Teaching Biostatistics in Medicine and Allied Health Sciences

Preface
Contents
A Survey of Biostatistics Teaching in Medicine and Dentistry in Higher Education in the UK
	1	 Introduction
	2	 Materials and Methods
	3	 Results
	4	 Conclusion
	References
Evidence-Based Practice Teaching for Undergraduate Dental Students
	1	 Introduction
	2	 Description of the EBP Theme Components
		2.1	 Year 1 EBP
			Introductory Asynchronous Session
			E-Lectures and Small Group Tutorials
			Summative Assessment
		2.2	 Years 2 and 3 EBP
			Systematic Reviews and Critical Appraisal of Randomised Controlled Trials (RCTs) (Year 2 Term 1)
			Using Clinical Guidelines (Year 2 Term 2)
			Qualitative Research Methods (Year 2 Term 3)
			Critical Appraisal of Qualitative Research Methods (Year 3 Term 1)
			Meta-analysis and Critical Appraisal of Observational Studies (Year 3 Term 2)
			Amalgamating and Disseminating Published Evidence (Year 3 Term 3)
		2.3	 Year 4 EBP
	3	 Experience of Running the EBP Theme
		3.1	 Assessment Results
		3.2	 Feedback
			Student Feedback
			Peer Feedback
			External Examiner Feedback
	4	 Conclusions
	References
Teaching Medical Statistics Within the Context of Evidence Based Medicine
	1	 History of Evidence Based Medicine
	2	 Purpose of Statistics Teaching
		2.1	 Doctors as Consumers of Research
		2.2	 Doctors as Producers of Research
	3	 EBM Teaching at Nottingham
	4	 Statistics Teaching Within the EBM Module
	5	 How Statistics Is Taught Alongside Other Disciplines
	6	 How Significant Is Our Teaching?
	7	 Future Challenges and Opportunities
	8	 Conclusion
	References
Teaching Null Hypothesis Significance Testing (NHST) in the Health Sciences: The Significance of Significance
	1	 Introduction
	2	 Teaching Statistics to Non-specialists
		2.1	 Teaching NHST: An Overview
	3	 Approaches to Teaching NHST and the P-Value
		3.1	 Teaching the Underlying Principles of Statistics
		3.2	 Sampling Distributions
		3.3	 Deriving the P-Value
		3.4	 What Does P < 0.05 Mean?
		3.5	 History of NHST
		3.6	 Developing a Culture of Uncertainty
	4	 Statistics Reform
	5	 Concluding Comments
	References
Teaching Conceptual Understanding of p-Values and of Confidence Intervals, Whilst Steering Away from Common Misinterpretation
	1	 Introduction
		1.1	 Concern That Statistical Inference Is Hard to Understand
		1.2	 Excessive Focus on p < 0.05 and Declaring Results as “Significant”
		1.3	 Evidence of Over-generalising Conclusions from p-Values
		1.4	 Concern Over Standards of Statistical Interpretation in the Applied Literature
		1.5	 Concerns That Odds Ratios (and Similar) and Are Ignored
		1.6	 Concerns That CIs Are Not Understood and/or Ignored
	2	 Proposal
		2.1	 Setting Where Teaching Was Delivered
		2.2	 Choose Language That Emphasises Calculated Results Amongst Study Participants
		2.3	 Confidence Intervals (CIs) Interpretations That Promote Understanding
		2.4	 Potential Misunderstanding When Focusing on the distribution of Sample Means
		2.5	 Plotting CIs for the Mean on Histograms of the Distribution of Data
		2.6	 Promoting the Interpreting of p-Values in Terms of Their Strength of Evidence
		2.7	 Promoting the Interpreting of p-Values on a Continuous Scale
		2.8	 Interpreting p-Values Using Language of Compatibility
		2.9	 Type I Error Rates and Quantifying Compatibility
		2.10	 Biased p-Values and CIs (and Biased Odds Ratios or Other Measures) When Analysis Plan Is Not Pre-specified
		2.11	 Student Feedback on MPH Course That Included Many Strategies to Aid Conceptual Understanding
		2.12	 Conclusions
	References
Using Directed Acyclic Graphs (DAGs) to Represent the Data Generating Mechanisms of Disease and Healthcare Pathways: A Guide for Educators, Students, Practitioners and Researchers
	1	 Introduction
		1.1	 Bridging the Divide Between Clinical and Statistical Expertise Using Directed Acyclic Graphs
		1.2	 Directed Acyclic Graphs as Epistemological and Educational Tools for Inferential Statistics
		1.3	 The Vulnerability of Directed Acyclic Graphs to Imperfect Aetiological Knowledge and Professional/Operational Experience
		1.4	 The Utility of Temporal Logic When Teaching, Learning and Using Directed Acyclic Graphs
	2	 Drawing Directed Acyclic Graphs Using Temporal Logic
		2.1	 Challenges Facing the Development of Sufficiently Comprehensive and Temporally Robust DAGs
		2.2	 Challenges Posed by the Omission of Measured, Unmeasured and Unacknowledged Variables
		2.3	 Challenges Facing the Temporal Positioning of Time-Variant Variables
		2.4	 Challenges Facing the Specification of Causal Relationships Amongst Non-asynchronous Variables
		2.5	 Challenges Posed by Intractable Uncertainties Regarding Temporally Obscure Variables
	3	 Using DAGs to Identify/Address Potential Sources of Analytical or Inferential Bias
		3.1	 Identifying Potential Errors and Biases Associated with Epistemological Assumptions
		3.2	 Identifying Potential Sources of Conditioning-Dependent Analytical and Inferential Bias
		3.3	 Mitigating Potential Sources of Bias When Selecting the Datasets and Statistical Adjustment Sets Used
		3.4	 Identifying Potential Sources of Bias During the Critical Appraisal of Published Analyses
	4	 A Role for Educators in Addressing the Limitations of Directed Acyclic Graphs
		4.1	 Transparent Reporting of Context-Specific Uncertainties, Assessments and Potential Errors
		4.2	 A Role for Educators in Optimising the Utility of Directed Acyclic Graphs
	Appendix 1: A Comprehensive List of Variables Considered Relevant by Escobar et al. [25]
	Appendix 2: Ten Recommendations to Improve the Application and Reporting of DAGs
	References
Statistics Without Maths: Using Random Sampling to Teach Hypothesis Testing
	1	 Introduction
		1.1	 Statistics and Anxiety
		1.2	 Recommendations on Teaching Statistics to Non-specialists
	2	 Our Teaching Approach
		2.1	 Statistics at Swansea Medical School
		2.2	 Our Teaching Journey
	3	 Teaching Without Maths
		3.1	 Statistics and Maths
		3.2	 Why Avoid Equations?
	4	 Statistics Using Random Sampling
		4.1	 Starting with the Core Concepts
		4.2	 Structure of the Session
		4.3	 Building on the Basics
	5	 The Response from the Students
	6	 Conclusions
	References
COVID-19: Online Not Distant—MSc Students’ Feedback on an Alternative Approach to Teaching ‘Research Methods and Introduction to Statistics’ at UCL Queen Square Institute of Neurology
	1	 Introduction
	2	 Materials and Methods
		2.1	 Setting
		2.2	 Participants
		2.3	 Module Design and Delivery
			Asynchronous, Independent Activities: Preparation and Recap
			Synchronous, Interactive Sessions: Q&As and Workshops
			Asynchronous, Interactive Activities: Discussion Forums
			Moodle Page
		2.4	 Collection of Feedback
		2.5	 Statistical Analysis
	3	 Results
	4	 Discussion
	References
Common Misconceptions of Online Statistics Teaching
	1	 Introduction
	2	 The Operating Model of CASC
		2.1	 Pre-COVID19 Era
		2.2	 Post-COVID19 Era
	3	 Top 10 Common Misconceptions About Online Statistics Courses
		3.1	 Infrastructure/Equipment
		3.2	 Teaching Methods and Classroom Environment
		3.3	 Planning Ahead
	4	 Conclusions
	References
Authentic Project-Based Assessment Using the Islands: Instructor’s View
	1	 Introduction
		1.1	 Project-Based Learning and Assessment
		1.2	 A Virtual Population
	2	 Context and Design
		2.1	 Class (Pre-project) Activities
		2.2	 Project Activities
	3	 Discussion
		3.1	 Strengths of Islands Based Project
		3.2	 Considerations for Implementing Project-Based Learning with a Virtual Tool
		3.3	 Will this Suit Me as a Lecturer?
	4	 Conclusions
	References
An Interactive Application Demonstrating Frequentist and Bayesian Inferential Frameworks
	1	 Introduction
	2	 Accessing the Application
	3	 Structure of the Application
	4	 Frequentist: Simulate
		4.1	 Inputs
			Binomial Distribution: Probability (θ)
			Sample Size (n)
			Number of Replications (k)
			Seed Value for the Random Number
		4.2	 Outputs
			Observed and Expected Proportions
			Sampling Distribution
			Sample Estimator
			Confidence Interval (CI)
		4.3	 Activities
	5	 Frequentist: Apply
		5.1	 Inputs
			Sample Size (n)
			Hypothesised Population Parameter (θ0)
			Significance Level or Type 1 Error (α)
		5.2	 Outputs
			Observed Data
			Sample Estimate and 95% CI
			Null Hypothesis Significance Testing (NHST)
		5.3	 Activities
	6	 Bayesian: Simulate
		6.1	 Inputs
			Parameter
			Sample Size (n)
			Prior: Parameters of Beta Distribution
		6.2	 Outputs
			Likelihood L(θ| y)
			Prior Distribution f(θ)
			Posterior Distribution f(θ| y)
		6.3	 Activities
	7	 Bayesian: Apply
		7.1	 Inputs
			Sample Size (n)
			Prior: Parameters of Beta Distribution
			MCMC Conditions
		7.2	 Outputs
			Observed Data
			Likelihood Function L(θ| y)
			Prior Distribution f(θ)
			Posterior Distribution f(θ| y)
			Convergence Diagnostics and Output Analysis (CODA)
		7.3	 Activities
	8	 Lesson Plan
		8.1	 Frequentist Framework
			Intended Learning Objectives
			Topics
		8.2	 Bayesian Framework
			Intended Learning Objectives
			Topics
	9	 Conclusions
	References
Teaching Data Analysis to Life Scientists Using “R” Statistical Software: Challenges, Opportunities, and Effective Methods
	1	 Introduction
		1.1	 Why Do (Life-) Scientists Need Statistics?
		1.2	 Teaching and Learning Statistics
		1.3	 What Is R?
		1.4	 Climbing the R Learning Curve
	2	 Teaching Statistics Using R
		2.1	 Methods for Teaching and Learning Statistics Using R
		2.2	 Undergraduate Teaching
			Undergraduates’ Responses to Learning Statistics Using R
		2.3	 Postgraduate Teaching
			Annotation by Learners of Template Analysis Scripts
			Guidebook with Generic Script Coding that Maps Onto Our Other Teaching Materials
			Informal, Learner-Led “Data Analysis Clinics”
			Friendly Online Support via Email
			“R-Space”: A Facebook Forum that Facilitates Peer-to-Peer Interaction as Well as Expert Input
			Video Tutorials Enabling Independent Learning
			Postgraduates’ Responses to Learning Statistics Using R
	3	 Conclusions
	References
Statistics in a World Without Science
	1	 Introduction
		1.1	 Problems
			Common Pitfalls of Exploratory Analysis Using NHST
				Statistical Hypothesis Testing Does Not Test a Scientific Hypothesis
				Data-Dependent Analyses Are Biased
				No Appropriate Consideration of the Sample Size Means No Appreciation of Error Rates
				The Experimental Null and Statistical Null May Be Different
				A Good Fitting Statistical Model Does Not Imply Validity
		1.2	 Simple Solutions
		1.3	 What About Discovery?
	2	 Conclusions
	References
Killing Me Softly with Your Stats Teaching: How Much Stats Is Too Much Stats?
	1	 Introduction
	2	 Setting the Scene
	3	 Aims of the Session
	4	 Process
	5	 Summary of Ideas and Discussion
		5.1	 Reflections on the Challenges of Teaching Statistics in Life-Sciences Degrees
		5.2	 Reflections on Content of Statistics Courses in Life-Sciences Degrees
		5.3	 Reflections on the Format of Statistics Courses in Life-Sciences Degrees
	6	 Final Reflections and Conclusions
	References
Life as a Medical Statistician
	1	 Introduction
	2	 Research
	3	 Teaching
	4	 Administration
	5	 Reflections
	References