Reliability

• What is reliability?

  Measure of how consistent findings from an investigation are. If researchers replicate study exactly, they will get similar results

  
  

  Two types-

  • Internal reliability
  • External reliability
• Internal Reliability

  Describes how consistent something is within itself. Way to assess:

  
  

  Split Half Method -

  • Split test into two parts
  • Participants complete both parts
  • Test strength of correlation between two parts
  • Strong correlation above 0.8 indicates internal reliability
• External Reliability

  Describes how consistent a study is when its repeated. Consistent regardless of when the investigation is used or who does it or different participants. Ways to assess:

  
  

  Test retest method-

  Repeat study using same procedures at different times and test correlation between the two versions

  
  

  Inter observer reliability-

  Two or more observers record behaviours during the same observation using set of behavioural categories. Then they test the correlation

  
  

  Formula - Total no.1 of agreements divided by total no.1 of observations. + 0.8 = high inter observer reliability
• Improving Reliability

  Experiments-

  • Used standardised procedures
  • Lab experiments

  Observation-

  • Ensuring behavioural categories have been properly operationalised
  • Categories should not overlap

  Questionnaires-

  • Closed questions to reduce range of possible answers

  Interviews-

  • Use structured interviews instead of unstructured interviews
  • Same interviewer