Does it measure what it was supposed to measured & can it be generalised beyond reseach setting ?
(measure of accuracy - you won't get a mark for this on exam
Internal Validity:
The ability of the study to test the hypothesis that it was designed to test
How accuately can it test?
External Validity:
How well the resulta of the study can be generalised outside the research setting
1 of 6
Factors Affecting Internal Validity
Operanaliseation of Varibales
Operationalise variable / behavioural categores to ensure the study is testing the behaviour that it was set out to test
(not fully operationalised = low internal val.
Control of Extraneous Variables:
Demand Characteristics:
Researchs want to measure REAL/NATURAL behaviour
If the participants change their behaviour because demand characteristics it's not natural behaviour ∴ Results not valid
Experimenter Bias
If researchers are biased in the way they record behaviour - results will reflect researchers interprtation rather than natural behaviour ∴ Results not valid
Participant Variables:
Participant varibale affecting results rather than variables being measured - IV not affecting DV ∴ Results not valid
Experimantal Desing ( Experiments only!)
Independent groups desgn - Participant variables can confound results
Repeated measures design - Order effects can confound results (although it can be reduced by counter-balancing)
2 of 6
Improving Internal Validity (experiment)
Single-Blind Technique:
Reduces demand characteristics
Participants don't know what group or condition they are in & ∴ cannot change their responses to suit / foil researcher
Double-Blind technique:
Reduces demand characteristics & researcher bias
Neither participnts nor experimenter know what each condition or group represents
Revise/Remove Questions:
Poor face validity - questions need to be revised to relate more obviously to topic
Low concurrent validity - reseacher should remove irrelevant questions
Improve Research Design:
To control / overcome:
investigator effects,
demand charcteristics,
confounding variables,
social desirablility bias,
poorly operationalised behaviour categories
3 of 6
Assessing Internal Validity
Face Validity:
Commonsense judgement
Does the test seem to be assessing what it claims to be assessing?
Where a behaviour appeares at 1st sight to represent what is being measured
Concurrent Validity:
If test produces similar findings to another existing measure
Content Validity:
A test has content validity is experts in that field agree that the methodology accuartely measures the desired behaviour
4 of 6
Types of External Validity
Population Validity:
Whether the findings can be applied to the target population from the limited sample of the study
Ecological Validity:
Whether the findings can be generalised to situations outside of the research setting
Temporal Validity:
Whether the findings can be ganeralised to other historical peroids / stay valid over time
5 of 6
Improving External Validity
Population Validity:
Uses large sample
Using a sampling techniqque that is more likely to produce a representative sample
Ecological Validity:
Give participants a task with high mundane realism
If possible, conduct a field experiment / natual experiment
Comments
No comments have yet been made