VOL 87 NO.5 PALFREY AND PRISBREY:PUBLIC GOODS EXPERIMENTS usually had a dominant strategy to free ride terns of behavior as well.These featu while here the subiects sometimes have a dom s in the Appendix jects repeat on per pe In earlier experiments,a subject who con- different group of three other sion error co rom a subject who have the Because was always bigger than.subicc in the first twe never had an incentive to contribute,and there. identify experience effects.The first sequence calle with each value of V is coded as inexperi alth mo ob- ession that lasts a thermore,it was impossible even to observe n wh In all our environments,subjects receive designs.In our ion in that a omly assigned ac 20 ng to a nated and contribution arisi nfrom confusion or decision can be differentiated from ken values.Each time a subject is to make a new ue to nonmonetary components he or she is in Thus,a key benefit of our design is that the do not kno resulting data allows the accurate and unbiased assignments ofr's,but the distribution is pub measurement of the experi And.directly fr The valu 0 is al the esti T come estimates of the amount of altruism and ns of the choice hehavior of each individ. ual at different values of r,and permit the can for the robustn ch。 gregate od V and incomplete information.that are endemic between ex riments.We ha ve an e to natural settings. ber of observations for each of the four differ values I.Experimental Design and Pro (3,6 10.15)(see Table edures There are specific features of our design that subjects contribute in every decision period.In ant to served pat- that condition,on average,40 percent of the commonly ame su bjects are assigne n value tha orth times individu tl.(1989 with he e are n f the Isa (1984) distr lar mtive ways of cnrience producem VOL 87 NO. 5 PALFREY AND PRISBREY: PVBUC GOODS EXPERIMENTS usually had a dominant strategy to free ride, while here the subjects sometimes have a dom￾inant strategy to contribute. In tbe past, sub￾jects repeated tbe decision witb tbe same incentives eacb period; bere tbe subject's in￾centives cbange eacb period. In earlier experiments, a subject who con￾tributed because of confusion or decision error could not be differentiated from a subject who contributed because of altruism or warm glow. Because r, was always bigger than V. subjects never had an incentive to contribute, and there￾fore every contribution could be called a de￾cision error. Bebavior motivated by altruism or a warm glow, although potentially ob￾served, could not be separately identified. Fur￾tbermore. it was impossible even to observe noncontribution when r, < V. Tbus there is an inherent limitation in past designs. In our design this problem is elimi￾nated and contribution arising from confusion or decision error can be differentiated from contribution due to nonmonetary components of tbe utility function. Tbus. a key benefit of our design is tbat tbe resulting data allows tbe accurate and unbiased measurement of strategies — measurement tbat controls for tbe possibility of subject error. And, directly from the estimated strategies come estimates of the amount of altruism and warm glow in the individual utility functions. We can also check for tbe robustness of exist￾ing results to environments tbat include im￾portant features, sucb as diverse preferences and incomplete information, tbat are endemic to natural settings. I. Experimental Design and Procedures There are specific features of our design tbat enable us to address issues that are relevant to understanding other commonly observed pat￾terns of behavior as well. These features are listed below. A sample copy of tbe instructions is in tbe Appendix. 1. Each subject participates in four se￾quences of ten p>eriods (one decision per pe￾riod), eacb ten-period sequence with a different group of tbree otber subjects.^ Tbe first two such sequences bave tbe same value of V. The last two sequences also have tbe same value of V. but different from the value in tbe first two sequences. Tbis allows us to identify experience effects. Tbe first sequence witb eacb value of V is coded as inexperi￾enced, and the second sequence as experi￾enced.'' All four sequences occur in a single session tbat lasts approximately 90 minutes. Eacb session includes 16 subjects. 2. In all our environments, subjects receive r,'s tbat are randomly assigned according to a uniform distribution between 1 and 20 in unit increments. We sometimes refer to these as to￾ken values. Each time a subject is to make a new decision, he or she is independently and ran￾domly assigned a new r, for that decision. Sub￾jects do not know tbe otber subjects' assignments of r/s, but the distribution is pub￾licly announced at the beginning of tbe experi￾ment. Tbe value of Vis also publicly announced. Therefore, the data contain multiple obser￾vations of tbe choice behavior of each individ￾ual at different values of r,. and permit tbe estimation of response functions at both tbe individual and aggregate levels. 3. We vary the value of tbe public good. V, between experiments. We have an equal num￾ber of observations for each of the four differ￾ent values of V e {3, 6, 10, 15} (see Table 1). One value. V = 3, bas tbe feature that group efficiency is not maximized when all subjects contribute in every decision period. In that condition, on average. 40 percent of tbe time subjects are assigned a token value tbat is worth more tban four times tbe individual line experiments with homogeneous preferences, Gerald Marwell and Ruth E. Ames (1980) and D. S. Brookshire et al, (1989) have also conducted experiments with het￾erogeneous preferences, but these are not comparable for other reasons. None of these experiments varied individual incentives across decisions, nor did they provide explicit information about the distribution of incentives in the pop￾ulation. Palfrey and Howard Rosenthal (1991) use an en￾vironment similar to the one explored here, but the public good technology is step-level, not linear. ' Fixing the groups for a sequence of ten periods was done lo maintain comparability with past experiments. We also conducted a replication of one of the Isaac et al. (1984) treatments, using our instructions, computer pro￾tocol, and subject pool. We obtained results, reported in Palfrey and Prisbrey (1993). that were similar to Isaac et al. (1984). ^ Alternative ways of coding experience produce sim￾ilar results