Hypothesis testing as risk behaviour with regard to beliefs

Abstract

In this paper hypothesis‐testing behaviour is compared to risk‐taking behaviour. It is proposed that choosing a suitable test for a given hypothesis requires making a preposterior analysis of two aspects of such a test: the probability of obtaining supporting evidence and the evidential value of this evidence. This consideration resembles the one a gambler makes when choosing among bets, each having a probability of winning and an amount to be won. A confirmatory testing strategy can be defined within this framework as a strategy directed at maximizing either the probability or the value of a confirming outcome. Previous theories on testing behaviour have focused on the human tendency to maximize the probability of a confirming outcome. In this paper, two experiments are presented in which participants tend to maximize the confirming value of the test outcome. Motivational factors enhance this tendency dependent on the context of the testing situation. Both this result and the framework are discussed in relation to other studies in the field of testing behaviour. Copyright © 2000 John Wiley & Sons, Ltd.

Effects of asymmetric questions on impression formation: A trade-off between evidence diagnosticity and frequency

“…The crucial point is that when people ask asymmetric queries, they look for highly diagnostic but rare information. A confirming answer to an asymmetric-confirming question strongly supports the focal hypothesis but is exceptional, whereas a disconfirming answer is not decisive but is common (Poletiek & Berndsen, 2000). Hence, if we obtain a yes-answer from a new acquaintance to the question "Have you ever risked your life to save somebody else?…”

“…Accordingly, unlike symmetric questions, asymmetric questions imply that highly frequent answers have less diagnostic value than highly infrequent answers whenever the prior probability of the focal hypothesis is not extreme; that is, when our knowledge about the target of our questions is limited or when we do not have stereotype-consistent expectations (e.g., McKenzie, 2006;Poletiek, 2001;Poletiek & Berndsen, 2000). It is worth emphasizing that, theoretically, diagnosticity and frequency are independent and that the trade-off is not certain to occur, as illustrated above for circumstances in which stereotypes influence the testers' information search (e.g., Trope & Thompson, 1997).…”

“…In fact, from a Bayesian perspective, whenever the focal hypothesis has the same a priori probability as its alternative (i.e., in the absence of strong a priori beliefs), the use of an asymmetric strategy when gathering information implies a trade-off between the diagnosticity and frequency of the anticipated answers (e.g., McKenzie, 2006;Poletiek, 2001;Poletiek & Berndsen, 2000;Rusconi et al, 2012). The crucial point is that when people ask asymmetric queries, they look for highly diagnostic but rare information.…”

“…is asked among a team of fire-fighters, a yes-answer will not be as rare as in the general population (e.g., Trope & Thompson, 1997). However, when people do not have strong a priori expectations and the prior probability of the focal hypothesis is approximately p = 0.5 (as in the case of information searching about unknown persons or about non-stereotyped social targets), a frequent outcome (e.g., an answer that occurs commonly, a frequent test result) is less diagnostic than an infrequent outcome from a Bayesian perspective (e.g., McKenzie, 2006;Poletiek, 2001;Poletiek & Berndsen, 2000). Note that the trade-off between diagnosticity and frequency always occurs whenever the degree of asymmetry of questions (A) is calculated using the aforementioned formula; that is,…”

“…However, other studies have revealed people's tendencies to maximize the diagnosticity of the acquired information in more contextualized tasks (e.g., Poletiek & Berndsen, 2000). In other words, in tasks using less abstract materials than those used by Skov and Sherman (1986), people would prefer to ask questions about features that are not so likely to confirm the focal hypothesis (i.e., ( ) 5…”