FBA: Correlation vs. Causation
"To demonstrate a functional relationship, the antecedents and consequences must be manipulated to show their influences on the problem behaviour" - Miltenberger (2008, p. 285-286)
Descriptive and Experimental Methods
There are three methods of carrying out a functional behaviour assessment; these are (1) direct observation, (2) informant methods and (3) functional analysis.
The first two methods (direct observation and informant methods) can be classed as “descriptive assessments” because they describe events that occur around the challenging behaviour.
The third method (functional analysis) is classed as an “experimental assessment” because it experiments with possible causes of the behaviour in order to identify the function of the behaviour (Miltenberger, 2008).
There is a very important distinction between the descriptive and experimental assessments, and it is one of “correlation versus causation”.
Correlation & Causation
Basically, descriptive methods can only provide correlational data. This means that all they can do is give a description about how X seems to happen before the challenging behaviour occurs or that X seems to happen after the challenging behaviour has occurred.
By describing what happens before or after the target behaviour these methods allow practitioners to create a hypothesis about what may be causing the behaviour, but they cannot allow a practitioner to say what is causing the behaviour. That said, these descriptive methods can be used prior to implementing a behavioural intervention and as a result the interventions can be effective (Miltenberger, 2008).
The experimental assessments on the other hand, allow practitioners to identify a causal link between the antecedent and consequence of the challenging behaviour. Once the causal link has been identified, a practitioner can then say that X is causing the behaviour (Cooper, Heron, & Heward, 2007).
This link can be established because experimental methods will deliberately manipulate (change) what happens before and/or after the behaviour to test how these manipulations affect the challenging behaviour.
Data would be recorded on the challenging behaviour throughout each manipulate and this data will allow practitioners to identify the actual cause of the behaviour.
We have written an article demonstrating how a functional analysis can detect the cause of a behaviour - you can read this hypothetical example of a functional analysis here.
References
- Cooper, J., Heron, T., & Heward, W. (2007). Applied Behaviour Analysis. New Jersey: Pearson Education.
- Iwata, B., Pace, G., Dorsey, M., Zarcone, J., Vollmer, T., Smith, R., Rodgers, T., et al (1994). The functions of self-injurious behavior: An experimental-epidemiological analysis. Journal of Applied Behaviour Analysis, 27, 215-240. DOI:10.1901/jaba.1994.27-215
- Miltenberger, R. (2008). Behaviour Modification. Belmont, CA. Wadsworth Publishing.
