Experimental Designs: Between-Subjects vs. Within-Subjects

Review of Experimental Research Strategy

  • Goal: Demonstrate cause-and-effect between variables.

  • Requirements: Manipulate IV, measure DV, compare scores, control extraneous variables.

Two Basic Research Designs

  • Within-Subjects Design: Same participants in all conditions.

  • Between-Subjects Design: Different participants in different conditions.

Characteristics of Between-Subjects Designs

  • Manipulate IV, measure DV.

  • Goal: Determine differences between treatment conditions.

  • Requirement: One score per participant.

  • Separate groups for treatments.

Structure of a Between-Subjects Experiment

  • Sample from population.

  • Assign participants to conditions to create equivalent groups.

  • Create separate treatment conditions (IV).

Structure of a Within-Subjects Design

  • Same participants in all treatments.

Advantages of Between-Subjects Designs

  • Scores are independent.

  • Not influenced by practice, fatigue, or contrast effects.

Disadvantages of Between-Subjects Designs

  • Requires many participants.

  • High score variability due to individual differences.

Individual Differences as Confounding Variables

  • Assignment bias: Confounding variable systematically differentiates groups.

  • Threatens internal validity.

  • Makes clear conclusions impossible.

Age as a Confounding Variable

  • Differences may be due to treatments or age differences.

Additional Confounding Variables

  • Individual differences (e.g., age, smartness).

  • Environmental variables (e.g., room size).

Equivalent Groups

  • Groups must be created and treated equally, composed of equivalent individuals.

Limiting Confounding by Individual Differences

  • Random assignment.

  • Matching groups.

  • Holding variables constant.

Individual Differences and Variability

  • Large individual differences = large variance.

  • Influences statistical interpretation.

Differences Between Treatments and Variance Within Treatments

  • Large differences between treatments = good.

  • Large differences within treatments = bad (hides patterns).

Minimizing Variance Within Treatments

  • Increase differences between treatments, decrease variance within treatments.

  • Standardize procedures, limit individual differences, random assignment, matching, sample size.

Other Threats to Internal Validity of Between-Subjects Designs

  • Differential attrition.

Threats to Internal Validity (cont’d.)

  • Communication between groups: Diffusion, compensatory equalization/rivalry, resentful demoralization.

Applications and Statistical Analyses of Between-Subjects Designs

  • Two-group mean difference (t-test).

Advantages and Disadvantages of the Two-Group Design

  • Advantage: simplicity, maximizes treatment difference.

  • Disadvantage: limited information, only two groups.

Comparing Means for More Than Two Groups

  • Single-factor multiple-group design (ANOVA).

  • Stronger evidence than two-group design but caution with too many groups.

Comparing Proportions for Two or More Groups

  • Nominal/ordinal DV (chi-square test).

Within-Subjects Experiments and Internal Validity

  • Single group, all treatments (repeated-measures).

Structure of a Within-Subjects Design

  • Same participants in all treatments.

Threats to Internal Validity to Within-Subjects Experiments

  • Environmental and time-related variables.

Other Threats to Internal Validity

  • History, maturation, instrumentation, regression, order effects.

Limiting Threats to Internal Validity

  • Shorten time between treatments (increases order effects).

  • Counterbalancing treatments (reduces order effects).

Comparing Within-Subjects and Between-Subjects Designs

  • Advantages of within-subjects: fewer participants, eliminates individual differences, reduces variance.

Differences Between the Within- and Between-Subjects Designs

  • Within-subjects: no individual differences between groups.

  • Each subject serves