A Rasch Analysis Approach to the Development and Validation of a Social Presence Measure

Social presence theory was developed by Short et al. (The social psychology of telecommunications. Wiley, London, 1976) to explain the impact of different media such as text, audio, or video on interpersonal communication. They defined social presence as “the salience of the other person in the interaction,” which was interpreted as the degree to which the other person is perceived as physical “real” in the communication. Social presence theory was first applied by Gunawardena (Int J Educ Telecommun 1:147–166, 1995) for online educational contexts. Since then it has become an important construct for summarizing the effects of mediated communication on the social interaction and the group dynamics that happen in distributed collaborative learning groups. However, a robust scale for measuring perceptions of social presence is still lacking. Although Short, Williams, and Christie did measure social presence by using four semantic differential scales, they never validated this scale. Indeed, other social presence instruments have come to existence but none of these instruments tapped physical realness of others as the single trait of interest. Furthermore, questions may arise about the psychometric qualities of these instruments as they, at best, used exploratory and confirmatory factor analyses but did not account for the nonlinearity of rating scale steps and other issues. To fill this gap, the current research aimed at developing a robust social presence measure by using the Rasch measurement model as a rigid construct validation method. The findings of the Rasch analyses (fit of items and persons, unidimensionality, category probability curves) in Winsteps version 4.4.1 revealed two dimensions of social presence: Awareness of others and Proximity with others. The first was measured with 15 items while the latter was measured with 12 items. The psychometric quality of the Awareness 15-item set was good to excellent whereas the quality of the Proximity 12-item set was moderate to good. Future research is aimed to improve the psychometric qualities even more and also to determine whether the two dimensions are actually inconsequential (Linacre in Detecting multidimensionality in Rasch data using Winsteps Table 23, 2018b).