Back in 2011, I published a scientific study on the psychological design of health behaviour change websites in the Journal of Medical Internet Research (JMIR).
With my co-authors Prof Thelwall and Prof Dawes, we designed this study with the goal of identifying how to build websites that could promote public health and also help people improve their wellbeing. For instance, we wanted to understand the psychology behind websites that could motivate people to quit smoking, exercise more, or eat healthier foods.
We were not interested in subjective opinions about what worked; we were not interested in subjective views on what theories were believed to work. Rather, we wanted solid evidence of what worked. But moreover, we wanted to isolate the psychology behind websites that could change a person’s health behaviours.
To achieve this goal, we used a scientific method called meta-analysis. This technique allowed us to pool trends across scientific experiments on behaviour changing websites.
Many of the interventions that we assessed appeared to be simple on the outside, but proved to be complex on the inside. They frequently used tailoring algorithms that, in some cases, contained libraries with hundreds of messages that could be combined into thousands of message combinations.
During our study, we first tested the effectiveness of online health behaviour change websites, to see if they worked. We found that that online interventions proved to be as good as sophisticated print interventions, but with the advantage of lower costs and larger reach.
Before we could assess the psychology of website design, we had to overcome one major obstacle. When we started the project, we could not find an influence system capable of describing the full range of psychological factors that could shape a person’s behaviour through online interaction. To overcome this challenge, we first developed a theoretical model called the Communication-based Influence Components Model.
The model is largely based on the premise that the same psychology employed by person to person interaction is the same as that between a computer and person.
We used this model to integrate persuasion research from several fields into a simple framework suitable to describing or designing persuasive online systems. However, we largely worked with two excellent frameworks by Abraham and Michie (2008) and Michie et al (2005).
Next, we assessed the factors behind successful interventions, and identified the psychological factors that could change users’ behaviours. During our study, we quantified the effectiveness of more than fifty psychological techniques known to influence users’ behaviours. This helped us identify the psychological architectures of health behaviour change websites. In other words, we identified the most common packages of psychological appeals that have been scientifically proven to influence behaviour.
The chart below shows some of the most popular and effective approaches, in terms of influencing users’ behaviours. The numbers represent a statistical unit with the larger number demonstrating the psychological techniques that contributed the largest impact on users’ behaviours.
The psychological architecture of online intervention proved to be a key component of persuasive websites. In our study, the most effective psychological architecture are goal-directed websites. These websites inform users about the consequences of their behaviour, prompt them to set goals, then encourage them to track their progress toward those goals while providing feedback on their performance.
During the process, the most effective approach is simply to provide knowledge about a problem. However, when doing this, it is important to demonstrate the consequences of a particular behaviour (much like the health belief model). Social influence (such as peer pressure or social norms) proved to be a powerful factor across many interventions. Next, motivation and goal setting are highly effective and often integrated with behavioural regulation where people are taught to control their behaviour.
Time is also a critical factor in intervention efficacy, with shorter interventions achieving the largest impacts. This is likely due to the trend where people rapidly drop out of health programs over time. Thus the longer the program, the larger the attrition, and the lower its impact.
The dominant psychological theory was Prochaska’s transtheoretical model, which describes how people progress through stages of change, from denying a problem, to thinking about changing, to planning for change, then changing, and possibly maintaining that change for a lifetime. However, during the process, people relapse continually, and a few of the more sophisticated systems were ready to help people when they relapsed.
Although it is not listed, self efficacy was not an effective psychological construct, but this may be because many of the goal directed interventions were designed for people who were already motivated to change.
This study took over two years to complete, consolidated 95 sources, and drew on over thirty scientific studies, representing 14,895 participants. When I initially published this research in January 2010, it was the first comprehensive statistical meta-analysis on the psychological design of websites. Now, this recent journal paper extends that work by showing the effects of each psychological tactic.
These are the key documents behind this study:
- The original study published in my doctoral thesis
- The full scientific paper is available in the Journal of Medial Internet Research
- The Communication-based Influence Model is described here and a summary on JMIR