In daily life, individuals often need to predict whether or not some outcome will transpire. In the context of gambling, they might need to predict the likelihood that some card will appear or some sporting club will prevail. In the context of work, they might need to predict whether the price of shares will rise or whether they are unlikely to be awarded some tender. In specific circumstances, these predictions are biased.
Specifically, individuals often overestimate the likelihood of outcomes they would like to transpire. To illustrate, if they would like a specific club to win or price to rise, they will overrate the probability of these events, called the desirability bias or wishful thinking effect.
An important insight, however, is the desirability bias is not universal but only surfaces in specific circumstances (Windschitl, Smith, Rose, & Krizan, 2010). In particular, when almost no valid information is available to decide which outcome will arise, this bias is pronounced (Windschitl, Smith, Rose, & Krizan, 2010). In most other instances, this bias dissipates.
The desirability bias is often observed in a specific paradigm, developed by Marks (1951). In a typical example, participants must predict whether or not the next card will be marked, with a particular symbol or color. They earn money if their prediction is correct. In addition, in some conditions, they also earn some money if the marked card appears, regardless of their predictions. In other conditions, they lose money if this marked card appears.
A common finding is that individuals tend to predict the marked card will appear when this card attracts money rather than incurs a cost. Accordingly, when individuals want this card to appear, they predict this outcome is more likely (e.g., Irwin, 1953).
Variations of this task have been developed. In most studies, cards either are marked or not marked. In the studies conducted by Windschitl, Smith, Rose, and Krizan (2010). however, some of the cards were one color and other cards wesre another color.
In many contexts, especially when the marked card paradigm is not applied, the desirability effect is not significant (e.g., Bar-Hillel, Budescu, & Amar, 2008). To explore this issue in more detail, Krizan and Windschitl (2007) conducted a meta-analysis.
In 14 of these studies, participants needed to specify which outcome they predict instead of the likelihood of these outcomes. In all of these studies, the outcome was stochastic--that is, entirely random. Tshat is, the outcome, which is whether or not a marked card will appear, was random. Indeed, in 12 of these 14 studies, the marked card paradigm was utilized.
In 13 of these 14 studies, the desirability bias was significant. The odds ratio, on average, was 2.26.
In 16 of the other studies, participants needed to specify the likelihood that each outcome would appear. Within this set, in 9 of the studies, the outcome was stochastic or random--and only two of the effects were significant. In the other 7 studies, the outcome was not entirely stochastic or random. For example, individuals might need to answer a question and then predict whether their answer was correct. In these studies, four of the effects were significant. Hence, these results were mixed.
In studies conducted by Windschitl, Smith, Rose, and Krizan (2010), participants completed a series of scales, such as measures of affectivity, need for cognition, reliance on rational or experiential processes, trait optimism, regulatory focus, and belief in good luck as well as other instruments. None of these factors, however, significantly moderated the desirability effect. Further research, thus, needs to be conducted to ascertain whether this effect varies across individuals.
Some researchers maintain the desirability bias might represent a demand characteristic and, thus, reflect a methodological artifact (see Rosenthal & Fode, 1963). To understand this possibility, consider the methodology. In one trial, the experimenter might highlight the marked card will earn money& in this trial, the marked card is conceptualized as valuable In the same trial, the experimenter then asked participants to predict which card will appear. Participants might not want to act dismissively, implying the valuable card is unlikely to appear. They might, thus, feel obliged to feign their faith in the experimenter, maintaining they believe this card will be selected.
Windschitl, Smith, Rose, and Krizan (2010). however, undertook a variant of the classical methodology to refute this possibility. Specifically, in this variant, the experimenter was not aware of whether the marked card was valuable or not. Instead, before each trial, the computer presented information, directed only to the participant, which indicated whether the marked card would attract money or incur a cost. These participants, thus, would not have felt obliged to predict the valuable card was more probable. Despite this measure, the desirability bias prevailed. Participants assumed the valuable card would be more likely to be selected.
According to the biased evaluations account, preferences or desires distorts the evaluation of information. If individuals want a specific event to unfold, their memory or attention is unduly directed to cues or hints that are consistent with this possibility, reminiscent of motivated reasoning (e.g., Balcetis, 2008).
In the context of the marked card paradigm, suppose that individuals would prefer an orange card to appear. Their memory might be biased towards previous instances in which the orange card appeared. As a consequence, they might overestimate the frequency of this event.
Nevertheless, this finding cannot explain the dissipation of the desirability bias in the trivia paradigm. Specifically, in one of the studies conducted by Windschitl, Smith, Rose, and Krizan (2010), participants received a series of trivial questions, like "What animal makes a louder noise: a blue whale or a lion". Participants were asked to predict which of these two alternatives were true. In addition, like the marked card paradigm, on some trials, they also received more money if the first alternative was true. On other trials, they received more money if the second alternative was true.
In this instance, the desirability bias was negligible. If a specific alternative would earn money, participants were no more likely to predict this option was correct.
The biased evaluations account would imply the desirability bias should extend to this context. Indeed, this account would imply the bias might be especially pronounced in this context. To demonstrate, in this instance, participants could reflect upon many possible sources of evidence to support either of the two alternatives. If one alternative is desirable, they should readily be able to shift their memory and attention to evidence of this option, as stipulated by the biased evaluations account.
Preferences or desires might not bias evaluation of the evidence. Instead, these preferences or desires might bias the threshold to predict an outcome. If individuals prefer a specific alternative, they might predict this option will unfold even if the evidence is modest. In contrast, if individuals do not prefer a particular alternative, they will predict this option will unfold only if the evidence is pronounced (cf., Price & Marquez, 2005, cited in Windschitl, Smith, Rose, & Krizan, 2010).
Again, however, this explanation assumes the desirability bias should persist, and perhaps be more pronounced, when trivia questions, with two options, are presented in lieu of cards. That is, regardless of whether participants need to decide whether a specific card will be selected or a specific answer to a question is correct, this mechanism should apply. Windschitl, Smith, Rose, and Krizan (2010), however, generated findings that contradict this hypothesis: The effect vanished when trivia questions were used in lieu of cards.
Windschitl, Smith, Rose, and Krizan (2010) proposed an alternative mechanism. According to these authors, when individuals need to predict some outcome, they can engage in one of two processes. First, they might deliberate over the evidence, methodically and systematically reflecting upon numerical and objective information. Second, they might trust their intuition, hunches, or guesses--especially if such numerical or objective information was unavailable. Only when they trust these intuitions or guesses will participants select the alternative they prefer or desire.
Windschitl, Smith, Rose, and Krizan (2010) conducted a series of studies, all variants of the marked card paradigm, that substantiate this proposition. First, in one study, they varied the percentage of cards that were marked with a specific color. That is, 30, 40, 50, 60, or 70% of the cards were marked with a specific color.
The desirability effect was particularly pronounced when 50% of the cards were marked with this color. In this condition, participants are likely to guess, which amplifies the desirability effect. In the other conditions, they have access to more objective information--notably, that one color is more or less frequent than is another color. These participants are thus less inclined to guess, which should curb the desirability effect.
Second, in one study, the marked card paradigm was utilized again. On some trials, however, participants were asked to trust their hunch or intuition--intended to enhance their reliance on guesses. On other trials, participants were intended to consider the statistical properties of marked cards to reach a decision.
When participants trusted their hunches, and thus might have guessed, the desirability effect was observed, even when the marked card was observed on 30 or 70% of trials. This finding is consistent with the proposition that guesses exacerbate the desirability effect.
Third, in another variant of the marked card paradigm, participants did not merely specify whether or not they assume a marked card--that is a card of a specific color--will be selected. Instead, they indicated the likelihood this card will be selected, using a analogue or slider scale. In this instance, participants do not need to guess which card will appear when both alternatives are equally likely. They can utilize objective or numerical information to estimate the likelihood of each option.
Thus, according to the biased guessing account, the desirability bias should dissipate in this condition. Indeed, as Windschitl, Smith, Rose, and Krizan (2010) showed, the perceived probability of each alternative was unrelated to the desirability of these options& the desirability effect was not observed.
Finally, as this account assumed, the desirability bias may be observed when participants need to predict which of two answers to a trivia question are correct--but only if they guess the answers. To assess this possibility, in one study, the two answers were almost identical. For example, participants were asked "The first police force in Paris was established in 1676 or 1667?" They needed to predict which alternative is true& in addition, on some trials, the first option would attract money regardless of their prediction.
In these example, when participants presumably guessed, the desirability effect was observed. They chose answers that attracted money and were thus desirable. However, when the responses could be differentiated, and thus more systematic processes were probably applied, the desirability effect dissipated (Windschitl, Smith, Rose, & Krizan, 2010).
Balcetis, E. (2008). Where the motivation resides and self-deception hides: How motivated cognition accomplishes self-deception. Social and Personality Psychology Compass, 2/1, 361-381.
Bar-Hillel, M., & Budescu, D. V. (1995). The elusive wishful thinking effect. Thinking & Reasoning, 1, 71-104.
Bar-Hillel, M., Budescu, D. V., & Amar, M. (2008). Predicting World Cup results: Do goals seem more likely when they pay off? Psychonomic Bulletin and Review, 15, 278-283.
Gordon, R., Franklin, N., & Beck, J. (2005). Wishful thinking and source monitoring. Memory & Cognition, 33, 418-429.
Harvey, N. (1992). Wishful thinking impairs belief-desire reasoning: A case of decoupling failure in adults? Cognition, 45, 141-162.
Irwin, F. W. (1953). Stated expectations as a function of probability and desirability of outcomes. Journal of Personality, 21, 329-335.
Irwin, F. W., & Metzger, M. J. (1966). Effects of probabilistic independent outcomes upon predictions. Psychonomic Science, 5, 79-80.
Krizan, Z., & Windschitl, P. D. (2007a). The influence of outcome desirability on optimism. Psychological Bulletin, 133, 95-121.
Krizan, Z., & Windschitl, P. D. (2007b). Team allegiance can lead to both optimistic and pessimistic predictions. Journal of Experimental Social Psychology, 43, 327-333.
Krizan, Z., & Windschitl, P. D. (2009). Wishful thinking about the future: Does desire bias optimism? Social and Personality Psychology Compass, 3, 227-243.
Marks, R. W. (1951). The effect of probability, desirability, and "privilege" on the stated expectations of children. Journal of Personality, 19, 332-351.
Rosenthal, R., & Fode, K. L. (1963). Psychology of the scientist V: Three experiments in experimenter bias. Psychological Reports, 12, 491-511.
Windschitl, P. D., Martin, R., & Flugstad, A. R. (2002). Context and the interpretation of likelihood information: The role of intergroup comparisons on perceived vulnerability. Journal of Personality and Social Psychology, 82, 742-755.
Windschitl, P. D., Smith, A. R., Rose, J. P., & Krizan, Z. (2010). The desirability bias in predictions: Going optimistic without leaving realism. Organizational Behavior and Human Decision Processes, 11, 33-47.
Windschitl, P. D., & Weber, E. U. (1999). The interpretation of "likely" depends on the context, but "70%" is 70%-right?: The influence of associative processes on perceived certainty. Journal of Experimental Psychology: Learning, Memory, and Cognition, 25, 1514-1533.
Last Update: 7/11/2016