Describing sets in terms of a two-valued variable, either value can be chosen: exam results may be referred to by pass rates or fail rates. What determines such framing choices? Building on work by McKenzie and colleagues on reference points in the production and interpretation of framed information, we investigate two determinants of frame choice. One is that speakers tend to focus on the component that has increased vis-à-vis a previous state, the other is the tendency to choose the component larger than 50%. We propose to view reference points as pointing to different kinds of communicative relevance. Hence the use of the previous state and the 50% reference points by speakers is not just a function of the information, but is co-determined by a communicative cue in the context: the question being asked about this information. This line of thought is supported by two experiments containing items offering two-sided distribution information at two points in time. Our first experiment employs a static task, requiring a description of the most recent situation. The second experiment uses a dynamic task, asking participants to describe the development between the two time points. We hypothesize that in static tasks the component size is the strongest frame choice determinant, while in dynamic tasks frame choice is mainly driven by whether a component has increased. The experiments consist of 16 different scenarios, both with symmetrical contrasts (i.e., dogs vs. cats) and with asymmetrical ones (i.e., winning vs. losing). Both experiments support the hypotheses. In the static task, the size effect is the only consistent effect; in the dynamic task, the effect of direction of change is much larger than that of size. This pattern of differences between size and change effects applies across symmetrical and asymmetrical contrasts. Our experiments shed light on cognitive and communicative regularities involved in the production of framed messages: people do tend to prefer larger and increasing components when choosing a frame, but the relative strength of both these preferences depends on the communicative task.
Read full abstract