Quasi-Experiments and Observational Studies

Final Exam Information

• Final exam date: Tuesday, December 10, 2024
• Time: 11:30 AM
• Location: Recreational Gym

Quasi-Experiments

• Definition: A quasi-experiment is an experiment that does not use random assignment.
• Reasons for using a quasi-experiment: Researchers might use a quasi-experiment instead of a true experiment when random assignment is not possible or ethical.
• Types of quasi-experimental designs:
  • Pre-experimental design
  • Nonequivalent groups design
  • Time-series design
  • Multiple time-series design
• Potential confounds: Confounding variables are variables that can influence the results of a study, which might be different in quasi-experiments compared to true experiments.

Observational Studies

• Distinction from people-watching: Observational studies are more structured and methodical than simply watching people. They are designed with specific research questions in mind.
• Types of observational studies:
  • Naturalistic observation: Observing subjects in their natural environment
  • Participant observation: Observing subjects while participating in their environment
• Disguised vs. Undisguised: Disguised observation means participants are unaware of being observed, while undisguised observation means they are aware
  • Lab experiments and field experiments
• Experimenter effects: The potential influence observers have on the behavior of the subjects.
  • Ways to lessen experimenter effects: Avoiding observer bias may lessen these effects

Data Collection

• Narrative records, recoding, data reduction
• Checklists, sampling methods
• Nomothetic approach: Focuses on general laws.
• Idiographic approach: Focuses on unique patterns of individuals.

Single-Subject Experimental Designs

• Distinction from case studies
• Circumstances for use

Baseline Data

• Characteristics: A clear baseline that shows consistent and stable performance
• Importance: Helps researchers to identify changes in behavior and measure the effectiveness of interventions
  • Withdrawal design
  • Reversal design
  • Alternating-treatments design
  • Multiple-baselines design
  • Changing-criterion design

Threats to Internal Validity

• Factors that can affect internal validity in single-subject experimental designs

Physical Trace Studies

• Benefits and dangers of using physical trace studies
  • Traces vs. products
  • Accretion vs. erosion
  • Natural vs. controlled measures
  • Continuous vs. discontinuous records
  • Records versus documents
  • Coding systems

"Big Data"

• Definition and examples of big datasets
• Research questions that are best suited for analyses of big datasets
• Difficulties of working with big datasets

Validity

• Discussion of internal validity and external validity
• Definitions of independent variables, dependent variables, and subject variables.
• How correlations might occur (e.g., X causes Y; Y causes X; Z influences both X and Y; spurious correlations)
• Mean, median, and mode
• Standard deviations and ranges, entropy

Error Types

• Type I error: False positive (looking for an effect that doesn't exist)
• Type II error: False negative (not detecting an effect that exists)
  • Random selection
  • Random assignment
  • Factorial designs: Within-subjects, between-groups, mixed designs (split plot).
• Describing factorial design: -Identifying main effects -Identifying interactions
  • Calculating marginal means
  • Describing cells

Research Designs

• Describing different research designs such as true experiments, quasi-experiments, correlational studies, observational studies.