Experimental Design

Experimental Design

1. Good experimental design

   • Confounding variables
     • Cannot determine whether an independent variable and an uncontrolled variable is responsible
   • Internal validity
     • Extent to which the results can be attributed to the independent variable

2. Basic experiment

   • An independent variable with two levels
     • Experimental and control group

3. Independent groups design

   • Also called between groups design
   • Different participants are assigned to each level of the independent variable
   • Pre post test design
     • Pre test scores are compared to check that both groups are equal in the critical variable
     • Pre test can be used to:
       • Determine highest and lowest scores
       • For small sample sizes
       • In longitudinal studies
   • Solomon four group design
     • Treats the pretest as a second independent variable
     • 1/2 receives only the dependant variable, others receive the pre/post test
   • Repeated measures design
     • Also the within subjects design
     • The same participants participate in both conditions
     • Less error variance b/c participants do not vary
     • Effects that may occur:
       • Order effect
         • Order of presenting the treatments affects the dependent variable
         • Counterbalancing
           • Complete counterbalancing
             • All possible orders of presentation are included
           • Counteracts order effect
           • Partial counterbalancing
             • Uses latin square
             • Each condition appears at each ordinal position and precedes and follows each condition
       • Practice effect
         • Improvement of performance as a result of practice
       • Fatigue effect
         • Deterioration in performance as the participant becomes tired
         • Time interval
           • Rest period can counteract a fatigue effect.
       • Contrast effect
         • Response to the second condition is alter b/c the 2 conditions are contrasted

4. Matched pairs design

   • Match people on a participant characteristic
   • Ensures that groups are equal