Experimental Design

Created by Stephanie Peabody, Psy.D. Types - Broad Distinction

Qualitative Quantitative Qualitative Research

Narrative – Subject’s own words – Summarizes behaviors Descriptive Methods – Interviews – Observation notes – Surveys Quantitative Research

Numerical data collected & analyzed Explores relationship between variables – Independent (single or multi-leveled) – Dependent (single or multiple) » Analysis may permit exploration of an between variable Experimental or quasi-experimental Experimental Designs

Characterized by complete of groups or subjects Groups are independent Usually employs strong control Quasi-Experimental Designs

Groups or subjects not randomly assigned – e.g., sample of convenience May not have a comparison group Typical of clinical research – e.g., within subjects repeated measures Less “subject-intensive” Broad Distinctions

Between subjects – Dependent measures taken one time – Data are independent (i.e., not correlated) Within subjects – A “repeated measures” design – Dependent measures taken multiple times – Data are dependent Mixed – Between and within Design Types

Single factor (one-way) – Studies one independent variable Multi-factor – Studies multiple independent variables » May have several levels – Examples: » Two-way (e.g., 2 x 2) » Three-way (e.g., 2 x 2 x 2) Time-series Single Factor Designs

Pretest-posttest (one-group) Pretest-posttest (control group) Posttest-only (control group) Pretest-Posttest (one group)

Quasi-experimental Analysis One set of measures – paired t test taken before and after Weakness treatment or – No comparison or intervention control group Compare pretest and posttest scores Pretest-Posttest (control group)

Experimental design - Significant random assignment differences in Two groups experimental group – Control not found in control – Experimental group attributable to treatment Measures on dependent variable Analysis made on both groups – difference scores pre- and posttest compared with independent t test – ANCOVA pretest score as covariate Multiple Factor Designs

Two-way factorial – e.g., 2 x 3 Three-way factorial – e.g., 2 x 2 x 3 Two-Way Factorial Design

Studies multiple Example: 2 x 3 independent variables ME2 – Main effects (ME) L1 L2 L3 – Each with a number of L1 levels (L) ME1 Permits study of L2 interactions Analysis – ANOVA Three-Way Factorial Design

Studies multiple Example 2 x 2 x 3 independent variables ME 2 L2 – Main effects (ME) L1 – Multiple levels (L) L1 – Interactions effects ME1 Analysis L2 – ANOVA L1 L2 L3 – Post hoc pairwise comparisons ME3 Counterbalanced Design

Possibility of order effects biasing data in a repeated measures design Solutions – Randomize order – Counterbalance trials - order systematically varied » Example - two treatments (T1 - T2) “Crossover design”

Half of subjects - T1 then T2

Half of subjects - T2 then T1 Design

Minimizes order effects Test session 1 2 3 Subject 1 A B C

Subject 2 B C A

Subject 3 C A B Single Subject Design

Permits analysis of effects of treatment in individual subjects (or groups) Elements – Subjects usually own control – Repeated measures – Design phases (times series analysis) Single Subject Design

analysis – Dependent measure is continuous – Establish baseline – Measure treatment effect over time

Baseline Treatment

Time Case Report

Subject a single individual Often uses a narrative format May be non-experimental or experimental – Develops a profile of the subject using: – Visual observation – Interviews/surveys/ – Objective data May provide generalizations about other subjects with similar conditions