Alternative Research Designs

Experimental Psychology

Lecture, Chapter 13

Importance of Internal Validity

nInternal validity evaluates whether or not your IV is the only possible explanation for any change in your DV; without it, conclusions cannot be drawn.

nConfounding is caused by uncontrolled extraneous variables that vary systematically with the IV.

nExtraneous variables may unintentionally operate to influence the DV.

nCause-and-effect relationships are established when we know that a particular IV (cause) leads to specific changes in the DV (effect).

How to Protect Internal Validity

nRandom assignment, a control technique that ensures that each participant has an equal chance of being assigned to any group in an experiment, protects internal validity.

nHaving larger groups may make group equality more difficult.

nRandom selection, choosing participants from a population in such a way that all possible participants have an equal opportunity to be chosen, is important to establish external validity.

Threats to Internal Validity

nSelection (controlled with random assignment)

nHistory (controlled with pretest and posttest for both groups)

nMaturation (controlled with pretest and posttest for both groups)

nTesting (controlled with pretest and posttest for both groups)

nStatistical regression & interactions with selection (controlled with random assignment from the same extreme pool)

nExperimental mortality (detected with 2 measurement times)

nInstrumentation

nDiffusion or imitation of treatments

Designs

nPretest-Posttest Control-Group Design – 2 groups of participants, randomly assigned, pretested and posttested (has potential testing effects due to pretest); described on p. 333

nSolomon Four-Group Design – same as above with 2 additional groups that have not been tested (controls for internal validity but no statistical test to analyze all 6 sets of data); p. 334

nPosttest Only Control-Group Design – 2 groups, randomly assigned, both posttested; can be extended to include additional groups or IVs; p. 335

Single Case Designs

nA single-case experimental design is an experiment that consists of one participant (N = 1 design).

nA case study approach uses observation to compile numerous datum about a single case.

nSingle-Case Experimental Designs, developed by B. F. Skinner, involve repeated measures, baseline measurement, and changing one variable at a time.

Single-Case Experimental Designs

n“A” = refers to the baseline measurement

n“B” = refers to the outcome measurement

n“A-B design” = measure baseline behavior, institute a treatment, and use a posttest.

n“A-B-A design” = measure baseline behavior, institute treatment, use posttest, then return to baseline condition

n“A-B-A-B design” = baseline, treatment, posttest, return to baseline, repeated treatment, and 2^nd posttest

nIssues: impracticality, impossibility, and/or ethical compromise

Quasi-Experimental Designs

nA quasi-experimental design is used when you cannot randomly assign your participants to groups but you do manipulate an IV and measure a DV.

nNonequivalent group design – 2+ groups not randomly assigned; a comparison group is compared to one or more treatment groups; nonequivalent because no random assignment

nIt is imperative to select comparison group as similar as possible to the control group

nInterrupted time-series designs involve a single group of participants, include repeated pretreatment measures, an applied treatment, and repeated posttreatment measures.

nHistory, the main threat to internal validity, can be controlled by testing frequently, including a comparison group, or removing the treatment after applied (if possible).