Reliability

• Reliability


  Refers to how consistent the findings from an investigation are
• Test-retest reliability


  A method of assessing the reliability of something by assessing the same person on two separate occasions and attempts to show whether results are same both times
• Inter-observer reliability


  The extent to which there is agreement between 2 observers in observing behaviour
• What is the number for having high-inter observer reliability?

  +0.8
• Test-retest method

  1. Is doing the same test or questionnaire but has to be done at a period of time after so they don't recall their answers
  2. Then the two test scores should be correlated to see if they are significant and if they are it is reliable
• Inter-observer reliability method

  1. Observational research introduces subjectivity, bias and unreliability
  2. Inter-observer involves two researchers collecting their own data on observing the same event and comparing results
• Measuring reliability 


  Reliability is measured using a correlation analysis. In test-retest and inter-observer reliability the 2 sets are correlated
• Improving reliability for questionnaires

  • Reliability of questionnaires should be measured using the test-retest method
  • A questionnaire with low test-retest may need to be rewritten to include less ambiguous questions that may be hard to answer
  • To fix this you could have more closed questions
• Improving reliability for interviews

  • Ensuring reliability for interviews is using the same interviewer
  • Interviewers should try not to ask leading questions
  • Interviews that are less structured are less likely to be reliable
• Improving reliability for observations

  • Reliability of observations can be improved by making behavioural categories operationalised
  • If categories are not operationalised well then observers may have to make their own judgement
• Improving reliability for experiments

  • In an experiment it is the procedures that are the focus of the reliability
  • In order to compare the performance of different PPs the procedure must be the same
  • Researchers are concerned with standardised procedures