📊Experimental Design Unit 9 – Repeated Measures Designs

What's the Deal with Repeated Measures?

• Involve measuring the same participants under different conditions or at different time points
• Reduce variability by using the same subjects across treatments, increasing statistical power
• Require fewer participants compared to between-subjects designs, making them more efficient
• Allow researchers to study changes within individuals over time or across conditions
• Particularly useful when individual differences are large relative to treatment effects
• Help control for individual differences that could otherwise obscure treatment effects
• Introduce the potential for carryover effects, where the impact of one treatment affects subsequent treatments
  • Counterbalancing the order of treatments can help mitigate this issue

Key Concepts You Need to Know

• Within-subjects factor: The independent variable that is manipulated within each participant (e.g., time or treatment condition)
• Between-subjects factor: An independent variable that varies between different groups of participants (e.g., gender or age group)
• Counterbalancing: Varying the order of treatments across participants to control for potential order effects
• Sphericity: The assumption that the variances of the differences between all pairs of treatments are equal
  • Violations of sphericity can lead to inflated Type I error rates
• Carryover effects: The influence of one treatment on the subsequent treatment, which can confound the results
• Practice effects: Improvements in performance due to familiarity with the task or measures over time
• Fatigue effects: Declines in performance due to boredom, tiredness, or decreased motivation over time
• Baseline measurements: Initial measurements taken before any treatments are applied, serving as a reference point

Types of Repeated Measures Designs

• Time-series design: Participants are measured at multiple time points, allowing the study of changes over time
• Crossover design: Each participant receives all treatments in a randomized order, with a washout period between treatments
• Matched-pairs design: Participants are matched based on a relevant variable and then randomly assigned to different treatment orders
• Randomized block design: Participants are divided into homogeneous blocks based on a nuisance variable, and treatments are randomized within each block
• Factorial design: Combines repeated measures with between-subjects factors to study the effects of multiple independent variables simultaneously
• Solomon four-group design: Incorporates both pre-tests and post-tests to control for the potential effects of pre-testing on the results
• Single-case design: Intensively studies a single participant or a small group of participants over time, often used in clinical settings

When to Use (and When to Avoid) Repeated Measures

• Use repeated measures when:
  • Individual differences are expected to be large relative to treatment effects
  • The sample size is limited, as fewer participants are needed compared to between-subjects designs
  • Studying changes within individuals over time or across conditions is of interest
• Avoid repeated measures when:
  • Carryover effects are likely and cannot be adequately controlled for
  • The task or measures are prone to practice or fatigue effects
  • The time between measurements is too short for the effects of one treatment to dissipate before the next treatment
  • The research question involves comparing different groups of participants rather than changes within individuals
• Consider using a combination of repeated measures and between-subjects factors when:
  • Both within-subject and between-subject comparisons are of interest
  • The effects of multiple independent variables need to be examined simultaneously

Setting Up Your Experiment

• Determine the research question and hypotheses, considering the appropriateness of repeated measures
• Select the within-subjects factor(s) and levels, ensuring they are relevant to the research question
• Choose between-subjects factors, if applicable, to create a factorial design
• Determine the sample size based on the desired statistical power and anticipated effect sizes
  • Account for potential attrition due to the repeated nature of the design
• Develop a counterbalancing scheme to control for order effects, ensuring that each treatment order is equally represented
• Establish a timeline for the study, considering the duration of each treatment and the washout period between treatments
• Pilot test the procedure to identify and address any potential issues or confounds
• Obtain informed consent from participants, clearly explaining the repeated nature of the study and any potential risks or discomforts

Dealing with Data: Collection and Analysis

• Ensure consistent data collection procedures across all measurements and participants
• Use appropriate methods to handle missing data, such as multiple imputation or maximum likelihood estimation
• Check for violations of assumptions, particularly sphericity, and apply corrections (e.g., Greenhouse-Geisser or Huynh-Feldt) if needed
• Conduct a repeated measures ANOVA to test for significant effects of the within-subjects factor(s) and any between-subjects factors
  • Use post-hoc tests (e.g., Bonferroni or Tukey) to compare specific treatment levels if a significant effect is found
• Consider using multilevel modeling or linear mixed models for more complex designs or when assumptions are violated
• Report effect sizes (e.g., partial eta-squared or Cohen's d) in addition to p-values to convey the magnitude of the effects
• Visualize the results using appropriate graphs, such as line plots or bar charts, to illustrate changes across treatments or time points

Common Pitfalls and How to Dodge Them

• Order effects: Use counterbalancing to ensure that each treatment order is equally represented across participants
• Carryover effects: Allow sufficient time between treatments for the effects of one treatment to dissipate before administering the next
• Practice effects: Include practice trials or sessions to familiarize participants with the task and reduce the impact of practice on the results
• Fatigue effects: Keep the study duration reasonable and provide breaks as needed to minimize fatigue
• Demand characteristics: Use blinding techniques (e.g., single or double-blind) to minimize participants' awareness of the study's purpose and hypotheses
• Violation of assumptions: Check for violations of assumptions (e.g., sphericity) and apply appropriate corrections or alternative analyses if needed
• Attrition: Anticipate and plan for potential participant dropout, particularly in longer studies with multiple sessions
• Generalizability: Be cautious when generalizing the results beyond the specific sample and conditions studied

Real-World Applications

• Clinical trials: Repeated measures designs are commonly used to assess the effectiveness of treatments over time (e.g., measuring symptoms at baseline and after treatment)
• Educational research: Repeated measures can be used to track student learning and progress across different time points or educational interventions
• Developmental psychology: Researchers often use repeated measures to study how cognitive, social, and emotional processes change across different developmental stages
• Usability testing: Repeated measures designs can assess how user performance and satisfaction evolve as they gain experience with a product or interface
• Sports science: Athletes' performance can be measured across different training phases or interventions to optimize training programs
• Environmental studies: Repeated measures can track changes in environmental variables (e.g., pollution levels) over time or in response to interventions
• Market research: Consumer preferences and behaviors can be assessed across different product variations or marketing strategies using repeated measures designs