Reliability and Validity Flashcards
What is validity?
Measuring what it is supposed to measure.
Concurrent validity
Highly and positively correlated with another construct measuring the same thing.
Predictive validity
Predicts the behaviour it is theoretically linked to.
Construct validity
Apropriate for construct being measured based on operational definition.
Convergent validity
Positive correlation between measures of the same construct.
Difference between concurrent and convergent validity?
Concurrent refers to multiple constructs and convergent looks at two ways of measuring same construct.
Divergent validity
No relationship between measurements of different constructs.
External validity
Generalisability of findings.
4 threats to generalisability
- Selection Bias
- Volunteer Bias
- Participant characteristics
- Cross-species generalisations
What is the novelty effect?
Limiting generalisability due to novel setting. eg lab vs real world
What is the practice affect?
Participants improve through the repetition required in some studies.
What is the fatigure affect?
Results inaccurate as participants get tired after repetition in a study.
What is sensitisation?
An enhanced response to stimuli that has been presented repeatedly.
How does time of measurement affect a study?
Different time of day may illicit a different response eg. morning vs afternoon
What is internal validity?
Limited to experimental design - factors that allow for interpretation of results
What are confounding variables?
Other influences on a study that systematically influence results
Extraneous variables
Environmental conditions that may or may not impact results.
What are the threats to external and internal validity?
Experimenter bias
Demand characteristics
Participant reactivity - hawthorne affect
What is reliability?
Consistency in a measurement
Sources of errors
Observer
Environmental changes
Participant changes
3 forms of external reliability
- Test-retest
- Parallel forms
- Inter-rater reliability
What is parallel forms reliability?
Two different tests asking the same thing
What is internal reliability?
Assessing the same construct using different measures
Split half reliability
Splits the scale into 2 random sets and scores on both should be roughly the same.
K-R Kuder-Richardson 20
Looks at all possible correlations and takes the average
Cronbach’s Alpha
Looks at correlations between items in a measure
Standardisation involves…..
Same delivery
same instructions
same scoring
