Modelling is one of the most popular behaviour change methods used in global development. It is especially popular in programming for young people. The typical application of modelling involves showcasing a relatable role model being reinforced for a desired behaviour. It is especially popular amongst NGOs because of its potential reach and impact. For example, a recent study by Riley (2022) showed that having girls watch the film Queen of Katwe, a 2016 film based on the true story of Phiona Mutesi who took up chess aged nine and went on to compete in international tournaments, led to better performance in exams.

Modelling is an interesting behaviour change method to understand because it seems that the scale of interventions that can be designed using it can make it very cost-effective, for example, due to the number of potential viewers of a film whose behaviour could be influenced. To help NGO practitioners take full advantage of modelling as a behaviour change method, this post provides an overview of its theoretical underpinnings and the conditions that make it work. 

It starts with ‘conditioning’

To understand the underlying mechanisms through which modelling works, we first need to understand operant conditioning. Operant conditioning is “a form of associative learning in which the behavioural act becomes associated with its outcome: either reward or punishment, which will result in more or less frequent occurrence of the behaviour” (Bartholomew et al., 2016).

At its core, behaviour change is about learning. Through operant conditioning we learn which behaviours we should engage in (the ones being positively or negative reinforced) and which ones we should stop doing (the ones being positively or negatively punished).

Vicarious learning

Modelling as a behaviour change method has its theoretical basis in Social Learning Theory (Bandura & Walters, 1977). It works through our ability to learn vicariously, that is, we are able to learn by observing the actions and outcomes of other people. This remarkable human ability was famously demonstrated in Albert Bandura’s Bobo Doll experiments. In this experiment, children in the experimental group saw a role model (an adult that resembled a teacher) who repeatedly hit, kicked and was generally unpleasant to the Bobo Doll. When the children were alone with the Bobo Doll in a room, they behaved just as unpleasantly with it, and re-enacted the same behaviours: kicking and punching the doll. For behaviours to be modelled, four key elements must be in place:

  • Attention: we see the behaviour being conducted and reinforced or punished.
  • Retention: we remember the sequence under which the behaviour must be performed.
  • Production: we can independently replicate the behaviour.
  • Motivation: we associate some positive outcome with the behaviour and want to do it again.

We learn best vicariously when (1) we like the social model being presented to us and (2) when we see that their behaviour is being positively reinforced. In the case of the film Queen of Katwe¸ we see the protagonist Phiona being positively reinforced for her accomplishment in chess. We could then hypothesise that because Phiona’s character is likeable and relatable to Ugandan girls, and that she is being positively reinforced, this will lead Ugandan girls wanting to model this behaviour and taking her example as a potential of what they could achieve.

Types of reinforcement

One of the key reasons that modelling works is that we are re-enacting an experience that the observer is not going through in real life. To encourage retention of how the behaviour is to be performed, we must induce an emotional state in the observer. In the case of Queen of Katwe, the reinforcement is associated with positive emotions. There are various types of positive reinforcement that can be used when designing behaviour change products:

  1. Social reinforcement: the model can get praise from people in their immediate social network.
  2. Vicarious reinforcement: we can see the model being reinforced after completing a specific behaviour.
  3. Self-reinforcement: we can see the model rewarding themselves with a gift or something else that is regarded as pleasant. 

Effective application of modelling

There are some parameters that are essential for the correct application of modelling in behaviour change products (Bartholomew et al., 2016). There are at least five key conditions that need to be fulfilled:

  • The learner has to be able to relate to the model. Although this is an obvious parameter, it can still be easy to get this wrong if the communication material or the intervention is not pretested. It is best to not rest on assumptions but rather make sure that the audience can identify with the model.
  • The model has to present a scenario where they are coping with a situation that did not start off well. It is important that the model exemplifies that there has been a struggle or some level of a learning curve that eventually led to them succeeding in whatever they want to do. The model or the storyline of the model must show that they tried multiple solutions before they finally reached the one that worked. This is important because it makes the model relatable to the person we are trying to influence by exemplifying that it is normal not to do the behaviour correctly right away and that some degree of trial and error can be expected.
  • The observer has to watch the model being reinforced. Reinforcement, as discussed earlier is a key ingredient of why modelling works. Therefore, an intervention that includes a model that is not being reinforced in some way is not going to be effective.
  • The social and physical environment have to be conducive to doing the behaviour. It is well and good to model a behaviour that programme developers think is desirable, but if the physical or social environments are not conductive to doing the behaviour (e.g. a key service is not available) then it is not good practice to be promoting the specific behaviour.
  • The person we are communicating to needs to know how to create a facilitative environment. For modelling to work, the learner or the persons we are designing the intervention for must have a minimum of some of the subskills that are needed to conduct the behaviour correctly.

Have you used modelling recently? Share your experience with me here!

Bandura, A., & Walters, R. H. (1977). Social learning theory (Vol. 1). Prentice Hall: Englewood cliffs.

Bartholomew, L. K., Markham, C., Ruiter, R., Fernandez, M., Kok, G., & G.S., P. (2016). Planning Health Promotion Programs: An Intervention Mapping Approach.

Riley, E. (2022). Role Models in Movies: The Impact of Queen of Katwe on Students’ Educational Attainment. The Review of Economics and Statistics, 1–48. https://doi.org/10.1162/rest_a_01153