Periodic Reporting for period 1 - NISIHealth (Digital Nudges, Incentives and Social Influence in Habit Formation of a Global Health Behaviour)
Periodo di rendicontazione: 2019-02-01 al 2021-01-31
The lack of physical activity is a leading cause of preventable death globally. For most people, failure to exercise is unlikely to be due to ignorance or lack of interest. While individuals are frequently motivated to start exercising in order to reduce and reverse different health risks, they often encounter difficulties in forming a sustained behavioural change--a habit. Habits are behaviours repeated in the same setting. NISIHealth aimed: A. to develop information-theoretic measures of regularity and predictability in fitness behaviour which can be used as new measures of habits that better predict future retention, B. to understand the role of digital nudges and economic incentives in fitness habit formation and C. to investigate how social influence enhances individual habit formation in the age social media. To assess these research questions, the fellow -- in collaboration with a global exercise tracking companies -- devised methodologies currently used in information theory and statistics on large scale historical individual datasets of fitness activity. In addition, the action made extensive use of surveys and high-fidelity randomized control experiments to validate empirical results and establish causal relationships. The complementarity between the expertise of supervision and the fellow’s qualification as well as the arrangements and environment provided by the host institution help towards the success of this ground-breaking and timely project as well as an effective dissemination and utilization of the outcomes. The conclusions of the proposed action could have important implications to health agencies in EU and around the world and may shape the way public health policy is imposed.
The action objectives are the exploration of the role of individual habits in the long-term fitness behaviour as well as the causal effect of digital nudges, incentives and social influence in motivating people’s healthy behaviour change. The action aimed to use big data analytics on individual-level historical data from a global fitness tracker as well as devise state-of-the-art digital field experiments to end up with high impact conclusions regarding human behavioural changes in exercise. The current project is potentially of major interest to public health agencies in Europe Union but also globally and can potentially shape the way health recommendation policy is imposed.
Under the course of the action we conducted a field experiement in collaboration with Runkeeper to show that social influence is an important motivation factor for people to exercise. We further build information theoretic measures of habits and we showed that habitual individual are less susceptible to social influence.
The action objectives could not be rigorously implemented without the help and collaboration of fitness trackers. The fellow set up a collaboration with the data science teams of Runkeeper and WeRun, two of the leading fitness trackers. Runkeeper is a smartphone application that has recently become one of the world’s most popular personal fitness programs, counting more than 50M users worldwide out of which 23M resides in the European Union region. This application tracks performance over time, allowing users to view a detailed history of activities; get notifications for new personal "bests" and milestones; measure progress against current goals and follow detailed plans. Runkeeper allows users to share their personal content with friends through posting activities, achievements and plans to the app-embedded running-buddy network as well as share their performance to other social channels like Facebook and Twitter. The fellow’s collaboration with Runkeeper, offers a tremendous research opportunity and a promising asset for Europe.
The action objectives are the exploration of the role of individual habits in the long-term fitness behaviour as well as the causal effect of digital nudges, incentives and social influence in motivating people’s healthy behaviour change. The action aimed to use big data analytics on individual-level historical data from a global fitness tracker as well as devise state-of-the-art digital field experiments to end up with high impact conclusions regarding human behavioural changes in exercise. The current project is potentially of major interest to public health agencies in Europe Union but also globally and can potentially shape the way health recommendation policy is imposed.
Under the course of the action we conducted a field experiement in collaboration with Runkeeper to show that social influence is an important motivation factor for people to exercise. We further build information theoretic measures of habits and we showed that habitual individual are less susceptible to social influence.
The action objectives could not be rigorously implemented without the help and collaboration of fitness trackers. The fellow set up a collaboration with the data science teams of Runkeeper and WeRun, two of the leading fitness trackers. Runkeeper is a smartphone application that has recently become one of the world’s most popular personal fitness programs, counting more than 50M users worldwide out of which 23M resides in the European Union region. This application tracks performance over time, allowing users to view a detailed history of activities; get notifications for new personal "bests" and milestones; measure progress against current goals and follow detailed plans. Runkeeper allows users to share their personal content with friends through posting activities, achievements and plans to the app-embedded running-buddy network as well as share their performance to other social channels like Facebook and Twitter. The fellow’s collaboration with Runkeeper, offers a tremendous research opportunity and a promising asset for Europe.
At the very beginning we started with a collaboration with Runkeeper. We designed and conducted a six-month long randomized field experiment involving 1.37M users of their online fitness tracker platform. In more details, the experimental design employs rate limit randomization to deliver social push notifications to application users. We randomly assign application users -- that have received at least one social push-notification in the 30-day period before the experiment starts -- to policies governing peer exercise exposure notifications, with rate limit of k hours, i.e. being able to receive a social notification within a k-hour window if a friend finishes a run. Out of this sample,~40% are allocated in the status quo experimental group (k = 24h), ~15% of the users are assigned to a holdout group (k = Infinity) and the rest of individuals (45%) are allocated equally at three groups of increased notifications with k equal to {6, 8 and 12 hours} . The treatment modulates the volume of social notifications users will receive from friends. Through the experimental analysis, we find that a single exposure to peer's fitness activity increases the probability of exercising by about 1%.
We further as proposed, we build up information theoretic measures of habits that are based on the entropy measure of context stability. We evaluate those measures by performing a survey of standard classical habit measures. We used those measures to show in our experiment that habitual individuals are less susceptible to social influence.
Just after we finished that work, the COVID-19 pandemic for couple of months turn our attention to a question with important implications during the first lockdown: Which locations should close first and which last during a progressive lockdown. In collaboration with MIT, we use mobility data from Safegraph and other openly available datasets to come up with a roadmap based on the socioeconomic benefit but also cumulative danger of transmissibility for each location. Among the locations we examine are gyms and other places of exercise that align well with the overall topic of the action. The results are now published in Risk Analysis (https://onlinelibrary.wiley.com/doi/abs/10.1111/risa.13800) and the project is acknowledge for financial support.
Using the same dataset and in conjunction with another mobility dataset from Veraset (under a Data Use Agreement) we study how exercise habits were affected by the different lockdowns during the different COVID-19 waves in Europe and United States. We found that for couple of weeks during and after the lockdowns individual shift from Gyms to parks for their exercise habits. In addition we study the research question that were promised in our proposal of how different exogenous events can affect long terms habit formation. We found that individuals that suspended their exercise habits due to lockdowns and other related policies, they recover quite fast after they policies are lifted.
We further as proposed, we build up information theoretic measures of habits that are based on the entropy measure of context stability. We evaluate those measures by performing a survey of standard classical habit measures. We used those measures to show in our experiment that habitual individuals are less susceptible to social influence.
Just after we finished that work, the COVID-19 pandemic for couple of months turn our attention to a question with important implications during the first lockdown: Which locations should close first and which last during a progressive lockdown. In collaboration with MIT, we use mobility data from Safegraph and other openly available datasets to come up with a roadmap based on the socioeconomic benefit but also cumulative danger of transmissibility for each location. Among the locations we examine are gyms and other places of exercise that align well with the overall topic of the action. The results are now published in Risk Analysis (https://onlinelibrary.wiley.com/doi/abs/10.1111/risa.13800) and the project is acknowledge for financial support.
Using the same dataset and in conjunction with another mobility dataset from Veraset (under a Data Use Agreement) we study how exercise habits were affected by the different lockdowns during the different COVID-19 waves in Europe and United States. We found that for couple of weeks during and after the lockdowns individual shift from Gyms to parks for their exercise habits. In addition we study the research question that were promised in our proposal of how different exogenous events can affect long terms habit formation. We found that individuals that suspended their exercise habits due to lockdowns and other related policies, they recover quite fast after they policies are lifted.
The work carried out enhance innovation capacity, create new market opportunities, strengthen competitiveness and growth of fitness tracking companies through the publication of the results. Furthermore, the project brings important benefits for society and health behavior issues. The methods learned from the program have been applied in fitness tracking industry successfully and the work carried out can contribute to public health policy implementation. Potential users of the project results are public health agencies and the fitness tracking industry. While we have communicated extensively the results with industry, we did not communicate a lot with public health agencies as yet.