

Poster #75 - COR for Kindergarte Validation Study

Thu, March 21, 12:30 to 1:45pm, Baltimore Convention Center, Floor: Level 1, Exhibit Hall B

Integrative Statement

We developed an assessment tool for use by kindergarten teachers. This tool, COR for Kindergarten, provides an observation-based authentic assessment that kindergarten teachers can successfully administer and reliably score.

Testing the COR for Kindergarten Assessments

We conducted three sequential studies to test and refine the assessment items and data system: an initial pilot, a revised pilot, and an implementation study. The results of the psychometric analysis of the implementation study follow.

Implementation Study

During the 2016–2017 school year, 79 kindergarten teachers from 28 schools in seven states participated in the COR for Kindergarten implementation study. These teachers rated 1,506 kindergarten children using the Kindergarten Entry Record (KER) and 738 kindergarten children using the Ongoing Record (OR).

After the teachers completed each assessment, the data was examined to determine whether the levels used to assess children’s progress indeed followed the developmental progression theorized and conducted reliability and validity studies. The data was also examined to determine whether the structure of the instrument was consistent in distinctly capturing the various developmental content areas. We analyzed the data on Conquest using the Multidimensional Partial Credit Rasch Model.

Implementation Study Results: KER

Scale Reliabilities (Chronbach's Alpha): There was an increase from the pilot study in four domains; only the ATL domain remained the same.

Domain Correlations (Rasch): Some of the scores are highly correlated and some are not, appropriately reflecting that young children progress at different rates in different areas of development.

Model Fit, Wright Maps, and Item Curve Characteristics (Rasch): The weighted and unweighted Mean-Square fit statistics for the items were all below 1.4 meaning that no misfits were observed for any item beyond that expected within the Partial Credit Rasch model. The Wright Maps and the polytomous Item Curve Characteristics across the items provide further evidence of an orderly successive progression of developmental levels.

Implementation Study Results: Ongoing Record
Scale Reliabilities (Chronbachis Alpha): Figure 1 shows the COR for Kindergarten OR Domain Scale reliabilities in Figure 1, along with the number of items per domain, for a sample of N=738 children attending kindergarten.

Domain Correlations (Rasch): Figure 2 shows the intercorrelations among the domains of the COR for Kindergarten OR in Figure 2. Again, some of the scores are highly correlated and some more moderately, appropriately reflecting that young children progress at different rates in different areas of development.

Model Fit, Wright Maps, and Item Curve Characteristics (Rasch): The weighted and unweighted Mean-Square fit statistics for the items were all below 1.4, meaning that no misfits were observed for any item beyond that expected within the Partial Credit Rasch model. The Wright Maps and the polytomous Item Curve Characteristics across the items provide further evidence of an orderly successive progression of developmental levels.

Summary

The implementation study results show that the COR for Kindergarten assessment provides reliable and valid psychometric measurement of the development of children entering kindergarten and throughout the kindergarten year for five key domains.

Author