Quasi-Experimental Designs For Evaluating MSP Projects: Processes & Some Results Dr. George N. Bratton Project Evaluator in Arkansas.

Slides:



Advertisements
Similar presentations
Ed-D 420 Inclusion of Exceptional Learners. CAT time Learner-Centered - Learner-centered techniques focus on strategies and approaches to improve learning.
Advertisements

Defining Characteristics
Designs to Estimate Impacts of MSP Projects with Confidence. Ellen Bobronnikov March 29, 2010.
ESTEEMS (ESTablishing Excellence in Education of Mathematics and Science) Project Overview and Evaluation Dr. Deborah H. Cook, Director, NJ SSI MSP Regional.
Using Statistics in Research Psych 231: Research Methods in Psychology.
Using Statistics in Research Psych 231: Research Methods in Psychology.
Using Statistics in Research Psych 231: Research Methods in Psychology.
Inferential Stats for Two-Group Designs. Inferential Statistics Used to infer conclusions about the population based on data collected from sample Do.
Lecture 9: One Way ANOVA Between Subjects
T-Tests Lecture: Nov. 6, 2002.
Today Concepts underlying inferential statistics
PISA Partnership to Improve Student Achievement through Real World Learning in Engineering, Science, Mathematics and Technology.
Statistical Analysis. Purpose of Statistical Analysis Determines whether the results found in an experiment are meaningful. Answers the question: –Does.
Testing Hypotheses.
Psy B07 Chapter 1Slide 1 ANALYSIS OF VARIANCE. Psy B07 Chapter 1Slide 2 t-test refresher  In chapter 7 we talked about analyses that could be conducted.
Chapter 8 Introduction to Hypothesis Testing
Statistical Analysis Statistical Analysis
Group Discussion Explain the difference between assignment bias and selection bias. Which one is a threat to internal validity and which is a threat to.
T-test Mechanics. Z-score If we know the population mean and standard deviation, for any value of X we can compute a z-score Z-score tells us how far.
The Evaluation of Mathematics and Science Partnership Program A Quasi Experimental Design Study Abdallah Bendada, MSP Director
Measuring Changes in Teachers’ Mathematics Content Knowledge Dr. Amy Germuth Compass Consulting Group, LLC.
Evaluating Outcomes Across the Partnerships Tom Loveless Director, Brown Center on Education Policy The Brookings Institution Saturday,
T tests comparing two means t tests comparing two means.
Evaluating the Vermont Mathematics Initiative (VMI) in a Value Added Context H. ‘Bud’ Meyers, Ph.D. College of Education and Social Services University.
Reaching for Excellence in Middle and High School Science Teaching Partnership Cooperative Partners Tennessee Department of Education College of Arts and.
T-test EDRS Educational Research & Statistics. n Most common and popular statistical test when comparing TWO sample means. n T-tests, though used often.
THE DRAGON CONNECTION March Who are we?  Jefferson City Schools  Small, rural school district 60 miles north of Atlanta, 18 miles north of the.
St. Cloud Partnership in Mathematics Grant Presented by: Jona Deavel, Math Coach/7-8 th Grade Math Teacher and Jenny Merriam, Grant Coordinator.
Research Strategies Chapter 6. Research steps Literature Review identify a new idea for research, form a hypothesis and a prediction, Methodology define.
Mathematics and Science Education U.S. Department of Education.
Educational Research: Competencies for Analysis and Application, 9 th edition. Gay, Mills, & Airasian © 2009 Pearson Education, Inc. All rights reserved.
U.S. Department of Education Mathematics and Science Partnerships: FY 2005 Summary.
Assisting GPRA Report for MSP Xiaodong Zhang, Westat MSP Regional Conference Miami, January 7-9, 2008.
Psychology 301 Chapters & Differences Between Two Means Introduction to Analysis of Variance Multiple Comparisons.
Statistics Psych 231: Research Methods in Psychology.
Understanding Research Design Can have confusing terms Research Methodology The entire process from question to analysis Research Design Clearly defined.
Mid-State Mathematics Partnership Excellence in Teaching and Learning in Middle Schools Dovie Kimmins, Ed.D. E. Ray Phillips, Ph.D. Tennessee Mathematics,
Research Study by Michela DeBari.  In many districts, at the middle school level, foreign language classes, often have advanced students, who have had.
Mathematics and Science Partnerships: Summary of the Performance Period 2008 Annual Reports U.S. Department of Education.
INTRODUCTION TO ANALYSIS OF VARIANCE (ANOVA). COURSE CONTENT WHAT IS ANOVA DIFFERENT TYPES OF ANOVA ANOVA THEORY WORKED EXAMPLE IN EXCEL –GENERATING THE.
Evaluating Impacts of MSP Grants Hilary Rhodes, PhD Ellen Bobronnikov February 22, 2010 Common Issues and Recommendations.
Essential Question:  How do scientists use statistical analyses to draw meaningful conclusions from experimental results?
Arkansas Capacity Building Science Partnership Grant: Beyond Traditional Professional Development Models 2008 Math Science Partnership Regional Conference.
Chapter 9 Three Tests of Significance Winston Jackson and Norine Verberg Methods: Doing Social Research, 4e.
Mathematics and Science Partnerships: Summary of the FY2006 Annual Reports U.S. Department of Education.
The Evaluation of Mathematics and Science Partnerships Program A Quasi Experimental Design Study Abdallah Bendada, Title II Director
NORMA WILT WILMINGTON UNIVERSITY The Connection Between Writing and Reading Comprehension.
Evaluating Impacts of MSP Grants Ellen Bobronnikov Hilary Rhodes January 11, 2010 Common Issues and Recommendations.
Three Broad Purposes of Quantitative Research 1. Description 2. Theory Testing 3. Theory Generation.
Mathematics and Science Partnerships: Summary of the Performance Period 2008 Annual Reports U.S. Department of Education.
Chapter 6: Analyzing and Interpreting Quantitative Data
Evaluating Impacts of MSP Grants Ellen Bobronnikov January 6, 2009 Common Issues and Potential Solutions.
T tests comparing two means t tests comparing two means.
Evaluation Requirements for MSP and Characteristics of Designs to Estimate Impacts with Confidence Ellen Bobronnikov February 16, 2011.
BHS Methods in Behavioral Sciences I May 9, 2003 Chapter 6 and 7 (Ray) Control: The Keystone of the Experimental Method.
Exercises. Exercise 1. A researcher uses his morning and afternoon history classes to test a new instructional strategy. In the morning class the students.
Educational Research Inferential Statistics Chapter th Chapter 12- 8th Gay and Airasian.
Inferential Statistics Psych 231: Research Methods in Psychology.
EVALUATING A MIDDLE SCHOOL MATH M.ED. PROFESSIONAL DEVELOPMENT PROGRAM. Knowledge, Pedagogy, Practice or Student Achievement:
Research and Evaluation
Issues in Evaluating Educational Research
Evaluation Requirements for MSP and Characteristics of Designs to Estimate Impacts with Confidence Ellen Bobronnikov March 23, 2011.
EXPERIMENTAL RESEARCH
Evaluation of An Urban Natural Science Initiative
Hypothesis tests Single sample Z
12 Inferential Analysis.
Chapter Eight: Quantitative Methods
Quantitative Research
Comparing Populations
12 Inferential Analysis.
Presentation transcript:

Quasi-Experimental Designs For Evaluating MSP Projects: Processes & Some Results Dr. George N. Bratton Project Evaluator in Arkansas

Purposes of Evaluation An evaluator collects, summarizes, and analyzes data to a) inform participants, b) inform project staff, c) inform other stakeholders/partners, and d) satisfy the reporting requirements of funding agencies/sources.

Math-Science Partnerships MSP’s conduct research based interventions which will directly impact teachers’ content knowledge. This impact will be reflected in improved teaching practices which result in improved student achievement as measured by either a norm-referenced or a criterion- referenced examination.

As a project evaluator I have used quasi-experimental designs which employ matched-pairs methodology as a control mechanism. Also, I have relied solely on the results of norm-referenced or criterion-referenced examinations for the quantification of changes. Doing so increases the probability of valid, replicable results.

Quasi – Experimental Designs Not a method of choice, rather a fallback strategy. Useful when the nature of the independent variable precludes randomization or randomization is simply not feasible.

Matched – Pairs Methodology Note that even nonequivalent comparison groups can bolster your design. For teacher-participants – pre-test and post-test For student achievement within a classroom – pre- project to current year comparison of subject area of interest to another subject area For school achievement – cluster analysis to identify a non-participating pair school, then pre- & post- project subject area of interest comparisons

Instruments Teacher Content Knowledge – Diagnostic Teacher Assessments in Mathematics and Science (DTAMS) Arkansas Comprehensive Testing, Assessment, and Accountability Program (ACTAAP) Iowa Tests of Basic Skills (ITBS)

Quantitative Impact on Teachers: Analysis of Gain Scores on DTAMS (MSMP) Participating teachers were given DTAMS examinations Algebraic Ideas, Geometry/Measurement, and Probability/Statistics in the summer of 2005 and again in the spring of During the period of time between these examinations those teachers engaged in the project’s professional development activities. For the n=45 individuals that took all six examinations the paired t-test procedure was employed on the overall score, the Knowledge Type scores, and the Subcategory scores for each pair of exams. For each examination the Knowledge Types are I-memorized/factual knowledge, II-conceptual understanding, III-reasoning/problem solving, and IV-pedagogical content knowledge. The Subcategories are mathematical subject dependent.

Paired t-test For a test – treat – retest experimental design without a control group, one must employ gain scores (Post-test score minus pre- test score) rather than a standard two group test of means because the two sets of scores (pre and post) are not stochastically independent of each other. This set of differences is then tested via the standard t-statistic with degrees of freedom = n-1 as the basis for judging the null hypothesis that the true difference is zero. The alternate hypothesis for these tests was that the true difference is not zero. This t-statistic is the sample mean difference divided by the standard error (standard deviation divided by square root of sample size).

Score TypeMean Gain Standard Dev T- value P-valueSignificance Know I High Know II <.00001High Know III Not Signif Know IV <.00001High Total <.00001High Patterns High Expressions <.00001High Equations Not Signif Algebraic Ideas Results Middle School Math Project

Student Achievement Analysis ACTAAP Math vs ACTAAP Language SEARK Mathematics Project Some test results were reported from n=58 participants, however there were 10 participants who reported all four desired scores (ACTAAP Math & Language 2006 and ACTAAP Math & Language 2007) The teacher is the experimental (aggregation) unit, not the student. Criterion is the percentage of teacher’s students scoring at advanced or proficient level.

Computational Details Null hypotheses is there is no change in the difference between 2006 and 2007 and the alternate hypothesis is that there is a positive change. The t-statistic is t=1.757 with 9 degrees of freedom. The level of significance is

Student Achievement Analysis MSMP ACTAAP Math Cluster Analysis Example Cluster Analysis is a category of procedures and measurements that result in a numerical taxonomy of a set of observations based on some set of quantitative variables. Such procedures are widely used in both the biological and social sciences as well as in marketing research. Using this approach a “match” was determined for each participating district. Note: In MSMP all grade level mathematics teachers in the 10 districts participated, so the district was the experimental unit.

I used the following district variables to base the clustering: enrollment, % of free and reduced lunch, % of gifted and talented, % minority, and millage (tax rate). All variables were standardized so that measurement scale differences wouldn’t play a part in distance computation.

For large sample sizes the difference in sample proportions can be assessed via a standard normal distribution employing the formula, where the t subscript refers to the treatment group, the c subscript refers to the control (non-participating group) and

. For both years, 2005 and 2007, a z-score was obtained for each pair of schools for each of grades 6-8. These z-scores were averaged across all schools for each year so that an average z-score of differences was determined for each year. The only statistically significant difference was at grade 8. This is extremely important since the 8 th grade students in 2007 had been 6 th grade students when the project began.

The instructional practices and assessments discussed or shown in this presentation are not intended as an endorsement by the U.S. Department of Education.