Internal Validity
Internal Validity
Revolves around the question of whether your IV actually caused any change that
you observe in your DV
Threats to Internal Validity
Any factor that allows for an alternative explanation
Extraneous Variables
Confounding Variables
The History Threat
Events that occur between the DV measurements in a repeated measures design
Distractions in the experimental environment
Experiences people have between measurements
Repeated measures designs are most vulnerable to this threat
The Maturation Threat
Changes in participants that occur over time during an experiment
Can refer to long-term changes (aging, cognitive development) during a
longitudinal study
Can refer to short-term changes (hunger, fatigue, boredom)
How much time it takes before maturation becomes a threat depends on the
demands placed on participants and other factors (amount of sleep participants
had before beginning, participant motivation, etc)
The Testing Threat
Internal validity is threatened when the process of measuring the DV itself causes
a change in the DV
Practice effect—scores are different when taking a test a second time simply due to have
taken the test before.
Reactive measures—DV measurements that change the DV being measured
The Instrumentation Threat
Occurs when the equipment or human measuring the DV changes the measuring
standard over time
Measuring equipment malfunction
Inconsistencies in researcher measurement behavior
Inter-rater reliability—comparing scores of several raters
Standardized Scoring practices
The Threat of Statistical Regression
Occurs when low scorers improve or high scorers fall on a second administration
of a test due solely to statistical reasons
Regression to the mean: extreme scorers are likely to move toward the mean
(mean magnetism)
Example: sample of people who scored very low on the SAT. Statistical
regression says they are likely to score higher next time despite the instructional
intervention given
Example: sample of basketball players with extremely high free throw accuracy.
Distract them. Lower accuracy. Due to distraction or regression?
The Threat of Selection
Choosing participants in such a way that the groups are not equal before the
experiment, thus one cannot be certain that the IV caused any difference we
observe after the experiment.
Selecting participants because they are already members of a certain group.
Differences between groups reflect differences that existed between them
BEFORE the presentation of the IV
Note: A validity threat for true experimental designs, but often done intentionally
in nonexperimental designs.
The Threat of Mortality
“Mortality” means death (animals) or dropping out of an experiment (humans)
Occurs if experimental participants from different groups drop out of the
experiment at different rates (if those from one group drop out at a higher rate
than another group)
Often happens when the experimental condition is noxious, unpleasant, or
demanding
Pre-experiment mortality can impact results.
Example: values-based education, comparing freshmen and seniors. Seniors
report holding stronger values. Due to values-based education or mortality
(people whose values are incongruent with the school’s leave)
The Threat of Diffusion or
Imitation of Treatments
Occurs if participants in one treatment group become familiar with the treatment
of another group and copy that treatment
If participants of one group share information with the other group, the groups
may behave similarly due to having this information
Example: teaching a study strategy to the 11:00 class and not the 1:00 class, but a
loudmouth in the 11:00 class tells a friend in the 1:00 class about the strategy and
that persons starts using it and telling others. (treatment diffusion—all participants
are using the same treatment)