ONLINE DISINHIBITION AS A FACILITATOR OF SOCIAL CHANGE: DEVELOPMENT OF A NOVEL EXPERIMENTAL PARADIGM

(1)

Research Master in Economics and Business

Master’s Thesis

ONLINE DISINHIBITION AS A FACILITATOR OF SOCIAL

CHANGE: DEVELOPMENT OF A NOVEL EXPERIMENTAL

PARADIGM

Author: Žan Mlakar

Supervisors: dr. J. W. Bolderdijk & dr. H. Risselada

(2)

ABSTRACT

Minority-led social change movements are a common phenomenon in our society and have received extensive attention in the academic literature. However, it is yet to be understood under what circumstances is minority influence more effective and how the minority-induced social change unravels over time. In order to partially address the given questions, I introduce multiple factors that could affect the efficacy of minority influence. Specifically, I propose that anonymity and physical distance—two main characteristics of the online environment— strengthen minority influence and facilitate social change. Furthermore, to test this hypothesis, a novel experimental paradigm that simulates over-time social change dynamics is designed and tested. The analysis of the pilot study shows that, with some minor improvements, the devised paradigm is an appropriate method for examining the effects of the online environment and other relevant situational factors on over-time social change dynamics.

(4)

1. INTRODUCTION

Social change is a commonly observed social phenomenon as minorities trying to achieve social change—with greater or lesser success—are readily noticeable throughout history. The presently most salient examples in public consciousness include the female voting rights struggle in the early 20th century, civil and LGBT rights movements in the second half of the 20th century, and the ecological movement of the past few decades. Furthermore, even more recent cases can be found—e.g. the currently ongoing #MeToo campaign addressing the prevalence of sexual misconduct in our society and the sudden rise of the alt-right political ideas in the past years.

In line with the prevalence of minority-induced social change phenomena, a substantial amount of literature has already studied the efficacy of minority influence. Moscovici, Lage, and Naffrechoux (1969) and Moscovici (1985) discovered that minorities have the necessary power to, through expressing differing opinions, convince other people. However, contrary to the pressures exerted by a majority, minority influence often evokes changes only on a personal (i.e. not publicly visible) level (Wood, Lundgren, Ouellette, Busceme, & Blackstone, 1994). Such is the case since, in an ordinary social setting, people inhibit their immediate reactions to external stimuli (Hirsh, Galinsky, & Zhong, 2011) and focus rather on fulfilling their inherently human goal of affiliating with others (Cialdini & Goldstein, 2004). This inhibition renders individuals unlikely to publicly disagree with the majority (Asch, 1952, 1956). Ultimately, since majority members are unlikely to express public support for the minority stances even when they agree with said stances, it is also difficult for minorities to successfully induce full social change.

(5)

Indeed, certain contextual factors have already been shown to liberate people through releasing their social inhibition, namely anonymity and physical distance. Specifically, anonymity makes people feel less accountable to others (Festinger, Pepitone, & Newcomb, 1952) and physical distance reduces peoples’ need to manage their impressions (Bond, 1982; Guerin, 1986). Through this, both factors facilitate peoples’ disinhibition and, consequentially, make it easier for people that privately agree with the minority standpoint to also adopt it publically. Simultaneously, these two factors are also the main building blocks that, especially in socio-psychological terms, distinguish the online environment from the offline environment. Therefore, I propose that minorities have a greater potential for successfully achieving social change in the online, relative to the offline, environment.

Multiple theoretical and practical advances arise from examining the effect of the online environment on social change. At the least, it can help us understand one circumstance in which minority influence could be stronger, namely the online environment. Already basic understanding is valuable as it can help us predict how social change unfolds in the modern world where a big part of all communication happens online (Newport, 2014; Pew Research Center, 2011). Beyond that, my study also adds a further explanation of how and why small political interest groups and other minorities have so effectively gained a stronghold in the online environment. One well-documented recent example of this phenomenon is the sudden rise of the alt-right movement in the past years (Hodge & Hallgrimsdottir, 2019).

Next to the mentioned theoretical implications, my study also advances the given topic methodology-wise. Past research in the field misses out on an important aspect of social change—i.e. its time component—and studies minority influence solely in the context of a single-round interaction (see e.g. Wood et al., 1994 for a review and Horcajo, Petty, & Briñol, 2010 for more recent work). Subsequently, no repeated-interaction experimental design that could assist me in testing my hypotheses exists in the past literature. Therefore, I design a novel experimental game that makes it possible to capture over-time social change dynamics across varying conditions.

(6)

paradigm can heighten the external validity of the empirical work tackling the given topic. Second, on a theoretical level, the devised paradigm can help researchers examine the dynamic aspects of social change. For example, it could be used to study how the speed of social change, the size of the tipping point and other similar social change phenomena unravel under varying conditions. Since the observability of such phenomena strongly depends on the over-time component of social change, the given paradigm can further our understanding of social change as a dynamic majority-minority dispute process.

(7)

2. THEORETICAL BACKGROUND

2.1. Online disinhibition decreases conformity

People’s social behaviour differs significantly between the online and the offline environment. When people are in the offline environment—i.e. when others are physically present and people know that they are not anonymous—their behaviour is often inhibited (Suler, 2004). This state of inhibition is represented by individuals not providing their direct behavioural reactions to external stimuli, but always considering multiple response options before expressing anything (Hirsh et al., 2011). On top of that, as social animals, people feel restrained by social desirability, consistency, and similar reputational concerns in the offline environment (Cialdini & Goldstein, 2004). Following these concerns, people constantly try to make favourable impressions on others (Leary & Kowalski, 1990). In combination, the constant consideration of multiple response options and being guided by social motivations increase the likelihood of individuals conforming to the majority in the offline environment and impede the progression of social change.

This inhibiting effect is jointly produced by the physical presence and lack of anonymity people experience in the offline environment. The lack of anonymity primarily makes the inherent human need for affiliating with other human beings more salient. This need for affiliation leads individuals to inhibit their direct responses to a stimulus and to publicly enforce the majority stance, even if they might not privately agree with it (Cialdini & Goldstein, 2004). More concretely, the enforcement of majority consensus occurs because when everyone can observe the actions of all other persons surrounding them, every individual is accountable for the actions they perform to the rest of the group. Thus, if one contradicts the local majority consensus, it can lead to social exclusion, awkwardness, and the affiliation goal not getting fulfilled (Festinger et al., 1952; Kiesler, Siegel, & McGuire, 1984). In an effort to avoid these unwanted social consequences, the best strategy for the individual is usually to conform.

(8)

relevant norms (Carver & Scheier, 2012; Guerin, 1986). These effects occur since, when others are physically present, people automatically start managing the impressions they make and adapting their behaviour in a way that facilitates gaining social approval from others (Ferris, Beehr, & Gilmore, 1978). And indeed, conformity is one of the most effective ways of obtaining social approval.

Contrary to the offline environment, when people are in the online environment they become disinhibited. This means that instead of considering multiple competing response options to a stimulus, they automatically choose the most readily available response and behave in line with it (Hirsh et al., 2011). In addition, the increased anonymity and physical distance that are present in the online environment lessen the conformity pressures. The decreased accountability provoked by anonymity causes the need for affiliation that drives conformity offline to disappear when people are online (Cialdini & Goldstein, 2004; Festinger et al., 1952; Kiesler et al., 1984). The physical distance between individuals reduces people’s need to seek social approval, making them less likely to actively manage the impression they make on others (Ferris et al., 1978). In sum, both anonymity and physical distance make it easier for individuals to resist conforming to the majority. In combination with the disinhibited state leading people to express their most readily available response, this heightened resistance to conformity pressures should make it more likely for individuals to enforce the minority stance in case they agree with it.

2.2. Implications for social change and hypotheses development

In my theoretical model, social change is defined as a wide-scale behavioural switch from a current societal consensus to an alternative one that was proposed by a minority. By definition, the speed with which social change occurs depends on how fast the majority members start publicly enforcing the minority stance. Therefore, social change is expected to be sped up whenever it is easier for majority members to publicly enforce the minority stance. As established in the previous section, the physical distance and anonymity generated by the online environment produce that exact effect—i.e. they increase the ease with which individuals can defy the majority and back the minority stance. Subsequently, it becomes evident that, on a larger scale, the online environment should speed up social change. Formally:

(9)

Following the reasoning established in the previous section, the mechanism through which the online environment fosters social change lies in the disinhibition that the online environment produces in individuals. Such is the case since, when disinhibited, individuals care less about the impressions they make on others (Ferris et al., 1978), feel less accountable to the rest of the group (Festinger et al., 1952), and are more likely to express their direct behavioural reactions to external stimuli (Hirsh et al., 2011). Together, these facets of disinhibition guide people towards conforming less to the majority and should thus constitute the underlying reason for why the online environment facilitates social change. Therefore, I further hypothesise:

Hypothesis 2: Experienced disinhibition mediates the positive effect of the online environment on the speed of social change.

However, beyond the mediating effect of experienced disinhibition, behavioural (dis)inhibition and need for affiliation also exist as people’s innate personality characteristics. On the one hand, the behavioural inhibition system (BIS) represents an individual chronic tendency to, when the situation provokes such a reaction, feel more disinhibited (Carver & White, 1994). On the other hand, trait need for affiliation represents a tendency to, when provoked by the environment, feel a greater need to affiliate with others (Wiesenfeld, Raghuram, & Garud, 2001). Since the online environment facilitates social change through a momentary increase in experienced disinhibition and decrease in the felt need for affiliation, I assume that these processes should be evoked to a greater extent in people naturally possessing higher susceptibility to disinhibition and trait need for affiliation. Consequentially, the effect of the online environment on social change should be more pronounced for individuals possessing lower levels of BIS and trait need for affiliation. Specifically:

Hypothesis 3a: The effect of the online environment on the speed of social change is more pronounced for people possessing a weaker behavioural inhibition system.

(10)

3. METHODOLOGY

3.1. Experimental paradigm development

The novel experimental paradigm is developed in a few separate steps with each taking into account multiple considerations. In this section, the paradigm is described in five segments, with a focus in every segment on what each aspect of the experiment is trying to accomplish and on arguing how each aspect of the experiment contributes towards this paradigm being an appropriate method for studying social change. First, I describe the basic structure of the experimental game and its introductory phase. Second, I depict the activity regular participants are asked to perform in the game. Third, I explain the payoff structure regular participants are subjected to and how it simulates people’s natural social motivations. Lastly, I describe how confederates are asked to behave, the logic behind the algorithm that makes their public choices, and how the combination of the confederate activity and the algorithmic structure is supposed to induce social change dynamics.

3.1.1. Basic structure

In order to capture over-time social change, the experiment is structured as a multi-round group game. Each experimental group consists of 16 participants, of which 12 are regular participants that play the game and 4 (i.e. 25% of the group) are confederates whose moves are pre-programmed to mirror the decisions a persisting minority would make in this game. Each participant’s role within their group, i.e. whether they actually play the game or just have the role of a confederate, is determined at random.

(11)

holds for a group where an individual member can only assess what is the group consensus based on an indirect and limited account of other people’s behaviour. However, in my game, individuals are shown direct reports of the consensus state after each round. This produces greater immediacy of the minority influence, which makes the minority more effective (Tankard & Paluck, 2016) and can decrease the needed size of the tipping point. Therefore, an approximately 25% persistent minority of confederates seems to be a safe choice for my game as it ensures that the tipping point is always reached within the given time frame.

Within each experimental group, the main game task for the participants is to collaboratively choose a name for a newly established company out of a pre-given set of two company names— those names being ‘Tao’ and ‘Eta’. This specific task is chosen since having two clear options that the participants can choose from makes it easy to observe and to communicate to participants what the state of the consensus is in each round. The final goal in the given task is for the group to reach a full consensus on the company name without directly communicating with each other. Therefore, the game continues until the round in which all participants choose the same company name, which at that moment becomes the winning company name (and is limited to a maximum of 24 rounds due to practical time limitations).

3.1.2. Pre-game procedure

(12)

they are offered a chance to see the basic game instructions and game point system pages again in case they still do not understand parts of it. When all participants indicate that they are confident in their understanding of the game and its underlying rules, the game begins.

3.1.3. Regular participant activity

The regular participants play the game as described above. Each of the maximum 24 playing rounds consists of three parts. In the first part, the participants choose one of the two company names. In the second part, they assess how likely they think it is for the name they chose to become the name everyone in the group chooses in the last round. In the third part, they observe the percentage and the nicknames of participants that chose each of the two company names in the given round (see Appendix C for the regular participant game pages). The most relevant to my main research question are the first and the third part of each round as these are the moments when regular participants make their central decision and receive information about the decisions other participants in their group made. The second part of each round is used only as an exploratory confidence measure. While this structure of the game and its rounds creates a situation where social change processes can potentially occur, it does not yet allow me to test the proposed research question cleanly. Such is the case since the structure of the game itself does not automatically induce any specific social change dynamics into the group or incentivise individuals to behave in certain ways. Therefore, to ensure that the social change process in the group does occur, I devised a special pay-off scheme for regular participants and an algorithm that determines the public choices of confederates. The pay-off structure artificially evokes and amplifies people’s social motivations, while the algorithm creates dynamic social activity that is necessary for social change to occur in the given limited experimental setting.

3.1.4. Game points system and payoff structure

(13)

each participant is proportionally rewarded points from the remaining group payoff pool. The individual points are calculated as:

𝐼𝑛𝑑𝑖𝑣𝑖𝑑𝑢𝑎𝑙 𝑝𝑜𝑖𝑛𝑡𝑠

𝑇𝑜𝑡𝑎𝑙 𝑝𝑜𝑖𝑛𝑡𝑠 𝑖𝑛 𝑔𝑟𝑜𝑢𝑝 𝑝𝑜𝑜𝑙

=

# 𝑖𝑛𝑑𝑖𝑣𝑖𝑑𝑢𝑎𝑙 𝑤𝑖𝑛𝑛𝑖𝑛𝑔 𝑛𝑎𝑚𝑒 𝑐ℎ𝑜𝑖𝑐𝑒𝑠

# 𝑡𝑜𝑡𝑎𝑙 𝑤𝑖𝑛𝑛𝑖𝑛𝑔 𝑛𝑎𝑚𝑒 𝑐ℎ𝑜𝑖𝑐𝑒𝑠

.

(1)

This incentive scheme accomplishes three main goals. First, it puts participants in a situation where their behaviour can be perceived as socially costly—i.e. where their actions can negatively affect the other group members. This costliness of one’s behaviour is something that naturally exists in the society and is one of the main drivers of peoples’ conforming behaviour (Cialdini & Goldstein, 2004). However, no such costliness of one’s behaviour is naturally present in a low-stakes environment of an experiment. Therefore, to heighten the external validity of the paradigm, the costliness of one’s behaviour needed to be present in the experimental setting as well.

Second, the incentive system makes it possible to observe ‘wide-scale social change’ in a limited time period. Since the context and argumentation of my theorising are shaped around societal-level change, it concerns a usually very long time frame (i.e. such processes often take years or decades to occur in society). However, the experimental setting constraints the available time frame to less than an hour. Therefore, in order to be able to make any conclusions about societal-level change, the paradigm needs to allow for this long process to occur in a way shorter time frame. This was achieved through incentivising people to reach a changed consensus fast by decreasing the available pool of points with each new round and through realising participants’ payoffs only if they reach a full consensus.

(14)

incentives are constructed in a way that does not bias individuals towards one type of behaviour, but rather makes them focused on all different social processes occurring in the group.

3.1.5. Confederate activity and public choice algorithm

For confederates, the rounds are structured in two parts (see Appendix D for the confederate game pages). On the first page, the confederates need to choose one of the two company names as well. However, they do not make the company name choice as themselves, but are rather asked to assess which choice a profit-maximising agent would have made. While their choices are tracked and might be used in some exploratory analyses, they do not have any impact on the actual game. Rather, their choices are replaced by the ones made by an algorithm. Thus, the main goal of the confederates making these choices is for regular participants to see that everyone in the room is staring at the screen and making clicks with a mouse. Ultimately, this should help make the regular participants believe that everyone in the room is indeed participating in the game. In the second part of each round, the confederates only observe the percentage of participants that chose each of the two options and not the nicknames like the regular participants. For their participation, the confederates receive a fixed payoff of 6€. The algorithm that makes confederates’ public choices works in two separate stages, namely the early-game stage and the post-consensus stage. The early-game strategy is used from the beginning of the game until the round in which the initial consensus among all regular participants occurs. The post-consensus strategy is used after the initial consensus round until the end of the game. In the early-game stage, one confederate chooses the option that was chosen by a majority of regular participants in the current round, while three confederates choose the minority option. In case the amount of regular participants choosing each of the two name choices is equal, two confederates choose each of the two company names. In the post-consensus stage, all four confederates keep on choosing the company name that is different from the one all regular participants chose in the initial consensus round. The confederates continue doing so until all regular participants switch to their side or until the rest of the 24 rounds pass by.

(15)

change in the amount of anti-consensus votes to not be too sudden after the initial consensus round, a majority of confederates already backs the anti-consensus option before that. Second, the algorithm provides a persistent disagreeing minority that provokes social change after the initial consensus has been reached. This second part is of great importance as the activity the confederates display here constitutes the actual process of social change induced by a persistent minority.

3.2. Manipulation of the online/offline environment in the paradigm

Two specific aspects of the online environment produce disinhibition in people, namely physical distance (Guerin, 1986) and anonymity (Kiesler et al., 1984). This was taken under special consideration when constructing the manipulation. In order to disentangle the effects of both contextual factors, they will be varied separately in future studies. However, in the pilot study, the sole focus lies on anonymity. The main mechanism through which anonymity is theorised to reduce the conforming behaviour of individuals lies in the loss of accountability. That is, when people are anonymous, others cannot link the actions they perform back to them and, thus, they hold no social responsibility for their actions (Festinger et al., 1952). Since anonymity consists of multiple aspects, I manipulate only the loss of accountability so as to exclude any other confounding aspects full anonymity has that could interfere with the main theorised effect.

(16)

traced back to them and they hold no social accountability for their actions. See figures in Appendix A for the illustration of the two conditions.

3.3. Pilot study characteristics

3.3.1. Experimental groups and participants

The goal of the pilot study is to investigate whether the designed paradigm fulfils its purpose— i.e. whether it creates a situation in which full social change can be observed. Fifty-one students from the University of Groningen participated in the pilot study, forty in the role of a regular participant and 11 in the role of a confederate. They were recruited through public invitations on different social media platforms, a post on the university lab website, and personally using flyer distribution. The final pilot study sample thus consisted of 24 (47.1%) female and 27 (52.9%) male participants. Their average age was 24, ranging from 18 to 32. A majority of participants held a university degree (n = 38; of which 16 held a Bachelor degree and 22 a Master degree), while there were also some participants with a post-secondary (n = 3) and a secondary (n = 10) education level.

The participants were separated into four experimental groups, two in the anonymous condition and two in the identifiable condition, based on the timeslots they signed up for. Two experimental groups consisted of fourteen participants, one consisted of twelve participants, and one of eleven participants. There are certain minor differences in how the paradigm was implemented in different groups. In the second and last group chronologically, the paradigm was implemented exactly as described. In the first group chronologically, the algorithm determining confederate activity failed to register that an initial consensus was reached and thus the whole game occurred in the early-game stage. In the third group chronologically, I decided to make a minor adaptation to the early-game confederate activity based on the insights from the previous two sessions. This adaptation will be discussed in section 4.1.

3.3.2. Procedure

(17)

of the experiment 10 minutes before the experiment started and were greeted in front of the experimental room by the experimenter. After having their attendance noted by the experimenter, they entered the room, sat down behind the assigned desk, and were instructed to prepare their laptop and wait in silence until everyone arrived. The experiment began exactly on time. Therefore, any participant that was late had their participation in the experiment denied. In order to deal with the drop in the number of participants, the computer code for the experiment was written in a way that from four to sixteen players could play the game, with the percentage of participants having the role of a confederate always kept around 25%.

To begin the experiment, all participants present in the room were sent an email containing the link to the study and were asked to open it on their computers. Then, they chose a nickname and, based on the condition, had to either 1) write the chosen nickname in the computer, on the nametag, and communicate it to others, or 2) only write the chosen nickname in the computer. The rest of the experiment was conducted fully on the computer and the experimenter was present only to make sure people are not communicating with each other. Thus, the experimenter did not interfere with the game in any other way. The sequence of the experimental activities on participants’ computers was as follows. First, the participants read the instructions for the company name game (see Appendix B). Second, the participants played the company name game (see Appendices C and D). Third, the participants answered some additional process measures (see section 3.3.3.). The full experimental session lasted on average for 45 minutes.

3.3.3. Measures

(18)

constructed items. The behavioural disinhibition system and trait need for affiliation were measured using already existing scales.

The experienced disinhibition measure was constructed by separating the concept of disinhibition into multiple facets (i.e. freedom from the self, freedom from social pressures, and freedom from a priori stances) and then constructing two items measuring each of these sub-dimensions. Participants were asked to assess each of the items using a 5-point Likert scale ranging from ‘1 – Strongly disagree’ to ‘5 – Strongly agree’. Freedom from the self was measured using items “During the game, I felt the need to be consistent.” and “I felt like the choices I made expressed certain information about my identity.” (both reverse-scored; Pearson’s r = 0.33). Freedom from social pressures was measured using items “I felt a pressure to conform to the majority in the game.” and “I wanted to be liked by other participants in my group.” (Pearson’s r = 0.45). Freedom from a priori stances was measured using items “I found it important that the company name I chose fit its industry well.” and “How nice the company name sounds played an important role in my choices.” (both reverse-scored; Pearson’s r = 0.31). The low between-item correlations that emerged in all three sub-dimensions will be problematized in section 4.1.

(19)

4. PILOT STUDY OUTCOMES AND DISCUSSION

Since my thesis is primarily focused on developing a novel paradigm, the goal of the pilot study is to investigate whether the devised paradigm produces the context needed to observe social change and is thus suited for testing my hypotheses. However, the hypotheses are not yet tested at this point. Accordingly, the pilot study was run with a low sample size and there is no classical results section in my thesis. Instead, I opted for a section that integrates the main insights from the pilot study with a discussion of these insights. However, in addition to discussing the insights, the current section also provides an outline of how my hypotheses will be tested once a sample with sufficient statistical power is obtained. The section is structured in two parts. In the first part, the paradigm and the experimental manipulation are evaluated based on the observations and feedback gathered in the pilot sessions. The insights from this evaluation are then used to inform the potential improvements of the experimental paradigm. In the second part, I describe the analytical strategy that I will employ when the paradigm is run on a sample with sufficient statistical power. This second part includes the descriptions of how the variables will be calculated from the experimental data and what kinds of regression models will be used for testing my hypotheses.

4.1. Evaluation of the paradigm and the experimental manipulation

(20)

manipulation was on a between-group level. In the company name game itself, an initial consensus was established in all four experimental groups. Additionally, after the initial consensus round, there was a certain number of participants slowly changing their choices to the minority-backed anti-initial consensus company name (see Figure 1 for examples of how this occurred in two of the groups). In sum, it seems that the experimental paradigm did manage to create a situation where over-time social change can occur and successfully mirrored a multitude of considerations individuals have during the process of social change. However, despite these favourable outcomes, there were two downsides to the paradigm that should be resolved in future studies for the paradigm to fully satisfy its purpose.

The first downside of the current paradigm lies in the fact that while the initial consensus was established in all of the experimental groups, it often happened rather late—i.e. in three out of four sessions it took almost half of the game for participants to reach the initial consensus. Since only the game rounds occurring after the initial consensus round are relevant for answering my research question, any delays in how early in the game that occurs compromise the comparability of the data that is obtained from the game. Three main solutions could be introduced to the game that would ensure that the initial consensus is reached earlier in the game, while not interfering with the rest of the game.

(21)

1994; Sugden, 1995). Therefore, making one of the company names a focal point could establish an overwhelming majority immediately in the first or second round and, through that, facilitate the establishment of an early initial consensus. Already in the current paradigm, the company name ‘Tao’ turned out to be the focal point given that in all four experimental groups choosing ‘Tao’ emerged as the initial consensus in the game and was chosen by a majority of regular participants in the first round. Besides, some participants even directly reported in their final comments that they saw ‘Tao’ as the focal point. In the current version of the game, ‘Tao’ unintentionally became the focal point since it was displayed as the choice on the left of the in-game page (see Appendix C) and since it was more often mentioned in the instructions pages. Therefore, to further facilitate ‘Tao’ as the focal point, it could be mentioned even more often in the instructions.

The third potential improvement could be decreasing the number of regular participants needed for establishing the initial consensus. In the current version of the game, all regular participants needed to choose a specific company name for that company name to become the initial consensus. However, the delay in the establishment of the initial consensus was usually caused by one or two regular participants that decided to oppose this majority consensus and kept choosing the contrary company name. Therefore, since only an overwhelming majority is needed for something to be seen as a consensus, the number of participants supporting a certain company name needed in a round could be decreased to two less than the total number of regular participants for that round to still be defined as initial consensus round. Combined with the first proposed adaptation—i.e. increasing the visible support for the majority stance by two confederates—the two changes would not produce any difference in terms of how the regular participants see the game, while also making it more likely for the initial consensus to be reached earlier. In sum, the proposed three changes should be able to lower the number of game rounds needed for establishing the initial consensus to less than 5 and, through that, extend the number of rounds in which social change can be observed.

(22)

participants in the late rounds. Therefore, participants decided it was worth more to them to punish the people opposing the group consensus rather than the payoff they would receive themselves. There seem to be two improvements that could be implemented to deal with this issue.

First, the issue shines a light on the fact that the incentive system as it was incorporated in the current paradigm also incentivises one unexpected type of behaviour—i.e. punishing the minority for resisting the group consensus. In order to deal with this problem, the incentive system could be changed in a way that makes the payoff the individuals receive seem worth more to them than punishing the minority. The simplest solution that can ensure this would be a change in the incentive system that produces a greater difference between reaching consensus in the late rounds and not reaching a consensus at all. In other words, as it stands now, the available pool of points for a group payoff decreases by 10 points (2€) each round and equals 0 points (0€) in the last round. If the decrease of the group pool in each round would have been decreased to 5 points (1€), then 120 points (24€) would remain for the group payoff in the last round. Since this would significantly increase the difference in the benefit a regular participant receives from reaching consensus in the late rounds of the game and not reaching a consensus at all, it would motivate participants to focus more on reaching consensus rather than on punishing the opposing minority.

The second solution to the problem of not observing the full social change could be made by changing the size of the persistent minority (i.e. the number of confederates). As it stands now, the size of the confederate minority was kept around 25% in different experimental groups. However, past research shows that the minimum needed size of the tipping point is 25% (Centola et al., 2018). While I argued in the methodology section that the display of minority influence in my paradigm is more direct and, thus, the tipping point could potentially be lower, this judgement was mostly speculative and does not seem to hold given the results of the pilot study. Therefore, to ensure that the tipping point is always reached—and thus the full social change can be observed—the 25% should be set as a strict minimum for the size of the tipping point.

(23)

2.05, SD = 1.00; Identifiable, M = 2.70, SD = 1.13; t (40) = 1.93; p = 0.06; d = 0.61). Therefore, it can be concluded that the devised manipulation is appropriate for varying the degree of accountability and, consequentially, anonymity individuals feel in the game.

However, there is no significant difference in experienced disinhibition between the two conditions, as participants in both conditions felt similarly free from the self (Anonymous, M = 3.10, SD = 1.14; Identifiable, M = 2.88, SD = 0.79; t (40) = 0.72; p = .47; d = 0.23), similarly free from social pressures (Anonymous, M = 2.90, SD = 1.13; Identifiable, M = 2.67, SD = 1.15; t (40) = 0.62; p = .54; d = 0.20), and held a similar degree of a priori commitment to the initial company name choice (Anonymous, M = 3.10, SD = 1.02; Identifiable, M = 3.12, SD = 1.07; t (40) = 0.08; p = .94; d = 0.02). Since the manipulation successfully induced the feelings of exposure, the failure of the manipulation to also induce feelings of disinhibition could be accredited to the measure of experienced disinhibition I constructed.

(24)

(25)

4.2. Analytical strategy for testing hypotheses

The following section describes the analytical strategy I propose to use to test my hypotheses on a sample with sufficient statistical power using the data obtained from the devised experimental paradigm. There are five variables in the hypotheses that need to be determined in order to test the hypotheses, namely Online Environment (OE), Speed of Social Change (SSC), Experienced Disinhibition (ED), Behavioural Inhibition System (BIS), and Trait Need for Affiliation (NFA). Online Environment is a dummy variable that represents whether the group each participant partook in occurred in the online or the offline environment condition (1 = online condition, 0 = offline condition). Speed of Social Change is operationalised as the number of rounds after the initial consensus in which a participant chose the company name that differs from the initial group consensus. For example, if the initial consensus is the company name “Tao”, then the speed of social change is operationalised as the number of rounds in which each participant chose “Eta”. Therefore, a higher value of the variable represents a participant choosing the anti-consensus option more often in the second part of the game and, consequentially, a higher speed of social change. Experienced Disinhibition, Behavioural Inhibition System, and Trait Need for Affiliation are calculated as the mean of a participant’s ratings on the items of each scale, with higher values of each variable representing a stronger presence of the given psychological state/trait.

To test the four hypotheses, the default choice of the estimator is the ordinary least squares (OLS) estimator. However, given that the data is structured in groups and that one of the variables occurs on the group level (i.e. Online Environment), the potentially nested structure of the data is investigated using the intraclass correlation coefficient 1 (ICC1). If the ICC1 turns out to be significant—meaning that the data indeed displays significant nesting—the OLS estimator is replaced with a multi-level linear mixed-effects model, which is based on the maximum likelihood estimator. Using the chosen estimation technique, a separate regression model would have been used to test each of the hypotheses.

The first hypothesis states that the speed of social change occurs faster in the online, compared to the offline, environment. The linear regression model for testing this effect is described as:

𝑆𝑆𝐶 = 𝛽

₀

+ 𝛽

₁

𝑂𝐸 + 𝜀.

(2)

The hypothesis is confirmed if the main effect of Online Environment (β1) turns out to be

(26)

The second hypothesis can only be tested if the first hypothesis is confirmed. The second hypothesis states that the effect of the online environment on the speed of social change is mediated by the experienced disinhibition. The structural equation model for testing this effect is described as:

𝐸𝐷 = 𝛽

₀

+ 𝛽

₁

𝑂𝐸 + 𝜀

_𝐸𝐷

,

(3)

𝑆𝑆𝐶 = 𝛾

₀

+ 𝛾

₁

𝐸𝐷 + 𝛾

₂

𝑂𝐸 + 𝜀

_𝑆𝑆𝐶

.

(4)

The hypothesis is confirmed if the indirect effect of Online Environment as mediated by Experienced Disinhibition (measured with the coefficient product β1γ1) turns out to be positive

and significant. To determine whether the model displays full mediation or only partial mediation, I further examine the direct effect of Online Environment (γ2). The direct effect

being insignificant signals full mediation.

Hypothesis 3A states that the effect of the online environment on the speed of social change is more pronounced for individuals with a weak behavioural inhibition system and hypothesis 3B states that this same effect is more pronounced for individuals that are low in trait need for affiliation. The linear regression model for testing hypothesis 3A is described as:

𝑆𝑆𝐶 = 𝛽

₀

+ 𝛽

₁

𝑂𝐸 + 𝛽

₂

𝐵𝐼𝑆 + 𝛽

₃

(𝑂𝐸 𝑥 𝐵𝐼𝑆) + 𝜀,

(5)

while the linear regression model estimated for testing hypothesis 3B is described as:

𝑆𝑆𝐶 = 𝛽

₀

+ 𝛽

₁

𝑂𝐸 + 𝛽

₂

𝑁𝐹𝐴 + 𝛽

₃

(𝑂𝐸 𝑥 𝑁𝐹𝐴) + 𝜀.

(6)

Hypothesis 3A is confirmed if the interaction effect of Online Environment x Behavioural Inhibition System (β3 in (5)) turns out to be negative and significant. Similarly, hypothesis 3B

is confirmed if the interaction effect of Online Environment x Trait Need for Affiliation (β3 in

(27)

5. CONCLUSION AND FUTURE RESEARCH

In the present paper, a new paradigm that offers the potential for investigating dynamic aspects of minority-induced social change was devised. The pilot study examining the given paradigm showed that, with certain adaptations, the paradigm can successfully simulate the complex process of social change in a limited experimental setting. Through this, it creates a possibility for testing the proposed research question—i.e. how does the online environment facilitate social change? However, beyond the proposed research question, there is a wide range of other questions that the theorising and the proposed paradigm shed a light on.

On the one hand, there seem to be four main kinds of research questions that could be examined using the devised paradigm. First, at the moment, I consider only the speed of social change as a dynamic social change outcome that differs depending on the situation, in this case, online or offline. The first extension could thus be to consider how the devised paradigm could also be used to examine how other outcomes in the dynamic process change under varying circumstances. For example, the size of the tipping point that I took as a given might differ depending on the situation (Centola et al., 2018) and could thus be smaller in the online environment.

Second, the two contextual factors that my study focuses on are anonymity and physical distance. Therefore, future research could extend upon the present paper by varying a multitude of different relevant factors that can facilitate or impede peoples’ experienced disinhibition, which drives the changes in social change dynamics. An example of relevant contextual factors that impede peoples’ experienced disinhibition are the salience of group identity considerations (Reicher, Spears, & Postmes, 1995), the expectancy of future interactions (Gergen & Wishnov, 1965), and goal-relevance of presentational concerns (Leary & Kowalski, 1990).

Third, the issue individuals seek consensus on as used in the present study holds no inherent moral aspects nor even any potential to be morally contentious. However, past research shows that moral differences between individuals can often be the hardest to resolve (Haidt, Rosenberg, & Hom, 2003). Hence, future research could incorporate differentially morally loaded issues in order to investigate how dynamic social change outcomes are differently influenced in situations with more or less morally loaded issues.

(28)

for why social change takes time. First, it might be necessary to have repeated interaction— both in terms of the number of interactions and in terms of the number of different minority members one interacts with—to be convinced to change. Second, it might be possible that an individual needs time to process the information they received from the minority. Especially if the information feels reasonable, but is in contrast with peoples’ important personal values, strong reluctance to change immediately could exist. Therefore, varying different aspects of the paradigm (e.g. the amount of time in between rounds, the number of confederates pushing the option in each round, and providing arguments for why individuals choose different stances) could shed light on the given question that is of high practical relevance.

On the other hand, there seem to be two further questions that the presented theorising opens, but cannot be addressed by the devised paradigm. First, in the theoretical background, the argumentation behind the proposed effect of the online environment on social change rests solely on the reduced salience of the need for affiliation that occurs due to the experienced disinhibition. However, experienced disinhibition does not only impact people’s public behaviour, but also how individuals process external information (Hirsh et al., 2011). At the same time, when people are in the ordinary inhibited state, there are certain significant differences in how they perceive minority- and majority-backed information. Specifically, individuals are more likely to blindly accept majority-backed stances and scrutinise minority-backed statements (Horcajo et al., 2010). Given these insights, there seems to also be a potential for the proposed effect of the online environment on social change to be partially driven by individuals focusing more on information at hand rather than source status when they are experiencing disinhibition.

(29)

(30)

6. REFERENCES

Asch, S. E. (1952). Group forces in the modification and distortion of judgements. In S. E. Asch (Ed.), Social Psychology (pp. 450–501).

Asch, S. E. (1956). Studies of independence and conformity: I. A minority of one against a unanimous majority. Psychological Monographs: General and Applied, 70(9), 1–70. https://doi.org/10.1037/h0093718

Baumeister, R. F., Ainsworth, S. E., & Vohs, K. D. (2016). Are groups more or less than the sum of their members? The moderating role of individual identification. Behavioral and Brain Sciences, 39(2016). https://doi.org/10.1017/S0140525X15000618

Bond, C. F. (1982). Social facilitation: A self-presentational view. Journal of Personality and Social Psychology, 42(6), 1042.

https://doi.org/http://dx.doi.org/10.1037/0022-3514.42.6.1042

Bond, C. F., & Titus, L. J. (1983). Social facilitation: A meta-analysis of 241 studies. Psychological Bulletin, 94(2), 265.

Carver, C. S., & Scheier, M. F. (2012). Attention and self-regulation: A control-theory approach to human behavior. Berlin: Springer Science & Business Media. Carver, C. S., & White, T. L. (1994). Behavioral inhibition, behavioral activation, and

affective responses to impending reward and punishment: the BIS/BAS scales. Journal of Personality and Social Psychology, 67(2), 319.

Centola, D., Becker, J., Brackbill, D., & Baronchelli, A. (2018). Experimental evidence for tipping points in social convention. Science, 360(6393), 1116–1119.

https://doi.org/10.1126/science.360.6393.1082-d

Centola, D., Willer, R., Macy, M., Centola, D., Willer, R., & Macy, M. (2005). The Emperor ’ s Dilemma : A Computational Model of Self-Enforcing Norms. American Journal of Sociology, 110(4), 1009–1040.

Chen, D. L., Schonger, M., & Wickens, C. (2016). oTree—An open-source platform for laboratory, online, and field experiments. Journal of Behavioral and Experimental Finance, 9, 88–97.

(31)

Annual Review of Psychology, 55, 591–621.

https://doi.org/10.1146/annurev.psych.55.090902.142015

Ferris, G. R., Beehr, T. A., & Gilmore, D. C. (1978). Social facilitation: A review and alternative conceptual model. Academy of Management Review, 3(2), 338–347. https://doi.org/https://doi.org/10.5465/amr.1978.4295344

Festinger, L., Pepitone, A., & Newcomb, T. (1952). Some consequences of de-individuation in a group. Journal of Abnormal and Social Psychology, 47(2), 382–389.

https://doi.org/10.1037/h0057906

Gergen, K. J., & Wishnov, B. (1965). Others’ self-evaluations and interaction anticipation as determinants of self-presentation. Journal of Personality and Social Psychology, 2(3), 348.

Gladwell, M. (2000). The Tipping Point. London, UK: Abacus.

Guerin, B. (1986). Mere Presence Eﬀects in Humans: A Review. Journal of Experimental Social Psychology, 22, 38–77.

https://doi.org/https://doi.org/10.1016/0022-1031(86)90040-5

Haidt, J., Rosenberg, E., & Hom, H. (2003). Differentiating Diversities: Moral Diversity Is Not Like Other Kinds. Journal of Applied Social Psychology, 33(1), 1–36.

Hirsh, J. B., Galinsky, A. D., & Zhong, C. B. (2011). Drunk, powerful, and in the dark: How general processes of disinhibition produce both prosocial and antisocial behavior. Perspectives on Psychological Science, 6(5), 415–427.

https://doi.org/10.1177/1745691611416992

Hodge, E., & Hallgrimsdottir, H. (2019). Networks of Hate: The Alt-right, “Troll Culture”, and the Cultural Geography of Social Movement Spaces Online. Journal of Borderlands Studies, 1–18. https://doi.org/10.1080/08865655.2019.1571935

Horcajo, J., Petty, R. E., & Briñol, P. (2010). The effects of majority versus minority source status on persuasion: a self-validation analysis. Journal of Personality and Social Psychology, 99(3), 498–512. https://doi.org/10.1037/a0018626

Kiesler, S., Siegel, J., & McGuire, T. W. (1984). Social psychological aspects of computer-mediated communication. American Psychologist, 39(10), 1123–1134.

(32)

Leary, M. R., & Kowalski, R. M. (1990). Impression management: A literature review and two-component model. Psychological Bulletin, 107(1), 34–47.

https://doi.org/10.1037/0033-2909.107.1.34

Mehta, J., Starmer, C., & Sugden, R. (1994). Focal points in pure coordination games: An experimental investigation. Theory and Decision, 36(2), 163–185.

https://doi.org/https://doi.org/10.1007/BF01079211

Moscovici, S., Lage, E., & Naffrechoux, M. (1969). Influence of a Consistent Minority on the Responses of a Majority in a Color Perception Task. Sociometry, 32(4), 365.

https://doi.org/10.2307/2786541

Moscovici, Serge. (1985). Innovation and minority influence. In S. Moscovici, G. Mugny, & E. Avermaet (Eds.), Perspectives on minority influence (pp. 9–52). Cambridge:

Cambridge University Press.

Newport, F. (2014). The New Era of Communication Among Americans. Retrieved March 18, 2018, from

http://news.gallup.com/poll/179288/new-era-communication-americans.aspx

Pew Research Center. (2011). Global Digital Communication: Texting, Social Networking Popular Worldwide. Retrieved March 18, 2018, from

http://www.pewglobal.org/2011/12/20/global-digital-communication-texting-social-networking-popular-worldwide/

Reicher, S. D., Spears, R., & Postmes, T. (1995). A Social Identity Model of Deindividuation Phenomena. European Review of Social Psychology, 6(1), 161–198.

https://doi.org/10.1080/14792779443000049

Sugden, R. (1995). A Theory of Focal Points. The Economic Journal, 105(430), 533–550. https://doi.org/https://doi.org/10.2307/2235016

Suler, J. (2004). The Online Disinhibition Effect. CyberPsychology & Behavior, 7(3), 321– 326. https://doi.org/10.1089/1094931041291295

Tankard, M. E., & Paluck, E. L. (2016). Norm Perception as a Vehicle for Social Change. Social Issues and Policy Review, 10(1), 181–211. https://doi.org/10.1111/sipr.12022 Wiesenfeld, B. M., Raghuram, S., & Garud, R. (2001). Organizational identification among

(33)

Journal of Management, 27(2), 213–229.

https://doi.org/https://doi.org/10.1177/014920630102700205

Wood, W., Lundgren, S., Ouellette, J. A., Busceme, S., & Blackstone, T. (1994). Minority influence: A meta-analytic review of social influence processes. Psychological Bulletin, 115(3), 323–345. https://doi.org/10.1037/0033-2909.115.3.323

(34)

7. APPENDICES

7.1. Appendix A: Experimental setup

(35)

7.2. Appendix B: Instruction pages and questions testing participant understanding of the game

Note: The first two figures depict the first two instructions pages that are the same for the

confederates and the regular participants. The first page welcomes participants to the game

(36)

(37)

Note: The following four figures depict the rest of the pages as shown to the regular

participants. The first page describes the game point system and the payoffs. The second page

(38)

(39)

(40)

(41)

7.3. Appendix C: Regular participant game pages

(42)

7.4. Appendix D: Confederate game pages

ONLINE DISINHIBITION AS A FACILITATOR OF SOCIAL CHANGE: DEVELOPMENT OF A NOVEL EXPERIMENTAL PARADIGM

Research Master in Economics and Business

Master’s Thesis