Observations

• Observation


  Researchers watching and recording behaviour as it happens
• Controlled observation


  • Controlling the situation the participants experience and recording their behaviours in a lab
  • Helps to control as many variables as possible, giving the participants the same experience
  • Allows for repeating the observation and getting reliable results, high internal validity
• Naturalistic observation


  • Observing participants in their normal environment
  • High realism, participants should behave as they normally would
  • Findings have high external/ecological validity
  • Lack of control means there may be unknown extraneous variables causing the behaviour
• Overt observation


  • Participants can see the researcher and know they are being observed
  • Important for informed consent, but participants may change their behaviour
• Covert observation


  • Participants don't know they are being observed
  • Observes natural behaviour, but can be unethical without informed consent
• Participant observation


  • Researcher becomes involved in the group they are studying
  • Researcher has first-hand knowledge and may build rapport, but risks losing objectivity
• Non-participant observation


  • Researcher stands back and records the group without becoming a part of it
  • Increases objectivity, but may miss important findings from being too removed
• Operationalising behavioural categories


  1. Clearly defining variables to be objectively measured

  2. Example: Defining "aggression" as recording every punch, push and kick
• Time sampling


  1. Recording all relevant behaviour at set points (e.g. 15 seconds every 10 minutes)

  2. Can miss important behaviour that happens outside of the short recording periods
• Event sampling


  1. Recording all the behaviour from the list of operationalised behavioural categories

  2. May need lots of observers to accurately record all participants, and may not record relevant behaviour not on the list
• Assessing reliability


  1. Conducting a test of inter-rater reliability

  2. Using two researchers conducting the same observation separately

  3. Comparing their data sets and expecting a correlation of at least 0.8 to show reliable results