reliability and validity Flashcards
what is validity
the extent to which results of a research are legitimate
3 ways to assess validity
face validity
concurrent validity
predictive validity
what is face validity (assessing validity)
-the measure is scrutinised to see whether it appears to measure what its supposed to
what is concurrent validity (assessing validity )(2)
-extent to which a psychological measure compares with an existing measure
-results obtained should match/be similar to the results of the existing measure
what is predictive validity (assessing validity)
-how well a test can predict future events/behaviours
eg. how childhood attachment measured using the strange situation is able to predict how the child will grow up to behave in adulthood
what is internal validity and what is it influenced by
-whether the outcomes are due to only manipulation of the IV
-influenced by:
extraneous + confounding variables
participant variables+demand characteristics
investigator bias
what is external validity and what are the 3 types
extent to which findings can be generalised to other settings, populations and times
-ecological validity, temporal validity, population validity
what is ecological validity
generalisability to other situations and settings
what is temporal validity
generalisability to other historical times and eras
what is population validity
generalisability to different populations (ages, genders, cultures)
how to improve validity in qualitative methods (2)
-use of triangulation (using a number of different sources of evidence)
-researcher demonstrates the interpretative validity of conclusions- the extent to which the researchers interpretations of events matches those of the p’s eg. include direct quotes from p’s
how to improve validity in questionnaires (2)
-use lie scale to assess consistency of responses and control effects of social desirability bias
-assure p’s that all data remains anonymous
how to improve validity in experimental research (3 different ways)
-use a control group- easier to assess if changes in the DV were due to IV
-use standardised procedures to minimise impact of participant reactivity+investigator effects on validity of results
-single blind- reduces demand characteristics. Double blind- reduces demand characteristics and investigator effects
how to improve validity in ovbservations (2)
-covert observations- behaviour more likely to be authentic- higher ecological validity
-behavioural categories shouldnt be broad, ambiguous and overlapping so they dont affect validity of results
what is reliability
reliability=consistency of results over time
what is internal reliability
how consistent something is within itself
what is external reliability
consistent results produced regardless of when the investigation is used
what is split half method (assessing reliability) (2)
-randomly select half of questions and put them in one form, then put the other half in another form
-these 2 forms of the same test are done separately- should yield the same score and have a coefficient of 0.80
what is test retest method (assessing reliability) (3)
-researcher administers the same test on the same person but at different times
-results should yield coefficient of 0.80
-sufficient time should be left between tests so p’s cant recall answers. Time shouldnt be too long as p’s attitudes may change
what is inter-observer reliability (assessing reliability) (3)
-extent to which there is agreement between multiple observers observing the same behaviours
-eliminated subjectivity bias
-should yield coefficient of 0.80- high inter-observer reliability
how to improve reliability in questionnaires (2)
-questionnaires with a low test-retest reliability should have rewritten questions
-replace open questions with closed questions
how to improve reliability in observations (2)
-have operationalised behavioural categories
-categories that arent operationalised well or observers are open to subjectivity can end up with differing, inconsistent results
how to improve reliability in experiments
standardised procedures
why are experiments often reliable
researcher has control over many aspects
-control allows for the experiment to be replicable for similar findings
how to improve reliability in interviews (4)
-use same interviewer each time
-is this isnt possible, all interviewers must be properly trained
-structure their interviewers in a certain manner so everything is similar
-easier in structured interviewers where interviewers behaviour is more controlled by fixed questions
