A choice is most tempting when it is both immediate and certain. Prevailing experimental measures of static present-bias rely on the random incentive scheme (RIS) to gather rich data from subjects. Recent studies show that RIS adds background risk to decisions, thereby increasing risk-taking behavior. Evidence has shown that the immediacy effect is significantly moderated (or even eliminated) by uncertainty. Thus estimates of the present-bias factor $\beta$ are possibly biased upward toward unity, underestimating present-bias intensity. I conduct a framed field experiment that determines whether uncertainty indeed diminishes the immediacy effect by using real-effort tasks to approximate immediate and certain utility flows. Subjects allocate tasks using convex time budgets under both RIS and certain implementation. I compare within-subject estimates of the quasi-hyperbolic present-bias factor under these two mechanisms, assuming convex effort costs. I type individuals as present- or future-biased under each mechanism, which provides data on the interaction between the immediacy and certainty effects within subjects. I also report how the introduction of uncertainty affects the present-bias parameter on average, which would be some of the first incentivized evidence of the effect of uncertainty on immediacy.
Is an option is especially tempting when it is both immediate and certain? To study the effect of risk on present bias, I conduct an online experiment in which workers allocate about thirty minutes of real-effort tasks between two weeks. I compare choices made two days before the first workday against choices made when work is imminent. In baseline treatments, one choice is randomly implemented; meanwhile, one treatment implements a particular allocation with certainty. By assuming that effort costs are not affected by the mechanism (and thus independent of risk preferences), my novel design permits estimation of present bias using a decision with a consequence both immediate and certain. I find the average intensity of present bias is far greater under certainty than under risk. I find no evidence that present bias is more pervasive among individuals, suggesting instead that present-biased individuals become more myopic.