reliability Flashcards
what is reliability
-a measure of consistency
how do we know if information is reliable
-can the study be replicated?(if so its reliable)
-if so will the results be consistent(if so results are relaible)
what is internal reliablity?
-extent to which the measure is consistent within itself(if one part of questionaire shows ur introvert so should other)
what is external reliability
-produces the same results each time the test is carried out (are the results consistent w other studies)
how to check if a test is externaly reliable?(test re test)
(test re test method)
-repeat the test w same Ps diff ocassions again if results are similar it must be reliable
-most common w questionnaires /psych tests/ experiments may be applied to interviews
-should be sufficient time between test and re test that Ps cant recall their answers but not so long that their attitudes and opinions change
how do we check if a test is externally reliable?
(inter rater reliability)
-two or more observers comapre their data by conducting a pilot study to check observers are applying behavioural categories in the same way
-observe the same event but record data independantly
what do you use for content analysis ?
-inter rater reliability
what do you use for interviews ?
-inter interviewer relaibility
how can we measure reliability
-in the test re test and inter observer reliability the 2 sets of scores are correlated
-to be deemed as reliable the correlation co efficient should exceed +.80.
how can you assess the internal reliability
(split half method)
-split in 2 and have the same Particpant do both halves of the test then comapre the results for both halves.
-if the 2 halves produce similar results then it has internal reliability
how can you improve reliability in questionaires
-repetiton of similar and very clear Qs
-a question that produces low re test reliability may need some items re written or removed
OPEN QS= may be misinterpreted so may be replaced w closed Qs so its less ambigious
how can u improve reliability in interviews
-use same interviewer and same questions
-if not possible use trained interviwer to avoid ambigious or leading qs
-easier to control in structured interview
-unstructured is more free flowing and less reliable
how to make experiments more reliable?
-use standardised procedures and instructions
-same and consistent procedures for accurate comparisons
how to make observations more reliable
-behaviourable catrgories should be measureable e.g operationalised (categories shouldnt overlap)
e.g hugging and cuddling and all behvaiours included
-if not operationalised then observers may make own judgements that can differ
-observers may need further training to discuss their decisions.
how to ensure tests are reliable?
1.pilot test
2.standardise
3.inter rater
