● Variables are observed but not controlled or

manipulated.

● Permits study of complex processes.

● Little control over phenomena of interest.

● Permits only weak causal inference.

Experimental studies

● Some variables are manipulated directly while

others are observed.

● High control over phenomenon.

● Permits causal inference.

● Accuracy and validity of the statistical

methods used in the experiment.

● Whether the “measurement device” measures

the psychological construct of interest.

● Applicability of the results beyond the

experiment.

Internal validity

● Methodological soundness and lack of bias in

the study.

Test-retest reliability

● Consistency of scores across multiple tests.

Internal consistency

Consistency of all parts of treatment.

Interjudge reliability

● Consistency of raters within experiment.

Matched group designs

● Groups are chosen at random, but matched on

one or more subject variables.

● Dependent variable is measured both before and

after treatment.

Repeated measures design

● Dependent variable is measured repeatedly

during (different levels of) the treatment.

Counterbalanced design

● Repeated measures design in which the

treatment order differs across subject groups.

● An observational experiment in which two or

more variables are measured but nothing is

manipulated.

Case study

● An experiment involving only one subject.

● Different groups must be sampled at random

from a single population!

● Since the experiment only uses one group,

differences between groups across treatments

are irrelevant.

double-blind experiment

● The person administering the treatment does

not know which treatment group the

participant is in.

● The person measuring the dependent variable

does not know which treatment group the

participant is in.

The Hawthorne effect

● Sometimes scores can improve merely because an

experiment is being conducted or because of a

trivial response to a change in the level

treatment.

Subject deception

● Usually manipulation checks are included in order

to detect this.