Validity & Reliability Flashcards
Define reliability
Extent to which a test produces consistent findings when repeated
Define external reliability
Reliability over time; produce same results each time test is carried out
How can external reliability be increased?
Test-retest method —> Do same test with same participants again at different time; must be sufficient enough time to prevent ppts remembering answers
Define internal reliability
Items in questionnaire or psychological test are consistent within themselves
How can internal reliability be increased?
Inter-observer reliability —> using multiple observers, avoids subjectivity bias + unreliability in the collection process
How can inter-observer reliability be improved?
- pilot study = ensures behaviour categories are measured in the same way
- behavioural categories = ensure they are looking at the same behaviour
- can also be applied to interviews (inter-interviewer) and content analysis (inter-rater)
How can reliability of different methods be assessed?
Split-half method (self-report techniques)
- splits a test in two, have ppts do both halves
- if both halves have similar results = high IR
Correlation (quantitative data)
- correlation of +0.8 and significant = reliable
Experiments
- replication
How can reliability of different methods be improved?
~ questionnaires = rewrite questions to reduce ambiguity of interpretation
~ interviews = same interviewer, training to avoid leading/ambiguous questions; structured interviews
~ experiments = keep conditions the same, standardise procedure
~ observations = 2+ observers, operationalise behavioural categories, ensure no overlap
Define validity
Measure of truth/accuracy; extent to which test measured what it claims to be measuring
Define internal validity
How much findings of the dependent variable are caused by IV and not other factors
Define external validity
How much results can be generalised beyond the study
Describe the three types of external validity
- ecological validity = can results be generalised to the real world and life
- population validity = can results be generalised to the general population
- mundane realism = degrees to which materials and procedures involved reflect events in the real world
Define face validity
Degree to which a procedure, particularly psychological test/assessments, appears to be effective in terms of stated aims
Define concurrent validity
Comparing a new procedure w a similar procedure done before, where validity has already been established —> must score strong positive correlation (+.80)
Define temporal validity
Whether results from a study or concepts in a theory remain true and valid over a period of time
How can validity be improved for experimental research?
~ using control group to compare
~ standardising procedure
~ use double-blind procedure
How can validity of an observation be improved?
~ covert observation, more authentic behaviour
~ ensure behavioural categories are discrete + unambiguous
How can validity of a questionnaire be improved?
~ lie scale; assesses consistency of ppts responses + reduces social desirability bias
~ anonymity, produces more truthful responses
How can validity of qualitative methods be improved?
~ triangulation; use many diff sources of evidence
~ researchers must demonstrate interpretive validity : interpretation of events matches ppts’
