AuthorsE. Løhre
TitleCommunication of predictions: effects of anchors, frames, and expressions of uncertainty
AfilliationSoftware Engineering, Software Engineering
StatusPublished
Publication TypePhD Thesis
Year of Publication2015
DegreePhD
Date Published01/2015
PublisherDepartment of Psychology, University of Oslo
Place PublishedOslo
Abstract

When you are reading the weather forecast, listening to experts on the radio predicting economical growth, or telling your friends that you think Barcelona will win the Champions League, you are involved in the communication of a prediction. This thesis investigates communicative influences on predictions, and the findings demonstrate that changes in the communicative situation (e.g., irrelevant numerical anchors, or the frame a prediction is presented in) can affect both the production and interpretation of predictions.

In Paper I, we explored potential communicative moderators of the anchoring effect, which occurs when an initial piece of information unduly influences subsequent judgments. Previous studies have found that the numerical preciseness of the anchoring value and the credibility of the source of the anchor can influence the anchoring effect. In three experiments with 381 software professionals as participants, we however found no evidence that numerical preciseness or source credibility affected the strength of the anchoring effect for predictions of software project effort. However, we found a strong anchoring effect, which demonstrates that unrealistic values introduced in communication can dramatically influence the prediction that is produced.

Paper II examined the context-dependency of performance time predictions and verbal probability expressions. Previous research has demonstrated that some verbal probability expressions (e.g., “possible”) are associated with maximum outcomes, while other expressions (e.g., “certain”) are associated with minimum outcomes. However, we found that the orientation of the time dimension can be influenced by using linguistic frames that focus on duration (“It is possible/certain that I will spend _____ hours on this task”) or on speed (“It is possible/certain that I can do this task in _____ hours”). In the duration frame, a high number of hours indicates the maximum outcome (the highest duration), while in the speed frame, a high number may represent the minimum outcome (the lowest speed). Three experiments showed that possible was associated with high estimates and certain was associated with low estimates of performance time in the duration frame, while the speed frame led possible to be associated with low estimates and certain to be associated with high estimates.

The topic of Paper III was the uncertainty of predictions. Some theorists prefer to describe uncertainty as due to stochastic processes (external uncertainty), while others define uncertainty as a subjective degree of knowledge or belief (internal uncertainty). This distinction is also found in natural language, with some expressions focusing on external uncertainty (“It is 70% certain”, “There is a 70% probability”) while other expressions are associated with internal uncertainty (“I am 70% certain”). Four experiments demonstrated that both speakers and listeners are sensitive to the external/internal distinction in communication of predictions. External expressions were seen as more informative and were thought to signal greater knowledge, while internal expressions were associated with personal interest and engagement. A lower numerical probability was deemed necessary to recommend an action when probability was framed as external rather than internal, perhaps due to the increased perceived “objectivity” of external expressions.

The findings of this thesis have important implications, both for speakers that want to ensure that their predictions are understood as intended, and for listeners that want to accurately understand predictions. The thesis draws on previous research on communicative influences on judgment and decision making to put the current findings into context. In particular, research on anchoring, framing, and communication of uncertainty are relevant, and some connections with research on psycholinguistics are also discussed.