EVOLUT COOPERATION ANTH 169
Popular in Course
Popular in anthropology, evolution, sphr
This 7 page Class Notes was uploaded by Kristoffer Nader on Thursday October 22, 2015. The Class Notes belongs to ANTH 169 at University of California Santa Barbara taught by M. Gurven in Fall. Since its upload, it has received 9 views. For similar materials see /class/227013/anth-169-university-of-california-santa-barbara in anthropology, evolution, sphr at University of California Santa Barbara.
Reviews for EVOLUT COOPERATION
Report this Material
What is Karma?
Karma is the currency of StudySoup.
Date Created: 10/22/15
Reports Experiments and Ethnography Combining Methods for Better Understanding of Behavior and Change1 CAROLYN K LESOROGOL George Warren Brown School of Social Work Washington University St Louis Campus BOX 1196 One Brookings Drive St Louis MO 63130 USA c1esorogolwustledu 24 V 0 Supplementary material appears in the electronic edition of this issue on the journal s web page httpwwwtjournalstuchicagot eduCAhomethtmlH There is much evidence that social institutions facilitate and sustain relations of trust reciprocity and coopera tion particularly in smallscale societies where they of ten prove pivotal to successful management of com muna y shared resources McCay and Acheson r987 Ostrom 2000 In many societies including the African herders considered here informal institutions enable community members to agree upon land use rules mon itor each other s actual practice and apply sanctions against violators Dahl and Hjort r976 McCabe 1990 These institutions work in part because individuals have a stake in retaining the multifaceted social relationships that ensure their survival in harsh and unpredictable en vironments and are willing to forgo some individual ben e ts in order to sustain these relationships over the long term Seabright r993 39 ey may be effective institutions are not static nor do they perfectly allocate bene ts across social groups Knight 1992 Given their role in shaping social relations it follows that an institutional change affects those relations in particular ways The challenge is to understand the nature and scope of such effects and this entails devising methods of measuring them at the level of individual or group behavior This article addresses 2005 by The WennerGren Foundation for Anthropological Re search All rights reserved 001 I7 0 A 2 t It Research for this article was supported by grants from the Na tional Science Foundation rant No 0090320 FulbrightHays and t e Center for New Institutional Social Sciences and the Grad uate School of Arts and Sciences of Washington University St Louis I thank the Of ce of the President Kenya for granting re search permission and the Institute for Development Studies Uni versity of Nairobi for research af liation I thank Iean Ensminger Colin Camerer and three anonymous reviewers for helpful com ments on the manuscript this challenge by combining ethnographic and experi mental methods to investigate changes associated with the conversion of communally owned rangeland to pri vate property among Samburu pastoralists in northern Kenya Speci cally I nd that privatization is associated with less cooperative behaviors and a reduction in ad herence to the authority of traditional elders in the com munity Further this research reveals both the potential and the challenges of conducting experiments in the eld and contributes to the emerging literature on the use of these methods in nonWestern sm sc e societies Henrich et al 2004 2001 Cardenas 2003 Tracer 2003 Ensminger 2000 EXAMINING BEHAVIOR CHANGE There are a number of methods for exploring individual behavior Quantitative survey data may serve as indi cators of general tendencies For example measures of school enrollments health facility usage wage labor rate or market transactions might be marshaled as ev idence that people in an African community are partic ipating more in modern society and by extension adopt ing modern attitudes and eliefs Sociologists and anthropologists often survey attitudes and values of their research informants to try to gauge their liker behavior However attitude surveys may be weak indicators of behavior since they reveal only what informants say they would do in a given situation not what they ac tually do Anthropologists also collect qualitative data from observations and interviews as evidence of behav ioral trends By closely examining particular situations and how individuals act and react they draw conclusions about behavior and sometimes generalize across a pop ulationZ Experimental methods provide a novel way to trian gulate with ethnographic methods such as observation interviewing and surveys Particularly when the evi dence from ethnography is mixed experiments have the advantage of focusing on particular types of behavior and enabling the researcher to observe numerous individuals faced with the same behavioral choice Developed pri marily in economics experimental games elicit behavior by asking people to make choices in bargaining situa tions Keeping the games and procedures standard and ab tractin the game from particular realworld situa tions makes comparison across eld sites possible The games generate behavior that can be analyzed statisti cally replicated and compared within and across pop ulations Camerer and Fehr 2000 The purpose of ex perimental economics is to test the assumptions of 2 Barth 1993 for example advocates such an approach to the understanding of behaviort r29 130 CURRENT ANTHROPOLOGY economic theory regarding human behavior and to de velop theories that more adequately account for observed behavior Camerer 2003 Ledyard 1995 While it is rel atively new in anthropology there is increasing interest in using experiments crossculturally Roth et al 1991 Ensminger 2000 Henrich et a 2001 Henrich and McElreath 2002 One of the most important ndings of experimental economics is that behavior in games diverges sharply from the llrational egoist assumption of neoclassical economic theory that individuals act to maximize their material selfinterest Ensminger claims that llit appears that even in highly developed societies individuals place some value upon sharing and cooperation even when they are given every opportunity not to 20002159 She notes that crosscultural experiments have found mem bers of smallscale societies in developing countries no more fair trusting or cooperative than people in the de veloped world In contrast to a moraleconomy perspec tive Scott 1976 she argues that it may be elements of the institutional environment that account for levels of trust fairness and cooperation rather than generalized levels of these factors in a given society 2000169770 If institutions in uence the levels of trust fairness and cooperation that we observe in society then we might expect an institutional change to be re ected in behaviors that reveal such norms Accordingly we might expect that a change in property rights among pastor alists would alter levels of fairness and cooperation among them and that these changes would be detectable in the behavior of individuals in experimental games ETHNOGRAPHIC BACKGROUND LAND PRIVATIZATION AMONG THE SAMBURU The Samburu are pastoralists living in semiarid north central Kenya They number approximately 200000 and herd cattle sheep goats and some camels Historically they were highly mobile moving several times a year in accordance with rainfall and pasture availability In re cent decades sedentarization has proceeded in response to government policies favoring settlement and a desire among many people to live in proximity to health ser vices schools and markets While subsistence produc tion of milk and meat are still mainstays of the Sam buru s lifestyle they are increasingly involved in market transactions of livestock and other commodities as well as in wage labor Sperling 1987 Holtzman 1996 Little et al 2001 Lesorogol 2002 Before the colonial period 189071963 land was com munally held and used and Sambaru had access to areas far beyond the borders of the current Samburu District They did not have xed clan territories but councils of local elders made decisions regarding resource sharing and land use Spencer 1965 The British regime took control of the land vesting decisionmaking authority in government hands While the Samburu appreciated certain aspects of British policy such as the weakening of the neighboring Turkana they generally objected to government efforts to limit the seasonal migrations dic tated by the exigencies of pastoral livestock production Lesorogol 2002 After independence in 1963 parts of the district un derwent an adjudication process to determine owner ship The primary goal of land adjudication in Kenya was to establish individual freehold title to land Okoth Ogendo 2000 However the semiarid lands inhabited by pastoralists were deemed unsuitable for individual own ership because of their lower productive potential In stead following the advice of donors such as the United States government and the World Bank the Kenya gov ernment established a system of llgroup ranches wherein title to land was transferred to groups of house holds Rutten 1 992 Galaty 1994 The idea was thatlocal residents would gain title to the land as a group and would furthermore jointly manage their herds of live stock The government and donors anticipated a shift from the existing system of production in which land was used communally but livestock were individually owned and managed to a system of group land ownership and group management of livestock geared increasingly toward market production Grandin 1981 Rutten 1992 Kimani and Pickard 1998 Most Samburu were not interested in such a radical change to their livestock production system Many op posed group ownership of land an idea that did not exist in their culture Lanyasunya 1990 Others remained un aware of the consequences of the land adjudication ex ercise largely because government 0 cers o ten con cealed the details of the process which was implemented in a topdown heavyhanded fashion Lesorogol 2002 Warned by their leaders that neighboring communities were likely to covet their lands many Samburu joined group ranches solely to preserve their claim to the land However a small group of Samburu mostly men de sired private land of their own These individuals had had experiences outside Samburu District such as for mal education military service or employment which had exposed them to agricultural societies in other parts of the country w ere land ownership was highly valued They associated land ownership with becoming part of modern Kenyan society in which owning land along with formal education Westernstyle clothes and hous ing and whitecollar employment was a maker of suc cess They also realized that land even in Samburu Dis trict was a valuable commodity and an investment opportunity The land adjudication process in Samburu District contained a littleknown provision enabling in dividuals to make claims for private land and these in dividuals took advantage of it In Siambu 37 individuals laid claim in 1978 to the community s most productive land a swat of at fer tile land on a high plateau Lesorogol 2003 This brought them into con ict with the elders who opposed priva tization and it was only in 1986 that a compromise was reached all the land on the plateau was equally subdi vided among resident households while the less desir able land on a rocky escarpment became a group ranch About 240 households received individual plots of ap proximately 23 acres each To date this is the only in stance of equal subdivision of land among the Samburu and it was implemented despite the fact that most res idents opposed privatization per se PRIVATIZATION AND BEHAVIOR ETHNOGRAPHIC EVIDENCE Since most Siambu people agreed to privatization in or der to avoid eviction by the 37 individuals who were trying to appropriate all the land one might not expect to see many changes in actual land use or ideas relating to land ownership During 18 months of eldwork une 20007December 2001 I conducted structured household surveys with 100 randomly selected households in Siambu as well as indepth interviews with more than 30 key informants These data as well as my observa tions of daily life in Siambu revealed that a number of aspects of and use remain essentially unchanged more than ten years after privatization there are few fences dividing property there are fences around cultivated elds cattle still move within and outside the bound aries of the privatized area and people still share some communal resources such as water However changes are also evident A few individuals are fencing their entire parcels not only their cultivated elds More people participate in farming than did so before privatization cattle movements are more limited and encroachments on cultivated areas are punished Timber and rewood have become privately owned com modities that are bought and sold Change in attitudes is manifested in pervasive rhetoric about the virtues of land ownership Practically everyone I spoke with agreed that owning land was a good thing because it freed in dividuals from the control of the community enabling them to make decisions about their land and develop ment unhindered by others These emphases on individual control and autonomy are rather surprising in a society that has relatively ro bust institutions of social control such as a male ageset system exogamous clans and councils of elders Many informants claimed to be immune to the authority of others even traditional elders at least regarding land use decisions Research in anthropology and ecology has demonstrated how pastoralist communities regulate land use on common land and the role of the council of elders is a feature of successful systems Dahl and Hjort I976 McCabe I990 Scoones 1994 It is arguably the authority of elders that often protects common property from the overuse and degradation that would result from an absence of rules and enforcement procedures By re moving their de jure authority over land use privatiza tion shifts bargaining power away from elders as an in stitution of communal land management and toward landowners who now possess greater rights in their in dividual parcels Seabright 1993 and this raises the question to what extent elders authority is eroding in other domains In sum this ethnographic evidence yields a somewhat contradictory picture regarding the extent to which so cial norms regarding land have changed and how behav Volume 46 Number I February 2005 13I ior has been affected In order to clarify the situation I turned to experimental games EXPERIMENTAL DESIGN Because the eldwork was conducted ten years after pri vatization a longitudinal approach measuring behavior in Siambu before and after privatization was not pos sible Instead I decided to compare the behavior of Siambu residents with that of residents of a comparable Samburu community where land had not been privati zed Bernard 1995268 points out that this type of two group posttestonly design can be very effective partic ularly where there are large measurable differences in results between the two groups and where these are sub stantiated by ethnographic evidence The comparison community Mbaringon is located 40 km southeast of Siambu I con ucted extensive eldwork in Mbaringon as well as Siambu and used results from Mbaringon to compare with those in Siambu not only in the experi mental games reported here but also regarding the pol itics of privatization and its socioeconomic outcomes Mbaringon and Siambu have similar cultures envi ronments and land use patterns but they are far enough apart geographically not to have had unusually signi cant in uences on each other for example from daily interactions While it is theoretically possible that the variations I observe in cooperation and adherence to au ority between Mbaringon and Siam u are due to in trinsic differences in the communities predating priva tization Ithink that there are good reasons to doubt this Interviews with many elders in both Mbaringon and Siambu reveal the same pattern of seminomadic land use relying on agreedupon rules of access to seasonally avail able resources prior to privatization in Siambu Fur thermore when a few individuals began to experiment with cultivation in Siambu in the early I960s the com munity under the direction of the elders punished them collectively The opposition to privatization of land in Siambu also involved collective action organized by el ders Lesorogol 2002 2003 These examples suggest that Siambu people were cooperative and that elders au thority was strong prior to privatization However be cause as in most anthropological studies privatization itself was a natural experiment and not a controlled lab oratory one the causal relation of privatization to be havior should be considered indicative not conclusive HYPOTHESES AND METHODS I hypothesized that given the high value they assigned to in ividual free om and control over land Siam u peo ple would behave more sel shly and display lower levels of cooperation in games that measured these variables A decline in cooperation was also a logical consequence of the breaking up of large homesteads and increasing agricultural activities in which labor sharing across households was minimal Lesorogol 2002 In addition I expected Siambu residents to show less respect for elders authority than people in Mbaringon because elders had I32 CURRENT ANTHROPOLOGY lost their authority to make decisions regarding land use as a result of privatization The methods used in these experiments were based on protocols developed by a MacArthur Foundation funded crosscultural project in which games were played in 15 smallscale societies across the globe Hen rich et al 20018 The games were played for real money with the stakes set at one day39s wage here Ksh 100 133 Except in one game players were kept anony mous to each other to reduce reputation effects and games were played in rapid succession to minimize col lusion Players were adults mostly from randomly se lected households that had participated in earlier sur veys A small proportion of players played more than one game though without a decisionmaking role in the sec ond game but no one played the same game twice Public meetings were held in each community to ex plain that they would have a chance to play fun games for real money without revealing the purpose of the games or any speci c procedures On game day people gathered in the morning at the primary school and the directions for the game were read to them4 A number of examples were given to ensure understanding of the game and the roles of the various players Participants received a showup fee of Ksh 40 50 that they retained regardless of what they received in the game Players were chosen at random by drawing their names from a hat and sent into another room to play5 In the game room the instructions were repeated and several ex amples were given to con rm the player39s understanding of the game including the implications of different choices The players easily understood most games and it was not necessary to eliminate anyone because of mis understanding of the game After playing people either went home with their takings or waited in a separate room for their payoff depending on the type of game THE DICTATOR GAME MEASURING FAIRNESS In the Dictator game each pair of players is given an endowment of Ksh 100 Player 1 then decides how to divide the money between himself and player 2 He may opt to keep all the money the choice predicted by neoclassical theory which assumes a pro tmaximizing rational individual or he may give any fraction of the money in 10shilling increments to the other player Player 1 keeps whatever he does not give player 2 Player 239s only role is to accept the amount offered by player 1 from the administrator Positive offers from player 1 in dicate a concern for fairness over selfinterest I con 3 This project now in its second phase is the largest crosscultural research project to use experimental economics games in small scale societies and I am one of the researchers in this phase 4 An exception to this was with the Public Goods game in which directions were read to players when they came into the playing room 5 In the rst games I strati ed the players by wealth However I found that random selection also resulted in a mix of wealth and other variables among the players and decided that it was unnec essary to continue to stratify the sample deliberately ducted this game with 30 pairs of players in Siambu and 32 pairs in Mbaringon Siambu offers in the game averaged 30 of the stake compared with 246 for Mbaringon This difference is not statistically signi cant at the 05 level suggesting that Siambu players are at least as fairminded as Mbar ingon players6 The combined Mbaringon and Siambu mean offer of 273 is within the range of developed country experiments of 20 30 Camerer 200357 58 Fewer players kept all the money for themselves 75 combining Siambu and Mbaringon than is common in developedworld experiments where up to 30 40 of players give offers of zero Camerer and Fehr 2000 The distribution of offers g 1 shows that 23 of Siambu players split the stake 50 50 with player 2 com pared with only 9 in Mbaringon In contrast 38 of players in Mbaringon gave 20 making this the modal offer there The distribution is quite similar to what En sminger 2000 found among the Orma another group of Kenyan pastoralists The spread of offers over the range from 0 to 50 implies that there is no consensus on a single norm of fairness whereas in US studies modal offers of 0 and 50 indicate the two competing norms of pure selfinterest and fairness Ensminger 2000 While I had hypothesized that privatization of land in Siambu would make individuals more sel sh the results of the crosscultural experiments which I learned of only after I had returned from the eld suggest another in terpretation Henrich et al 2001 found that the degree of market integration7 correlated with higher offers in Ultimatum and Trust games These games are similar to the Dictator game in that player 1 decides how to split a pot of money with player 2 In the Ultimatum game player 2 has the option of rejecting player 139s offer in which case both players get nothing In the Trust game 6 I used the nonparametric MannWhitney test to compare the mean offers in Siambu and Mbaringon because of the nonnormal distribution of offers in the games The results of the MannWhit ney test for the Dictator game were 2 164 p 010 7 A rough index of market integration considering degree of wage labor and general involvement in market transactions was used in the rst round of the crosscultural project More precise measures have been developed for the subsequent round of experiments now under way Relative Frequency 00 01 02 03 04 05 06 07 Offers as Fraction of Stake Size 08 09 10 FIG 1 Distribution of offers in the Dictator game Black Mbaringon N 32 white Siambu N 30 the experimenter triples the amount player I allocates to player 2 and then player 2 has the option of returning money to player 18 It may be that greater interaction with markets instills norms of fairness consistent with a 5050 split as individuals realize that the advantages of anonymous transactions depend on one39s reputation for fairness even among strangers Ensminger 2004 Both Mbaringon and Siambu are integrated into markets they trade livestock and other goods purchase food and other necessities and participate in wage labor The of fers in both communities are consistent with those in other societies with similar levels of market integration studied in the crosscultural project supporting the mar ketintegration hypothesis Further privatization of land in Siambu has brought people there into greater contact with markets for land and with government institutions more generally perhaps accounting for their relatively high offers THE PUBLIC GOODS GAME COOPERATION The Public Goods game allows four players to contribute to a quotcommunity projectquot Each payer is given Ksh 50 and may allocate any amount in Ioshilling increments to the project Their contributions are combined and dou bled by the experimenter and the total amount is divided equally among the players Clearly the game involves the danger of free riding by players who contribute noth ing but share equally in the takings It differs from the others discussed here in that the players see each other9 However they are not allowed to talk to each other nor do they know the offers made by the other players I hypothesized that Siambu players would exhibit lower levels of cooperation lower contributions to the community project then Mbaringon The average offer of players in Mbaringon 57 exceeded that of players in Siambu 48 but this difference was not statistically signi cant MannWhitney test 2 124 p 21 Par ticularly striking however was the fact that 31 of players in Mbaringon contributed their entire endow ment to the community project while only 6 did so in Siambu g 2 These results indicate that the mutual trust required for successful cooperation was achieved to a greater extent among Mbaringon players than among Siambu players In fact the Siambu community has had dif culty co operating in building a new school Funds were raised in the mid I 990s for a new school building using barambee a Kenyan practice whereby members of a community contribute to public projects Midway through construc tion the project stalled and many people believed that the community members in charge of the project had 8 In the rst round of the crosscultural project very few of the experimenters used the Dictator game However in the current phase all researchers are using the Dictator game and thus results comparable with mine will soon be available 9 There is experimental evidence that participants are more co operative make higher offers when they know with whom they are playing especially if they are allowed to communicate with them Cardenas 2003 Camerer 2003 Ostrom and Walker 2003 Volume 46 Number I February 2005 133 Relative Frequency 00 02 04 06 08 10 Contributions as Fraction of Endowment FIG 2 Distribution of offers in four person Public Goods game Black Mbaringon N 84 white Siambu N 80 The endowment was Ksb 50 with doubling of contributions by the experimenter misused the funds In this connection Ensminger 2000 points out the possible cueing effect of the Public Goods game Since the game rather closely resembles baram bee it is likely that when people play the game they behave as if they were participating in a barambee She surmises that Orma offers in this game were higher than in other games because of this association with bar ambee in which there are sanctions against free riding The fact that Siambu offers were lower may re ect their negative experience with the school barambee sanc tions had failed to rein in the corruption that marred the school building attempt leading many people to become disillusioned with the whole process Another element of the Siambu ethnographic context is salient here When land was privatized households were forced to move to their new parcels While efforts were made to keep extended families in proximity there was inevitably some separation Even when parcels were geographically contiguous large homesteads were bro ken up as each household moved onto its parcel Settle ment structures re ect this change virtually all Siambu homesteads are composed of a single household not an extended family as is customary and prevalent in Mbar ingon This physical separation has reduced the dayto day cooperation normally involved in pastoral livestock production and domestic work When neighbors are far away it is harder for women to share daily tasks For example I witnessed fewer informal gatherings of women in Siambu than in Mbaringon where women came together frequently to share child care gather re wood or fetch water It was also more dif cult to as semble men for meetings in Siambu than in Mbaringon or in many other communities where I had worked pre viously Further Siambu people39s increasing reliance on householdlevel crop cultivation for food and income may have reduced the incentives to cooperate given the constraints on time and labor and the rising costs of cooperation resulting from physical separation I34 I CURRENT ANTHROPOLOGY ELDERS39 AUTHORITY IN A DOUBLEBLIND DICTATOR GAME Privatization removed elders legal authority over land in Siambu They can no longer make binding decisions about how individuals use land as they do on communal or groupranch land by for example closing off certain areas for grazing and imposing nes on violators Legally any individual owner can assert the right to keep livestock off his land A few have done so by fencing their parcels Individual rights to grow crops are recognized and even protected by the elders the reverse of past sanctions against cultivation Even if elders39 de facto authority over land has not been completely lost it has been seriously compromised and transformed Does this relative loss of authority extend to other realms of daily life To help answer this question I designed a version of the Dictator game to measure adherence to elders39 au thority The game was to be played exactly the same way as the normal Dictator game but rst I convened a meet ing of elders and asked them to agree on an offer that they felt was the most appropriate for player I to make that is to establish a norm for the offer In both com munities the elders agreed that a 5050 split was the best offer to make When giving instructions to the play ers I informed them of the elders39 norm but reiterated that they were free to make any offer they chose To ensure anonymity this game was played doubleblind so that even I would not know who made which offer10 In this version of the Dictator game the mean offer in Mbaringon was 263 slightly higher than in the rst game 246 While there was still a mode at 2 the whole distribution was shifted to the right g 3 In Siambu however the mean offer dropped signi cantly from 30 to 206 and there was a mode at 1011 Offers of 50 dropped from 23 to 7 The difference in means between Mbaringon and Siambu in this version of the Dictator game was nearly signi cant MannWhit ney test 2 I83p 06 These results indicate that players in Siambu did not adhere to the norm set by the elders On the contrary players apparently flouted the authority of the elders by giving particularly low offers There is a possibility that this drop was an effect of the doubleblind treatment in this experiment However if that were the case we would expect a similar drop in Mbaringon which did not occur12 By contrast in Mbaringon the rise in offers of 10 To do this players made their offers in a separate room from the playing room placing their offers in unmarked envelopes and depositing the envelopes in a box After all the offers were made the box was retrieved and the offers were recorded and distributed randomly to the recipients II The drop in offers was signi cant in Siambu MannWhitney testz 249 p 01 12 The doubleblind treatment aims at controlling for experi menter effects Thus the fact that offers dropped signi cantly in Siambu could be due to experimenter effects there aside from any effect of the elders setting a norm for the game Since I am some what less well known in Siambu than in Mbaringon where Ihave lived for over ten years it is possible that people made higher offers in the rst Dictator game out of concern for what I would think of them At the same time the fact that I am very well known in 045 04 035 03 025 02 015 01 005 Relative Frequency 00 01 02 03 04 05 06 07 08 09 10 Offers as Fraction of Stake Size FIG 3 Distribution of offers in double blind Dictator game With norm set by elders Black Mbaringon N 30 white Siambu N 30 40 and the absence of zero offers suggest some effort to respect the norm although the mean only rose slightly The problems in cooperation noted above may also signal a decline in respect for elders39 authority During my eldwork I observed two cases of intraclan marriage in Siambu a practice strictly prohibited by Samburu In both instances men of the warrior agegroup challenged their elders arguing that there was no meaningful reason to deny these couples the right to marry They even made arguments from modern science that the prospective bride and groom were not genetically closely related in order to support their case Most elders privately opposed the marriages but were unable to stop them These ex amples suggest a reduction of elders39 authority in Siambu that is con rmed by the game results THE IMPORTANCE OF OWNERSHIP IN SAMBURU NOTIONS OF FAIRNESS In games in which player one decided how to split the endowment given to both player I and player 2 Dictator Dictator with Norm18 there was a mode of 20 in Mbaringon Was this just chance or was something more systematic at work Other information gleaned during the games encouraged me to examine this issue more deeply First a few players in the Dictator game told me that they gave 20 because this was fair They pointed out that retaining money for themselves and their fam ilies was their main priority Second when discussing the Dictator game with the elders in Mbaringon some took the position that 20 was a reasonable offer since family needs should take precedence over sharing the money with a stranger Others claimed that since the money had been given to both players it was only fair Mbaringon could also be a reason for people to worry about how I perceived their offers It is dif cult to speculate on the direction of experimenter effects My general feeling was that people were not overly concerned about my knowing their offers but I cannot en tirely rule out possible experimenter effects 13 This was also the case in two other games not reported here Dictator with Third Party Punishment and Trust to split it evenly These arguments appeared to hinge on the issue of ownership Who really owned the money in the game Further investigation revealed that concepts of fairness depended crucially on the notion of ownership or enti tlement After the games I had individual discussions with about 30 people in both Siambu and Mbaringon in which I posed several hypothetical situations involving sharing that are common occurrences in daily life I asked women llIf you had a kilogram of sugar at home and someone came by asking for sugar how much would you give her Or llIf you were butchering a goat to eat and someone came by how much meat would you give I asked men llIf you were hunting which is unusual but not entirely unheard of and killed a gazelle and were butchering it and someone came by how much meat would you give Or llIf you walking with a friend in the forest and came upon a dead gazelle how would you share the meat Answers to these questions were very uniform across informants regardless of age gender and wealth In cases where the individual clearly owned the resource the amount they would give away was about 20 In the sugar example women said that they would give the person a glass of sugar about 200 grams 20 if a kilo gram They might give slightly more or less depending on their relationship to the person but a glass was the norm Similarly when butchering their own animal ei ther because it was in their herd or because they ha hunted it they would give a passerby a hind leg there is even a saying to this effect about a quarter of the beast However when ownership was clearly joint as in the case in which the two friends discovered a dead ga zelle in the forest then a 5050 split was considered fair since neither friend had more right to the dead animal than the other When ownership was clear norms of fairness were also clear In all these discussions informants from both Mbaringon and Siambu gave answers almost immedi ately Very 39ttle thought was required because the norms were known Considering the offers made in the games in light of these norms I think that the question of ownership remained ambiguous enough to enable multiple interpretations Even though the wording of the instructions was clear that the money was being given to both players the fact that player 1 gained physical possession of the money and was given the right to de cide how it was divided created a situation of de facto ownership by player 1 A player could reason that he actually owned this money and in that case an offer of 20 was perfectly fair This is consistent with other experimental results showing that subjects who ha earned the right to be player 1 by scoring high on a generalknowledge test gave lower offers in the Dictator game Hoffman et al 1994 Hoffman and Spitzer 1985 Bolton Katok and Zwick 1998 note that the speci c instructions given to players make a difference in how players perceive their roles and how they play In this case it seems that when individuals felt entitled to the money they behaved in more selfinterested ways Con Volume 46 Number 1 February 2005 135 fusion about ownership may also account for the spread of offers in both communities Two conclusions follow from this analysis First the Dictator game may be a more accurate measure of fair ness if the ownership issue is rendered unambiguous One way to test this would be to play the game using one of the hypothetical situations and see if offers were more consistent Second if concepts of fairness vary across cultures and are situationally dependent even ex periments may not be an accurate basis for comparisons Mbaringon people appeared to be less fairminded in these games than players in the United States or else where but they were making equally fair offers accord ing to their norms of fairness and their understanding of ownership in the game While this insight may enable us to alter the game to make it more accurate in Sam buru this will come at a cost of less comparability across u tures CONCLUSIONS HOW USEFUL WERE THE GAMES In general experimental economics strives to increase understanding of the bases of human behavior Here my intention was to combine experimental methods with ethnographic data to draw conclusions regarding the ex tent to which behavior has changed following an insti tutional change the privatization of land Beyond the speci c ndings from the experiments the approach was useful in at least two ways First the results of the games triangulated with ethnographic observations and inter view data to reinforce certain conclusions less cooper ation and less adherence to authority in Siambu and cast doubt on others more selfinterested behavior Using standard games enabled comparison of my results with those of a larger project with many more communities leading to new hypotheses about behaviorin Siambu and Mbaringon the marketintegration hypothesis Second the games themselves raised new research questions about cultural norms of fairness leading to an interesting new line of enquiry which I pursued employing quali tative interviews Combining experiments with anthro pological insight into culture and social context has the potential to yield stronger analyses and greater under standing of the links between institutions change and behavior References Cited EARTH F 1993 Balinese worlds Chicago University of Chi cago Press BERNARD R 1995 2d edition Research methods in anthror pology Qualitative and quantitative approaches Walnut Creek AltaMira Press BOLTON G E KAT K AND R zwrcx 1998 Dictator game giving Rules of fairness versus acts of kindness Internar tional Journal of Game Theory 27269799 CAMERER 200 Behavioral game theory Experiments in strategic interaction Princeton Princeton University Press AMERER C AND E 000 Measuring social norms and preferences using experimental games A guide for anthro pologists MS MacArthur Foundation Anthropology Project