Can Social Media Predict Employee Personality?

Before Cambridge Analytica, technology has had a substantial impact on assessment methods. While research in this area is in early phases, some work has found that using data like Facebook “Likes” can predict personality. Data mined from other social media networks has been linked to other facets of personality. In fact, new technologies have the potential to completely change the structure of the assessments market because of enhanced user experience and crowd sourcing as a method for employee selection. There are, however, several ethical issues with new assessment methods. One relates to the purported validity of such new tools. The other issue relates to transparency and anonymity concerns. While some predictive technologies are in place to revolutionize the assessments market, using data without explicit user consent may undermine the process from an ethical point of view. Our aim in this post is to examine changes in the assessments field in light of new technologies and to provide information on the emerging use of big data-based assessments for organizations. 

New technologies have opened doors in order to use assessments for HR purposes. One study, published in 2013, demonstrated that correlations between personality traits, cognitive ability, and demograpics could be predicted with surprising accuracy simply based on what users “Liked” on Facebook. Two other studies have found differences in narcissism between Twitter & Facebook users. Yet another study has found links between words bloggers use and personality with significant accuracy. Additionally, metadata has the potential to predict personality. Clearly, there is incredible potential for individuals in HR roles in organizations to harness this data to assess and profile individuals. 

Recent changes in assessment tools that utilize new data methodologies have the potential to change the marketplace significantly because of their potential to reduce costs. According to experts, the gap between organizations hiring people who do not fully fit their needs and professionals who want jobs that fit them may close. Analytics utilized for HR purposes have the potential to create a situation in which organizations can directly "crowd source" the talent they need. This has been referred to as the “Uberization” of talent – after the on-demand car service. Accordingly, these new methods could drastically cut costs for organizations in the hiring process. Put simply, it is significantly less expensive to configure and run an algorithm than it is to use an established test to find talent. 

User experience may be enhanced in the future with new methodologies compared to tried-and-true assessments. Ten to 15 years ago, consumer experience did not matter as much with assessments for HR purposes. Now, it is more important to potential applicants to use tools that are enjoyable to complete rather than tedious and time consuming. The case of gamification has emphasized this point. Gamification has involved the creation of various “games” whose outcomes can be correlated with measures of personality or other factors. One example is Knack.com that provides a game called “Wasabi Waiter” that is a situational judgment-type task. When all potential talent need to do is “Like” something on Facebook or play a game, clearly their enjoyment with these types of methods will be greater than with previous assessment methods such as filling out an irksome job application online. 

While these new advances have occurred in the field, the research to substantiate the validity of such assessments has not kept apace. Incremental validity (the extent to which a new measure is valid over and above previously established measures) of certain algorithms has yet to be measured. Incremental validity is important to assess because, for example, in the Facebook “Likes” study, we need to know if this type of assessment algorithm might offer something better over and beyond traditional assessment methods (as of this writing, the Facebook algorithm study and other types of big data-based methods only offer about 50% of the predictive validity of traditional assessments). While the validity is not yet there, does the fact that such algorithms are cheaper and easier to administer have an impact on decision-making from an HR point of view? Clearly, as this field opens up, researchers need to partner with those who are commercializing these tools in order to make sure they are actually valid and rigorous. 

In addition to understanding incremental validity of various new tools, the fact that some assessments have not been substantiated in a rigorous way can create ethical issues for organizations and for those commercializing various measures. The potential exists for manipulation and poor quality of assessments. For example, one new tool, has faced criticism. In this particular case, Lumosity, a product utilizing “brain games” is targeted toward two groups: the elderly hoping to stave off Alzheimers and to individuals hoping to benefit their cognitive abilities in general. Many neuroscientists have criticized Lumosity and similar tools for making unsubstantiated and inaccurate claims about their effectiveness. While this assessment has not made it into the HR sphere, similar tests have the potential to. For the most part, the big data-based assessments industry has gone relatively unregulated and almost anyone can make claims about their tools. Corporations preying off the ignorance of the elderly or anyone wanting to “improve” their cognitive abilities pose substantial ethical concerns that need to be investigated further. 

The emerging field of big data-based assessments may also have other ethical issues. According to an article in The Atlantic, “By one estimate, more than 98% of the world’s information is now stored digitally, and the volume of that data has quadrupled since 2007”. With such a vast amount of data at our disposal, how do we use it in an ethical way? Much of this data has been collected outside of our conscious awareness – web browsing, meta-data collected from when, where, and to whom we make telephone calls. Using social media has unwittingly helped to unleash this new database of information. Of course, after the Edward Snowden revelations, we now know that there is a database of such information that the government is actively collecting. After Snowden revealed the NSA’s data collection practices, the public was outraged and clearly felt violated. A quarter of US citizens changed their patterns of social media platforms. Additionally, 52% of US citizens surveyed said they were either “very concerned” or “concerned” about US government surveillance. Given the reach of current dissatisfaction among US citizens, large-scale programs like the NSA clearly make individuals feel uncomfortable with the fact that their data can be stored and recalled, and used for unknown purposes at any point in time. Similarly, using data on a smaller scale within the realm of a single organization for assessment purposes might leave participants feeling violated if their data is being used without their consent. Therefore, when using data to profile individuals, these data need to be used in a transparent way. 

The use of large-scale big data-based assessments for hiring and development practices might have another downside although organization change efforts might be able to reduce ethical concerns. As discussed previously, monitoring people at work electronically is dubious from an ethical point of view. However, when people do know they are being monitored at work, their engagement from work-related tasks can suffer, which may have an impact on performance – although at this point, more research needs to be done to substantiate these claims. Other research has shed light on this potential point as an ethical violation. If people in a given workplace are being electronically monitored, negative outcomes can be reduced by providing them with a compelling reason for using their data and giving them input in the process. When these conditions are met, they feel more at ease and may be more in line with a sense of procedural justice. This could have practical implications for ethics in using assessment tools. If companies were to scrape data from emails, for example, and try to make links between the content of those emails and individual performance, individuals might be more likely to buy into this process if they have had some input on the process to understand why the policies are in effect. Granted, our explanation here is simplistic as such a change in organizational policies would likely involve massive organizational change efforts, but there still is evidence that some of the effects of using big data in a way that invades privacy could be negated if employees participate in the process. 

New advances in psychological assessments that are open to HR professionals and organizations are exciting and changing quickly. Suddenly, we live in a world in which the algorithms from social networking sites and metadata from our smartphones and Internet usage can predict things like personality, cognitive ability, and demographics. Due to the shifting nature of assessments, the market is also shifting. Finding talent through assessments has the potential to become on-demand and crowd sourced like Uber. Furthermore, user experience will likely be enhanced through gamification. These exciting new directions in the assessments field are not without their problems, however. Several concerns exist regarding the use of such data. Major issues still remain regarding validity. Products have the potential to be sold (and are) without being properly scientifically vetted. Furthermore, gaining explicit user consent with the myriad sources of data available to researchers, data scientists, and employers alike, is important. If explicit consent is not obtained and employees find out their data is being used, they may lash out against the company in destructive ways and organization-wide performance could suffer. One way around this may be to instill major organizational change to gain buy-in from employees so they do not feel uncomfortable that their data is being used. Clearly, the direction big data-based assessments are currently moving in is exciting, but it also implies many shifts – in assessments themselves, in the marketplace, in ethics, and in how we work with organizations and accept the use of our data in a major way. These changes are not minor – instead, they are radical and like the revelations of Edward Snowden, are likely to involve impassioned large-scale societal discussions on all sides of the issue. 

Selected references: 

Alge, B. (2001). Effects of computer surveillance on perceptions of privacy and procedural justice. Journal of Applied Psychology, 86(4), 797-804.

Asman, M., & Essex, P. (2001). Electronic monitoring of employees at work. Ohio CPA Journal, 60(1), 25-28.

Chamorro-Premuzic, T. (2014a, March 21a). The Uberisation of talent: Can the job market really be optimised? Retrieved April 5, 2015, from http://www.forbes.com/sites/tomaspremuzic/2014/03/21/the-uberisation-of-talent-can-the-job-market-really-be-optimised/

Chamorro-Premuzic, T. (2014b, July 16b). Should we get excited about new assessment tools? Retrieved April 5, 2015, from http://www.hrexaminer.com/should-we-get-excited-about-new-assessment-tools/

Kosinski, M., Stillwell, D., & Graepel, T. (2013). Private traits and attributes are predictable from digital records of human behavior. Proceedings of the National Academy of Sciences, 110(15), 5802-5805.

McKinney, B., Kelly, L., & Duran, R. (2012). Narcissism or openness?: College students’ use of Facebook and Twitter. Communication Research Reports, 29(2), 108-118.

Nerney, C. (2014, March 28). How HR analytics can transform the workplace. Retrieved April 5, 2015, from http://www.citeworld.com/article/2137364/big-data-analytics/how-hr-analytics-can-transform-the-workplace.html

Panek, E., Nardis, Y., & Konrath, S. (2013). Mirror or Megaphone?: How relationships between narcissism and social networking site use differ on Facebook and Twitter. Computers in Human Behavior, 29, 2004-2012.

Peck, D. (2013, November 20). They're watching you at work. Retrieved April 6, 2015, from http://www.theatlantic.com/magazine/archive/2013/12/theyre-watching-you-at-work/354681/

Quarter of US citizens changed technology use after Snowden revelations. (2015, March 16). Retrieved April 7, 2015, from http://sputniknews.com/us/20150316/1019572149.html

Tucker, P. (2014, March 28). How the NSA can use metadata to predict your personality. Retrieved April 7, 2015, from http://www.defenseone.com/technology/2014/03/how-nsa-can-use-metadata-predict-your-personality/81538/

Underwood, E. (2014, October 22). Neuroscientists speak out against brain game hype. Retrieved April 6, 2015, from http://news.sciencemag.org/brain-behavior/2014/10/neuroscientists-speak-out-against-brain-game-hype

Yarkoni, T. (2010). Personality in 100,000 words: A large-scale analysis of personality and word use among bloggers. Journal of Research in Personality, 44, 363-373