Which statistical formats facilitate what decisions? The perception and influence of different statistical information formats

Research suggesting both a superiority of frequencies over single-event probabilities and of smaller reference classes over very large reference classes have utilized tasks that vary in computational complexity. The present research sought to simply and directly evaluate if-apart from simplifying statistical inference tasks-frequencies and small reference classes are clearer than other formats. After eliminating possible computational confounds, simple frequencies (based on small reference classes, e.g. 1/3) and to some extent relative frequencies (percentages, e.g. 33%) were perceived as clearer than absolute frequencies (based on very large reference classes, e.g. 90 million Americans) and single-event probabilities (e.g. 0.33). Concurrently, these different formats were evaluated in terms of their relative influence. Absolute frequencies were relatively more persuasive for smaller magnitudes (e.g. 2.7 million) but less persuasive for larger magnitudes (e.g. 267 million), as compared to analogous presentations. Single-event probabilities were judged to minimize the significance of information. Copyright (C) 2002 John Wiley Sons, Ltd.