reliability + validity

• reliability:

  • how consistent the findings from the investigation/ measuring device are
  • measuring device - reliable if produces consistent results everytime its used
• reliability of observational techniques:

  • assessing - repeat observation e.g rewatching + inter rate reliability - comparing with observer
  • improving - behavioural categories have to be operationalised, observers need to be equally experienced
  • graph - observer a on x axis, b on y (scattergram = +0.8 good reliability)
• self report techniques:

  • assessing - test/retest reliability e.g give test again, inter-interviewer reliability e.g extent to which interviewers agree
  • improving - reduce ambiguity e.g rewrite questions, use same interviewer each time + good training to reduce leading questions
• experiments:

  • DV often measured using rating scale/ behavioural categories - method used to measure should be consistent
  • improving - standardisation, instructions, procedures same each time, experiment is repeated then compared
• validity:

  • producing a result that is legitimate
  • internal - study measures what its meant to measure
  • external - findings can be generalised beyond research setting
• factors that affect internal validity:

  • investigator effects
  • not operationalised
  • cofounding variables
  • social desirability bias
  • demand characteristics
• ecological validity:

  • to do with how DV is measured
  • example: field experiment godden + baddeley learning word lists - low mundane realism + aware theyre being assessed low ecological validity
• assessing validity:

  • face validity - extent to which test items look like what the test is meant to measure
  • concurrent validity - comparing existing method/test with the one you want to use e.g existing questionnaire, one you want to use
• improving validity:

  • poor face validity - rewrite questions, poor concurrent - remove irrelevant questions
  • demand characteristics + investigator effects - double blind
  • experiment - control group + standardisation
  • questionnaires - anonymity, no social desirability bias
  • observations - covert + behavioural categories operationalised
  • triangulation - number of different sources used as evidence