Future directions in personality assessment: Data-scraping personality from your smartphone activity, fitness band information and social media posts

When thinking about the future of personality assessment, one could be reminded of the hoverboards from Back to the Future Part II. That is, researchers and innovators are certainly making strides, but they’re still a while away from becoming as common as a bike (as portrayed in the movie’s vision of what 2015 would look like).

The same can be said for innovations in personality assessment. Research published in peer-reviewed journals are starting to show that the Big 5 personality traits can be accurately predicted from:

• How people use their smart-phones in real-time (Chittaranjan et al., 2013), including the language they use in text messages (Mairesse et al., 2007)
• The types of images uploaded or filters applied on Instagram and Facebook (Ferwerda et al., 2016; Segalin et al., 2017)
• The intensity of a workout as measured by smart watches or fitness bands (Gao et al., 2019)
• A person’s face, even when it’s non-expressive (think facial recognition technology; Kramer et al., 2011)

For example, people who use more negative than positive words in text messages are more likely to be introverted; people who use Instagram filters that are lower in brightness and higher in saturation are more open to experience; and those who do not set their phones to silent mode (and instead turn on the “Ring Once” function) are more likely to be extraverted.

So how do we make predicting personality via data scraping a reality, particularly in light of the public’s negative sentiment from Cambridge Analytica’s data breach? Below are just some of the many questions that need to be asked and answered.

Are we even assessing the right things?

The Big 5 factors are consistently criticised in the literature for lacking a robust theoretical foundation, and yet the majority of research on data-scraping continues to focus on the Big 5. These new approaches to assessing personality are progressive but need to measure personality traits that are grounded in personality theory (e.g., socioanalytic theory) and still need to be subjected to validation tests.

What are the main ethical considerations?

Universities have established formal and stringent ethics committees, but who will police and monitor ethical issues outside of academic environments. How can we ensure that those who are responsible for ethical considerations have no financial interests in data-scraping technologies?

Technological advancements (particularly when it comes to personal data) and ethics will always clash. But how can we increase ease of access to data whilst ensuring appropriate use? For example, is it possible to make consenting to data-scraping as common as Seek asking for permission to access your LinkedIn profile to create an online resume, or how smartphones request “crash” statistics to help inform diagnostics?

What about adverse impact?

There have been many real-life cases of artificial intelligence demonstrating bias towards certain groups of people (see here for some recent examples). Unlike more established methods of personality assessment (particularly ones that have conducted adverse impact analyses), artificial intelligence and data-scraping methods have yet to overcome exhibiting biases.

Personality assessed by data-scraping has many potential and significant advantages (e.g., personality assessment at the click of a button versus answering hundreds of questions, and the potential for reduced “faking” since actual behaviour is analysed in the form of real-time phone activity data). Until the above questions (and countless others) are answered, we will need to continue using evidence-based personality assessments that are predictive of performance and that are bias-free. But the question remains, when (not if) will hoverboards and “data-scraped” personality assessment become the norm?

This topic was recently presented at the biennial Industrial and Organisational Psychology conference in Adelaide and was titled, “Contemporary trends in personality and leadership assessment: Where are we now and where are we headed?”