1. Discovery Learning Projects in Introductory Statistics
PI: Dianna Spence Co-PI: Brad Bailey
Overview
NSF DUE
Objective: Develop materials to help instructors combine real-world, discovery-learning projects in introductory statistics courses; measure the impact on student performance and beliefs.
Design of Pilot Test:
Eight participating instructors (nationwide) test the methods and materials developed to facilitate discovery projects in their classes.
CONTROL GROUPS: Each instructor teaches one or more sections without using the methods and materials designed to facilitate projects. (Completed AY )
TREATMENT GROUPS: During subsequent academic terms, each instructor teaches one or more sections using the methods and materials developed. (Completed AY )
Curriculum Development
Materials (Online/Print)
Student Guide
Instructor Guide
Technology Guide
Contents include
Scoring rubrics
Proposal forms
Guidance at each phase
Condensed and extended project timelines
Resources for developing variables/constructs
Student Project Description
Students generate their own research questions, define their own variables, and draft a research proposal. Upon approval, students collect data; organize, analyze, and interpret results, and report findings in a formal paper and an oral presentation. Below are project types and examples.
Linear Regression – Examples:
NBA Player Salaries and Points per Game
Car Engine Horsepower and Average Miles per Gallon
t-Tests – Examples:
Comparing Males and Females: Attitudes about Tattoos
Comparing Northern and Southern States: Divorce Rates
Instruments: Student Outcomes and Subscales
Content Knowledge (CK) – 17 multiple choice items
Statistics Self-Efficacy (SE) – 16 Likert scale items
Perceived Usefulness of Statistics (PU) – 11 Likert scale items.
Subscales of Student Outcome Measures
Mixed Methods Research
Qualitative Inquiry
Five instructors participated in qualitative data collection.
Quantitative Analyses
Control vs treatment: compare student outcomes
For individual instructors
For overall groups
All 8 instructors (N = 353 Control, 441 Treatment)
Preferred protocol only: 5 instructors (N = 198 Control, 344 Treatment) Instructors omitted for protocol exceptions:
Treatment section taught in summer mini-mester format
Target student population substantially different in treatment group
Treatment section was different course than control
Repeated treatment sections after initial trial: 3 instructors (N = 122 Control; N = 122 / 103 / 43 in 1st, 2nd, 3rd Treatments)
Multivariate models: predict student outcomes by treatment, instructor variables, other factors (esp. for instructors with repeated treatment sections)
Results (Qualitative)
Each theme is explored with respect to codes (left). Participants are compared, with commonalities, differences, and specific manifestations noted.
Sample Observations
Nature of interaction between instructor and students is changed when projects are incorporated (Engagement/communication)
Student misconceptions are easier to identify earlier in learning cycle when carrying out projects (Conceptual learning/project purpose)
Students appear more invested in the outcome of studies they have designed themselves (Real-world application/student dispositions)
Instructors vary widely in how they choose to guide students conceptually through projects (Conceptual learning/Instructor pedagogy)
Results (Quantitative)
Control vs. Treatment – Overall Group Gains: All Instructors
(Overall group declines: none)
Control vs. Treatment – Overall Gains: Instructors within Protocol
Significant Gains and Declines: Individual Instructors (*Protocol exceptions; omitted from table directly above)
The Experience Factor: Instructors Repeating Treatment Sections
Guiding Questions
How do instructors vary in their implementation of the discovery statistics projects?
How do the instructors’ teaching philosophies pervade the project implementation?
What are the implications of the observed instructor differences for student learning?
Data Collection
Written prompts
Course syllabi/materials
Phone interviews
Classroom observations
Analysis
Code and Triangulate (team process)
Identify and describe themes
Synthesize with quantitative findings
Scale
Group N
Mean (% of max) p
Content Knowledge – Subscale: Identifying Analysis C T
353
441
1.33 (44.3%)
1.54 (51.3%)
.001
Self-Efficacy Main Scale
78.51 (81.8%)
80.20 (83.5%)
.021
Self-Efficacy – Subscale:
Hypothesis Testing
23.41 (78.0%)
24.43 (81.4%)
.002
Self-Efficacy – Subscale: Data Collection
19.12 (79.7%)
19.98 (83.3%)
.000
Scale
Group N
Mean (% of max) p
Content Knowledge Main Scale C T
198
344
7.52 (44.2%)
8.63 (50.7%)
.000
Content Knowledge – Subscale: Linear Regression
2.50 (35.7%)
3.12 (44.6%)
Content Knowledge – Subscale: Identifying Analysis
1.23 (41.0%)
1.61 (53.7%)
Self-Efficacy Main Scale
76.00 (79.2%)
80.13 (83.5%)
Self-Efficacy – Subscale:
Hypothesis Testing
22.08 (73.6%)
24.38 (81.3%)
Self-Efficacy – Subscale: Data Collection
18.86 (78.6%)
20.04 (83.5%)
Primary Coding Categories
Emerging Themes
Conceptual Framework
Project Purpose
Instructor Pedagogy
Student Dispositions
Communication
Engagement
Real-world Application
Conceptual Learning
Scale #1 #2 #3 #4
#5*
#6* #7
#8*
CK – Content Knowledge (Main) 
CK Subscale: Linear Regression 
CK Subscale: Hypothesis Testing
CK Subscale: Identifying Analysis
CK Subscale: Sampling
SE – Self Efficacy (Main)
SE Subscale: Linear Regression
SE Subscale: Hypothesis Testing
SE Subscale: Data Collection
SE Subscale: General
Scale / Subscale #1 #2 #7
Group
Content Knowledge 
CK: Linear Regression
CK: Hypothesis Testing
CK: Identifying Analysis
CK: Sampling
Self Efficacy
SE: Linear Regression
SE: Hypothesis Testing
SE: Data Collection
SE: General
Contact Us
PI Dr. Dianna Spence
Department of Mathematics
University of North Georgia
Co-PI Dr. Brad Bailey Dahlonega, GA
Phone:
Project Website
Content Knowledge
Statistics Self-Efficacy
Perceived Utility
Linear Regression
Hypothesis Testing
Identifying Appropriate Type of Analysis
Sampling
Data Collection
General (Learning and Understanding Statistics)
Real-World Relevance
Personal Benefit to Understanding
N = N = N = N = 43
Experience with Treatment: Significant Gains