Sarah Osman
When researchers tracked what happened after girls in Uganda watched Queen of Katwe, the 2016 film based on Phiona Mutesi, who picked up chess at nine in a Kampala slum and went on to compete internationally, the girls who saw it went on to perform measurably better in their exams (Riley, 2022). One film. A measurable shift in educational attainment. You can see why the sector loves this.
If you have worked in youth programming, you have almost certainly reached for a role model at some point. It is the intuitive move. Find someone the audience admires, show them succeeding, and let aspiration do the rest. And because a single film or radio drama can reach millions, modelling looks like the rare intervention that is both effective and cheap. That combination is exactly why it gets chosen. It is also exactly why it so often quietly fails.
The failure is not in the method. Modelling works, and we understand why it works. The failure is in what gets skipped. Because modelling looks cheap, the temptation is to treat the film, the drama, the celebrity endorsement as the intervention itself. Commission the content, secure the broadcast, count the viewers, report the reach. But the evidence is clear that modelling only changes behaviour when a specific set of conditions is in place, and the cost-efficiency that makes the method attractive is precisely what tempts designers to ignore those conditions.
The five-figure film with a beloved national actor playing a transformed character will not move behaviour if the conditions around it are not designed for. A higher production budget does not compensate for a missing condition. The visibility of the campaign is not evidence that the campaign works.
The method runs on vicarious learning, the ability to acquire behaviour by observing other people perform it. Albert Bandura demonstrated this decades ago when children who watched an adult behave aggressively toward a doll went on to copy that behaviour precisely when left alone with it (Bandura, 1965). The mechanism sits on top of operant conditioning, which is the more basic process by which we learn that certain behaviours are followed by reward or punishment and adjust accordingly (Bartholomew et al., 2016). Vicarious learning lets us absorb that pairing without going through the experience ourselves. We watch someone else do the thing, watch what happens to them, and update our own behaviour as if we had been through it.
For this to actually shift what someone does, four things have to happen in sequence. The observer has to pay attention to the modelled behaviour, retain the sequence of how it was performed, be able to produce the behaviour themselves, and be motivated to do so by some positive outcome they associate with it. Miss any of these and the chain breaks.
We learn best vicariously under two specific conditions. We have to like or identify with the model, and we have to see them being reinforced in some way for the behaviour. The reinforcement can be social (praise from people around the model), vicarious (a visible reward following the behaviour), or self-administered (the model rewarding themselves). What matters is that the observer registers the pairing of behaviour and positive outcome.
There are five conditions that determine whether a modelling intervention will work in practice (Bartholomew et al., 2016).
The audience has to genuinely identify with the model, which is an assumption that needs pretesting rather than asserting. A model who looks aspirational to the programme team but distant to the audience will not produce identification, and without identification there is no vicarious learning.
The model has to be shown being reinforced. A likeable character who succeeds with no visible reward gives the observer nothing to want. This is one of the most common failure points in earnest campaign material that wants to avoid appearing materialistic or transactional. The instinct to strip out reward is precisely what removes the active ingredient.
The social and physical environment has to actually support the behaviour. Modelling a health-seeking behaviour when the clinic does not exist, or modelling a savings behaviour when the financial product is not available, is not a programme. It is a frustration. The behaviour cannot be vicariously learned and then performed if performance is structurally blocked.
The audience needs to already hold a minimum of the subskills required to act on what they have learned. Vicarious learning compresses experience, it does not replace foundational capability. If the modelled behaviour assumes literacy, digital fluency or social permission the audience does not have, the intervention has skipped a step it cannot recover.
And the model has to be shown coping with a situation that did not start well. This is the one I want to spend the rest of the post on, because it is the condition that most consistently gets dropped, and it is the one that does the most work.
The model cannot simply succeed. They have to be shown struggling first. The story has to make the failure visible, the false starts, the trying of things that did not work, before the eventual success arrives. This runs against almost every instinct we have when we produce communication material in the development sector. The instinct is to show the polished, triumphant, aspirational version, the winner at the top of the podium, the smiling success story in the donor report. But that version is the weaker intervention.
When the audience only sees the triumph, the distance between their own life and the model’s becomes a reason to disengage rather than a path to follow. The unblemished success story communicates, however unintentionally, that this person is fundamentally different from you. That they had something you do not have. That the outcome is for them and not for you. The struggle is what closes that distance. It is what tells the viewer that getting things wrong at first is normal, expected and survivable. It is what makes the eventual reinforcement believable.
This is why Queen of Katwe worked: because it showed a girl from Katwe failing, doubting, leaving, returning, and continuing anyway. The reinforcement landed at the end because the struggle made it credible.
The cost-efficiency of modelling is real, but it is conditional, and the conditions are where the actual work lives. The film is not the intervention. The conditions around it are. Skip them and you have not run a cheap programme. You have run no programme at all.
References
Bandura, A. (1965). Influence of models’ reinforcement contingencies on the acquisition of imitative responses. Journal of Personality and Social Psychology, 1(6), 589–595.
Bartholomew, L. K., Markham, C., Ruiter, R., Fernandez, M., Kok, G., & Parcel, G. S. (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.