Social community seamless pattern of different group of people in modern style, crowd loop colorful background with mixed men and women.

What can we learn from failed attempts to change people's behavior?

By Layal Liverpool

Failed behavioral interventions often share common characteristics


A study of interventions to modify human behavior suggests that those who fail share common characteristics.

Identifying these traits could help predict possible avenues in which future interventions could fail and provide an opportunity to prevent it from happening, says Magda Osman of Queen Mary University of London.

Osman and her colleagues analyzed 65 articles published between 2008 and 2019 that identified failed behavioral interventions, including nudges – subtle suggestions for influencing people's behavior.


They found that behavioral interventions based on social comparisons and social norms, such as encouraging people to adopt a behavior by pointing out that it is general or normal behavior in society, made up the majority – 40 percent – of the failed interventions examined.

Other strategies that emerged among the failed interventions were those that sent messages through letters or texts (24 percent) or by labeling products (12 percent), and those that relied on default settings, such as opt-in or opt -Out strategies (15 percent).

The researchers also categorized different ways in which interventions failed, such as by having no effect at all or by backfiring and causing an undesirable side effect. Considering both the type of behavioral intervention and the possible ways in which interventions might fail beforehand could help in designing more successful interventions, says Osman.

Osman and her team develop models that, based on their analysis of failed interventions, can predict how a particular behavioral intervention could play out. "You can simulate different outcomes before doing a behavioral intervention that might fail," which could save time and money, she says.

Journal reference: Trends in Cognitive Sciences, DOI: 10.1016 / j.tics.2020.09.009

More on these topics: