TEORIJA IN PRAKSA let. 59, 2/2022 487 UDK 392.73(492+497.4) Žiga VELKAVRH, Aljaž ULE* INDICATORS OF HUMAN SOCIALITY IN SLOVENIA AND THE NETHERLANDS: EVIDENCE FROM EXPERIMENTS WITH STUDENTS** Abstract. In the article we report a series of experiments with volunteers designed to detect differences in behav- ioural characteristics among Slovenian, Dutch and international students. Using eight standard tasks from experimental economics, we investigate the differences using experimental measures of solidarity, trust, coop- eration, positive and negative reciprocity, competition, honesty, and risk attitudes. No significant cohort effects in any of the eight decisions are found when we com- pare the Slovenian and international cohorts. Still, when comparing the Dutch and Slovenian cohorts, Dutch students are found to exhibit lower levels of solidarity, generosity and honesty. This points to differences in soci- ality between institutionally similar yet ideologically dis- tant countries like Slovenia and the Netherlands. Keywords: cross-national study, experimental econom- ics, game theory, sociality Introduction Experiments are the original way of doing research in the natural sci- ences. In contrast, since the mid-20th century in the social sciences experi- ments have only been the key method of research in psychology. The Asch * Žiga Velkavrh, MSc, Assistant and Young Researcher, Faculty of Mathematics, Natural Sciences and Information Technologies and Andrej Marušič Institute, University of Primorska, Slovenia; Aljaž Ule, PhD, Associate Professor, Faculty of Mathematics, Natural Sciences and Information Technologies, University of Primorska, Slovenia and Faculty of Eco nomics and Business, Universiteit van Amsterdam, The Netherlands. ** This research was financially supported by the Slovenian Research Agency (ARRS) through research project J7-6828 and by the University of Amsterdam through the Speerpunt Behavioural Economics. Ule’s research is supported by the Slovenian Research Agency (research program P1-0285). The research of Velkavrh is supported by the Slovenian Research Agency (research program P1-0285, research projects J1-9110, J1-9186 and J1-2451 and Young Researchers Grant). The content of this article is part of a broader analysis presented in a chapter of Velkavrh's doctoral dissertation entitled "Behavioral traits across time and countries: essays on indirect reciprocity, deception and nationality", that will be defended at the Faculty of Mathematics, Natural Sciences and Information Technologies at the University of Primorska. DOI: 10.51936/tip.59.2.487-508 Žiga VELKAVRH, Aljaž ULE TEORIJA IN PRAKSA let. 59, 2/2022 488 conformity experiments and Milgram experiments about hierarchical sub- mission had a strong impact on both the expert and general public in the 1960s (M. Ule, 2004). In economics, the experiments initially focussed on testing the standard assumptions about the efficiency of free markets (V. L. Smith, 1962), but eventually covered the general area of human decision- making, overlapping with fields such as psychology, social psychology, and evolutionary biology. Experiments (the laboratory in particular) are a much more marginal method in sociology and anthropology. This is despite them offering an important tool for both basic and auxiliary research that is able to yield impor- tant, systematic, controlled and highly replicable insights into social human behaviour. Instead, the method of choice for quantitative research in sociol- ogy is the vignette study, which surveys human attitudes in hypothetical situa- tions. In contrast, experiments can position subjects in real social or economic situations where each decision holds real social or economic consequences for all involved. When these situations mimic real-life conflicts and trade-offs, they raise real moral dilemmas, which offers an insight into non-hypotheti- cal values and actual human decision processes. A typical example of such an approach is experimental economics using game theory to design sim- ple versions of actual economic dilemmas, and offering performance-based monetary incentives for the realism of decisions. A design of this nature can increase both the internal and external validities of laboratory experiments for the social sciences (Hertwig and Ortmann, 2001; Schram, 2005). The main advantage of the experimental method is a clear, efficient, transparent, replicable and reliable tool for detecting causal relationships (e.g., Ule and Živoder, 2018). In the social sciences, it facilitates exact analysis of phenomena up to a medium scope such as interpersonal relations, con- formism, biased judgement, and social exchange. Controlled experimenta- tion has in recent decades thus led to substantial conceptual revolutions in several social disciplines. Economists have developed theories of pro-social motives that are not driven by individual market success, political scientists have developed and then criticised the theory of rational electoral choice, while communication scientists have engaged in a systematic exploration of the influence process (Webster and Sell, 2014). One particular advantage offered by experiments is replicability of the decision environment across different locations like cities, countries and cohorts. This facilitates cross- cultural research that is low on noise and confounds. In this article, we report the results of one such experimental study that replicated the same experiment in two cities in Slovenia and one in the Netherlands. Our experiment was built around a fixed sequence of eight economic tasks that induced different moral or economic phenom- ena: solidarity, cooperation, dominance, positive and negative reciprocity, Žiga VELKAVRH, Aljaž ULE TEORIJA IN PRAKSA let. 59, 2/2022 489 trust, honesty, and risk aversion. In all three locations, we recruited local and international students in order to compare the behavioural characteris- tics of Slovenian students with those of international students and students from a distant European society: the Netherlands. Theoretical concepts and contexts Sociality is “fundamentally dynamic and dialectical, subject to extension and contraction, and having both positive and negative valences, it is not only a resource but also a burden” (Sillander, 2021: 1–2). The general social sciences aim to describe how the most common behavioural characteristics of human sociality vary within and across different contexts, which offers an insight into their drivers and evolution across history and geography. Heterogeneity in sociality may, for instance, help explain differences in the dominant responses by people to social crises and conflicts. Culture is one source of contextual variance, and it is important to understand the extent to which it impacts the heterogeneity in sociality. Societies are often compared on dimensions like trust, cooperation, hon- esty, fairness, reciprocity, and risk attitudes (e.g., Boer and Fischer, 2013; Thöni, 2019). These are among the key characteristics of human sociality and commonly viewed as positive. None is simple or one-dimensional, and we can find a wealth of related concepts in sociology and psychology. Trust, for instance, has emotive, behavioural and communication elements, even if it cannot be commanded, but only offered and accepted. Trust is not simply a rational act; it always contains an element of faith, but not blind faith. Trust therefore presupposes risk and may lead to disappoint- ment and regret if expectations are unmet (Luhmann, 1988). Similarly, the display of solidarity or reciprocity in relationships spans positive and nega- tive orientations. Solidarity may require social exclusion, while positive reci- procity often emerges in relationships that understand negative reciprocity. Demonstrations of solidarity, honesty and reciprocity in relationships also depend on the expressed strength of prosocial orientations and the wider social context (Smith and Sorrell, 2014). Moreover, it is not merely the behaviour that varies situationally for the same person; the core motivations to act also vary situationally within the same individual (Ross and Nisbett, 1991). For example, while the reciprocity of prosocial individuals does not strongly depend on the impressions of the other (honesty, intelligence and unintelligence, in particular), that of the pro- self individuals is chiefly promoted by impressions of honesty/sincerity and less by intelligence/unintelligence (Van Lange and Semin-Goossens, 1998). In this article, laboratory experiments are used to explore whether the influence of dominant cultural patterns and national traditions can be Žiga VELKAVRH, Aljaž ULE TEORIJA IN PRAKSA let. 59, 2/2022 490 detected over the variance in social behaviour from personal aspects. We control for social and institutional factors by creating similar experimental incentives and environments in all geographical locations under study, and by controlling for our subjects’ demographic characteristics. This would be difficult to control in a conventional public opinion survey. Survey responses are also often subject to prevailing stereotypes and prejudices in given national or social settings. One’s personal sense about the basic characteristics of sociality may be especially driven by prejudices and ste- reotypes that affect the social categorisation of individuals or groups, such as those describing what is typical or atypical for the social functioning of people from one’s own groups or from some foreign, especially marginal group (M. Ule, 2004). While every nation possesses stereotypes about how it compares to others, they can be misleading (Scheuch, 1993). There are hence few cross-national comparative studies of sociality, for example the regional analysis of 30 European countries by Koster (2013) and a compre- hensive cross-cultural study by Hofstede et al. (2010) that compares over 75 countries and regions on several dimensions, including individualism/ collectivism, power distance, and uncertainty avoidance. Since sociality is most clearly expressed in practice in people’s actual behaviour in various social situations, we decided to conduct a comparative incentivised experimental study among Slovenian, Dutch and international students with respect to eight indicators of sociality: solidarity, trust, coop- eration, positive and negative reciprocity, competition, honesty, and risk attitudes. These indicators were measured with eight standard tasks from experimental economics. In so doing, we are aware that “individual and cultural differences in game behaviors can reflect both the ways in which people perceive game situations and their general social preferences” (Yamagishi et al., 2013: 260). Cross-cultural experimental comparative research is more commonly employed for individual tasks, although some studies have a larger scope. For example, Henrich et al. (2005) implemented three experimental eco- nomic tasks in 15 small-scale societies around the world, testing assump- tions about economic rationality in the social behaviour of people from dif- ferent social and cultural backgrounds. The key results of this research were: first, there is no society in which experimental behavior is fully consist- ent with the selfishness axiom; second, there is much more variation between groups than previously observed, although the range and patterns in the behavior indicate that there are certain constraints on the plasticity of human sociality; third, differences between societies in market integration and the local importance of cooperation explain a substantial portion of the behavioral variation between groups; fourth, Žiga VELKAVRH, Aljaž ULE TEORIJA IN PRAKSA let. 59, 2/2022 491 individual-level economic and demographic variables do not consist- ently explain behavior within or across groups; and fifth, experimen- tal play often reflects patterns of interaction found in everyday life. (Henrich et al., 2005: 797–798) The scope of our study is broader as we cover eight classic economic tasks, yet it is narrower in geographic comparison given that our subjects come overwhelmingly from various European countries, primarily Slovenia and the Netherlands. Our working hypothesis is therefore that the sociality patterns in our samples are mostly similar, with the variance driven more by demographic characteristics than nationality. Research methods To gather the data, we organised a series of experiments with volun- teers recruited from among students at various faculties in Koper and Ljubljana in Slovenia, and in Amsterdam in the Netherlands. In total, 128 subjects participated in the experiment, each once. Our sample contains 49 Slovenian students who study in Slovenia, 23 Dutch students who study in the Netherlands, and 56 international students who study in Slovenia or in the Netherlands but are neither Slovenian nor Dutch. All the Slovenian and Dutch students in our sample study in their home country. The experi- mental sessions were conducted between May 2017 and February 2018. The experiment was conducted through computers, using the Z-tree experimen- tal software (Fischbacher, 2007). Statistical analysis was performed in the statistical software Program R (R Core Team, 2019) using stats and vgam packages (Yee, 2010). Each subject participated in an identical sequence of eight experimen- tal tasks at a laboratory dedicated to economic experiments at their uni- versity. After the experiment, the subjects completed a brief questionnaire that included demographic and background information. Communication between subjects was not allowed during the experiment. Anonymity was assured throughout the experiment by placing subjects randomly in private cubicles and making it obvious that the experimenters could not connect their decisions to their names. In each task, the subjects could obtain points with their decisions. At the end of the experiment, we randomly selected one task and paid each sub- ject 10 eurocents for every point they had obtained in the selected task. In this way, the decisions were not hypothetical but held real consequences for the subjects’ earnings. Performance-based earnings are the key element of economic experiments, intended to increase the realism of every deci- sion the subjects make (e.g., Hertwig and Ortmann, 2001). The subjects had Žiga VELKAVRH, Aljaž ULE TEORIJA IN PRAKSA let. 59, 2/2022 492 the payment procedure explained to them before the experiment yet did not know which task would be paid, inducing them to consider each of the eight tasks as if it were one that would determine the earnings for all sub- jects. A participation fee and earnings from a disconnected post-experimen- tal task were added on top of the money earned from the decisions and the total paid to the subject anonymously and in private before they left the laboratory. The average earnings were EUR 12 for an average duration of 50 minutes, a substantial premium over the average student wage. No other benefits were accrued from participation, except for the money earned from fees and decisions and this was advertised during the recruitment. Each session began with instructions about laboratory conduct and then the subjects participated in the eight experimental tasks as described below. For each task, they first received the description written in a neutral lan- guage to avoid framing, and then everyone simultaneously submitted their decision. Subjects did not learn about the decisions of the other subjects until the end of the experiment to avoid any spill-overs between the tasks and to assure we could analyse each task separately. All interactive tasks were therefore translated into simultaneous games. The complete instruc- tions and raw data are available upon request from the authors. The experiment comprised of six interactive tasks (two 3-player games, four 2-player games), and two individual tasks. Everyone completed the tasks in the sequence presented below, starting with the Solidarity game and finishing with the Risk task. In the interactive tasks, the subjects were randomly grouped in pairs or triplets. Identities of group members were not revealed to protect anonymity. We derive a simple prediction for each task using standard economic theory. We do not consider that subjects may randomise (use mixed strategies). Task 1: Solidarity game The “Solidarity game” investigates prosocial attitudes of fortunate indi- viduals with regard to less fortunate others. It was developed by Selten and Ockenfels (1998) to measure the “willingness to help people in need who are similar to oneself but victims of outside influences such as unforeseen illness, natural catastrophes, etc.” (Selten and Ockenfels, 1998: 518). In this game, donations are one-sided and there is no mechanism for explicit reciprocity. The specific setup is as follows. Each subject in a group of 3 will play a lottery that gives either 60 points (“rich”) with a 2/3 probability, or 4 points (“poor”) with a 1/3 probability. Before a subject is told the outcome of any- one’s lottery, they make two decisions that only become relevant if they later receive 60 points in their private lottery. First, they decide how many of their 60 points they would donate to a poor subject if there were just one Žiga VELKAVRH, Aljaž ULE TEORIJA IN PRAKSA let. 59, 2/2022 493 in their group. Second, they decide how many points they would donate if both of the other subjects in their group were poor. The final payoffs are as follows. If all three subjects are rich (poor), each gets 60 (4) points. If just two subjects are rich and donate x 1 ϵ {0, …, 60} and y 1 ϵ {0, …, 60} to the third poor subject, the former end with 60-x 1 and 60-y 1 points whereas the third ends with 4+x 1 +y 1 points. If only one subject is rich and donates x 2 ϵ {0, …, 30} to each other subject, she ends with 60-2x 2 points, and the other two with 4+x 2 points each. A rich donor does not benefit financially from helping the poor. The standard prediction for the game is therefore that no donations will be made. However, a donor might donate some points if they dislike large inequalities (Fehr and Schmidt, 1999). Indeed, evidence from previous experiments sug- gests that many subjects commit to positive donations, leading to substantial average solidarity (Selten and Ockenfels, 1998). Solidarity can be affected by culture, however. For example, Ockenfels and Weimann (1999) found that subjects in the western part of Germany donate significantly more often (79% vs 52%) and higher average amounts (25%–31% vs 16%–20% of points) than those in the eastern part of Germany. As shown by Brosig-Koch et al. (2011), these differences between West and East Germans were still visible in 2009 even after controlling for other variables such as education and gender. Task 2: Public goods game The “Public goods game” models a problem of cooperation where the selfish interests of individuals conflict with the collective interest of the group as a whole (e.g., Andreoni, 1988). It exposes the free-riding problem that occurs when selfish individuals use and enjoy the benefits of publicly provided work, like clean environment and public facilities, but do not pro- vide any work themselves. Widespread free-riding may destroy public good provision by the others (e.g., Marwell and Ames, 1979). Collective problems investigated with this game include teamwork, public space organisation, donations to charities, and global pollution. In our setup, the subjects are placed in groups of 3. Each must allocate 9 tokens between two projects: private and common. Any token allocated by any subject to the common project yields 2 points to each subject. Each token allocated by a subject to their private project yields 4 points to the sub- ject and no points to the other two. A token in the common project is less profitable for the contributor, but more profitable for the group. Subjects can earn 54 points each if they invest all tokens in the common project. Yet, every subject can earn more by allocating their own tokens to their private project. They thereby earn points from both the common and private pro- jects. Still, if everyone free-rides like this, there is no public good and the Žiga VELKAVRH, Aljaž ULE TEORIJA IN PRAKSA let. 59, 2/2022 494 subjects earn just 36 points each. If three group members contribute (x 1 , x 2 , x 3 ) to the common project, subject i earns π i = 2 (x 1 + x 2 + x 3 ) + 4 (9 - x i ) points. The standard theory predicts no contributions to the common project, which is interpreted as an example of a free market failing to lead to effi- cient economic outcomes. In contrast, experimental evidence shows that many subjects contribute considerable amounts to the common project (40%–60% on average; see, e.g., Ledyard, 1995; Chaudhuri, 2011). Average contributions are similar in countries with highly integrated market econo- mies (Brandts et al., 2004), yet vary from 22% to 65% in small-scale societies (Henrich et al., 2005). Task 3: Trust game The “Trust game” is a simultaneous variant of the dynamic investment game that is used to measure both trust and trustworthiness among experi- mental subjects (Berg et al., 1995; Bohnet and Zeckhauser, 2004). The idea behind the model is that trust increases social welfare but may be prone to abuse and is therefore risky. The standard example is of two traders who can avoid lawyer fees if they trade without any contracts. One sends money to the other and the other should send goods back to the first after receiv- ing the money. This exchange can be enforced with a contract. However, if one trusts the other to return the goods, the two can avoid the contract- associated costs. Related dilemmas emerge in many daily interactions and trust is an essential element of functional societies. The trust game measures the fundamental level of trust in a society: towards anonymous strangers. Our game involves a pair of subjects, a trustor and a proxy (trustee). Each has two available actions. The trustor (she) is given 40 points that she can either hold or transfer. The proxy (he) gets 0 points if the trustor holds. Yet, if the trustor transfers, the proxy takes the trustor’s 40 points and turns them into 120 points that she can either keep or share equally with the trustor. In our task, the proxy decides whether to share without yet knowing the deci- sion of the trustor. The final payoffs are shown in the table below. Each cell corresponds to a pair of actions and shows the resulting payoffs for the trus- tor (first number) and the proxy (second number). Table 1: TRUST GAME (proxy) keep share (trustor) hold 40 , 0 40 , 0 transfer 0 , 120 60 , 60 Source: own analysis. Žiga VELKAVRH, Aljaž ULE TEORIJA IN PRAKSA let. 59, 2/2022 495 The standard prediction is that the proxy will keep the points, to which the rational response of the trustor is to hold her points. This is obviously inefficient because both can earn 60 points if they transfer and share their points. The trustor would transfer her points only if she trusts that the proxy will share. A transfer therefore indicates trust and sharing indicates trust- worthiness. The common experimental finding from the sequential version of the trust game is that people generally show a substantial amount of trust, even to complete strangers, and that trust is often rewarded (Berg et al., 1995). This indicates that trust and reciprocity are both important economic primi- tives. There is some experimental evidence that trust varies across similar countries. For example, Willinger et al. (2003) found that in Germany sub- jects trusted more than in France, despite trustworthiness being similar in the two countries. Survey questionnaires, for comparison, may suggest greater variation in trust than what is observed in incentivised experiments. For instance, Holm and Danielson (2005) found similar levels of experimen- tal trust between subjects in Tanzania and Sweden, despite significantly dif- ferent responses to the survey’s trust questions. Survey results concerning trust may measure social stereotypes or private trustworthiness rather than actual trust and depend on how respondents understand and interpret the questions as well as their subjective reference point (Glaeser et al., 2000; Sapienza et al., 2013; Banerjee, 2018). The fact that Eurostat (2013) and the World Values Survey (Inglehart et al., 2018) both found that respondents in Slovenia had less trust than those in the Netherlands makes it interest- ing to gather evidence about their actual trust in incentivised experimental exchanges. Task 4: Ultimatum game The “ultimatum game” is a simple model of bargaining (Güth et al., 1982). A proposer suggests a division of 100 points, while the responder then either accepts or rejects this division. In case of rejection, the two earn nothing. This game is used to investigate the prosocial attitudes of propos- ers and the negative reciprocity of responders. By rejecting a positive offer, the responder sacrifices a positive earning to indicate displeasure and pun- ish the proposer. At the same time, a high offer indicates that the proposer understands the possibility of such negative reciprocity among the people in their community. Our setup considers the simultaneous version of the originally sequen- tial decision game (like, e.g., in Harrison and McCabe, 1996). In our pairs, the proposer (she) offers a number of points P ϵ {0, …, 100} to the responder (he) who at the same time indicates the minimum number of points X ϵ {0, Žiga VELKAVRH, Aljaž ULE TEORIJA IN PRAKSA let. 59, 2/2022 496 …, 101} he is willing to accept. Here X = 0 means “accept any proposal”, while X = 101 means “reject every proposal”. Offer P is then compared to the mini- mum X. If P ≥ X, the offer is accepted, the proposer earns 100-P points, and the responder earns P points. If P < X, the offer is rejected and both subjects earn 0 points. The standard prediction for the dynamic game is that the responder will accept any positive offer and the proposer will offer either 0 or 1 point. Although the theory is less narrow for our simultaneous version of the game, the most plausible theoretic predictions are like those above. Choosing X ≤ 1 means the responder will earn points whenever the proposer makes a posi- tive offer (as they mostly do). Choosing X > 1, on the other hand, risks los- ing positive earnings from low offers. A rational responder should therefore choose a higher minimum X > 1 only when she is willing to incur a cost up to X to punish the proposer for an unfair offer. In ultimatum game experiments across industrialized societies, the aver- age offers are typically between 30% and 45% of the total, which are usually accepted. Offers below 20% are rare and often rejected (Camerer, 2003). Still, rejection patterns and the notion of a fair division might be country- specific (Roth et al., 1991 compared ultimatum bargaining in Israel, Japan, Slovenia, and the United States; see also Oosterbeek et al., 2004). Henrich et al. (2005) found larger differences between small-scale non-industrialised societies, with average offers ranging between 26% and 58% and a related variance in rejection patterns. Task 5: Chicken game This simple game measures subjects’ tendency to compromise and adopt a submissive role in society, which promotes hierarchical ranking. Subjects are paired and each chooses either option A (dominant) or option B (com- promise). If one chooses A and the other chooses B, they earn 70 and 30 points, respectively. If both choose to dominate with A, they both earn 0 points. If both choose to compromise with B, they each earn 40 points. The table below shows how again the payoffs correspond to the chosen options. Table 2: CHICKEN GAME A B A 0 , 0 70 , 30 B 30 , 70 40 , 40 Source: own analysis. It is best to choose A when the other chooses B, and to choose B when the other chooses A. The standard prediction is therefore that, despite facing a Žiga VELKAVRH, Aljaž ULE TEORIJA IN PRAKSA let. 59, 2/2022 497 symmetric social situation, the subjects will make asymmetric choices, with the dominant subject earning much more than the compromising subject. While subjects may agree that specialisation is efficient, they would disagree on who should profit from domination. In the absence of communication, like in our experiment, choosing A suggests a willingness to compete for a leading social position. This game was recently experimentally studied in the Netherlands by de Heus et al. (2010) who found that compromise B is chosen by up to 87.5% of the subjects, but cross-country comparisons are scarce. Carment (1974), for example, found that in a repeated similar experiment Indian males ini- tially compromise slightly more than Canadian males but the latter compro- mise more in the end. Task 6: Reward game In this task, we investigate positive reciprocity. Our “reward game” mod- els an exchange of favours between two subjects in a pair, the sender (she) of a gift and its recipient (he). The sender’s wealth is at risk of partial destruc- tion. She can gift some of their wealth to the recipient who holds the power to prevent the destruction of the sender’s remaining wealth. The recipient must pay to protect the sender but may do this as gratitude for the sender’s gift. A sender may then send a positive gift if she expects such positive reci- procity from the recipient. This “gift-exchange” was proposed by Akerlof (1982) as a model to explain why wages are often above the bare minimum. Well-paid workers make a bigger effort which, through positive reciprocity, benefits workers and employees alike. Low wages may on the other hand be perceived as unfair and lead to low productivity and high unemploy- ment (Akerlof and Yellen, 1990). In our game, we pair the subjects and then each sender is given 90 points and their recipient is given 10 points. The sender chooses a number of points G ϵ {0, …, 90} to give to the recipient, while the recipient chooses the minimum gift X ϵ {0, …, 91} for which he will protect the sender’s (remaining) points. Here, X = 0 means “always protect the sender”, and X = 91 means “never pro- tect the sender”. Sender’s gift G is then compared to the recipient’s demand X. When G ≥ X, the recipient pays 10 points and earns gift G, while the sender earns 90-G points. If G