The problem with assumptions: Revisiting "The dark figure of sexual recidivism".

What is the actual rate of sexual recidivism given the well-known fact that many crimes go unreported? This is a difficult and important problem, and in "The dark figure of sexual recidivism," Nicholas Scurich and Richard S. John (2019) attempt to make progress on it by "estimat[ing] actual recidivism rates . . . given observed rates of reoffending" (p. 171). In this article, we show that the math in their probabilistic model is flawed, but more importantly, we demonstrate that their conclusions follow ineluctably from their empirical assumptions and the unrepresentative empirical research they cite to benchmark their calculations. Scurich and John contend that their analysis undermines what they call the "orthodoxy in academic circles" (p. 173) of low sexual recidivism rates among individuals convicted of sexual offenses, but we underscore that their article does not analyze data in the traditional sense; instead, it just interprets past scholarly work through the use of strong assumptions in a way that, for practitioners, is likely to be opaque and misleading (and, for us, strays into speculation, argument, or advocacy and away from objective research). Our simple calculations show that their findings are highly sensitive to their assumptions, and we conclude that courts and others should recognize Scurich and John's work for what it is-a set of complex hypotheticals that are no more reliable than what judges and lawyers accomplish on their own by simply recognizing the basic problem that not all sexual offenses are reported.

Effect of advertising on awareness of symptoms of diabetes among the general public: the British Diabetic Association Study.

OBJECTIVE: To determine the impact of posters advertising symptoms of diabetes on public knowledge of these symptoms. DESIGN: Structured street interviews of members of the general public before, at the end of, and 10 weeks after a campaign advertising the main symptoms of diabetes. SETTING: Basingstoke and Wolverhampton. SUBJECTS: Three samples of 1000 members of the general public were interviewed. Samples were selected randomly but stratified to match the local population's age (20-75), sex, social class, and racial characteristics. MAIN OUTCOME MEASURES: Knowledge of symptoms of diabetes; perceived seriousness of diabetes; and induction of anxiety about symptoms in the target population. RESULTS: Advertising significantly raised knowledge (without prompting) of symptoms: thirst, 245 before v 411 at end of campaign (P < 0.0001) v 341 after (P = 0.0012 v before); polyuria, 72 v 101 (P = 0.0211) v 92 (P = 0.5169); lethargy, 180 v 373 (P < 0.0001) v 298 (P < 0.0001); knowledge of weight loss and visual disturbance was unaffected. The number of subjects lacking knowledge of any symptoms was reduced from 550 to 388 (P < 0.0001). The perceived seriousness of diabetes was unaffected (mean 7.6 in each phase on a scale of 1 (not) to 10 (very). Before advertising, 449 (45%) claimed to have one or more symptoms of diabetes, but this number fell at the end of the campaign (403; P = 0.0419) and 10 weeks afterwards (278; P < 0.0001). CONCLUSIONS: An advertising campaign raised public knowledge of diabetes symptoms without inducing fear of diabetes or anxiety about symptoms. Its potential for achieving earlier detection of non-insulin dependent diabetes should be evaluated.