*Names in bold indicate Presenter
The PD program was evaluated using an experimental design. Within each of 12 districts, selected middle schools were randomly assigned to treatment and control conditions. In those schools, the study focused on teachers of the core seventh grade mathematics course. Those in the treatment schools received the PD program described above; those in the control schools continued with business-as-usual. In the second year of the study, the sample was consolidated to six districts, and treatment schools in these “2-year” districts continued to receive the PD program for a second year.
The study monitored the delivery of the PD program and teachers’ participation in the PD program. In addition, the study used surveys to measure the extent of teacher participation in any PD in the treatment and control groups. By the end of the second year, the treatment had elicited a treatment-control difference in the amount of mathematics-related PD received of 64 hours. The total duration of the PD program, as delivered, was 114 hours, and teacher participation in the PD was high. The exposure to the PD program accumulated by the those present in the spring of the second year was attenuated by teacher mobility. Forty-nine percent of the treatment teachers teaching the core seventh grade mathematics course at the end of the second year were not teaching it at the beginning of the first year.
After one year of implementation, the study found statistically significant impacts on one of three measures of classroom practice, and no statistically significant impacts on teacher knowledge or student achievement. Student achievement was measured using a customized, computer-adaptive student achievement test focused on rational number topics.
After two years of implementation, the study found no statistically significant, cumulative impacts of the PD on teacher knowledge or student achievement. Classroom practice was not measured at the end of the second year.
To supplement these findings, a number of exploratory analyses were conducted. Controlling for student and teacher covariates, the estimated association between the teacher knowledge measure and student achievement was statistically significant.
To increase the statistical power of the impact analyses, we pooled all 1-year impacts from both years of the study. That is, we pooled: the 1-year impacts from the first year in the six 1-year districts; the 1-year impacts from the first year in the six 2-year districts; and the 1-year impacts from the second year in the six 2-year districts. Pooling the three 1-year impacts, we concluded that each year of the PD had a statistically significant impact on one of two subscales for teacher knowledge (ES = 0.28) and no statistically significant impacts on achievement.