Decisions from experience and statistical probabilities: Why they trigger different choices than a priori probabilities

Abstract

The distinction between risk and uncertainty is deeply entrenched in psychologists ' and economists' thinking. Knight (1921), to whom it is frequently attributed, however, went beyond this dichotomy. Within the domain of risk, he set apart a priori and statistical probabilities, a distinction that maps onto that between decisions from description and experience, respectively. We argue this distinction is important because risky choices based on a priori (described) and statistical (experienced) probabilities can substantially diverge. To understand why, we examine various possible contributing factors to the description-experience gap. We find that payoff variability and memory limitations play only a small role in the emergence of the gap. In contrast, the presence of rare events and their representation as either natural frequencies in decisions from experience or single-event probabilities in decisions from description appear relevant for the gap.

Fear shapes information acquisition in decisions from experience

“…sizes in all conditions were substantially larger than those observed in Study 1 and in previous studies (see Hau, Pleskac, & Hertwig, 2010, their Table 1). One key difference between this and previous implementations of the sampling paradigm is that presentation of the decision problems was interrupted several times by the emotion induction procedure.…”

How choice ecology influences search in decisions from experience

“…Consequently, we have recently learned much about the psychology of search in decisions from experience. We now know, for instance, that in the sampling paradigm (a) people tend to rely on fairly small samplesranging mostly from 11 to 19 draws, amounting to nearly 7 ± 2 draws from each distribution-thereby attenuating the impact of rare events (see Hau, Pleskac, & Hertwig, 2010, Table 1); (b) people respond to incentives such that increasing the monetary stakes substantially boosts sampling efforts (Hau, Pleskac, Kiefer, & Hertwig, 2008); (c) small samples amplify the difference between the expected average earnings, thus making the options more distinct and choice easier ; (d) people's short-term memory capacity is positively correlated with the size of the drawn sample (r = .36; Rakow, Demes, & Newell, 2008); (e) people endowed with high numeracy draw larger samples than those with low numeracy (Lejarraga, 2010); (f) people who report to have a high ability in rational thinking draw larger samples than those with low ability (Lejarraga, 2010); and (g) people tend to adopt one of two idealized search policies (piece-wise vs. round-wise search strategy, with the former involving more switching between options), with switching frequency being negatively correlated with sample size (r = À.44) and positively associated with a decision strategy conducive to rare events being underweighted (Hills & Hertwig, 2010).…”

How Representations of Knowledge Shape Actions

“…Both kinds of decision can be understood as opposite poles on a continuum of uncertainty about what one is choosing between. In Knight's ( 1921 ) terminology, decisions from descriptions involve a priori probabilities , whereas decisions from experience involve statistical probabilities , which one must assess "if at all, by tabulating the results of experience" (p. 215), so they invariably fall short of the standards of accuracy set by a priori probabilities (Hau, Pleskac, & Hertwig, 2010 ).…”

“…However, across numerous studies (reviewed in Hau et al, 2010 ), respondents typically proved restrained in their information search, with a median number of samples per choice problem typically ranging between 11 and 19. These results suggest that reliance on small samples is one factor that contributes to the attenuated impact of rare events (Hertwig et al, 2004 ).…”