Journal of Economic Psychology 27 (2006) 98–116 www.elsevier.com/locate/joep
Adolescent–adult interactions and culture in the ultimatum game Robert Hoffmann *, Jin-Yee Tee Nottingham University Business School, The University of Nottingham, Jubilee Campus, Nottingham NG8 1BB, United Kingdom Available online 2 August 2005
Abstract We report on ultimatum game experiments conducted with same-age as well as mixed-age groups of Asian adolescents and adults. The findings indicate that compared with adults, adolescents make significantly larger offers to adult responders and are less likely to reject. The findings are explained with reference to theories of childhood development and (culturespecific) socialisation. Ó 2005 Elsevier B.V. All rights reserved. JEL classification: D64; C72; C91 PsycINFO classification: 2300; 2800; 2930 Keywords: Adolescents; Ultimatum game; Culture; Fairness; Experiments
1. Introduction Experimental economics provides a powerful tool for the analysis of the rational choice foundations of traditional economics. Over the last few decades, much has been achieved in establishing to what extent economic agents conform to traditional *
Corresponding author. Tel.: +44 115 846 8070; fax: +44 115 846 6667. E-mail address: robert.hoff
[email protected] (R. Hoffmann).
0167-4870/$ - see front matter Ó 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.joep.2005.06.014
R. Hoffmann, J.-Y. Tee / Journal of Economic Psychology 27 (2006) 98–116
99
assumptions, and what alternative views may be fruitfully explored. Recently, a number of authors have begun to extend this type of approach to the investigation of the economic behaviour of children and adolescents. Two of the studies in this area report on experiments with children and adolescents playing the ultimatum game. The ultimatum game has been popular in experimental economics generally because it pits economic rationales of monetary gain against social preferences over alternative allocations of payoffs over players (Camerer & Fehr, 2004, p. 2). These include issues of fairness as well as self-respect. Experiments with non-adult subjects may shed light on how these types of preference develop during childhood. Both Murnighan and Saxon (1998) and Harbaugh, Krause, and Liday (2003) conducted ultimatum game experiments within various age groups of children ranging between 3 and 17 years and found significant differences between these both in terms of offer size and rejection rate. While each study reports different relationships between age and offer size, both found that younger responders are less likely to reject. These results have been interpreted as evidence that bargaining behaviour may have a social, rather than a genetic foundation (Camerer, 2003; Harbaugh, Krause, & Liday, 2003). Our current work is intended to contribute to this research in two ways. First, Harbaugh, Krause, and Liday (2003) as well as Murnighan and Saxon (1998) argue that much of the economic significance of childrenÕs behaviour is due to their interactions with adults in mutual bargaining. However, this type of process cannot be directly observed through the bargaining experiments within same-age groups that these studies report. As a result, we designed experiments in which the ultimatum game is played between adults and adolescents in mixed-age pairs. Our experimental design may shed light on the nature of such inter-age group bargaining, for example between parents and their children. We suspected that this age difference between proposer and responder may generate effects on player behaviour over and above those observed between games among children and games among adults respectively. The reason is that bargaining raises issues of distribution and fairness which intuition suggests may be sensitive to asymmetries in player characteristics. Secondly, in our experiments, we study South-East Asian subjects, a large proportion of whom are of Chinese descent. As a result, our experiments enable us to compare the behaviour of Asian adolescents as well as adults with those Western subjects studied in previous experiments on the ultimatum game. This type of comparison contributes to the growing literature in cross-cultural behavioural economics and psychology (Cooper, Kagel, Lo, & Gu, 1999; Hemesath & Pomponio, 1998; Henrich, 2000; Henrich et al., 2001; Kachelmeier & Shehata, 1992). There has been a recognition that decision making specifically and psychological processes generally may be culture sensitive (Nisbett, 2003). A number of authors have argued for the need of establishing culture-specific bodies of psychological insights. This case has been made strongly for the Chinese (Bond, 1996b). Our experiments may therefore contribute to this agenda by shedding additional light on the differential processes of socialisation and development amongst Asian, and in particular Chinese, adolescents. The remainder of our paper is organised as follows. Section 2 discusses in more detail the theoretical background to this study. Our own methods and experimental
100
R. Hoffmann, J.-Y. Tee / Journal of Economic Psychology 27 (2006) 98–116
procedures are outlined in Section 3. Experimental results are presented in Section 4. Section 5 concludes the paper with a discussion of the main findings.
2. Economic socialisation, moral development and ultimatum bargaining Economic socialisation describes the maturation of economic thinking as well as behaviour over the course of childhood (see Webley (2004) for an up-to-date overview). It is becoming increasingly popular as an area of research both within economic psychology and behavioural economics for a number of reasons. First, we may be interested in the decision making of children and adolescents to the extent that they engage in economically significant behaviour (Harbaugh, Krause, & Berry, 2000; Harbaugh, Krause, Liday, & Vesterlund, 2003; Roland-Le´vy, 1990; Webley, 2004). This may include their independent consumption decisions (Krause & Harbaugh, 1999) as well as their influence on the economic decisions of adults (Harbaugh, Krause, & Liday, 2003; Murnighan & Saxon, 1998). In addition, the study of children and adolescents may reveal much about the origins of adult decision making and their economic arrangements generally (Harbaugh, Krause, & Berry, 2000; Webley & Lea, 1993). In particular, Krause and Harbaugh (1999) as well as Camerer (2003) argue that research of this type may contribute towards a nature versus nurture debate with respect to economic decision making, i.e. to what extent the reasoning processes and preferences which economists study are the product of genetic inheritance or learning. The burgeoning literature on economic socialisation has so far mainly focussed on cognitive economic development, i.e. the childÕs understanding of the concepts and institutions of the adult economic world (Roland-Le´vy, 1990, for instance). In contrast, childrenÕs efforts to solve their own economic problems, either in interactions with other children or as participants in the adult economy, have been relatively neglected (Webley, 2004; Webley & Lea, 1993). However, a small but growing number of contributions to this latter agenda have investigated issues such as pocket money, child saving behaviour and the Ôplayground economyÕ (Webley, 2004). Additional contributions to this latter agenda have been made through the study of child decision making in laboratory experiments. To date, such experiments have been conducted for the trust game (Harbaugh, Krause, Liday, et al., 2003), to test for endowment effects (Harbaugh, Krause, & Vesterlund, 2000), to examine cooperation in the prisonerÕs dilemma and contributions to public goods generally (Fan, 2000; Krause & Harbaugh, 2000), attitudes towards risk (Harbaugh, Krause, Liday, & Vesterlund, 2002) and transitivity in choice (Harbaugh, Krause, & Berry, 2000). In addition, there have been two studies of childrenÕs behaviour in the ultimatum game (Harbaugh, Krause, & Liday, 2003; Murnighan & Saxon, 1998). This is a two-player sequential game in which player 1 (the proposer) selects a division of a constant stake, which is then offered to player 2 (the responder, see Fig. 1). If player 2 accepts, both players receive the proposed shares. In the case of a rejection, neither receives anything. As the acceptance of any positive share dominates rejection, the proposer should rationally offer the minimum share to the responder. Since it was first studied
R. Hoffmann, J.-Y. Tee / Journal of Economic Psychology 27 (2006) 98–116
rational fair
accept
3, 1
reject
0, 0
accept
2, 2
reject
0, 0
accept
1, 3
reject
0, 0
101
Proposer altruistic
Responder
Fig. 1. A simplified ultimatum game in extensive form. The proposer decides between offering a fair, rational or altruistic split of a stake of 4, and the responder can accept or reject the proposal. The subgame-perfect equilibrium is [rational, accept].
by Gu¨th, Schmittberger, and Schwarz (1982), the ultimatum game has been a popular vehicle for studying fairness considerations in economic decision making. The experimental evidence is summarised in Roth (1995) and Camerer (2003). In short, most offers tend to be in the region of 40–50% of the stake, and tend to be accepted. The rejection rate rises to about 50% for offers below 20% of the stake. The growing experimental literature on the ultimatum game has identified a host of factors that affect both offer levels and rejection rates in experiments. Camerer (2003) categorises these effects as methodological (player anonymity and game repetition), cultural, descriptive (framing of the game), structural (information, communication and strategy set) as well as demographic (gender, race, academic specialisation, physical appearance and age). Murnighan and Saxon (1998) investigated the differences in ultimatum bargaining generated by age, gender as well as incomplete information regarding the stake size. As far as age is concerned, their results seem to suggest an inverted U-shape in the relationship between age and offer size. Young children and adults tend to make lower offers than early adolescents. The authors interpret this finding as a ‘‘developmental discontinuity’’. A similar set of ultimatum game experiments was conducted by Harbaugh, Krause, and Liday (2003) within four classes of 7, 9, 14 and 18 year old boys as well as girls respectively, and yielded a similar relationship between age and offer behaviour. Camerer (2003) describes these findings as ‘‘the result of the swing of a pendulum from pure self-interest (at young ages), to obsession with strict equality (in 3rd grade), to an adult compromise’’. In both studies, younger children were less likely to reject offers. The results of these two studies under experimental conditions comparable to our current work reported below are presented in Fig. 2. What is the explanation for these findings? In addition to their own payoffs, ultimatum game players in general consider fairness in the distribution of payoffs among both players (Camerer, 2003; Camerer & Fehr, 2004). Proposals reflect expectations of rejection as well as altruistic considerations. In the dictator game, a variant of the
102
R. Hoffmann, J.-Y. Tee / Journal of Economic Psychology 27 (2006) 98–116 50 47.5 45
Offer (% of stake offered to responder)
42.5 40 37.5 35 32.5 30 27.5 25 22.5 20 17.5 15 12.5 10 7.5 5 2.5 0 2
3
4
5
6
7
8
9
10
11
Adult
School Grade Murnighan and Saxon (1998)
Harbaugh et al. (2003)
current study
Fig. 2. Comparative experimental results for same-age ultimatum games reported in Murnighan and Saxon (1998), Harbaugh, Krause, and Liday (2003) as well as in the current study. Age is given in terms of US school grades. In this system, grade 2 children are on average between 7 and 8 years old.
ultimatum game where offers are not subject to acceptance or rejection, even anonymous proposers make non-zero offers. Moreover, responders value monetary rewards as well as fairness in the offers they receive (Forsythe, Horowitz, Savin, & Sefton, 1994; Gu¨th et al., 1982), evidenced by systematic rejections of non-zero offers, both in theory (stated minimum acceptable offers) and game play. Fairness considerations are based on moral preferences which mature during childhood. The predominant approach to moral development, the cognitive-structural model, studies stages in independent development based on a childÕs innate abilities to respond to experiences made (Windmiller, 1980). PiagetÕs seminal work in this area was developed further by Kohlberg (1976, 1986) as well as Damon (1975, 1980).1 In DamonÕs stage theory of positive justice, young children tend to make single-criterion distributive decisions based initially on self-seeking rationales, and in subsequent stages on strict equality, and later merit. By the age of 10, a host of criteria are considered based on the characteristics of the situation at hand. In addition, by this age, authority is seen as a reversible and consensual agreement, rather than based on physical or psychological attributes. This theory provides an explanation of the experimental results discussed above. Support for this theory was provided by Enright et al. (1984) and Enright et al. (1980) for American,
1
The Piagetian notion of a universal and invariant sequence of developmental stages is controversial and has attracted significant criticism (Abrahamsen, 1999) as well as support (Dasen, 1994; Eckensberger, 1994) over the years.
R. Hoffmann, J.-Y. Tee / Journal of Economic Psychology 27 (2006) 98–116
103
European as well African children using new experimental methods allowing greater measurement precision and standardisation. As distributional decisions involve logical operations, there is a link between moral and cognitive development. However, it is less clear whether stages in logical reasoning precede (Kohlberg & Gilligan, 1971), parallel (Piaget, 1932) or potentially succeed stages in moral development (Damon, 1975). An alternative to the cognitive-structural approach to moral development is provided by cultural and social learning theories (Windmiller, 1980). This approach has antecedents in behaviourist thought and sees the child essentially as a tabula rasa, acquiring morality through processes of instruction, internalisation and reinforcement. The particular moral values acquired depend on the norms of the childÕs environment. Cultural differences in ultimatum game behaviour have been reported in a number of studies (Henrich, 2000; Henrich et al., 2001; Hoffman, McCabe, & Smith, 1982; Roth, Prasnikar, Okuno-Fujiwara, & Zamir, 1991). We are interested here particularly in the differences between Western and Chinese subjects, the latter of which make up the majority of our sample. Although Chinese culture has been the subject of previous work in experimental economics (Cooper et al., 1999; Hemesath & Pomponio, 1998; Kachelmeier & Shehata, 1992), no evidence has so far been collected in the context of the ultimatum game. An examination of Chinese subjects in the ultimatum game is of particular interest owing to the distinct cultural heritage of the Chinese people. The importance of childhood socialisation is a key feature of Chinese culture and has over many millennia sustained a distinct system of values among the Chinese (Wu, 1996), many of which are relevant for ultimatum game play. Two central dimensions are filial devotion, as well as harmony in interactions with peers (Bond, 1996a; Bond et al., 1987). Filial piety is one of the most prevalent, enduring as well as binding of all Chinese values, and justifies authority based on rank both within the family (in terms of age) and in interpersonal relationships generally (Ho, 1996). Lee (1974) found that similar to moral development, filial cognition matures in distinct developmental stages. While filial devotion governs relationships with superiors, the second important class of Chinese values, based on the teachings of Confucius, guides relationships with individuals of the same rank or seniority. This set of peer values includes cooperation, reciprocation, harmonious social interaction, tolerance of others, kindness, forgiveness and compassion (Bond et al., 1987). Based on these considerations, one would expect Chinese children to defer to adult opponents in the ultimatum game. Moreover, cooperative peer values may generate larger average offer levels in Chinese same-age ultimatum games compared with those involving Western subjects. This would be in line with experimental findings relating to Chinese and Western subjects in the prisonerÕs dilemma (Hemesath & Pomponio, 1998). It is important to note that these traditional values are still reflected in contemporary Chinese societies both in mainland China as well as in overseas Chinese communities such as the one studied in the present work. Amongst Singaporean Chinese parents for instance, education and adherence to Chinese values are reported to be the top two goals of childhood socialisation. Discipline and respect are frequently
104
R. Hoffmann, J.-Y. Tee / Journal of Economic Psychology 27 (2006) 98–116
reported values. Singaporean Chinese parents were found to be more traditional in their attitudes on these issues than their counterparts in Shanghai and Taiwan (Wu, 1996). We may expect to find the same in Malaysian Chinese parents, similar to Singaporeans both culturally and in terms of a shared history. While both independent development and socialisation contribute towards the formation of moral preferences, their effects on moral maturation are different. Children as young as 3–4 years distinguish between absolute moral issues concerning the welfare of others, and relative ones based on convention and consensus. This is the case both in Western societies as well as in Chinese cultures, where socialisation is comparatively pronounced (Yau & Smetana, 2003).
3. Experimental method Ultimatum game play is governed both by cognition and moral preferences regarding fairness. We have seen how both of these are formed during childhood through both independent developmental processes and deliberate socialisation. We conducted experiments to ascertain how these processes manifest themselves in ultimatum games played in same-age as well as mixed-age groups of adults and adolescents respectively, as well as between Chinese and non-Chinese adults and adolescents. Our research agenda was to investigate whether (and how) offer levels and rejection rates differ significantly between the various age and ethnic conditions we study. In particular, we wanted to compare behaviour in our mixed-age games to those adult–adult and child–child games studied by previous authors. In addition, we wanted to ascertain whether cultural differences would become manifest in the behaviour of our Chinese subjects. A total of n = 160 subjects for four separate experimental groups were recruited from three English-spoken private educational institutions in Kuala Lumpur, Malaysia. Details of the groups are given in Table 1. The institutions are in close geographical proximity and have student populations relatively homogeneous in socio-economic terms. Our set of subjects reflects the ethnic diversity of Malaysian society to some extent, which in 2000 was composed of Malay and other indigenous groups (58%), as well as of people of Chinese (24%), Indian (8%) and other (10%) ethnicity. These ethnic groups are approximately associated with Muslim, Buddhist-Taoist, Hindu and Christian religious beliefs respectively. Our subject pool
Table 1 Description of the four experimental groups Group
Institution Subjects Chinese subjects Average age
1
2
3
4
Monash University 40 37 20
Sri KL 40 28 11
KDU College 40 22 21
Sri KL 40 30 12
R. Hoffmann, J.-Y. Tee / Journal of Economic Psychology 27 (2006) 98–116
Group 1 Adults
AC
AA
Group 3 Adults
105
Group 2 Adolescents
CC CA
Group 4 Adolescents
Fig. 3. Matching of the four experimental groups for ultimatum game play. Solid arrows indicate the directions of offers, unfilled arrows denote responses to them. The four experimental conditions are denoted AA for games with adult proposers and responders, AC for games with adult proposers and adolescent responders, CA for games with adolescent proposers and adult responders, and CC for games with adolescent proposers and responders.
comprised all ethnic groups, although the majority of subjects were of Chinese descent (about 75%).2 The remainder comprised Malay, Indian and mixed-ethnicity individuals. Ultimatum games were played between these four groups in four separate experimental sessions (see Fig. 3). In session 1, held at Monash University Malaysia, each group 1 subject was asked to make two separate ultimatum game offers, one to a subject from group 2, and the other to a group 3 subject. Proposers were only told the age cohort and institution type of responders, who were otherwise anonymous and not present at the session. Their offers were submitted using questionnaires.3 The stake size was 10 Ringgit Malaysia (RM).4 Session 2 was conducted at Sri KL School and involved subjects from experimental group 4, each of whom was asked to make two separate proposals, to a group 3 and to a group 2 subject respectively, following the same procedure observed for the first group. In the third session, conducted subsequently at Sri KL, each group 2 subject was randomly allocated two proposals from among those made to group 2 subjects in the first two sessions, one from a group 1 proposer, and one from a group 4 proposer. Responders were only told the age cohort and institution type of the proposers of each of the two offers, who were otherwise anonymous and not present at the session. The offers were presented on questionnaires, which were also used to record responses. In the final session, conducted at KDU College, this process was repeated for group 3 subjects, who were each asked to respond to two offers made to this group, one randomly chosen from 2 For both economic and political reasons, Chinese students dominate private educational institutions in Malaysia generally. 3 The questionnaires used for both proposers and responders are given in the appendix. 4 At the time of the experiments, RM 10 was worth 2.63 US Dollars. In order to aid comprehension on the part of the adolescent proposers, possible offers were constrained to the nearest RM and listed on the questionnaire. The average hourly wage for casual labour in Kuala Lumpur was in the region of RM 4. An indication of sufficient incentivisation is that 95.66% of subjects across all groups indicated a willingness to participate in future experiments.
106
R. Hoffmann, J.-Y. Tee / Journal of Economic Psychology 27 (2006) 98–116
group 1 proposals, and one from group 4 respectively. Throughout all sessions, a nocommunication environment was strictly enforced. Experimental instructions were given on the questionnaires and administered verbally as well as through audio– visual support material. SubjectsÕ pay-outs were determined after the sessions on the basis of proposals, responses and matchings, and delivered after all sessions were completed. Three features of our experimental design should be noted. First, offers were communicated via questionnaires, and opponents remained anonymous throughout. The reason is that we wanted to control for a number of effects known to arise in faceto-face bargaining, such as the effects of player attractiveness and other physical features (e.g. Schweitzer & Solnick, 1999). However, the effects of player age differentials may also partly develop through face-to-face interactions and may therefore elude our design. On the other hand, to the extent that we do find age effects, we may be sure that they are not caused by intervening physical factors. For instance, as adults tend to be taller than adolescents, there is a possibility that apparent age effects found in face-to-face experiments may be partly attributable to an unrelated Ôheight premiumÕ identified in labour market remuneration (Persico, Postlewaite, & Silverman, 2001). A similar argument applies to our cultural variable. Our subjects had no knowledge of the ethnic background of their opponents. In this experiment, we were interested in cultural effects in bargaining decisions with anonymous opponents, rather than in the dynamics of bargaining when the ethnicity of opponents is known. Secondly, in our experiments, the stake size was held constant at RM 10 for both adolescent and adult players. It may be argued that this sum of money may provide different degrees of incentivisation for the two age groups and therefore shape their behaviour differently. In response, Harbaugh, Krause, and Liday (2003) use tokens as the stake, which subjects could later exchange for money or toys at rates depending on their age. However, the effect of differential stake valuation is likely to be small as stake size effects have been shown to be small in the ultimatum game generally (Camerer, 2003, pp. 60–62). In addition, a problem in devising suitably age-adjusted stake sizes arises in terms of adequately measuring the difference in incentivisation a monetary sum provides for different age groups. Finally, every one of our 160 experimental subjects made two decisions, either two proposals or two responses. This raises two issues, a theoretical one, and a practical one. The first is to what extent our results may be sensitive to this design feature. For instance, it might be argued that as a result of it, players were sensitised to age difference in their opponents and may have been influenced accordingly. Without further clarifying experiments, it is unclear how substantial this type of effect is in practice. In addition, it is also not obvious whether this or alternative designs are more representative of realistic economic phenomena which these and other experiments are designed to shed light on. Many adult–child negotiations potentially involve comparisons with experience of same-age bargaining on the part of those involved. These considerations should be borne in mind in the interpretation of our results. The second, practical issue is one of data analysis. As a result of the two-decisions design,
R. Hoffmann, J.-Y. Tee / Journal of Economic Psychology 27 (2006) 98–116
107
our experimental data set comprises a mixture of paired as well as independent observations. Care needs to be taken in the choice of appropriate analytic method. This choice is explained in detail in the following.
4. Results The general issue under consideration is whether the ages and ethnic backgrounds of both proposer and responder have an impact on offer size as well as propensity to reject. Both offers and responses were made under the four experimental conditions outlined in Fig. 3: group 1 adults to group 2 adolescents (denoted AC), group 1 adults to group 3 adults (AA), group 4 adolescents to group 2 adolescents (CC) as well as group 4 adolescents to group 2 adults (CA). As discussed previously, there are underlying general factors which govern the behaviour of proposer and responder in the ultimatum game which we need to account for in our experiments. In general, responder behaviour may be influenced by considerations of fairness of the offers they receive, and consequently, proposers form expectations of the responderÕs monetary valuation of fairness. We asked responders to state the minimum offer they would be willing to accept (RMAO). In addition, we asked proposers to indicate the minimum offer size they expect their opponents to accept (PEXP). In addition, responders were asked to assess offer fairness (FAIR) on a Likert scale between 1 (= strongly disagree offer is fair) and 5 (= strongly agree offer is fair). The variables in our dataset are summarised in Table 2. Table 3 contains a summary of the experimental results. The data show average offer levels just below the fair 50–50 split of the stake for offers under AA, AC and CC, and slightly above for condition CA. The modal offer was RM 5 in all four. While the mean offer levels are relatively similar between groups, the rejection rate differs noticeably between 5% and 30%. In order to test for any significant effects, we may compare both offer levels and rejection rates under the four experimental conditions. For each dependent variable, there are therefore six possible comparisons which need to be conducted (see Table 10). In addition, potential differences in offers and rejection rates for Chinese
Table 2 The experimental variables Variable
Description
O R PAGE RAGE RMAO PEXP FAIR PCHI RCHI
Share of stake offered to respondent (0 6 O 6 10) Percentage of offers rejected Proposer age (in years) Responder age (in years) Responder minimum acceptable offer Proposer expectation of responder minimum acceptable offer Responder (Likert scale) evaluation of offer fairness = {1, 2, 3, 4, 5} Dummy variable for proposer ethnic origin (1 = Chinese, 0 = other) Dummy variable for responder ethnic origin (1 = Chinese, 0 = other)
108
R. Hoffmann, J.-Y. Tee / Journal of Economic Psychology 27 (2006) 98–116
Table 3 Summary experimental data by experimental condition and ethnic origin Condition
Offer O
PEXP
R
FAIR
RMAO
Mean
Mode
StDev
Mean
StDev
Mean
StDev
Mean
StDev
Mean
StDev
AA AC CA CC
4.585 4.800 5.450 4.800
5 5 5 5
1.448 1.506 1.358 1.265
3.825 3.525 5.475 4.625
1.583 1.797 1.485 1.125
22.50 17.50 30.00 05.00
0.423 0.383 0.464 0.221
3.500 3.775 3.375 4.250
1.198 1.493 1.314 1.171
4.825 4.150 4.400 4.350
1.647 1.929 2.275 2.119
Chinese Other All
4.837 5.692 4.906
5 5 5
1.400 1.494 1.422
4.313 4.923 4.363
1.691 1.553 1.684
18.37 23.08 18.75
0.389 0.439 0.392
3.694 4.077 3.725
1.322 1.320 1.332
4.380 4.517 4.438
2.083 1.891 2.002
Means and standard deviations are given for offer size, proposer expectation of responder minimum acceptable offer, rejection rate, responder perceived fairness of offer, and responder minimum acceptable offer.
versus non-Chinese subjects suggested in Table 3 are tested. Following previous work, we used non-parametric tests throughout as experimental data tends not to be distributed normally (Davis & Holt, 1993, p. 526). In the following, the results of all these tests are reported in turn (Table 4). First, we consider offer levels. The experimental data regarding proposal behaviour are summarised in Figs. 4 and 5, which respectively display the offers made by adults (group 1) as well as adolescents (group 4) to both adult and adolescent responders. While both figures attest the preponderance of equal-share offers, they also suggest differences between the offers individual subjects make contingent on responder age. The possibility of such differences is evident especially in the relatively tall grey bars in Fig. 5. To identify significant effects, we compared offer levels under the four experimental conditions. Out of the six possible pairwise comparisons, two involve offers made by the same proposers to different responder types (comparing AA with AC, and CA with CC, respectively). The remaining four entail independent observations of offers made by different experimental subjects. For the former related samples comparisons, we conducted Wilcoxon Signed Ranks tests (see Siegel & Castellan (1988, p. 87)). Table 5 reports the results of testing for differences Table 4 Rejection rates by condition and offer level category O Category
Condition AA R
Altruistic FAIR Rational
16.67 20.83 30.00
All
22.5
AC % of O 15.00 60.00 25.00 100
R 14.29 00.00 42.86 17.50
CA % of O 17.50 47.50 35.00 100
R 25.00 07.69 63.64 30.00
CC % of O 40.00 32.50 27.50 100
R 00.00 03.33 16.67 05.00
% of O 10.00 75.00 15.00 100
Altruistic offers are O > 5, fair: O = 5, rational: O < 5. Percentages of total offers in each category (% of O) given for all conditions.
R. Hoffmann, J.-Y. Tee / Journal of Economic Psychology 27 (2006) 98–116
109
Offers made by Adults
100 90 80 70 60 Offer (% of stake offered to 50 responder)
40 30 20 10
39
35
37
33
31
29
25
Subject number
27
21
23
17
19
15
11
Offers to adults Offers to adolescents
13
9
5
7
1
3
0
Fig. 4. Offers made by group 1 proposers (Monash University) to both adult (KDU College) and adolescent (Sri KL Class 2) responders (conditions AA and AC). For each subject along the horizontal axis, the grey bar represents the offer made to an adult responder, and the black bar the offer made to an adolescent responder. Offers made by Adolescents
100 90 80 70
Offer 60 (% of stake offered to 50 responder) 40 30 20 10
39
35
37
33
31
27
29
23
25
19
Subject number
21
17
15
11
Offers to adults Offers to adolescents
13
7
9
5
1
3
0
Fig. 5. Offers made by group 4 proposers (Sri KL Class 1) to both adult (KDU College) and adolescent (Sri KL Class 2) responders (conditions CA and CC). For each subject along the horizontal axis, the grey bar represents the offer made to an adult responder, and the black bar the offer made to an adolescent responder.
110
R. Hoffmann, J.-Y. Tee / Journal of Economic Psychology 27 (2006) 98–116
Table 5 Wilcoxon Signed Ranks test for difference in offer levels of group 1 proposers to group 2 (OAC) and 3 (OAA) responders respectively OAC OAA
Negative ranks Positive ranks Ties Total
Z Asymp. sig. (two-tailed)
0.776 0.438
a b c
N
Mean rank
Sum of ranks
13a 11b 16c 40
9.50 16.05
123.50 176.50
OAC < OAA. OAC > O AA. OAC = OAA.
between the offers group 1 adults made to group 2 adolescents and group 3 adults respectively (conditions AA and AC). They show that adult offers are not influenced by responder age (two-tailed p = 0.438). Table 6 reports Wilcoxon test results of differences between group 4 proposals to group and 2 and 3 responders respectively (conditions CA and CC). The results indicate that the adolescents of group 4 made significantly larger offers to adults than to adolescent responders (two-tailed p = 0.012). The remaining four comparisons are between offers made in different groups and therefore involve independent samples. They were analysed using Mann–Whitney tests which reveal significant differences at the 10% level between the offer levels in conditions AA and CA as well as AC and CA, respectively (Table 7). Our second dependent variable is the rejection behaviour of responders. Table 3 suggests potential differences in rejection rates across the four groups. These are broken down by the level of offers made (altruistic, fair or rational) for each experimental condition in Table 4. In two conditions, AA and CC, the rejection rate rises with diminishing offer levels, as would normally be expected. However, in the mixed-age Table 6 Wilcoxon Signed Ranks test for difference in offer levels of group 4 proposers to group 2 (OCC) and 3 (OCA) responders respectively OCC OCA
Negative ranks Positive ranks Ties Total
Z Asymp. sig. (two-tailed)
2.503 0.012**
The symbol ** denotes significance at the 95%-level. a OCC < OCA. b OCC > OA. c OCC = OCA.
N
Mean rank
Sum of ranks
21a 10b 9c 40
17.96 12.45
371.50 124.50
R. Hoffmann, J.-Y. Tee / Journal of Economic Psychology 27 (2006) 98–116
111
Table 7 Mann–Whitney tests of differences between offer levels under different experimental conditions Conditions
U
OAA OAA OAC OAC The symbol
OCA OCC OCA OCC *
618.50 758.00 615.00 724.50
Z
p-Value 1.850 0.486 1.862 0.830
0.064* 0.627 0.063* 0.407
denotes significance at the 10%-level.
conditions AC and CA, altruistic offers are rejected more frequently than fair ones. We may use data collected on subjectsÕ fairness evaluations of offers (FAIR) as well as minimum acceptable offers (RMAO) to shed light on this. In condition AC, a single adolescent rejected an altruistic offer (O = 6). This observation may possibly be an outlier due to miscomprehension, as the subject concerned gave the offer the highest fairness rating and stated a RMAO of 5. In condition CA, 16 of 40 adolescents made altruistic offers, 4 of which were rejected by their adult responders. On average, these responders gave their offers a low fairness rating of 2 out of 5. Unfortunately, this rating is unable to differentiate between two possible interpretations of lacking fairness as either Ôtoo generousÕ or Ôtoo miserlyÕ. The average RMAOs of these subjects is 3.25, lower than the average for their group of 4.40 (see Table 3), indicating the possibility of misguided altruism towards their adolescent proposers. However, the small number of observations involved must be borne in mind when interpreting these data. The next step is to assess whether the differences in rejection rates across the four conditions are statistically significant. We used Fisher Exact Probability tests (see Davis & Holt, 1993, p. 546; Siegel & Castellan, 1988, p. 103) to examine differences in the proportions of rejections between the following pairs of conditions involving independent observations: AA and AC, AA and CC, AC and CA as well as CA and CC. Conversely, McNemarÕs tests (see Siegel & Castellan, 1988, p. 75), appropriate for related samples, were used for comparisons of responses made by the same subjects in two conditions, i.e. comparing AA and CA, as well as AC and CC, respectively. The results for all tests are presented in Table 8. Significant differences exist between rejection rates in the conditions AA and CC, as well as between CA and CC, respectively. Finally, we are also interested whether subjectsÕ cultural backgrounds had an effect on ultimatum game play. Table 3 shows slightly higher average offers and rejection rates for the non-Chinese subjects. We tested whether either offer levels or rejection rates displayed significant differences along cultural lines. Care must be taken as the number of non-Chinese subjects is small in some of the four groups and normality may not be assumed safely (see Table 1). Again, we employed Mann–Whitney tests to investigate offer levels, and Fisher Exact Probability tests to examine rejection rates. However, as previously, tests need to be constructed on the basis of the mix of independent and paired observations in our data set. Our approach here was to split the data set such as to ensure that any two behavioural
112
R. Hoffmann, J.-Y. Tee / Journal of Economic Psychology 27 (2006) 98–116
Table 8 Tests of differences between rejection rates under different experimental conditions Conditions AA AA AA AC AC CA
AC CA CC CA CC CC
Test
p-Value
FISHER MCNEMAR FISHER FISHER MCNEMAR FISHER
0.781 0.629 0.048** 0.293 0.125 0.006***
FISHER are Fisher Exact Probability tests. MCNEMAR are McNemar Related Samples tests. The symbols ** and *** denote significance (two-tailed) at the 5% and 1%-levels respectively.
Table 9 Mann–Whitney tests of differences between offers made by Chinese and non-Chinese proposers respectively under (a) conditions AA and CA, and (b) AC and CC Conditions PCHI = 1 (a) PCHI = 1 (b)
U PCHI = 0 (a) N-CHI = 0 (b)
328.00 380.50
Z
p-Value 1.485 0.819
0.138 0.413
observations made by the same experimental subject were never in the same test. This split generated two sub-samples for proposals which were tested for cultural effects separately. Table 9 shows the test results for (a) independent offers in the AA and CA conditions, and (b) for independent offers in the conditions AC and CC. No significant differences were found. Similarly, to ensure independence, we tested differences in Chinese and non-Chinese responses for two sub-samples using Fisher Exact Probability tests. The first generated a two-tailed p-value of 0.387 for responses to offers in conditions AA and AC. The associated CramerÕs V value was 0.129, indicating little correlation between ethic background and rejection rates for this sample. The second, conducted for offers in the conditions CA and CC, also demonstrated no significant differences (two-tailed p = 0.552, CramerÕs V = 0.085). As none of the tests yielded significant statistics, there is no evidence for cultural differences between the particular ethnic groups in our experiment.
5. Discussion The results of our statistical tests are summarised in Table 10. In this final section, we discuss reasons for and implications of these results. First, we can compare the results from our two same-age group experiments with results pertaining to comparable groups in previous experiments. Fig. 2 displays offer levels in our groups 1 and 4 relative to comparable groups in Murnighan and Saxon (1998) and Harbaugh, Krause, and Liday (2003). In contrast to this previous work, there is no significant difference between the offers made by our adults and adolescents playing same-age opponents. One possible explanation is that the age effect
R. Hoffmann, J.-Y. Tee / Journal of Economic Psychology 27 (2006) 98–116
113
Table 10 Summary of results comparing offer levels (O) and rejection rates (R) in the four experimental conditions and on the basis of subject ethnicity Condition
Difference in
1
2
O
R
AA AA AA AC AC CA
AC CA CC CA CC CC
No Yes* No Yes* No Yes**
No No yes** No No Yes***
PCHI = 1 RCHI = 1
PCHI = 0 RCHI = 0
No No
No No
The symbols *,
**
and
***
denote significant test statistics at the 10%, 5% and 1%-levels respectively.
found previously may be culture specific. For instance, a more pronounced childhood socialisation process among Asians (and Chinese in particular) may cause their adolescent behaviour to conform more closely to that of adults at an earlier age. As mentioned in the introduction, a comparison of our same-age groups of Malaysians with those studied in previous experiments sheds some light on the issue of cultural differences in utimatum game play. The offer averages in both our sameage groups are higher than those for comparable age groups in the two previous experiments. Our Malaysian children and adults offered 5% more of the stake to their peers compared with the US subjects of Harbaugh, Krause, and Liday (2003). This type of finding may be interpreted in terms of Asian, in particular Chinese values of cooperation and harmony. However, a definitive comparison cannot be made at this stage because (a) significance tests on these differences cannot be performed; and (b) some experimental conditions (gender, player anonymity, incentivisation) in these studies differ from our own. Properly controlled experiments may shed further light on the existence of cultural differences in ultimatum game play. Such work is currently underway. Our first research question concerned the effect of player age differences on offer levels in the ultimatum game. Table 3 indicates relatively high offers in condition CA, which were shown to be significantly different from offers made in the other conditions AA, AC and CC. No significant differences exist between offer levels in these other conditions (Table 10). First, adult proposers make no allowances for responder age. In addition, as mentioned, adolescent and adult proposers make the same level of offers to same-age responders. Why do adolescents make significantly higher offers to adult proposers? Previous research has shown that ultimatum offers are shaped by a combination of fairness and strategic considerations. Both potential explanations have implications for adolescent thinking in their encounters with adults. Adolescents may feel adults have greater entitlements to the stake in mixed-age bargaining, and may therefore expect adults playing adolescents to reject more readily. Our results support the previous finding that adolescents are less likely to reject generally. However, our mixed-age design provides additional insight into this
114
R. Hoffmann, J.-Y. Tee / Journal of Economic Psychology 27 (2006) 98–116
phenomenon. In particular, this difference is significant only when comparing condition CC with both AA and CA, respectively. In other words, adolescents reject less than adults only when responding to fellow adolescents. In games with adult proposers, adolescents reject as often as adult responders. As result, the lower rejection rates in children found here and in other experiments appear not to be a constant feature of child ultimatum game responses, but rather depend on proposer age. Our second objective concerned the existence of cultural effects within our experiments. In our ethnically mixed set of subjects, there were no differences in rejection or proposal behaviour between the two groups. However, these conclusions must be treated as extremely tentative due to the small proportion of non-Chinese subjects, and their cultural heterogeneity. A more explicit examination of Chinese cultural and ultimatum game behaviour is therefore called for.
Acknowledgement The authors would like to thank KDU College, Monash University and SRI KL School for their permission to conduct the experiments as well as students for their participation. We also thank the guest editors, two anonymous reviewers, SweeHoon Chuah, Martin Jones and Geoffrey Williams for helpful suggestions, and Carolyn You for assistance conducting the experiments. Financial support for this project by the British Academy (grant no. SG-36080) is gratefully acknowledged.
References Abrahamsen, A. (1999). Cognitive and linguistic development. In W. Bechtel & G. Graham (Eds.), A companion to cognitive science. Blackwell companions to philosophy (vol. 14, pp. 146–156). Oxford, UK: Blackwell. Bond, M. H. (1996a). Chinese values. In M. H. Bond (Ed.), The handbook of Chinese psychology (pp. 208–226). Oxford, UK: Oxford University Press. Bond, M. H. (Ed.). (1996b). The handbook of Chinese psychology. Oxford, UK: Oxford University Press. Bond, M. H., Shtin, N. T., et al. (1987). Chinese values and the search for culture-free dimensions of culture. Journal of Cross-Cultural Psychology, 18(2), 143–164. Camerer, C. F. (2003). Behavioral game theory. Princeton, NJ: Princeton University Press. Camerer, C. F., & Fehr, E. (2004). Measuring social norms and preferences using experimental games: A guide for social scientists. In J. Henrich, S. Bowles, R. Boyd, C. Camerer, E. Fehr, & H. Gintis (Eds.), Foundations of human sociality. Oxford University Press. Cooper, D. J., Kagel, J. H., Lo, W., & Gu, Q. L. (1999). Gaming against managers in incentive systems: Experimental results with Chinese students and Chinese managers. American Economic Review, 89(4), 781–804. Damon, W. (1975). Early Conceptions of positive justice as related to the development of logical operations. Child Development, 46, 301–312. Damon, W. (1980). Structural-developmental theory and the study of moral development. In M. Windmiller, N. Lambert, & E. Turiel (Eds.), Moral development and socialization (pp. 35–68). Boston, MA: Allyn and Bacon. Dasen, P. R. (1994). Culture and cognitive development from a piagetian perspective. In W. P. Lonner & R. S. Malpas (Eds.), Psychology and culture (pp. 145–149). Boston, MA: Allyn and Bacon.
R. Hoffmann, J.-Y. Tee / Journal of Economic Psychology 27 (2006) 98–116
115
Davis, D. D., & Holt, C. A. (1993). Experimental economics. Princeton, NJ: Princeton University Press. Eckensberger, L. H. (1994). Moral development and its measurement across cultures. In W. P. Lonner & R. S. Malpas (Eds.), Psychology and culture (pp. 71–78). Boston, MA: Allyn and Bacon. Enright, R. D., Bjerstedt, A., Enright, W. F., Levy, V. M., Lapsley, D. K., Buss, R. R., et al. (1984). Distributive justice development: Cross-cultural, contextual, and longitudinal evaluations. Child Development, 55, 1737–1751. Enright, R. D., Franklin, C. C., & Manheim, L. A. (1980). ChildrenÕs distributive justice reasoning: A standardized and objective scale. Developmental Psychology, 16(3), 193–202. Fan, C.-P. (2000). Teaching children cooperation – an application of experimental game theory. Journal of Economic Behavior and Organization, 41, 191–209. Forsythe, R., Horowitz, J. L., Savin, N. E., & Sefton, M. (1994). Fairness in simple bargaining experiments. Games and Economic Behavior, 6, 347–369. Gu¨th, W., Schmittberger, R., & Schwarz, B. (1982). An experimental analysis of ultimatum bargaining. Journal of Economic Behavior and Organization, 3, 367–388. Harbaugh, W. T., Krause, K., & Berry, T. (2000). GARP for kids: On the development of rational choice behavior. American Economic Review, 91(5), 1539–1545. Harbaugh, W. T., Krause, K., & Liday, S. G. (2003). Bargaining by children, working paper. Harbaugh, W. T., Krause, K., Liday, S. G., & Vesterlund, L. (2002). Risk attitudes of children and adults: Choices over small and large probability gains and losses. Experimental Economics, 5(1), 53–84. Harbaugh, W. T., Krause, K., Liday, S. G., & Vesterlund, L. (2003). Trust in children. In E. Ostrom & J. Walker (Eds.), Trust, reciprocity and gains from association: Interdisciplinary lessons from experimental research (pp. 143–154). New York: Russell Sage Foundation. Harbaugh, W. T., Krause, K., & Vesterlund, L. (2000). Are adults better behaved than children? Age, experience, and the endowment effect. Economics Letters, 70(2), 175–181. Hemesath, M., & Pomponio, X. (1998). Cooperation and culture: Students from China and the United States in a prisonerÕs dilemma. Cross-Cultural Research, 32(2), 171–184. Henrich, J. (2000). Does culture matter in economic behavior? Ultimatum game bargaining among the Machiguenga. American Economic Review, 90(4), 973–979. Henrich, J., Boyd, R., Bowles, S., Camerer, C., Gintis, H., McElreath, R., et al. (2001). In search of homo economicus: Experiments in 15 small-scale societies. American Economic Review, 91(2), 73–79. Ho, D. Y. F. (1996). Filial piety and its psychological consequences. In M. H. Bond (Ed.), The handbook of Chinese psychology (pp. 155–165). Oxford, UK: Oxford University Press. Hoffman, E., McCabe, K., & Smith, V. L. (1982). On expectations and monetary stakes in ultimatum games. Intertnational Journal of Game Theory, 25, 289–301. Kachelmeier, S. J., & Shehata, M. (1992). Culture and competition: A laboratory market competition between China and the West. Journal of Economic Behavior and Organization, 19(2), 145–168. Kohlberg, L. (1976). Moral stages and moralization: The cognitive developmental approach. In T. Lickona (Ed.), Moral Development and Behaviour (pp. 31–53). New York: Holt, Rinehart & Winston. Kohlberg, L. (1986). A current statement on some theoretical issues. In S. Modgil & C. Modgil (Eds.), Lawrence Kohlberg. Philadelphia, PA: Falmer. Kohlberg, L., & Gilligan, C. (1971). The adolescent as philosopher. Daedalaus, 100, 1051–1086. Krause, K., & Harbaugh, W. T. (1999). Economic experiments that you can perform at home on your children, working paper. Krause, K., & Harbaugh, W. T. (2000). ChildrenÕs contributions in public good experiments: The development of altruistic and free-riding behaviors. Economic Inquiry, 38(10), 95–109. Lee, B. (1974). A cognitive developmental approach to filiality development. MasterÕs thesis, University of Chicago, Chicago, IL. Murnighan, J. K., & Saxon, M. S. (1998). Ultimatum bargaining by children and adults. Journal of Economic Psychology, 19, 415–445. Nisbett, R. (2003). The geography of thought: How Asians and Westerners think differently . . . and why. London: The Free Press. Persico, N. G., Postlewaite, A., & Silverman, D. (2001). The effect of adolescent experience of labor market outcomes: The case of height. pIER Working Paper no. 01-050.
116
R. Hoffmann, J.-Y. Tee / Journal of Economic Psychology 27 (2006) 98–116
Piaget, J. (1932). The moral judgement of the child. New York, NY: The Free Press. Roland-Le´vy, C. (1990). Economic socialization: Basis of international comparisons. Journal of Economical Psychology, 11(3), 469–482. Roth, A. E. (1995). Bargaining experiments. In J. H. Kagel & A. E. Roth (Eds.), The handbook of experimental economics (pp. 253–348). Princeton, NJ: Princeton University Press. Roth, A. E., Prasnikar, M., Okuno-Fujiwara, V., & Zamir, S. (1991). Bargaining and market behavior in Jerusalem, Ljubljana, Pittsburgh and Tokyo: An experimental study. American Economic Review, 81, 1068–1095. Schweitzer, M., & Solnick, S. (1999). The influence of physical attractiveness and gender on ultimatum game decisions. Organizational Behavior and Human Decision Processes, 79, 199–215. Siegel, S., & Castellan, N. J. (1988). Nonparametric statistics for the behavioral sciences. New York: McGraw-Hill. Webley, P. (2004). ChildrenÕs understanding of society. In M. Barrett & E. Buchanan-Barrow (Eds.), ChildrenÕs understanding of economics (pp. 43–67). Hove, UK: Psychology Press. Webley, P., & Lea, S. E. G. (1993). Towards a more realistic psychology of economic socialization. Journal of Economic Psychology, 14(3), 461–472. Windmiller, M. (1980). Introduction. In M. Windmiller, N. Lambert, & E. Turiel (Eds.), Moral development and socialization (pp. 1–33). Boston, MA: Allyn and Bacon. Wu, D. Y. H. (1996). Chinese childhood socialization. In M. H. Bond (Ed.), The handbook of Chinese psychology (pp. 143–154). Oxford, UK: Oxford University Press. Yau, J., & Smetana, J. G. (2003). Conceptions of moral, social conventional, and personal events among Chinese preschoolers in Hong Kong. Child Development, 74(3), 647–658.