Task 5: Are you sure? Flashcards
Internal validity is ..
the ability of your research design to adequately test your hypotheses.
Confounding happens when..
two or more variables combine in such way that their effects cannot be separated.
The 7 threats to external validity:
- History (event other than treatment)
- Maturation (age or fatigue)
- Testing (prior testing affects post-treatment testing)
- Instrumentation (instrument calibration)
- Statistical regression (extreme scores move closer to the mean)
- Selection bias (subjects exposed to diff. treatments)
- Experimental mortality (loss of subjects).
You can only enhance the internal validity during the..
design phase of your study.
External validity is..
the degree that the results of a study can be generalised beyond the limited research setting and sample they were obtained.
Threats to external validity (4):
1- Reactive testing (pre-testing affects participants reaction)
2- Interactions between participant selection bias and the independent variable (effects only apply to participants)
3- Reactive effects of experimental arrangements (artificial experimental situations and participants knowledge that they are tested)
4- Multiple treatment interference (early + later treatment)
The steps you take to increase one type of validity tend to ..
decrease the other.
Research settings :
1- Laboratory setting
2- Field setting
Characteristics of laboratory setting are:
- setting that is artificial relative to the natural environment ;
- researcher gains important control
- lose a degree of generality.
The process of recreating the world in the laboratory is called
simulation. (observe -> identify -> reproduce)
Characteristics of the field setting:
- experiment conducted in the participant’s natural environment ;
- high external validity
- low internal validity.
Random sample:
every person in the population has an equal chance of being chosen for the study.
Non-random sample:
usually individuals from a highly specialised subpopulation.
Volunteer bias explains that..
a sample made up entirely of volunteers is biased, because the external validity of your experiment may be affected.
What are the types of validity?
- Content (questions cover the range of behaviours)
- Construct (results agree with theory)
- Criterion-related (correlation = concurrent / predictive)