Distinguishing trust and gullibility through Twitter language

In this project, we combine survey research with text mining on Twitter. In brief, we administered psychological tests to participants measuring their dispositional trust and gullibility. Subsequently, we scraped tweets from the participants’ Twitter profiles and correlated their language paaterns with their trust and gullibility scores.

We found that trust and gullibility correlated only weakly (r = .2) with each other. Further, we found that the two personality traits were characterized by different and sometimes even opposing language patterns. Especially analytical thinking seems to be quite common among trusters, whereas it is quite rare among gullible people. Thus, defining gullible people as “very trusting” is not very accurate.

For extended results, visualizations, and discussion, see our conference poster below.