Stacks Image 15

12. Single System Research Design

This section introduces ‘single-system research designs’, sometime referred to as single-case or single-subject research designs (NB: Not to be confused with a case study). They involve key research strategies in behaviour analysis where the fundamental research agenda involves searching for functional relations between dependent and independent variables. A system in this context can be the behavioural repertoire of an individual person, a family, or a community. For example, if you want to know how independent variables in the environment (e.g., in clinical, educational, sport, autism settings) affect a particular behaviour, you monitor that behaviour before these interventions are introduced, then you continue to monitor the behaviour while they are in place, and finally you check how the behaviour change is maintained after the environmental changes have been removed.

We are talking here about ‘intra-’ rather than ‘inter-systemic’ changes. In other words, the focus is not on comparing one individual with another, but rather on comparing the same behaviour of one individual (system) before, during and after independent variables are systematically introduced.

‘... experiments to evaluate treatment effects involve directly observing and measuring one or more specific behaviors of an individual repeatedly for a period of time while a particular treatment is not in place (the control or baseline condition), and while it is (the experimental or treatment condition). Events other than the treatment that might affect the target behavior(s) are eliminated (controlled) as much as possible.’ (Green, 2008, p.74. In O'Brien, Dennis M., et al. Autism in Pennsylvania: What Lies Ahead? [Harrisburg, Pa.]: Pennsylvania House of Representatives).

This approach is firmly anchored in the scientific method that does not rely on inferential statistics. Students in Psychology classes seldom get taught how the scientific method can be used in this way. They come to study Psychology because they have an interest in working with people and they end up being taught exclusively about population statistics. In fact, they also get taught that any other approach to the scientific method other than comparing groups of people using inferential statistics is to be regarded as a ‘quasi’ experimental method, a method that only looks like science in action (Movie 12.1). Bombarded by this indoctrination, they end up with virtually no skills appropriate to the various needs of monitoring the behavioural stream of an individual in clinical or educational settings (Movie 12.2).

There are a couple of interesting facts about the value of population statistics in relation to the needs of individuals. For example:
No one is ‘normal’. Everybody is made up of a multitude of individual characteristics. If you take an average of each of them (height, shoe size, length of fingers etc), you won’t find any individual who is average in all respects. This is known as the ‘jaggedness principle’. 

In the 1940s the Jaggedness Principle forced the US Air Force to refit fighter planes with adjustable seats and other fixtures. The cockpits were originally designed around the average range of 10 body measurements taken from a population of 4,063 pilots. But because no single pilot met all those criteria, they ended up with a seat which actually didn’t fit anybody. (

Here is a second one:
In 2011 the Australian Bureau of Statistics used the national census to find the average Australian. They announced that she is a 37-year-old woman with a son and a daughter aged six and nine. She is 162 cms (5’4”) tall, and weighs 71.1 kgs (11st). She lives in a three-bedroom house, has about $200,000 still to pay on her mortgage and her family originally came from the UK. However, when they checked this description against the census data they couldn’t find a single person in the whole country that fits that description.

Finally, the third one:
The average length of human gestation is 280 days, or 40 weeks, from the first day of the woman's last menstrual period. The medical term for the due-date is Estimated Date of Confinement (EDC). However, only about 4% per cent of women actually give birth on their EDC.

Clearly, there are serious limitations with relying on population averages to make decisions that are relevant to the individual. These problems, however, are more subtle when it comes to devising an educational or clinical intervention. One interesting and obvious facts really drives it home about the need to be aware of the dangers of averages and it is this:

“The average human has one breast and one testicle.” (Des MacHale)

Research strategies to investigate influences on the behavioural stream are best served by recording characteristics of the stream and then systematically adding and withdrawing independent variables. There are a range of basic single-system designs that have been used to address various issues in the analysis of behaviour (Movies 12.3-12.7). Which one you should use depends on the context of your research question. However, they all have a number of things in common:

  • Data from each repeated measurement of individual behaviour are plotted on a graph for immediate visual analysis;
  • Internal validity is achieved by each participant experiencing both the control and treatment conditions (intra-system comparison);
  • External validity is achieved through replication, an essential ingredient of science built into single-system designs;
  • The focus is on clinically and educationally important improvements in individual behaviour in comparison to baseline, not on statistically significant differences between group mean scores;
  • Individual differences and variability in behaviour are viewed as natural features to be studied further, not as unwelcome “noise”;
  • Independent variables are tightly controlled and their effects on behaviour directly observed. (adapted from Green, 2008)
Visual inspection of data is particularly useful as it allows the ‘client’, pupil or caregiver immediate access to information about treatment progress. Basically, if the data do not change in the desired direction, everyone can see it very quickly and clearly. This means that the intervention can be adjusted pretty much immediately in response to the data that are being recorded; a truly client-centred approach. This immediate feedback loop of single-system design also makes its use highly desirable from an ethically point of view, because it offers immediate accountability for the effectiveness of the intervention. In essence, this means that when behavioural outcomes are not being achieved, the focus of attention is placed squarely on the design of the intervention, a conclusion that students and parents of children with autism find easy to understand. The tactics used in the search for functional relations also helps guard against mentalistic interpretations of the causes of behaviour.

Additional video resources