Validity

1
Validity

• Internal validityevidence of cause and effect
• Confoundstwo variables cannot be separated
• False positives and false negatives
• Subject variables
• Construct validityoperational definitions
  correspond to underlying constructs
• Convergent and divergent validity

2
Validity

• External validitygeneralizability
• Population and ecological
• Internal validity a prerequisite fro external
  validity
• Statistical validityresults not a product of
  chance
• Statistical versus experimental hypothesis
• Probabilities

3
Threats to Internal Validity

• Pretestposttest designs
• Historyexternal events
• Maturationinternal changes in participants
• Testingpretest effects
• Practice effects
• Test sensitization
• Statistical regression
• Score True score Error
• Regression to the mean
• Instrumentation
• Observer drift

4
More confounds

• SelectionBiased assignment to conditions
• Lack of random assignment
• Intact groupsquasiexperimental research
• Selfselection
• Cohort effects
• Mortalitydifferential attrition
• Selectionmaturation interaction
• Diffusion

5
Threats to Construct Validity

• Loose connection between theory and method
• Subject roles
• Response sets
• Hawthorne effect
• Good subject
• Negativistic subject
• Apprehensiveevaluation apprehension, social
  desirability

6
Threats to External and Statistical Validity

• External validity
• Other subjects
• Species specific mechanisms
• Other timeshistorical effects, e.g., cohort
  effects
• Other settings
• Lab versus field
• Internal versus external validity
• Statistical validity
• Inappropriate choice of analysis
• Lack of statistical power
• Small n
• Chance produced results

7
Demand Charcteristics

• Totality of situational cues
• Role demands
• Subject compliance
• Avoiding demand characteristics
• Deception
• Manipulation checks
• Field studies
• Unobtrusive measures
• Multiple measures
• Debrief
• Placebo effects
• Psychological and physiological changes
• Placebo control
• Double blind

8
Experimenter Bias

• Biased data collection
• Unconscious bias
• Fraud
• Differing interpretations of data is not bias
• Biased interactions with subjects
• Experimenter characteristics
• Rosenthal studies
• Self-fulfilling prophecies
• Avoiding experimenter bias
• Blind and double blind
• Strong protocol
• Remove experimenter