How well does your phone know you? Can it describe your personality better than you can? CREST Doctoral researcher Kris Geyer investigates.
Getting a sense about someone’s personality is difficult: are they easy to engage with, competitive, sociable, open to different ideas, or neurotic? And if you could accurately identify someone’s personality then could you provide them with insights into how they could best contribute to society? For instance, what types of roles would they be adept at adopting; what group situations would they thrive in; and what pursuits would they find most engaging? If this was done on a larger scale, society could operate more cohesively and efficiently. But what information best informs us about how one person differs from another?
Personality questionnaires vs observing behaviour
For decades psychologists have treated established questionnaires as the gold standard. The Myers-Briggs questionnaire was popular within the industry, whilst the NEO personality inventory remains well used in academia.
These tests typically involve asking a person about their perception of themselves or how they are likely to respond in a hypothetical situation. But although these tests can reliably predict future behaviours, they can also be manipulated, and critics argue that how people see themselves is not always a true reflection of reality.
In the 1950s, prominent psychologists  of the day argued that personality is best established when observing how a person acts in their daily life. In their view, data from questionnaires and tests in experimental settings are less valuable because people will attempt to portray themselves in the best light. They concluded that the quality of a personality test can be ranked relative to how difficult it would be to cheat the test.
The quality of a personality test can be ranked relative to how difficult it would be to cheat the test.
We regularly see in society how observing behaviour in daily life can provide vital data that a test or questionnaire cannot. For example, if a job applicant is asked “Will you make significant contributions to the company if you are employed here?” they will likely respond affirmatively because they are incentivised extensively to do so. Organisations understand this and will therefore hire workers for probationary periods in order to test out whether new employees are a good fit. It is a lot harder for the new employee to cheat this system, and therefore this data is of higher quality.
So, in order to learn about someone’s personality, seeing how they act in real-life situations is incredibly valuable. However, until recently these approaches have been almost impossible in practice. Observing how someone behaves in their daily life would require someone to suspend their privacy and tolerate the continual presence of a researcher taking notes. But perhaps a solution to these problems can be found in your pocket.
Establishing personality through smartphone use
My PhD focuses on how psychologists can utilise smartphones in order to better understand and predict personality.
One of the biggest challenges I faced in my research was finding an existing smartphone app that would allow me access to sensitive digital data in a safe, reliable and ethical way. It turned out that many applications (apps) I looked at using for my study could not guarantee user privacy. Others were inappropriate for psychological research and some were not sufficiently reliable.
To overcome these issues, I developed my own apps. These easy-to-use apps operate securely and ethically to obtain data for personality assessments. I built multiple applications to target different data sources: location, smartphone usage and neurological activity. I then carried out multiple experiments to investigate how valuable these data sources are at providing insights into personality and psychology.
We found large discrepancies about how people claim to use their smartphones and how, in reality, they use them.
We found large discrepancies about how people claim to use their smartphones and how, in reality, they use them. For example, we found that when people use their phone often, but for small periods of time, (i.e. constantly checking their phone or sending a message rather than being in a conversation) they are more likely to underestimate their usage. Using their smartphone less often, but for longer periods of time, however, typically results in people overestimating their usage.
For one of our studies, we found that this explanation could account for 25.3% of the variation in how people estimated their smartphone usage. This has wide ramifications for psychological research into smartphone ‘addiction’.
From utilising the lessons learnt from developing the smartphone-usage logger, we were also able to develop other apps with very different data collection features. We created an Android app that could continually monitor participants’ location over an extended period of time (weeks). The data is yet to be analysed but could provide insights into a person’s behaviour and aspects of a smartphone user’s personality. For example, people who spend a lot of time at home may be more agreeable (one of the Big Five personality traits), whilst those who spend more time outdoors may be more extroverted. High definition location logging capabilities have previously been useful to detect experiencing depressive symptoms. 
Another app that we are currently developing will manage the data collection of electrical brain activity via an EEG (openBCI). This will enable researchers to reliably monitor the neurological activity of individuals outside of the laboratory setting. Researchers will be able to study how the neurology of individuals changes from day to day, which will provide significant implications for psychological research as a whole.
 Cattell, R. B. (1958). What is” objective” in” objective personality tests?”. Journal of Counseling Psychology, 5(4), 285.
 Saeb, S., Lattie, E. G., Schueller, S. M., Kording, K. P., & Mohr, D. C. (2016). The relationship between mobile phone location sensor data and depressive symptom severity. PeerJ, 4, e2537.