The coupled impact of conflict and imprecision from multiple forecasts
People often rely on forecasts from multiple experts to make decisions about uncertain quantities in domains as diverse as health, finance, and politics. Previous research has differentiated between perceptions of conflicting and imprecise forecasts showing greater conflict aversion. However, conflict and imprecision are rarely well-differentiated and are often correlated. This series of studies tests the hypothesis that the joint influence of conflict and imprecision on estimates of most likely values and judgmental ratings is a function of two key attributes of the forecast sets, overlap and symmetry. ^ Study 1 established the general response pattern in two interval forecast sets by varying the type and degree of these two key set factors. Study 2 generalized the results by extending the design to include three interval forecast sets. Results indicate that responses are task-dependent. Symmetry has a stronger impact on estimates, and overlap has a stronger influence on ratings. People center their estimates around the mean of the set as expected, but there is also a consistent shift toward lower values (a status quo bias). Range estimates correspond with the internal endpoints. Ranges are generally narrow when attention is focused on the center of the forecast set and wider when focus shifts to the extremes. Disjoint sets lead to greater shifting in and around a forecast set than intersecting sets. Attempts to model best estimates indicate that people focus on a local subset of the full region to make estimates. ^ Ratings are driven by the structure of the forecast set. Imprecision is considered the most indeterminate set and all hybrid sets are least indeterminate. The hybrid sets fall in between imprecision and conflict for compatibility ratings. Ratings follow a clear pattern of preference. Overlap drives preferences more than symmetry. People are more confident when the overlap is small and the set is symmetric, but the impact of symmetry depends on whether the set is centrally- or outwardly-focused when there are three forecasts. Disjoint sets generally lead to lower ratings, but defining disagreement is difficult since intersecting sets are sometimes rated as conflicting and disjoint sets sometimes are rating as agreeing.^
Benjamin, Daniel, "The coupled impact of conflict and imprecision from multiple forecasts" (2015). ETD Collection for Fordham University. AAI3727412.