A Reply to Bennett and Fritz's Critique of Marino and Mahan

Abstract

Bennett and Fritz (2005, this issue) raised several questions about the approach to display design that we used to create the graphic nutrition labels for our study. Their primary concerns appear to involve the theoretical principles they think should undergird an efficient display. Although their comments are informative e and constructive, they are largely misplaced because they relate not to our work but, rather, to comparative studies that have not yet been performed. Specifically, the maintain that (a) our results would have been stronger had our display been based more explicitly on display design principles that they advocate, (b) relevant domain semantics were not included, and (c) the task-display mapping could have been improved. We comment here on each of these points. Display Proximity The argued that the results we obtained using polar coordinate displays, although consistent with the theoretical underpinnings of our hypothesis (proximity compatibility principle), actually occurred for other reasons (the power of the perceptual systems) and that our results would have been even stronger had the task-display mapping been more explicit. According to the authors, only object configural displays are high proximity under our interpretation of the proximity compatibility principle, and a composite label design would have supported integrated and focused task performance more effectively. Display proximity refers to the perceptual interaction between stimulus dimensions in an information display. High-proximity displays have significant interactions between stimulus dimensions, whereas low-proximity displays have little such interaction. Configural displays are regarded as high-proximity displays because of their emergent features: object configural displays are simply a subset of this category. The proximity compatibility principle does not entail the idea that only object configural displays can exhibit the property of high proximity (Wickens & Andre, 1990: Wickens & Carswell, 1995). For example, bar graphs can exhibit high proximity if they contain emergent configural properties (such as bar-top alignment) that are directly related to task performance (e.g., Sanderson, Flach, Buttigieg, & Casey, 1989). In contrast, alphanumeric displays exhibit low proximity because individual indicators represent separate channels of information, and there is no interaction between stimulus dimensions. A polar coordinate display satisfies the definition of high proximity, and a tabular display satisfies the definition of low proximity. These displays were therefore reasonable choices for our study, the purpose of which purpose was to compare high--and low-proximity displays. Our evidence (Table 1 in Marine & Mahan, 2005; this issue) indicated that the polar coordinate display was more effective than the tabular display on the integration task. The contend that other high-proximity displays having different task-display mappings could have generated stronger results--perhaps so. We concede that the polar coordinate display may not have been optimal. However, the issue can be resolved only on the basis of a prospective, controlled comparison involving the putatively more efficient display. The authors assertion that a composite label (annotating a graphical format with digital values) would have more effectively supported integrated and focused task performance is similarly an empirical question that we did not address. Our prediction was that proximity-based matched task-display tandems would show superior performance as compared with mismatched tandems, and the prediction was supported by our results. Our study was an application of the proximity compatibility principle, not an evaluation of it. We agree that executing mathematical operations on data visually extracted from a graphical display is a less exact process than is reading a digital value. …