Control without deception

Lying to participants offers an experimenter the enticing prospect ofmaking "others' behaviour" a controlled variable,but is eschewed by experimental economists because it may pollute thepool of subjects. This paper proposes andimplements a new experimental design, the Conditional InformationLottery, which offers all the benefits of deceptionwithout actually deceiving anyone. The design should be suitable formost economics experiments, and works by amodification of an already standard device, the Random Lotteryincentive system. The deceptive scenarios of designswhich use deceit are replaced with fictitious scenarios, each ofwhich, from a subject's viewpoint, has a chance ofbeing true. The design is implemented in a public good experimentprompted by Weimann's (1994) result, from adeceptive design, that subjects are more sensitive to free-ridingthan cooperation on the part of others. The experimentprovides similar results to Weimann's, in that subjects are at leastas cooperative when uninformed about others'behaviour as they are if reacting to high contributions. No deceptionis used and the data cohere well both internallyand with other public goods experiments. In addition, simultaneousplay is found to be more efficient than sequentialplay, and subjects contribute less at the end of a sequence than atthe start. The results suggest pronounced elements ofoverconfidence, egoism and (biased) reciprocity in behaviour, whichmay explain decay in contributions in repeatedplay designs. The experiment shows there is a workable alternative todeception.