- In a population a person may not necessarily copy the behaviour that’s popular, but the behaviours they interact with.
- Populations have a kind of memory that keeps them locked into the same behaviour, making it difficult to change even if the size of the social group competing with it is reduced.
- With data a model could test the effectiveness of public health campaigns or other policy interventions to reduce unhealthy behaviours, such as smoking or eating junk foods, or to increase healthy behaviours like exercise.
Certain human behaviours tend to spread rapidly like a virus. This is observed in anything from habits in the kinds of food people eat to clothes they buy simply by who they interact with. It tends to be a given that human behaviours are socially determined, as they are usually learned through social interactions with others. Yet the ways in which they spread are not dissimilar to how some diseases spread – through social contact.
If there is an unhealthy behaviour that you want to reduce in a population, such as smoking, then in theory you could model it in a similar way to the spread of disease in an epidemiological model. However, while the term contagion usually connotes something harmful, healthy behaviours, such as choosing not to smoke, can also spread in this way.
When you model the behaviours of different groups in a population you can not only find out how the unhealthy behaviours spread, and discourage them, but also how to promote healthy behaviours so they stick. An epidemiological model allows you to do this because you’re reducing a population to a few key characteristics relevant to the infection, or in this case smoking behaviour.
To smoke or not to smoke
Using smoking as an example, as in an epidemiological model you would break up the population into two different groups: smokers and non-smokers, who would each be susceptible to changing their behaviour. Depending on the size of the groups and the span of their influence one would likely dominate over the other. The chance that a smoker will stop smoking or a potential smoker to start is determined by background rates setup by the modeller.
However, to match the model closer to reality it would be better to include multiple peer influences, such as former and potential smokers. This is because people tend to be exposed to multiple influences that affect whether or not they choose to smoke. Add these groups to the model and you will likely get results that reflect the actual acceptance or rejection of smoking behaviour. The model can then be used as a guide for understanding the behavioural causes of smoking.
Multiple peer influence
For any behaviour that is socially determined like smoking, the behaviour you copy usually involves someone converting from a non-smoker to a smoker or vice versa. A person influencing you to smoke or not depends on your susceptibility to smoking behaviour. If you were a former smoker you may be more likely to take up smoking compared to a potential smoker, who has never smoked before. These factors can be translated into parameters for a multiple peer influence model.
The rates at which people take up smoking and the size of the groups you’re modelling can all be included in the model. Once multiple peer influences are accounted for things start to get more interesting. Small changes in the different groups tend to make surprisingly big differences in determining whether a population takes up smoking or not, and with population data, these models could show which factors are most sensitive to the population outcome.
Point of no return
In research published in the recent book Tipping Points: Modelling Social Problems and Health, it was found that a smoking-free population could potentially stay smoke-free. In the model smokers could make up to 40 per cent of a population, but if the number of former smokers that encouraged others to quit crossed a certain threshold the number of smokers would drop massively. Even if you decreased the number of former smokers (the censure rate) in the model that puts less pressure on smokers to quit, the smoking population continues to stay low. This is known as a ‘hysteresis effect’ — a kind of memory or inertia that keeps the population in the same state.
In the case of smoking behaviour, when you have a lower number of smokers there will also be a lower rate of recruitment to smoking. The system is highly coupled, with multiple effects going on all things become amplified, allowing small changes to have large impacts on the population.
Although population data are needed to compare the results of the model’s results, the finding could be potentially useful to health policymakers because it demonstrates that in a population there may be ways to keep a population from smoking.
In the model, non-smokers dominated the population by encouraging current smokers to quit and successfully maintained a smoking free population. However, the opposite could also occur, if the pressure from former smokers to get current smokers to quit drops far enough then the smoking rate would rise, but as in the previous case, it is likely that no small increase in the pressure to quit from former smokers would change the system.
Data not included
While the model shows some interesting consequences of behavioural change when a population is subjected to multiple peer influences, data is needed to verify how well it resembles reality. However, what it does provide is a useful mathematical description of how human behaviours such as smoking spread in a population. Not only health researchers need apply. A similar approach could be used to model all kinds of behaviour, but learning how to reduce the spread of unhealthy behaviours in society is a clear application, as it is possible to include in the model actions of government policies and health campaigns to test their effectiveness before they are applied.
References and Further Reading
Bissell, J. J., Caiado, C. C. S., Goldstein, M. & Straughan, B. (2014). Compartmental modelling of social dynamics with generalised peer incidence. Mathematical models and methods in applied sciences 24(4): 719–750 (free to read online)
Straughn, B., Bissel J.J., Caiado, C., Goldstein, M., Curtis, S. Tipping Points: Modelling Social Problems and Health. Wiley