Integers are better: Adding decimals to risk estimates makes them less believable and harder to remember

Purpose: To determine whether the number of decimal places in a personal health risk estimate influences the extent to which people believe and remember the estimate. Methods: 3422 adults in a demographically diverse US-based online sample (mean age 50, 52% female, 74% white, 56% no college degree) were asked to imagine they were visiting an online risk calculator hosted by a prominent university’s medical school. We designed a mock calculator similar to existing calculators available online. The calculator asked a series of health questions relevant to kidney cancer and returned a hypothetical estimate of lifetime risk of kidney cancer. In this between-subjects experiment, participants were assigned one of seven risk estimates close to the average lifetime risk of kidney cancer in the US. Participants who were randomized to the no decimals condition received an estimate of 2%. Those in the one, two or three decimals conditions received an estimate of 2.1% or 1.9% (one decimal), 2.13% or 1.87% (two decimals), or 2.133% or 1.867% (three decimals). Participants were asked to indicate how believable they found the estimate to be on a six-point scale anchored by labels, “not at all,” and, “extremely.” Then, after completing a second, unrelated survey (median time for this task was 8 minutes), they were asked to recall to the best of their ability the kidney cancer lifetime risk estimate they had been given earlier. Results: Risk estimates expressed as integers were judged as the most believable (F(3, 3384)=2.94, p=.03). Compared to estimates with decimal places, integer estimates were judged as highly believable (defined as the top two points of the six-point scale) by 7 to 10% more participants (Chi-squared(3)=17.82, p<.001). Recall was highest for integer estimates. Odds ratios for correct approximate recall (defined generously as being within 50% of the original estimate) were, for one decimal place, OR=0.65 (95% CI 0.49, 0.86), for two decimal places, OR=0.70 (95% CI 0.53, 0.94), and for three decimal places, 0.61 (95% CI 0.45, 0.81). Exact recall showed a similar pattern, with larger effects. Conclusions: Using decimals in risk calculators offers no benefit and some cost. Rounding to the nearest integer is likely preferable for communicating risk estimates so that they might be remembered correctly and judged as believable.