7.2.6 Observational Design

Observational Design

When designing the observation, the observer has to decide the exact behaviour that they are looking for and how it will be recorded
There are different ways in which data can be gathered including: audio recordings, visual recordings (videos and photographs) and on-the-spot note taking
Observation Design: the choice of behaviour to record and how they are measured

These are operationalised behaviours that the observers will be looking for (we're not entirely sure why they are called 'categories' because it has to be very specific and not quite fit a whole category!)
The researchers will agree on the behaviours that should be recorded, creating these behaviour categories
The behaviour categories need to be able to reflect what is being studied, and then the researcher would develop codes for behaviours relevant to that:
- For example;
  - ABL = Aggressive body language
  - AGL = Aggressive language
  - AGB = Aggressive behaviour
  - NABG = Non-aggressive behaviour
The behaviour categories have to be operationalised to make sure they are specific and the behaviour that is recorded cannot be confused with anything else other than what we are looking for
- For example:
  - 'Aggressive behaviour' is too vague: What constitutes as aggressive? Is shouting as aggressive as punching?
  - We should create specific categories from this. E.g.: 'punching' 'kicking' 'pushing'
  - These three examples are much more specific and aren't easily confused with each other, which makes recording and analysing them also much easier

It can be difficult to observe all behaviour at all times during an observation, especially if it is a continuous observation
Researchers therefore can choose a sampling procedure to help this, such as time sampling and event sampling

The researcher records all behaviours for a set time frame, at a set point, e.g. they record each participant behaviour for 20 seconds every 15 minutes over a 2 hour observation
- Or, they could observe behaviour for 15 minutes every 3 hours of a two day observation
The researcher would choose the time sample relevant to the research
Time sampling allows the researcher flexibility to record behaviour and has the opportunity to record unexpected behaviours
However, it can also miss behaviours that do not happen within the set time

The researcher records/tallies every time a behaviour (from the behaviour categories) happens
The researcher observes these behaviours on target individuals
All behaviours that are included in the behaviour categories are on a list then that behaviour will be recorded
However, behaviours that are not on the categorised list are not recorded, which means relevant behaviours could be missed

Even with clear behaviour categories, interpreting behaviours can be affected by bias
Researchers can test the reliability of their of their own observation by comparing it with another researcher's observation and measuring the level of consistency between the two
Inter-observer reliability is the level of consistency between 2 or more trained observers when they conduct the same observation
- They must agree on the list of behaviour categories and how they are going to record them before the observation begins
- The observation is conducted separately by each observer
- They then compare the two independent data sets and test the correlation using a statistical test (e.g. Spearman's Rho)
- This produces a correlation coefficient result (e.g., 0.63)
- To be considered a good/high level of inter-rater reliability, it must pass a threshold of +0.8 or 80% consistency
Inter-observer reliability means it is less like to have observer bias and therefore, can be a valid representation of behaviour
A high correlation coefficient shows the behavioural categories are reliable