In a laboratory experiment, the voluntary provision of public goods is investigated when there is probabilistic uncertainty about the monetary return from production of the public good. After group members make their provision decisions, the return is drawn from an exogenously determined probability distribution. In a linear decision setting, voluntary provision of the public good is contrasted across three treatments. In the "uncertainA" treatment, the return is randomly drawn from a discrete probability distribution. In the "uncertainB" treatment, the return is drawn from a discrete probability distribution that is a mean-preserving spread of the distribution in the uncertainA treatment, but has larger variance. In the “certain†treatment, the return is known with certainty and equal to the expected value of the return in the uncertainA and uncertainB treatments. The data reveal that average provision of the public good is lower in treatments with uncertainty. However, the negative impact of uncertainty on provision only occurs when subjects experience the certain treatment prior to experiencing an uncertain treatment, suggesting an order effect to uncertainty. Also, there is evidence that subjects in treatments with higher uncertainty (variance of the public-good return) display a version of the "gambler's fallacy".
