Psy 232 Class notes GF07: Experimental research strategies

Manipulate an IV in order to determine effects on a DV

If done correctly, then one can infer a causal relation between changes in the IV and changes in the DV

Major issues in experimental methods:

            internal validity - Can causality be correctly inferred?

            reliability - Can the observed changes be replicated in similar experiments?

            sensitivity - Can effect of IV on DV be detected, even when effect is small?

            external validity - Can findings be generalized beyond the experimental situation?

Uses:

            test hypotheses generated by theories, thus improving understanding of behaviors

            test effectiveness of treatments or interventions

Major characteristic of experiments:  CONTROL

   How achieved:

            via correct manipulation of relevant IVs

            via holding potentially confounding variables constant

            via random assignment of participants to groups, or balancing condition in repeated measures designs

  Purpose:

            to ensure that changes in IV caused changes in DV

  Internal validity:

            covariation of changes in DV with changes in IV

            time order relation changes in IV always precede changes in DV

            elimination of confounding vars (plausible alternative explanations) that covary with the intended IV

            random assignment of participants to groups VERY IMPORTANT

                        Note:  distinguish this from random selection.

                        Block random assignment:  keep # of participants in each condition constant, and ensure that

                                    each participant potentially could be placed in any particular condition

            Major threats even when the foregoing are done well --

               Differential loss of participants from some groups seriously threatens internal validity

               Demand and Experimental effects

                        placebo control groups help to eliminate demand characteristics

                        double-blind experiments control both demand characteristics and experimenter effects

External validity:

            Do findings generalize to other, similar people or similar situations, locations, etc?

            random assignment applies ONLY to internal validity; it does not ensure external validity

            use of variations of stimuli, of participants may help

            replication with other locations, other people, etc. = major technique for establishing external validity

            random selection difficult to achieve when studying humans

            experiments specifically for theory-testing may not be immediately concerned with external validity

            meta-analyses

Special problem:  naturally occurring groups thus pseudo-experiment