1 Edge Hill University (UNITED KINGDOM)
2 Colorado State University (UNITED STATES)
3 University of Colorado Denver (UNITED STATES)
About this paper:
Appears in: ICERI2010 Proceedings
Publication year: 2010
Pages: 1275-1283
ISBN: 978-84-614-2439-9
ISSN: 2340-1095
Conference name: 3rd International Conference of Education, Research and Innovation
Dates: 15-17 November, 2010
Location: Madrid, Spain

In response to the declining mathematics performance of American students compared to their international counterparts and to a national call for high quality professional development for teachers, the National Science Foundation established the Math and Science Partnership (MSP) initiative. The charge was to create partnerships between higher education institutions and primary/secondary schools with the goal of improving student achievement in mathematics and science.

As part of this initiative, the Rocky Mountain Middle School MSP (RMMSMSP) targeted primary and early secondary teachers in seven partner school districts, offering long-term professional development (15 – 36 months, on average) in the form of mathematics content courses and pedagogical follow-up units. One of the goals of the RMMSMSP was to increase student mathematics achievement by increasing teacher content knowledge. This presentation focuses on the effects of the RMMSMSP on student mathematics outcomes on the state standardized test.


We used the quasi-experimental cohort control design (Shadish, Cook, and Campbell, 2002) to maximize the validity of our findings. The final sample consisted of 2319 students who came from the classes of 23 teachers for whom we could construct two groups of students: one group who had the teachers before they had taken any courses and another group who had the same teachers after they had taken their final number of courses. Students were equated statistically for academic ability and special education status.

After exploring several statistical models for understanding the student outcomes of the RMMSMSP, we analyzed the standardized test outcomes as a dichotomous variable, which represented either a proficient outcome or better (i.e. “advanced” or “proficient”) or a non-proficient outcome or worse (i.e. “partially proficient” or “unsatisfactory”) on the state standardized mathematics test. We used a logistic regression model with a random teacher effect to account for the clustering of students within teachers. Such a model is an example of a hierarchical generalized linear model (Raudenbush and Bryk, 2002).


Although our results were not statistically significant, a fact which is likely due to the small sample of only 23 teachers, they merit consideration. We found that students who received instruction from a participant teacher after the teacher took one mathematics course had odds of at least a proficient outcome which were approximately 1.5 times higher than students who received instruction before the teacher took a math course. Students who received instruction after the teacher took two or more mathematics courses had odds of at least a proficient outcome which were approximately 2 times higher than students who received instruction before the teacher took any math courses.

This study presented many methodological challenges, especially related to teacher movement, which drastically reduced the number of teachers available for sampling, and related to the use of standardized test data, which can be quite variable. While the results presented here are not statistically significant, we believe they are practically significant. They contribute to filling the gap in the literature concerning the association of teacher content knowledge professional development and student achievement, as well as highlighting methodological issues commonly present in professional development evaluations.
Mathematics achievement, professional development, middle level.