Sampling
Our target respondents were those who had the right to vote for the 2012 president in Taiwan. The sampling frame of this study is 244 courses of 37 departments and core programs in 9 colleges which mainly offered undergraduate students of the third (junior) or fourth (senior) grades to take in National Chengchi University (NCCU) in the first semester, 2012. We used stratified random sampling method to choose 63 courses as our sample courses. After obtaining the agreement of the teachers, the interviewers went to the classes to conduct the survey. A total of 1550 respondents in 49 courses filled out the questionnaires in the end.
22
From 1550 subjects, we elect subjects voting for Ma Ying-jeou and Tsai Ing-wen, and contact them by phone and e-mail. There are 208 students willing to participate in the trust game experiments. Due to the occurrence of experimental errors, we discarded 16 failure examples. The remaining 192 subjects include 121 subjects voting for Ma Ying-jeou and 71subjects voting for Tsai Ing-wen. We only select data of the first five rounds for the analysis each stage1.
Research Design
In this paper, we use z-Tree which was created by Fischbacher ,Department of Economics, University of Zurich, to perform our trust game experiments. We designed the ‘Trust game’ which is a two- players game in which player A(Trustor) is given 10 points. First, player A(Trustor) has the option to give part of that to player B(Trustee) and keeps the rest for herself. Whatever points player A(Trustor) chooses to give, the game
organizer trebles those points and gives them to player B. So if player A keeps 4 points to herself and gives 6 points to player B, then player B actually gets 18 points. Second, player B has the option of giving any of what he receives back to A. So, for example, if A gives 10 points and keeps nothing for herself and B gives back 25 points to A, then in the end, player A gets 25 points and B gets 5 points.
This experiment includes three stages, and each stage at least run 5 rounds. From the sixth round, the probability of each round proceeding is 50%. Random stopping rule is designed to mitigate the influences of the ending effect. Benchmark return is calculated for each stage of the last five rounds, and three stages add up.
One point is equate to two NTD. This experiment includes three stages. In stage 1, the computer randomly assigns trustors and trustees and does not show subjects’ political identity to 2012 presidential candidates (Ma Ying-jeou or Tsai Ing-wen). In stage 2, the computer re-assigns trustors and trustees randomly but displays their political identity logo. In stage 3, the subjects can choose their preferred objects of the same or different
political identity logo by themselves. Each stage represents different experimental conditions
We assume people in each round of three different stages of experiments have different degree of trustfulness and trustworthiness. We also assume the political identity logo will influence the degree of trustfulness and trustworthiness.
Results
Descriptive Analysis
Table 1 shows the behavior of trustfulness and trustworthiness for all participants in fifteen rounds of three
1 From the sixth round, the probability of each round proceeding is 50%. Random stopping rule is designed to mitigate the influences of the ending effect.
23
stages. Social psychologists think voting is a group behavior, and voters’ choices are affected by social groups surrounding them.
Table 1. Summary Results by Experimental Condition and Participant Political Identity Condition
condition Stage 1 Stage 2 Stage 3
Participant political identity Ma Tsai Ma Tsai Ma Tsai
Mean(SD) Mean(SD) Mean(SD) Mean(SD) Mean(SD) Mean(SD) Total Profit(point) 74.22(20.060) 75.62(28.695) 90.84(22.162) 87.06(23.670) 86.93(22.129) 79.22(20.626) Final Profit(point) 64.98(14.182) 64.70(20.601) 66.16(12.012) 64.85(14.186) 69.53(7.882) 66.62(12.784) Average entrustment to trustee(Ma) , per
Table 2 reports the regression coefficients explaining the transaction amount which trustor entrusts trustee in each transaction through the random-effects regression models. We used subject’s political identity,
homophily of political identity between trustor and trustee, and stage of experiment to test all the hypotheses mentioned before. Political identity was coded 1 voting for Ma and 0 voting for Tsai in 2012 Taiwan
presidential election. Homophily was coded 1 if the transaction partners were both voting the same president candidate and 0 otherwise. Three stages of experiments were coded 1, 2, 3 respectively.
Model 4 does not show the significant effect of different stages on trustfulness (Hypothesis 1a and 1b).
Trustiness of participants are no significant difference between stage 2 and stage 1. Trustiness of participants are also no significant difference between stage 3 and stage 1. However, in Model 5, if we do not control the variable that people in the first round give all ten points to their partner, stage is significant. Trustiness of participants are significant difference between stage 2 and stage 1 ,and between stage 3 and stage 1. Especially, in stage 3, trustiness of participants are more significant. Because participants can choose their preferred objects of the same or different political identity logo by themselves in stage 3, we can find stronger identity effect.
Through the cross tabulation(stage * frist_10_points ), there are 32.5％ people who in the first round give all ten points to their partner in stage 1. But, there are 52.1％ in stage 2 and 62.5％ in stage 3. There are most of the proportion of people who in the first round give all ten points to their partner in stage 3. Therefore, when we control the variable people who in the first round give all ten points to their partner, we can not see the
differences between stage 2 and stage 1, and between stage 3 and stage 1.
24
Model 2 does not show the significant effect of homophily principle on trustfulness (Hypothesis 2a). The major reason is that the average entrustment of political identity group for Tsai-Tsai (Tsai-Tsai is that both trustor and trustee vote for Tsai Ying-wen) is much lower than that of political identity group for Ma-Ma in stage 2 to stage 3 of Table 1. Evidently, homophily effect varies by Ma-Ma and Tsai-Tsai groups. After the homophily variable is classified into four combinations of political identity in Model 3, the significantly strong effect of Ma-Ma than those of other groups. In Model 3 (Hypothesis 2b), Tsai-Tsai and Tsai-Ma are significant less than Ma-Ma. It means homophily hypothesis only happens in people voting for Ma but not happens in people voting for Tsai. Participants voting for Ma trust members of a salient similar political identity more than members of dissimilar.
The results strongly support Hypothesis 3a. In Model 1 & Model 2, subject’s political identity is significant. It means people voting for Ma Ying-jeou are more trustful than those voting for Tsai Ing-wen.
According to structural theory of trust transaction, people voting for Ma Ying-jeou emphasize opportunity cost in each transaction. The opportunity cost of trust transaction is foregoing a potentially lucrative transaction outside of parochial exchange relations with familiar partners. The willingness to trust strangers minimizes this opportunity cost. Therefore, it means people voting for Ma Ying-jeou are more willing to trust strangers for missing better exchange opportunities.
In contrast, people voting for Tsai Ing-wen emphasize transaction cost of possible exploitation.
Commitment are alternative solutions to the problem of uncertainty in social and economic exchanges.
Commitment reduces this transaction cost by limiting transactions to durable relations in which the risk of being cheated is relatively low from previous literature. Therefore, it means people voting Tsai Ing-wen are not easy to trust strangers because of the risk of being cheated is high when they play with their partners in
temporary relationship. In Model 7, we add interaction effects of political identity and mutuality. A negative coefficient for this interaction effect indicates that the mutuality effect on trustfulness is stronger for people voting for Tsai than people voting for Ma. People voting for Tsai seem to pay the trustfulness according to commitment principle. Even though they are more difficult to trust others in the beginning, they tend to return more once they gain more from others.
In Model 6, the results provide support for Hypothesis 4. The effects of mutuality on trustfulness are significant. It means people with higher mutuality are more likely to be trustful than those with lower mutuality.
Therefore, mutual agreement between the participants must exist, and mutual agreement can facilitate this “act of faith” is generally regarded as “trust.”
25
Table 2. Random-effects regression analysis Predicting trustfulness
Mode l Mode 2 Model3 Model 4 Model 5 Model 6 Model 7
Variables Coef.(S.E.) Coef.(S.E.) Coef.(S.E.) Coef.(S.E.) Coef.(S.E.) Coef.(S.E.) Coef.(S.E.)
Main effects
Political Identity 0.74**(0.24) 0.65**(0.25)
Homophily 0.40(0.25)
Ma-Ma Ref.
Ma-Tsai -0.40(0.32) -0.39(0.33) -0.79(0.41) -0.36(0.26) -0.37(0.26)
Tsai-Ma -1.05***(0.32) -1.03***(0.32) -1.27***(0.40) -0.95***(0.25) -0.94***(0.25)
Tsai-Tsai -0.65*(0.32) -0.66*(0.30) -0.86* (0.40) -0.59*(0.25) -0.58*(0.25)
Stage 1 Ref.
Stage 2 0.22(0.29) 0.92** (0.35) 0.19(0.22) 0.20(0.23)
Stage 3 0.28(0.29) 1.23*** (0.35) 0.28(0.23) 0.29(0.23)
Relational commitment
Mutuality 2.40***(0.23) 3.12***(0.38)
Interaction effects
Political
Identity*mutuality
-1.15*(0.47)
Control
Frist ten points 3.17***(0.23) 3.13***(0.24) 3.13***(0.24) 3.06***(0.25) 2.95***(0.20) 2.93***(0.20)
Lower Middle Class Ref.
Middle Class 0.46(0.36) 0.46(0.36) 0.46(0.36) 0.48(0.36) 0.39(0.44) 0.46(0.28) 0.46(0.28)
Upper Middle Class 0.34(0.37) 0.33(0.37) 0.33(0.37) 0.34(0.37) 0.20(0.46) 0.32(0.30) 0.34(0.30)
General trust 0.58*(0.25) 0.56*(0.25) 0.56*(0.25) 0.55*(0.25) 0.25(0.31) 0.54**(0.19) 0.53**(0.20)
Father’s education 0.26(0.24) 0.28(0.24) 0.28(0.24) 0.28(0.24) 0.51(0.30) 0.25(0.19) 0.25(0.19)
26
Gender 0.23(0.26) 0.27(0.26) 0.27(0.26) 0.31(0.26) 1.31*** (0.31) 0.32(0.21) 0.34(0.21)
intercept 4.80(0.42) 4.60(0.43) 5.66(0.44) 5.51(0.47) 6.48(0.57) 4.70(0.38) 5.57(0.37)
R-sq: within 0.00 0.00 0.00 0.00 0.00 0.05 0.05
R-sq: between 0.45 0.45 0.45 0.45 0.16 0.54 0.55
R-sq: overall 0.31 0.32 0.32 0.32 0.11 0.39 0.39
Wald Chi-square
224.65*** 228.55*** 227.73*** 227.86*** 51.45*** 473.65*** 482.09***
Number of observations 1440 1440 1440 1440 1440 1440 1440
Number of groups 288 288 288 288 288 288 288
Observations per groups:
minimum
5 5 5 5 5 5 5
Observations per groups:
average
5.0 5.0 5.0 5.0 5.0 5.0 5.0
Observations per groups:
maximum
5 5 5 5 5 5 5
Notes：We measured mutuality as the probability of trustworthiness between two partners in previous round. Define trustor’s trustfulness as x, trustee’s trustworthiness as y, and
Model 4 does not show the significant effect of different stages on trustworthiness (Hypothesis 1c and 1d).
Trustworthiness of participants are no significant difference between stage 2 and stage 1. Trustworthiness of participants are also no significant difference between stage 3 and stage 1. However, in Model 5, if we do not control the variable that people in the first round give all ten points to their partner, stage is significant.
Trustworthiness of participants are significant difference between stage 2 and stage 1 ,and between stage 3 and stage 1. Therefore, trustworthiness of participants in stage 2 and stage 3 are still different from stage 1. When we control the variable people who in first round giving their partner all ten points, we can not see the
differences between stage 2 and stage 1, stage 3 and stage 1.
Model 2 does not show the significant effect of homophily principle on trustworthiness (Hypothesis 2c).
Just like trustfulness, the average returned entrustment of political identity group for Tsai-Tsai are also much lower than that of political identity group for Ma-Ma in stage 2 to stage 3 of Table 1. After the homophily variable is classified into four combinations of political identity in Model 3, the significantly strong effect of Ma-Ma than those of other groups. In Model 3 (Hypothesis 2d), Tsai-Ma is significant less than Ma-Ma but Tsai-Tsai is not significant. Although we can not provide a support for Hypothesis 2c directly, there are different
27
patterns of trustiness and trustworthiness for people voting for Ma and voting for Tsai when they play with the same identity group.
The results provide support for Hypothesis 3b. In Model 1, subject’s political identity is significant. People voting for Ma Ying-jeou are more trustworthy by others than those voting for Tsai Ing-wen. People voting for Ma are more willing to trust strangers for meeting better exchange opportunity. Therefore, according to mutual principle, people voting for Ma will give others more and will be returned more. They will be more trustworthy by others. In contrast, people voting for Tsai Ing-wen are not easy to trust strangers because of the risk of being cheated is high when they play with their partners in temporary relationships. Therefore, according to mutual principle, people voting for Tsai will give others less and will be returned less. They will not be more
trustworthy by others.
Table 3. Random-effects regression analysis Predicting trustworthiness
Mode l Mode 2 Model3 Model 4 Model 5
Variables Coef.(S.E.) Coef.(S.E.) Coef.(S.E.) Coef.(S.E.) Coef.(S.E.)
Main effects
Political Identity 0.98* (0.48) 0.78(0.49)
Homophily 0.91(0.49)
Ma-Ma Ref.
Ma-Tsai -0.94(0.64) -0.96(0.66) -1.68* (0.78)
Tsai-Ma -1.68**(0.63) -1.68**(0.64) -2.09** (0.76)
Tsai-Tsai -0.80(0.64) -0.80(0.64) -1.16(0.76)
Stage 1 Ref.
Stage 2 0.37(0.57) 1.64** (0.67)
Stage 3 0.30(0.59) 2.02** (0.67)
Control
Frist ten points 5.70*** (0.48) 5.61*** (0.48) 5.60*** (0.48) 5.52*** (0.50)
Lower Middle Class Ref.
Middle Class 0.83(0.71) 0.83(0.71) 0.83(0.71) 0.84(0.71) 0.68(0.85)
Upper Middle Class 0.85(0.74) 0.83(0.74) 0.82(0.74) 0.82(0.74) 0.58(0.89)
General trust *0.98(0.50) 0.95(0.50) 0.95(0.50) 0.93(0.50) 0.40(0.59)
28
Father’s education 0.33(0.48) 0.40(0.47) 0.39(0.48) 0.40(0.48) 0.81(0.57)
Gender 0.20(0.52) 0.28(0.52) 0.28(0.52) 0.33(0.53) 2.14*** (0.60)
intercept 6.15(0.83) 5.72(0.86) 7.42(0.88) 7.22(0.93) 8.97(1.09)
R-sq: within 0.00 0.00 0.00 0.00 0.00
R-sq: between 0.37 038 0.38 0.38 0.12
R-sq: overall 0.26 0.27 0.27 0.27 0.08
Wald Chi-square
172.29*** 177.32*** 176.71*** 176.23*** 37.97***
Number of observations 1440 1440 1440 1440 1440
Number of groups 288 288 288 288 288
Observations per groups: minimum 5 5 5 5 5
Observations per groups: average 5.0 5.0 5.0 5.0 5.0
Observations per groups: maximum 5 5 5 5 5
Discussion
This study tested a set of hypotheses derived from the approaches of homophily ,structural theory of trust and mutuality. The results offer some support for our main claim. First, when mutual agreement exists between each other, people with similar political identity are more likely to establish trustful relations than those with dissimilar political identity. Homophily is one of the most important theories that explain why people establish trustful relations with each other. According to social exchange theory, the mutual exchanges are important mechanisms to format trust relations between trustor and trustee. Trust cultivation requiring reciprocity between two agents is conceived as a necessary behavior to develop trust (Dariusz G. Mikulski, 2013).
Second, we found people voting for Ma Ying-jeou are more trustful than those voting for Tsai Ing-wen, and people voting for Ma Ying-jeou are also more trustworthy than those voting Tsai Ing-wen. We know most people voting for Ma Ying-jeou are bourgeoisie, with higher degree of general trust, would rather take risk trusting strangers in advance. They more focus opportunity cost, and expect to expand their return through trust more strangers. In contrast, most people voting for Tsai Ing-wen are working class, with lower degree of
general trust, would not easily take risk trusting strangers in advance. They more focus transaction cost in order to reduce transaction cost of possible exploitation. These results show in line with Yamagishi's structural theory of trust.
29
Third, we found people with higher mutuality are more trustful than those with lower mutuality. Therefore, once mutual agreement between the participants exists, and mutual agreement can facilitate this “act of faith” is generally regarded as “trust.” The trustfulness according to payoff value of trust between participants can be interpreted as a universal means for participants to mutually share the value of their trustworthy relationships.
References
Basu, K. (2000), Prelude to Political Economy: A Study of the Social and Political Foundations of Economics, Oxford and New York: Oxford University Press.
Basu, K. (2006), ‘Gender and Say: A Model of Household Decision-making with Endogenous Balance of Power,’ Economic Journal, vol. 116.
Brewer, Marilyn B. and Rodrick M. Kramer., 1986, “Choice Behavior in Social Dilemmas: Effects of Social Identity, Group Size, and Decision Framing.” Journal of Personality and Social Psychology 50: 543-540 Brewer, Marilynn B., and Norman Miller. 1996. Intergroup Relations. Pacific Groove, CA: Brooks/Cole
Publishing Co.
Carter, Michael R. & Castillo, Marco, 2003. "An Experimental Approach to Social Capital in South Africa," Staff Paper Series 448, University of Wisconsin, Agricultural and Applied Economics.
Chaim Fershtman & Uri Gneezy, 2001. "Discrimination In A Segmented Society: An Experimental Approach," The Quarterly Journal of Economics, MIT Press, vol. 116(1), pages 351-377, February.
Cook, Karen, 2005, “Networks, Norms, and Trust: The Social Psychology of Social Capital.” Social Psychology Quarterly 68(1):4-14.
Cook, Karen S., Margaret Levi and Russell Hardin (eds.) ,2010, The Complexities and Limits of Trust . New York: Russell Sage Foundation Publications.
Dariusz G. Mikulski(2013). Book Chapter 11: Cooperative Trust Games. Game Theory Relaunched /InTech March 27.
Francois, P. (2002), Social Capital and Economic Development, New York: Routledge.
Granovetter, M. (1985), ‘Economic Action and Social Structure: The Problem of Embeddedness,’ American Journal of Sociology, vol. 91, 481-510.
Gold, T. B. 1987. "The Status Quo is Not Static: Mainland-Taiwan Relations." Asian Survey 27(3): 300-315.
H. Liu, E.-P. Lim, H. W. Lauw, M.-T. Le, A. Sun, J. Srivastava, Y. A. Kim, Predicting trusts among users of online communities: an epinions case study, in: Proceedings of the 9th ACM conference on Electronic commerce, EC ’08, ACM, New York, NY, USA, 2008, pp. 310–319.
doi:http://doi.acm.org/10.1145/1386790.1386838. URL http://doi.acm.org/10.1145/1386790.1386838 Johnson, Noel D. and Alexandra Mislin, 2008, “Cultures of Kindness: A Meta-Analysis of Trust Game
Experiments (Dec.) Avalilable at SSRN: http://ssrn.com/abstract=1315325.
Kuwabara, Ko, Robb Willer, Michael W. Macy, Rie Mashima, Shigeru Teri, Toshio Yamagishi, 2007, “Culture, Identity, and Structure in Social Exchange: A Web-Based Trust Experiment in the United States and Japan.”