On the modified measure of illusion of control

Abstract

Two potential sources of the illusion of control phenomenon are discussed in the literature, prior knowledge or beliefs on the exercised level of control and the control process itself. The first source causes biases due to prior assumptions about the levels of control; the second source generates biased Bayesian processing, due to exerting some degree of control and thus having engagement in the process that generates outcomes. The illusion of control (IOC) phenomenon is re-examined by means of comparison with a proposed decision model using elements of Bayesian updating. In this paper we consider an agent that for a certain period of time is limited to merely observing, yet for another period of time also may control the results of a random process (a series of successes and failures). Namely, an agent may observe the frequency of successes in both cases: when merely observing a process without any degree of control and also when controlling the process. Agents have prior beliefs about their degree of control over the probability of a successful outcome. The agent under consideration is assumed to be partly rational in nature, as understood by the following assertions. We assume that an agent uses a Bayesian updating process to take into account the prior beliefs when asked about the perceived probabilities of success - when simply observing the process and when actually controlling it. However, we weaken the assumption about perfect rationality by assuming that the perceived probability of success may be further biased by exerting some measure of control over the process (but not when only observing the process). In our paper we propose an analytical method for identifying this part of the observed IOC which is caused by the prior beliefs held by such an agent.The main outcome of the paper is a proposed modified measure of illusion of control. The main idea behind this modification is to compare the perceived level of control with the real measure of control as modified by prior beliefs (of the assumed partly rational agent), instead of only comparing it with an unmodified measure of real control, as described in the literature. The application of such a method provides new insights into the phenomenon of illusion of control. The application of the method is validated through conducting an experiment and analysed at the individual level.