Seminar Series

Gus - LivePaper - Augmenting traditional publishing with modern functional requirements Pang - Probability Weighting in Text Entry Interaction


Augustus Ellerm & Pand Suwanaposee


University of Canterbury

Time & Place

Thu, 19 May 2022 12:00:00 NZST in John Britten HP 117 Seminar room


Pang - Abstract: The probability weighting function (PWF), in which small probabilities are over-weighted and large probabilities under-weighted, is evident in many psychology studies. However, research on the PWF in human-computer interaction (HCI) is scarce. Therefore, we conducted a study focusing on how the probability of a correct word appearing in text entry suggestions influences users' decision to use them. The main task of the experiment required the user to enter five short phrases using five different text entry suggestion systems that differ in the probability of showing a correct suggestion. We used an eye tracker and the users' mouse movement to determine suggestion usage. Findings correspond to the PWF in which users overuse suggestions at low probabilities but underuse them at high probabilities. This study adds to our understanding of the human perception of probability in HCI which is vital when designing user experiences involving uncertainty, e.g. auto-correct and speech recognition.

Recording of Gus Ellerm and Pang Suwanapossee