Observational designs (data collection techniques)

• Event sampling


  Behavioural checklist used to record each time a behavioural category is exhibited by ppt -> tally made of all observed 'events' of each behavioural category that take place during observation period
• Event sampling strengths:

  • Researchers have specific behaviours to look for = high replicability = objective data
• Event sampling weaknesses:

  
  

  • Done in one fixed time period = changes in behaviour occurring over a longer time period may be missed
  • Several behavioural categories to count = behaviours may be missed
  • Chronological order of 'events' ignored = insights into behaviour may be missed
• Time sampling


  Record of behaviours made at set intervals within given time frame with behavioural checklist used to record behaviours exhibited
• Time sampling strengths:

  • Better for situations in which ongoing interactions/behaviours can change over time
  • Time intervals can be designed to catch behaviour at different times of day/week/etc
  • Enables understanding of behaviours observed
• Time sampling weaknesses:

  • Important behaviours may be missed in unobserved/unrecorded time periods
• Inter-rater/inter-observer reliability


  A way to assess if the way you have chosen to measure the DV/co-variables is consistent across different people
• How should inter-rater/inter-observer reliability be assessed?

  • Different people independently doing observation using same categories/people/time + place
  • Compare results of raters/observers by putting both sets of results through inferential statistical test measuring correlation to see degree of similarity in results
  • Should have a strong 0.8 positive correlation
• How can inter-rater/inter-observer reliability be improved?

  • Training data collectors by giving specific instructions on how to evaluate + record behaviour/characteristics, offering chance for discussion of certain problems while monitoring data quality over a period of time
  • Ensure behavioural categories are clear + objective -> give training so that expectations are standardised
  • Use inferential statistical test to check reliability of categories/training -> if correlation is less than 0.8, behavioural categories need to be tightened