- Slides: 6
Internal Validity • All about whether the research design (and data analysis) warrants the conclusions. • Concerned with: – Causal relationships – Various threats to internal validity • • Single group Multiple group Social effects Experimenter effects
Causal Relationships • Three conditions are needed to establish a casual relationship: – Temporal condition: Presumptive cause must precede the effect. – Statistical relationship between cause and effect. • Correlation • t test, ANOVA, or other statistical test of a null hyp. • Elimination of plausible alternative explanations. – This is usually the most difficult to esbablished.
Plausible Alternative Explanations • Single group threats Posttest only design: X-------->O or Prettest-postest design: O-------X------>O Note X is a treatment O is an observation or measurement Threats include: History, maturation, testing, instrumentation, mortality, regression (to the mean)
Plausible Alternative Explanations • Multiple group threats Typically involve interactions of single-group threats and selection (of participants). Selection-history Selection-maturation Selection-testing Selection-instrumentation Selection-regression
Social Effects • Effects that result from social interaction Diffusion of treatment Imitation of treatment Compensatory Rivalry (John Henry Effect) Resentful Demoralization