Research Methodology Quiz

Designing a Study: Methodology and Practical Considerations

• The research question drives the study design
• A good research question is specific, focused, feasible, relevant, and answerable.
• Weak questions are too broad, vague, or lack specificity.

Examples of Weak Research Questions

• "How does diet affect health?" - Too broad; lacks specificity; needs to specify diet type, health outcome, and population
• "Why do people get sick?" - Vague and untestable; too broad; needs to focus on a specific factor or population
• "Is exercise good?" - Too simplistic and subjective; needs to define "good" and provide context

Common Pitfalls to Avoid

• Being too broad or vague
• Unclear or unmeasurable outcomes
• Assuming causation in associational studies

The PICO Framework

• Population: Who is being studied?
• Intervention/Exposure: What is the main factor being examined?
• Comparison: Is there a control or comparison group? (RCT, Case Control)
• Outcome: What result is being measured?

Examples Using PICO

• Does a mindfulness program improve anxiety in college students compared to no program?
• Is air pollution associated with increased asthma in children compared to low pollution exposure?

Linking Questions to Study Design

• The research question guides the choice of study design
• Randomized Controlled Trials (RCTs) are best for causal questions testing interventions
• Cohort Studies are best for studying effects of exposures over time
• Case-Control studies are best for studying associations with rare diseases or outcomes
• Cross-Sectional studies are best for descriptive studies assessing prevalence and associations at a single point in time

Choosing a Study Design

• RCTs: Strongest evidence for causation, minimizes confounding, establishes temporal sequence, can study multiple outcomes
• Cohort Studies: Efficient for rare outcomes, suitable for common exposures
• Case-Control Studies: Efficient for rare outcomes, can examine multiple exposures

The Study Population

• A group of individuals (or entities) who will be measured
• Must match your research question
• Define the target population
• Specify inclusion and exclusion criteria
• Example: Adults aged 30-65 with diabetes in a specific region

Target Population vs Study Sample

• Target population: Broader group, results apply to
• Study sample: Subset you actually study

Inclusion and Exclusion Criteria

• Inclusion: Required characteristics for participation
• Exclusion: Characteristics that disqualify participants

Recruitment vs Secondary Data

• Recruitment: Actively enrolling participants meeting study criteria (Surveys, questionnarie, interviews)
• Secondary data: Utilizing existing data for other purposes (e.g., health records, census data)

Sampling Methods (Probability)

• Simple random sampling: Equal chance for selection
• Stratified sampling: Dividing the population into subgroups
• Cluster sampling: Selecting naturally occurring groups

Sampling Methods (Non-Probability)

• Convenience sampling: Easily accessible participants
• Purposive sampling: Based on specific criteria
• Snowball sampling: Participants recruit others

Ensuring Valid Population Selection

• Align population closely with research question
• Use clear, specific inclusion/exclusion criteria
• Consider representativeness and generalizability

Practical Exercise: Defining Your Population

• Choose a research question
• Define the target population
• List inclusion and exclusion criteria
• Describe the ideal sampling method

Introduction to Variables

• Variables are characteristics or factors measured in a study
• Can be independent (cause), dependent (effect), or confounding
• Identifying variables is crucial for robust research

Levels of Measurement

• Nominal: Categories with no order (e.g., eye color)
• Ordinal: Ordered categories (e.g., education level)
• Interval: Ordered, equal intervals, no true zero
• Ratio: Ordered, equal intervals, true zero

Operationalizing Variables

• Translating abstract concepts into measurable terms
• Example: "Stress" as "cortisol levels in saliva"
• Crucial for clarity and consistency in measurement

Selecting Measurement Tools

• Surveys and questionnaires
• Physical measurements (e.g., blood pressure cuff)
• Observational techniques
• Digital tracking (e.g., smartphone apps)

Addressing Measurement Error

• Random error: Unpredictable fluctuations
• Systematic error: Consistent bias
• Strategies: Calibration, Multiple measures

Handling Missing Data

• Types: Missing completely at random, at random, not at random
• Approaches: Listwise deletion, mean imputation, multiple imputation, need to consider that on analysis

Measurement in Different Research Designs

• Experimental: Controlled manipulation of variables
• Observational: Measuring naturally occurring phenomena
• Important to consider and choose the appropriate measures

Ethical Considerations in Measurement

• Informed consent for data collection
• Privacy and confidentiality of measurements
• Cultural sensitivity in tool selection and administration

Choosing the Right Statistical Test

• Match test to data type and research question
• Consider parametric vs. non-parametric tests

Main Analysis Approaches by Study Design

• RCTs: T-tests, ANOVA, Chi-square, Regression
• Cohort Studies: Relative risk, Hazard Ratios, Kaplan-Meier, Logistic Regression
• Case-Control Studies: Odds Ratio, Conditional Logistic Regression, Chi-square

Power Analysis

• Determines sample size
• Considers effect size, alpha level, and desired power

Assumptions of Statistical Tests

• Normality, homoscedasticity, independence

Sensitivity Analysis

• Tests robustness of results to different assumptions
• Important for ensuring confidence in findings
• Used if missing data are an issue in the data

Subgroup Analysis

• Examines effects within specific population subsets
• Can reveal heterogeneous treatment effects
• Increased risk of false positives

Reporting Results

• Presentation of key findings: effect sizes, and confidence intervals
• Appropriate use of p-values and statistical significance

Interpretation Pitfalls/Limitations

• Correlation vs. causation confusion
• Over-reliance on p-values
• Ignoring practical significance

Reproducibility in Analysis

• Documentation of all analysis steps
• Use of version control for code
• Sharing data and analysis scripts

Ethics & Feasibility

• Rights and well-being of research participants
• Key principles: respect for persons, beneficence, justice
• Informed consent
• Protecting participant privacy
• Ethical review processes
• Assessing study feasibility: Resources, time, funding, personnel, and practical constraints

Creating a Study Timeline

• Break the study into stages: recruitment, data collection, analysis
• Use Gantt charts to visualize milestones and deadlines
• Account for potential delays

Budgeting for Research Costs

• Participants’ incentives
• Equipment costs
• Staff/personnel costs

Addressing Recruitment Challenges

• Finding eligible participants
• Broaden inclusion criteria
• Using multiple recruitment methods
• Consider pilot testing

Managing Data Collection Issues

• Unreliable tools, timing issues, low compliance
• Validating instruments
• Offering flexible scheduling
• Backup plans

Balancing Ethics and Practicality

• Ethical considerations impact study design
• Resource limitations affect ability and ideal procedure
• Aim is to have a balance between riguorous ethical design and practical constraints