The illusion of control: influencing factors and underlying psychological processes.

Abstract

The illusion of control refers to the overestimation of the probability of a win following a personal action in a gambling game. This thesis identifies gaps in the body of literature on factors influencing the illusion, uses a theoretically motivated methodology to address them, and tests the theory underlying the methodology. The thesis consists of a literature review and three papers. The review focuses on factors found to influence the illusion – factors such as the number of response options available in the gambling task, the degree of need for money, the average frequency of successes/wins in a sequence of rounds, and success-slope (i.e., whether wins are concentrated at the beginning or end of the sequence). The review draws attention to problems with the way the illusion of control has been measured in studies of success-frequency and success-slope. This observation, in turn, raises questions as to whether success-frequency and success-slope are, indeed, factors that influence the illusion. The review goes on to discuss the psychological processes underlying the effects of various influencing factors. Two relatively unexplored arguments are advanced. The first is that people in gambling tasks engage in problem-solving. Problem-solving involves searching for actions that bring about the desired outcome, which, in gambling settings, is a substantial monetary win. The greater the number of available response options and the need for money, the more likely the player is to still be searching for effective actions at the time that her perceived control is measured. Such a player is, in turn, less likely to report having 'no control' over the task. A second and related argument is that the actions people consider during problem-solving are influenced by their beliefs about the task at hand. In gambling, beliefs in the gambler's fallacy (Oskarsson et al., 2009) and beliefs about supernatural agents such as luck and God (Atran & Norenzayan, 2004) are particularly relevant. In line with terminology used by Rothbaum, Weisz and Snyder (1982), it is proposed that the illusion of control has two variants, primary and secondary, influenced by the gambler's fallacy and beliefs in supernatural agents respectively. The first two papers describe re-examinations of the effects of success-frequency (N = 97) and success-slope (N = 334) using a methodology consistent with the above explanation. Like most studies of these two factors, the experiments involved a gambling session under a particular success-frequency or success-slope condition, followed by a post-experimental questionnaire about the degree of perceived control over task outcomes. The novel aspects of the methodology included, for example, the separate measurement of the illusion's two variants. Success-frequency was found not to influence the illusion of control when it was measured in this way, while the influence of success-slope was confirmed, in that an 'ascending slope' (a concentration of wins at the end of the sequence) was found to be associated with higher illusory primary control. The finding regarding success-slopes suggests that people expected to learn the correct way of playing through trial-and-error, which is consistent with the above argument that people engage in problem-solving when gambling. The third paper describes a confirmatory factor analysis of a survey about erroneous gambling-related beliefs (N = 329). Items were based on interviews with people who gamble regularly, and, therefore, represented illusions of control – problem-solving solutions based on some playing experience. Consistently with the second argument presented above, the factor analysis showed that the items could be described in terms of two latent factors reflecting the gambler's fallacy and beliefs about supernatural agents, respectively.