reliability

• reliability


  consistency of measurement - findings of study are reliable if when the same measurement or test is repeated, it produces similar results
• Assessing reliability - test-retest reliability 


  extent to which the same test or measurement given to the same participants on 2 separate occasions produces same results - same participants tested then retested 2nd time in same way to see if scores correlate strongly - usually short interval between tests like a week so participants don't remember answer, interval shouldn't be too long that opinions or abilities can change - e.g. giving participants IQ test and repeating same test with same participants two weeks later before scores from two occasions correlated
• Assessing reliability - inter observer reliability 


  extent which there's agreement between two or more observers involved in observations of a behaviour - whether different observers watching same even independently collect scores that correlate strongly -e.g. 2 observers independently count number of aggressive acts shown by group of schoolchildren in playground using same behavioural categories, comparing results to see if they're consistent - scores from both observers correlated, correlation coefficient can be calculated - result of 0.8 or more suggests high inter-observer reliability
• improving reliability 


  inter-observer reliability low - behavioural categories might not be operationalised clearly enough, removed revised and rewritten - some observers need more practise using categories so can respond more quickly

  test-retest low - test items or questions might be ambiguous, removed, revised and rewritten - test items or questions overly complex or too broad, revised and simplified - test conducted slightly differently each time, ensure all aspects are standardised