- Observations take place in particiapant's natural environments and no variables are controlled. EG: seeing how many people are on mobile phones at a bus stop.
- High levels of ecological validity as participants don't know they're being watched - more natural behaviour.
- Allows otherwise impossible or hard-to-examine behaviours to be investigated.
- As variables are not controlled, extraneous variables may become confounding so the results are not due to what we expected.
- Ethical issues: informed consent, confidentiality. Ethical guidelines suggest that they take place in public places where people expect to be observed.
- Some variables are controlled by the researcher. May take place in a lab or items placed in location of interest. EG: observing children in a playroom in a lab via a one-way mirror and labelling the toys male, female or neutral.
- Cause and effect can be established as there are higher levels of control over the EVs.
- Participant's behaviour may not be natural if they know they are being watched so may change how they act, meaning that ecological validity is reduced.
- All relevant behaviour is recorded from direct observation.
- Provides rich, detailed data which can pave the way for future research to be developed.
- Some behaviour recorded may not be relevant and may only be recorded because it was visible or eye-catching.
- Hard to generalise.
- Behavioural categories and sampling procedures are pre-detemined. Data is then collected into pre-prepared collection grid - it is mainly quantitative data.
- Requires the researcher to join in the group or take part in the situation they are studying.
- Observees' behaviour is easier to understand.
- Researcher has to rely on memory which is unreliable.
- Presence of the researcher may change participant's behaviour.
- The researcher may become emotionally involved so wil be subjective.
- Based on observations made from a distance or outside of the location they are studying.
- Participants know they are being observed.
- Participants don't know they are being observed.
- Less demand characteristics as their behaviour will be more natural.
- Observer bias - expectations of the researcher can influence the outcome. For this reason, inter-observer reliability should be calculated.
- Ethical issues - deception, confidentiality. Ethical guidelines suggest doing this in a public place where people expect to be observed.
- Used when we see a continuous stream of behaviour that needs to be broken into categories of different behaviour we are most interested in. Can be a behavioural checklist or a coding system
- Categories should be operationalised - should be made as specific, objective and unambiguous as possible, EG: crying, hitting.
- All behaviour should be covered in the categories and there should be no waste basket category or guessing at meaning.
- Categories should be mutually exclusive and we shouldn't have to mark 2 categories at the same time.
- The researcher counts the number of times a certain behaviour occurs in the target individual. EG: if a behavioural checklist is used, a tick is made every time the behaviour occurs.
- The researcher records behaviours within a given time frame. EG: every 30 seconds, the researcher makes a note of what the target individual is doing. Or, if a coding system is used, a code for the behaviour occuring is recorded every 30 seconds.
Observer bias and inter-observer reliability
- Observer expectations can influence observations. For this reason, inter-observer reliability should be calculated.
- To be a reliable measure, observations should be replicable.
- To test this, usually more than one observer is used.
- Their records of the observation are later collected then compared for consistency and the extent of agreement is calculated.
- More than 80% agreement suggests good inter-observer reliability.