eric.ed.gov har udgivet:

This report presents the results of an experiment conducted in Alabama beginning in the 2006/07 school year, to determine the effectiveness of the Alabama Math, Science, and Technology Initiative (AMSTI), which aims to improve mathematics and science achievement in the state’s K-12 schools. This study is the first randomized controlled trial testing the effectiveness of AMSTI in improving mathematics problem solving and science achievement in upper-elementary and middle schools. AMSTI is an initiative specific to Alabama and was developed and supported through state resources. An important finding is the positive and statistically significant effect of AMSTI on mathematics achievement as measured by the SAT 10 mathematics problem solving assessment administered by the state to students in grades 4-8. After one year in the program, student mathematics scores were higher than those of a control group that did not receive AMSTI by 0.05 standard deviation, equivalent to 2 percentile points. Nine of the 10 sensitivity analyses yielded effect estimates that were statistically significant at the 0.025 level, consistent with the main finding. The estimated effect of AMSTI on science achievement measured after one year was not statistically significant. Based on the SAT 10 science test administered by the state to students in grades 5 and 7, no difference between AMSTI and control schools could be discerned after one year. Changes in classroom instructional strategies, especially an emphasis on more active-learning strategies, are important to the AMSTI theory of action. Therefore, a secondary investigation of classroom practices was conducted, based on data from survey responses from teachers. For both mathematics and science, statistically significant differences were found between AMSTI and control teachers in the average reported time spent using the strategies. The effect of AMSTI on these instructional strategies was 0.47 standard deviation in mathematics and 0.32 standard deviation in science. Two years of AMSTI appeared to have a positive and statistically significant effect on achievement in mathematics problem solving, compared to no AMSTI. Two years of AMSTI appeared to have a positive and statistically significant effect on achievement in science. AMSTI appeared to have a positive and statistically significant effect on reading achievement as measured by the SAT 10 test of reading administered by the state to students in grades 4-8. AMSTI did not appear to have a statistically significant effect on teacher-reported content knowledge in mathematics or science after one year. AMSTI did not appear to have statistically significant differential effects on student achievement in mathematics problem solving or science based on racial/ethnic minority status, enrollment in the free or reduced-price lunch program, gender, or pretest level. Appended are: (1) Explanation of primary and secondary confirmatory outcome measures; (2) Explanation of exploratory research questions; (3) Selection and random assignment of schools; (4) Statistical power analysis; (5) Data collection procedures and timeline; (6) Description of program implementation data collected but not used in report; (7) Alabama Math, Science, and Technology Initiative (AMSTI) teacher survey #3; (8) Data cleaning and data file construction; (9) Attrition through study stages for samples used in the confirmatory analysis; (10) Description of degree rank; (11) Equivalence of Year 1 baseline and analyzed samples for confirmatory student-level and classroom practice outcomes; (12) Internal consistency and validity of active learning measures; (13) Number of students and teachers in schools in analytic samples used to analyze Year 1 confirmatory questions; (14) Attrition through study stages for samples used in Year 1 exploratory analysis; (15) Tests of equivalence for baseline and analytic samples for Year 1 exploratory outcomes; (16) Statistical power analyses for moderator analyses; (17) Derivation and motivation of the Bell-Bradley estimator when measuring estimated two-year effect of the Alabama Math, Science, and Technology Initiative (AMSTI); (18) Attrition through study stages for samples contributing to estimation of two-year effects; (19) Examination of equivalence in baseline and analytic samples used in the estimation of two-year effects; (20) Estimation model for two-year effects of the Alabama Math, Science, and Technology Initiative (AMSTI); (21) Topics and instructional methods used at the Alabama Math, Science, and Technology Initiative (AMSTI) summer institute; (22) Parameter estimates on probability scale for odds-ratio tests of differences between Alabama Math, Science, and Technology Initiative (AMSTI) and control conditions in Year 1 (associated with summer professional development and in-school support outcomes); (23) Descriptive statistics for variables that change to a binary scale used in the Alabama Math, Science, and Technology Initiative (AMSTI) and control conditions in Year 1; (24) Comparison of assumed parameter values and observed sample statistics for statistical power analysis after one year; (25) Parameter estimates for Stanford Achievement Test Tenth Edition (SAT 10) mathematics problem solving after one year; (26) Parameter estimates for Stanford Achievement Test Tenth Edition (SAT 10) science after one year; (27) Parameter estimates for active learning in mathematics after one year; (28) Parameter estimates for active learning in science after one year; (29) Sensitivity analyses of effect of the Alabama Math, Science, and Technology Initiative (AMSTI) on Stanford Achievement Test Tenth Edition (SAT 10) mathematics problem solving achievement after one year; (30) Sensitivity analyses of effect of the Alabama Math, Science, and Technology Initiative (AMSTI) on Stanford Achievement Test Tenth Edition (SAT 10) science achievement after one year; (31) Sensitivity analyses of effect of the Alabama Math, Science, and Technology Initiative (AMSTI) on active learning instructional strategies in mathematics classrooms after one year; (32) Sensitivity analyses of effect of the Alabama Math, Science, and Technology Initiative (AMSTI) on active learning instructional strategies in science classrooms after one year; (33) Tests for violations of factors associated with assumption of equal first year effects on students in Alabama Math, Science, and Technology Initiative (AMSTI) and control schools; (34) Post hoc adjustment to standard error for estimate of two-year effect of the Alabama Math, Science, and Technology Initiative (AMSTI) on mathematics achievement after two years; (35) Parameter estimates for effect of the Alabama Math, Science, and Technology Initiative (AMSTI) after two years; (36) Parameter estimates for effect of the Alabama Math, Science, and Technology Initiative (AMSTI) on student reading achievement after one year; (37) Parameter estimates for teacher content and student engagement after one year; (38) Estimates of effects for terms involving the indicator of treatment status in the analysis of the moderating effect of the three-level pretest variable; (39) Parameter estimates for the analysis of the moderating effect of racial/ethnic minority status on the impact of the Alabama Math, Science, and Technology Initiative (AMSTI) on reading after one year; (40) Parameter estimates for analysis of average effect of the Alabama Math, Science, and Technology Initiative (AMSTI) on reading by racial/ethnic minority students after one year; and (41) Parameter estimates for effect of the Alabama Math, Science, and Technology Initiative (AMSTI) on reading for White students after one year. (Contains 26 figures, 136 tables, 1 box and 130 footnotes.)