Key points are not available for this paper at this time.
An Internet newsmagazine was created, involving all features of online interactivity. The text of all available reports was held constant. The imagery of a subset of articles was manipulated, however. The manipulated articles were presented either without images, with text-related innocuous images, or with text-related threatening images in both their headline displays and their text bodies. During a fixed period of time, readers were free to sample articles and to read as much of them as they pleased. Unbeknownst to them, their selective exposure behavior was automatically recorded. It was observed that the incorporation of threatening images fostered more frequent selection of the associated articles and markedly increased reading times of the corresponding texts. The incorporation of innocuous images had similar but more moderate effects. Retrospective accounts of reading were consistent with the recorded exposure behavior.
Knobloch et al. (Sat,) studied this question.