Master of Natural Sciences (MNS)



Document Type



This study took place at a medium-sized suburban high school. It was designed to determine the usefulness of certain teacher-made tests in predicting students' end-of-course (EOC) tests. The teacher taught the students the skills in which their performance was weakest on the previous state test. The students were tested after each skill on a four-point quiz (teacher-made test). Students who scored 3—4 moved on to the next lesson or enrichment, while those who scored 0—2 were re-taught and re-tested. The procedure was repeated throughout the school year. At the end of the course, students took the state-mandated end-of-course test. The results on the end-of-course tests were compared to the results of the teacher-made test scores. We used linear regression to interpret the findings. We found that 13 out of the 39 teacher-made tests had 0 correlation with the end-of-course test; 17/39 had 0.01 ≤ R2 ≤ 0.05; 7/39 had 0.06 ≤ R2 ≤ 0.15; one (1) had 0.17 and one (1) had 0.20. We also examined combinations of the teacher-made tests and the highest correlation was 0.24—generally regarded as a weak relationship. The regression analysis indicates that none of the teacher-made assessments and no combination of them had any practical value as a predictor of EOC. Some statistically significant correlations were observed, but even these would probably not provide teachers with information that could guide instructional decisions that would raise EOC test scores.



Document Availability at the Time of Submission

Release the entire work immediately for access worldwide.

Committee Chair

Madden, James