The e-ATLAS (Evaluation and Assessment of Teaching and Learning About Statistics) project is establishing a much needed evidence-based research culture in the statistics education community to better allow it to judge the effectiveness of its past and on-going efforts. Inspired by the internet's vastly expanded reach, resources, and accelerated development of the statistics education community, the project's innovative design seeks to provide customizable instruments that assess how teachers teach and what students learn. These instruments are administered on-line as part of a globally shared digital library of resources and linked to databases supporting assessment of course innovations, project evaluation, and research informing materials. Using these instruments, the e-ATLAS project is conducting a randomized nationally representative sampling of college statistics instructors and their classrooms. This provides a baseline of information on college teaching and student learning enabling investigators to triangulate the global position of their work. Through webinars and well-indexed exemplars the e-ATLAS project then supports new investigators in their drive for effective evaluation of their work.

Project Report

e-ATLAS: Evaluation and Assessment of Teaching and Learning About Statistics 1.Project Goals To better judge the effectiveness of past and on-going efforts to reform the teaching and learning of statistics To provide customizable instruments administered online as part of a globally shared digital library of resources. To provide evaluation models and support for statistics curriculum projects 2. Instruments Developed, Validated, Revised, and Field Tested A. For statistics instructors: Statistics Teaching Inventory (STI), 4 versions: Face-to-face classes (two versions): - One instructor per section - Lecture/recitation format (lecturer plus TA) Online classes (two versions) 100 % onlinr Hybrid classes (online and face to face combined) Survey format: Seven Sections: Pedagogy, Curricular Emphasis, Technology, Assessment, Beliefs, Course Characteristics, Additional Information B. For assessing students’ statistical reasoning: Goals and Outcomes Associated with Learning Statistics (GOALS) 20 forced-choice items which assess statistical reasoning in a first course in statistics Five sub-areas of reasoning, which include reasoning about…. the role of study design in drawing a statistical inference. characteristics of distribution interpretations of statistical inferences making inferences about group differences the relationship between sample size and sampling variability Data Collecting: STI Administered Statistics Teaching Inventory to a sample of instructors of introductory statistics classes – 98 instructors from a national stratified random sample – 400 instructors from the Consortium for the Advancement of Undergraduate Statistics Education (CAUSEWEB) listserv Instructors received an e-mail invitation to take the survey online What we learned about teaching statistics: Many instructors are using recommended practices, for example: More than half (63%) of instructors indicated most or all of the data used is real data Although parametric methods are most commonly emphasized, over 50% of instructors sampled are using randomization methods to some extent In assessments, instructors are emphasizing the ability to reason about statistical concepts more heavily than the ability to use formulas for computation Some evidence of misalignment between beliefs and practices: – Over 90% of instructors use technology, but almost half believe students should learn to read tables (e.g. t-tables, z-tables) – Less than 1/3 of instructors believe lecture should be the predominant method of content delivery, but 80% of instructors predominantly use lectures (or lecture-type materials) – 98% of instructors agree that students should learn connections between the nature of the data and inferences that are made, but only 1/3 are emphasizing the importance of collecting good data. Data Collection: GOALS Instructors recruited through Causeweb.org mailing list and from STI follow-up Administered GOALS test to introductory statistics students across different institutions in the U.S. – 1,153 students taught by 33 instructors (Fall 2013) – 932 students taught by 29 instructors (Spring 2014) GOALS test administered online What we learned about student learning: Students struggled most with reasoning about: the purpose of random assignment (26.7% correct) how correlation does not imply causation (17.3% correct) how, given two distributions with the same range, the one with less mass in the center has the larger standard deviation (30.1% correct) Next steps: wide use of STI and GOALS for research and evaluation project.

Agency
National Science Foundation (NSF)
Institute
Division of Undergraduate Education (DUE)
Type
Standard Grant (Standard)
Application #
1043141
Program Officer
John Haddock
Project Start
Project End
Budget Start
2011-06-01
Budget End
2014-12-31
Support Year
Fiscal Year
2010
Total Cost
$91,970
Indirect Cost
Name
University of Minnesota Twin Cities
Department
Type
DUNS #
City
Minneapolis
State
MN
Country
United States
Zip Code
55455