Predicting self-monitoring skills using textual posts on Facebook

Predicting self-monitoring skills using textual posts on Facebook

By
Michal Kosinski, Qiwei He, Cees A.W. Glas, David J. Stillwell, Bernard P. Veldkamp
Computers in Human Behavior. April
2014, Vol. 33, Pages 69-78

The popularity of the social networking site Facebook (FB) has grown unprecedented during the past five years. The research question investigated is whether posts on FB would also be applicable for the prediction of users’ psychological traits such as self-monitoring (SM) skill that is supposed to be linked with users’ expression behavior in the online environment. We present a model to evaluate the relationship between the posts and SM skills. The aim of this study is twofold: first, to evaluate the quality of responses to the Snyder’s Self-Monitoring Questionnaire (1974) collected via the Internet; and secondly, to explore the textual features of the posts in different SM-level groups. The prediction of posts resulted in an approximate 60% accuracy compared with the classification made by Snyder’s SM scale. The variable “family” was found the most significant predictor in structured textual analysis via Linguistic Inquiry and Word Count (LIWC). The emoticons and Internet slangs were extracted as the most robust classifiers in the unstructured textual analysis. We concluded that the textual posts on the FB Wall could partially predict the users’ SM skills. Besides, we recommend that researchers always check the validity of Internet data using the methodology presented here to ensure the data is valid before being processed.