Quasi-Experiments

• Cross-Sectional Design


  A method for studying developmental changes without running an experiment over time
• Cross-Sectional Design

  1. Select groups of people from different age groups

  2. Compare groups and make conclusions about developmental changes over time
• Potential concern:

  Cohort effects - Individuals belonging to a certain age group may have unique personal experiences based on the state of the world they grew up in
• Older adults may have fewer memories that involve technology. You may want to conclude that technology becomes less important as one gets older. However, life experiences with technology were dramatically different for these groups. Thus, it might not reflect a change in age but a cohort effect.
• Quasi-Experimental (QE) Design


  • Lacks random assignment to groups or conditions
• Internal Validity


  Potential threat to internal validity - Are inferences about cause-effect valid, can you be sure that what you manipulated (IV) caused the changes in what you measured
• Types of QEs

  • Non-Equivalent Groups Design (NEGD): Frequently used design in social research, two groups we think are similar: one is control, other is used treatment
• Single Case Experimental Designs

  Standard single case design:

  First: Measure baseline symptoms

  Second: administer treatment and measure symptoms
• Standard single case design


  • Cannot rule out that client was getting better (or worse) anyway, if treatment included multiple techniques, may not be able to determine which technique was the causal agent
• "Reversal" design


  First: Measure baseline symptoms

  Second: administer treatment and measure symptoms

  Third: remove treatment and monitor for return of symptoms
• "Reversal" design


  • Can rule out if client was changing anyway, ethical considerations
• Multiple-baseline design


  First: Measure baseline symptoms in multiple clients in different settings, or in same client in different settings

  Second: administer treatment and measure symptoms
• The Experimental Method


  Main advantages: Minimizes the effect of individuals, gives good estimates of "typical" individuals in population
• The Single Case "Experimental" Method


  Main advantages: Applicable to clinical settings, gives good estimates of specific individuals in population
• The Experimental Method


  Main Disadvantages: Less applicable to clinical settings, assumptions often violated
• The Single Case "Experimental" Method


  Main Disadvantages: Limitations to external validity (i.e., generalizability), not suitable for many traditional statistical techniques
• Critical Feautures

  Researcher ramdonly assigns PS

  • purposely change something in Ps world
  • attempts to hold all others factors constant
  • measures effect of change
• example?

  2 similar schools or communities