Research Methods - Observational methods + design

Observational methods

Different methods of observing participants in a piece of research. Some can co-exist but NOT ALL. They are:

Naturalistic OR controlled, Overt OR covert, Participant OR non-participant

Naturalistic Observation

No variables are controlled or manipulated by the researcher. Behaviour is often observed in a natural setting.

+) High ecological validity - likely behaviour is natural

-)Low internal validity - extraneous variables cannot be controlled

Controlled Observations

Variables are controlled and manipulated by the researcher. Behaviour is often observed in a controlled laboratory setting.

+) High internal validity - extraneous variables can be controlled

-) Low ecological validity - likely behaviour is unnatural

Overt Observations

Participants are aware they are being observed.

+) Ethical issues are less likely - Participant is able to give consent to be observed

-) Low ecological validity - likely behaviour is unnatural

Covert Observations

Participants are unaware they are being observed.

+) High ecological validity - likely behaviour is natural

-) Ethical issues - participant cannot give consent to be observed

Participant Observations

When the researcher becomes a member of the group they are observing.

+) Extreme/small details that would have been missed can be picked up by researcher

-) Greater risk of observer bias - researcher may become subjective and influence observations - low internal validity

Observer Bias

When a researcher becomes subjective and bias due to being influenced by their observations. This occurs often in participant observations (when observations recorded are influenced by being in close proximity to pts) as well as when unstructured observation is used (as the researcher may only record behaviours they are expecting).

Non-Participant Observations

Researcher remains outside of the group that they are observing.

+) Less risk of observer bias, more likely to remain objective - high internal validity

-) Researcher may miss small details

Observational Design

There are two types: Unstructured and Structured. Both consider how observations are recorded.

Unstructured Observation

When a researcher does not use any systems to record their observations - they simply write down everything that they see (which is relevant).

Unstructured Observation - Evaluation

+) Tends to produce accounts of behaviour that are rich in detail

-) Greater risk of observer bias as the observer may only record behaviours they are expecting to see

-) Tends to produce qualitative data which is difficult to analyse objectively and relies on subjective interpretation. This reduces internal validity of the findings.

Structured Observations

When a researcher uses systems to organise their observations. These systems includeL Behavioural categories, Time Sampling and Event sampling. They tend to produce quantitative data.

Behavioural Categories

When the researcher breaks down a target behaviour (the broad behaviour(s) they are interested in) into a set of behavioural categories. These categories should be: operationalised/clearly defined, exclusive/without crossover, cover only specific observable behaviours, cover all possible forms of a target behaviour.

Behavioural Categories - EXAMPLE

Target Behaviour = affection

Behavioural categories = kissing, holding hands, cuddling, touching shoulders, shoving etc

Time Sampling

When an observer records behaviours at predetermined time intervals (eg every 30 seconds). Useful when many relevant behaviours are likely to occur throughout the observation as it reduced number of observations that need to be made, so those observations can be made in more detail. However, behaviours between the time intervals are not recorded, therefore findings may not be representative.

Event Sampling

When the Observer records the number of times a certain (singular) behaviour (the ‘event’) occurs in a target individual/group. This is useful if a behaviour is infrequent and could be missed if time sampling is used. Cannot be used alongside behavioural categories + time sampling like other systems.

Structured Observations - Evaluation

+) There is less risk of observer bias when using systems

+) Produces quantitative data which is easier to analyse objectively, there is increased internal validity of findings

-) Use of Systems, such as event sampling and time sampling, can mean that data lacks detail as important behaviours might be missed (either due to lacking a category or not being observed due to time intervals).