Whenever people use technology, they leave behind a digital trace that documents their behavior.
Professor David Ellis, School of Management, explains how these data were used to study—at scale—the question of whether people behave consistently in their digital lives but in a way that is context dependent.
By analyzing 28,692 days of smartphone app usage across 780 individuals, it was found that it is possible to profile a person’s day-to-day use of different apps and showed that this profile remains consistent over time. The study found that a single day of data from an anonymous user can be matched to the correct originating user’s profile with greater than 70% accuracy when success is taken as the user appearing in the first 10 (top 1%) of all candidates. Thus, people show distinctive patterns of digital behavior even when compared with hundreds of other individuals. This has implications for security and privacy in the digital age.
Ethical issues to consider
- Working with secondary data
- Anonymity of participants
- Consent of participants
- Data security/storage
Mitigating actions
- Seeking full ethical approval from the host institution
- Ensuring that secondary data were given by consenting participants for their smartphone usage patterns to be analysed for academic purposes
- Ensuring that permission to use the data was obtained from the original authors
- Ensuring anonymity and confidentiality of data
- Formalising a data management plan. Plans on how data would be processed and analysed were preregistered with the Open Science Framework
- Abiding by any data sharing agreements put into place by the original data controllers
Lessons Learnt
A growing number of psychology journals including Psychological Science now require or encourage authors to share materials, raw data, and analysis scripts. In this instance, and to comply with data sharing agreements, researchers were unable to share the majority of raw (unprocessed) data. These can be requested from relevant data controllers. However, they were able to share processed data as well as annotated analysis scripts on the Open Science Framework. These allow for the replication of all reported confirmatory and exploratory analyses. An interactive data-visualization website was developed to help demonstrate how we generated key results.
The article received the badges for Open Data and Preregistration. More information about the Open Practices badges can be found here.