The Relationship Between Personality and Online Social Network (OSN) Behaviors: Why Predicting Personality Through OSN Data Is Possible
Besides the innumerable conveniences brought by the internet, the cyberspace also provides us an unprecedented window through which we are able to see the disparity among many millions of individuals. When we mention the difference among individuals, we are usually talking about personality, which is the visible consistent behavior patterns and interpersonal processes originating within the individual (Burger, 2008), as well as the internal distinct traits of cognition and emotion (Mischel, Shoda & Ayduk, 2007). To a great extent, it is the different personalities of internet users which lead to their various, individualized online behaviors.
There have been a lot of studies revealing the correlations between personalities and variety of online behaviors. Baker and Moore (2008) investigated the blogging behavior of Myspace users and reported that the intending bloggers scored higher on psychological distress, self-blame, and venting and scored lower on social integration. Kosinski et al. (2012) found psychologically meaningful relationships between personality and preferences to website and website categories. Psychologists have revealed that the personality characteristics of internet users were relevant to the overall pattern of choosing online services (Hamburger & Ben-Artzi, 2000), as well as the usage of certain functions or aspects of the internet, such as using Wikipedia (Amichai–Hamburger et al., 2008), online dating (Hall, 2010), playing online game (Charlton & Danforth, 2010), and online shopping (Huang & Yang, 2010).
Along with the enormous popularity of Online Social Networks (OSN) in recent years, such as Facebook, Twitter, Instagram and LinkedIn, many researchers put their attention on the OSN behaviors, partly because the OSN services often integrated many functions into a single platform and the use of OSN is bound with personal identity. Series of studies revealing the relationships between the users’ personality and their OSN behaviors largely enhanced our knowledge about how online behavior data reflecting personality characteristics. Correa, Hinsley and De Zuniga (2010) found that the Big-Five personality scores could predict the engagement of social media. Gosling et al. (2011) found that the Big-Five personality traits were correlated with both self-reported Facebook-related behaviors and observable profile information, and OSN users extended their offline personalities into the domains of OSNs. Lee, Ahn and Kim (2014) reported that the user’s personality may influence several aspects of Facebook usage, including photo uploads, status updates, and number of friends at Facebook Wall, and Like, Comment, and Share at Facebook News Feed. Some other researchers paid attention to the content posted on OSN, and found that the topics updated on Facebook reflected personality characters including Big-Five dimensions, self-esteem and narcissism (Marshall, Lefringhausen & Ferenczi, 2015).
The large amount of empirical evidences of the relationships between personality (and some other psychological profiles) and online behavior records clearly suggests the possibility of recognizing the user’s personality through their online behavior data. However, it is not easy to make judgment on the individual’s characteristics as to disclose the correlations between their personality and their behavioral indices. Schwartz et al. (2013) analyzed about 700 million words and phrases from Facebook massages, and calculated the correlations between more than 50 different LIWC (Linguistic Inquiry and Word Count, Pennebaker et al., 2015) categories and Big-Five personality. There showed many significant correlations between language use on Facebook and personality, confirming the connections between them, but the correlation coefficients were fairly low, which means that each single behavioral index (LIWC category) cannot effectively predict personality by itself. So one plausible idea seems to be integrating many personality-relevant online behavioral parameters into a predictive model, as the common practice in the big data mining.