Participatory practices at work change attitudes and behavior toward societal authority and justice

Generalized attitudes toward authority and justice are often conceptualized as individual differences that are resistant to enduring change. However, across two field experiments with Chinese factory workers and American university staff, small adjustments to people’s experience of participation in the workplace shifted these attitudes one month later. Both experiments randomly assigned work groups to a 20-minute participatory meeting once per week for six weeks, in which the supervisor stepped aside and workers discussed problems, ideas, and goals regarding their work (vs. a status quo meeting). Across 97 work groups and 1,924 workers, participatory meetings led workers to be less authoritarian and more critical about societal authority and justice, and to be more willing to participate in political, social, and familial decision-making. These findings provide rare experimental evidence of the theoretical predictions regarding participatory democracy: that local participatory experiences can influence broader democratic attitudes and empowerment.

expectations during the first meeting, and briefly repeat the expectations at the start of each subsequent meeting. The protocol for discussion leaders goes: "My name is Zhang Xiaohong, and you can call me Xiaohong. I'm a student from Soochow University, and I am helping a professor with a project on work experience. From this week on, I will come every Monday to lead a discussion with you on work related issues during your regular morning meeting time, for a period of six weeks. We will discuss the problems you have experienced in work, and the aim is for you to work better! Our meetings are easy-going.
We encourage everyone to speak up! Just voice out whatever's on your mind about your work, such as issues yesterday or in the past week, the difficulties you have at work, or things you think will help you and others. I may ask some questions, and there's no right or wrong answers.
Whatever you share will be helpful for the group and for us. I will take some notes during the discussion for research purposes, but I will not show my notes or talk with anyone who's not in our project team, including the factory people." In subsequent meetings, discussion leaders repeat the expectation, "as we all know, it's a meeting for us to share our opinions on production related issues. I'm here to discuss with you on how to work better, rather than testing you. No worry about being right or wrong. Just say whatever you think of about work and participate!" As a warm-up for problem solving, discussion leaders can start with easy questions such as "what type of order are you all working on today?" and "what steps are each of you in charge of?". Discussion leaders prepare and facilitate two questions for the workers to discuss. The number of discussion questions is secondary to the depth of the discussion. Though the content of discussion is flexible as long as it is work-related, we do have a module of focus for each the workers about the orders placed by customers and specifically how many pieces each order requires and how long they have before the suggested deadline. For example, if an order placed asks for 10,000 pieces within 20 days, then a worker or a group is expected to produce around 500 pieces daily if they spread production evenly across days. The calculation is simple enough to do for workers with a Chinese elementary school education. Instead of being assigned a fixed production goal daily, workers will be given all the relevant information and encouraged to come up with a daily production goal themselves. Each worker is given a small piece of paper to do simple calculation and asked to voice out their goals in front of their group members.
In the end, discussion leaders wrap up the meeting and remind them about the following week's participatory meeting.

Supplementary Note 2 -Qualitative Observations
About the factory.
The study took place in the Chinese branch of a multinational apparel manufacturer, which is the largest in employee size among all branches and is located in the eastern coastal area of China. Our study population, the factory workers, were mostly young women in their twenties or thirties 1 who migrated from rural China to the city. The factory is built on the edge of the city and far from the city center, in a location that is relatively inconvenient to reach by public transportation. Around half of the workers live in adjacent factory dorms and another half commute to work on a daily basis. From field interviews prior to the experiment, we find that these workers are eager to work, but have little education (most of them have not finished high school) or training that would allow them to get a high-skilled job in the city. Hence, they enter manual work in apparel manufacturing, which is regarded as labor-intensive and low in skill requirement. Compared with its competitors, the factory pays very well and its workers, although extremely busy, mostly receive a salary in the rank of the lower-middle class 2 in the city.

Qualitative findings from the pilot intervention,
We recount the lessons learned from the pilot intervention (labeled a to d), followed by a series of field-note episodes within these experimental meetings, which briefly sketched parts of the meeting flow and group dynamics. Rather than a unified narrative, the following entries are a series of unfolding actions taken from different meetings across time. We understand that 1 The fact that the workers are mostly young village women with little education is very similar to the personnel composition in the Harwood factory in Lewin' time. Whereas Harwood had around 300 workers, the current factory has thousands of workers, with quite different manufacturing scale. 2 The monthly wage of a typical worker in the experimental factory ranges from 3,000 Yuan ($483.82) to more than 7,000 Yuan ($1128.92). Some reference statistics: In 2013, the minimum wage per month in Beijing was 1,400 Yuan, about 24 percent of the 5,793 Yuan average monthly wage, as calculated by the municipal bureau of statistics. In Shanghai, the 2013 minimum wage was 1,620 Yuan, or 32.2 percent of the 5,036 Yuan average wage. Those are two of the highest income cities in China. The living standard of Suzhou is very close to Shanghai. observer biases are hard to avoid in qualitative data collection (Emerson, Fretz, & Shaw, 1995), but tried to be as impartial as possible about our observations and writings.
(a) Workers were very quiet in the first treatment meetings, and gradually talked more and more as they got used to this participatory style of meetings.
In the first meeting with the quality control group, the workers automatically formed two strictly straight lines close to each other, one in front and one back, like soldiers in an army. The researcher told them to feel more relaxed and form a circle so that everyone could face each other. In response to this, nobody moved. The same thing happened in the first meeting with the embroidery group, where workers automatically lined up, with a sizeable distance (at least 6 feet) away from the researcher. The supervisor later told us that the workers were never required to stand in lines during meetings, but they had formed such a habit for some unknown reasons. We speculate that workers have internalized the authoritarian work style of the factory, including the hierarchical arrangements, and were ready to follow any rule of an authority without question.
Workers were reluctant to speak up in the first meetings, and refrained even from saying their names, as demonstrated in the following excerpts from field notes: [1 st meeting with the embroidery group, July 1] I 3 started the meeting with self-introduction and went around the circle asking for their names. A silence. They were reluctant to talk. The girl standing in the middle of the workers facing me said "let's start with supervisor Wang." Wang said I had already met her. A short silence and I asked the person standing right next to me to start the self-introductions. The girl looked down to the ground and quietly said her name. Then one after another. I repeated each of their names, making sure I got them right.
[1 st meeting with the packing group, July 7] "What are you working on today as a group?" I asked.
The workers looked at each other and nobody spoke. I encouraged them to speak up. A woman raised her hand and said they were packing clothes. The others murmured.
However, workers adapted to the participatory style at a faster rate than we had expected.
For example, at the second meeting with the embroidery group, the workers started to smile and greet the researcher when she came in, and actively discussed the June salary payment 4 . During the third and fourth meetings, most group members voluntarily expressed their opinions on production-related issues such as difficulty encountered with a new order, and how to develop good gestures with a machine.

(b) A participatory meetings intervention was the most effective when the tasks involved some level of collaboration between workers.
One question that remained unclear in Lewin's work was whether a participatory meetings intervention was equally applicable to every work group. As a large part of the intervention treatment focuses on eliciting information exchange among workers on production related issues, we suspected the intervention would be most relevant for the groups which require collaborative tasks between group members, or for which an individual worker's performance is affected by and will affect the performance of her coworkers. It was indeed what we found in our pilot intervention. In groups that require collaboration between workers, like the sewing, packing, and cutting groups, workers expressed many constructive suggestions on what they need from the workers next to them to help themselves work faster and better, and what they could do to help their coworkers work faster and better. They had never thought of or had the chance to openly discuss working strategies like those. However, the discussion about group collaboration was not very successful for the embroidery and quality control groups, whose tasks did not necessitate collaboration between workers, as illustrated in the following episodes: [1 st meeting with the embroidery group, July 1] "Are you working on short-sleeve or long-sleeve clothes? What kind of things are you all working on as a group today?" I asked.
"Each of us works on different things." A worker said.
"What are each of you working on today?" I asked a second question. "We all work on different things. Different things every day. We follow the supervisor's assignments." "What are some of the strategies you've used for this task? Do you mind sharing with others? You know as a group, we need collaboration." A silence.
"Collaboration is not needed. We work on different things." A woman quietly responded.
(c) The discussion flow was much more fluent for the groups whose workers' jobs were interdependent. Any work problem could be easily turned into a discussion after the first meeting.
The sewing division is the factory's largest and most labor-intensive. Workers are organized into 20-30 person groups which work on a specific order placed by companies all over the world. For example, a group may specifically work on a purple baby one-piece while another group works specifically on a blue dress during a certain time period. Each worker in a group is assigned a step in the apparel production processes and tends to repeat the same step until the whole order is completed. As a group is vertically integrated, an individual's work performance might affect the workers after her, even though their salary earnings only depend on individual piece rates. Nevertheless, the optimal strategy is for the group to achieve its maximum productivity so that everyone can have a stable high level of output, rather than for each individual to maximize her own profit (which results in a fluctuation of individual productivity because an individual cannot produce faster when prior steps are not finished). Coordination issues become more prominent when there is a production order switch. One problem came with new cloth patches, and inefficient coordination between workers and between different divisions.
Workers complained about frequent order switches because they thought their salaries would suffer. Workers expressed grudges in the discussions. As one put it: "I can't work fast with new tasks. And I can't work fast unless the person before me works fast." Workers looked surprised when they heard that actually everyone shared the same problem. A solution might be as simple as help to unpack patches: [2 nd meeting with Friday's sewing group, July 10] A girl standing in the middle, who is in charge of the first step of the work process, said: "It would be great if the person after me or someone else will lend me a hand to carry the materials from carts to my working desk. The materials are too heavy for me and it slows me and also the group down." The woman after her nodded immediately and said she never noticed that the materialmoving was slowing their performance. Another worker said she hoped others would help her to do a few pieces when she could not finish all of them in time. 5 For the sewing groups, any production-related issue could be developed into a discussion. Another example follows: [2 nd meeting with Thursday's sewing group, July 9] These responses were prompted by the question, "what are some of your production problems that you've encountered this week?". It turned out that the sewing departments just changed new machines the previous week. The new machines were supposed to be better and safer than the old ones. But workers did not like them. Even though the factory had organized a lecture series from a technician on using the new machines, workers still had many problems unsolved. In the meeting workers discussed the problems they had encountered with the new machines. Hearing their voice, the supervisor focused on these problems in her supervisor summary part of the meeting, such as how to avoid leaking machine oil and how to communicate problems to the technicians.

(d) Putting questions in context and activating social roles elicited more responses than asking about individual experiences alone. Asking a question in an utmost concrete way was the most effective in getting responses from workers.
At the first several meetings, we encountered a problem that the workers found it difficult to describe "working strategies" in detail. Some workers thought on a very abstract level and said there were no specific strategies and everything came with some working experience. Several women mentioned that people would know the right gestures with experience but were unable to describe the process further. The researcher asked what a good gesture looked like and how to develop it. Again, the workers were unable to describe it in detail. However, when we put the question in the current context of the group and asked the question in another way, workers were able to understand the question in a concrete level and start the discussion: [4 rd meeting with Saturday's sewing group, July 25] The group switched order from producing a summer dress to a winter baby outfit at the beginning of this week. In the discussion, everyone said something about why switching tasks was hard for them. " A woman then walked close to the machine and showed how she used a mold to help sew a squared patch onto the front of an outfit seamlessly.
Apparently, the second way of asking the same question on "good gestures" was more effective as it created a concrete scenario for the workers to act upon, in particular when the social role of a "more experienced worker" in relation to a "new worker" was activated.
Similarly, when we discussed why product defects occurred and how to avoid them, the workers' first reactions were: "I should be more careful," "It comes with working experience. With more working experience, workers know how to avoid the defects," as two workers said during the discussions. But when the researcher pointed to a specific defective piece and asked how it occurred and how to repair it, workers focused the discussions on working strategies to fix that specific piece as well as other more general issues on product defects.
Workers were not used to thinking analytically about problems unless provided a concrete example or scenario. It may reflect a cross-cultural difference in people's thinking style (Nisbett, Peng, Choi, & Norenzayan, 2011). The factory workers are used to thinking holistically rather than analytically. Thus in the actual intervention, we always asked a question in the most concrete way possible and activated the context and social relations surrounding the target question to help workers engage in extended discussions.

Supplementary Note 3 -Randomization
We conducted Study 1 at a multinational textile factory in China. We sampled all work groups (N = 65) from the factory's sewing departments where workers are organized in groups.
Employees in the sewing work groups work on their own tasks, which are related to their coworkers' tasks. For example, one worker may be in charge of sewing the sleeves of a hoodie while another is in charge of sewing the hood pieces. Each work group has its own supervisor who oversees group members' work. The factory requires all work groups to have a 20-minute morning meeting before the start of each workday, in which the supervisor summarizes the previous day's work performance, recommends individual and group work strategies, and announces goals for individual workers. Workers rarely transfer to a different group after they are hired. Individuals in all groups provided informed consent during a recruitment phase one month before the experiment's commencement (no refusals were observed). 6 We randomly assigned the 65 work groups (N workers = 1,752; 93.6% female; mean age = 32.5 years, ranging from 18 to 53) to participate in a weekly morning participatory meeting (referred to as participatory meetings condition or treatment condition), or to have an observer attend the usual morning meeting (referred to as observer condition or control condition) once per week for six weeks. 7 To randomize, we used a non-bipartite matching scheme (Beck, Lu, & Greevy, 2015) to balance and minimize observable differences between the groups ex ante.
Within each department, work groups were matched based on their group size, average worker 6 We made oral public announcements in the sewing departments to invite all sewing workers to a study called "worker experience in the factory," with the cooperation of the factory's human resource department. Workers were specifically told that "researchers are not part of the factory but are coming to learn management practices and offer new technologies on work-related issues. All of you are invited to take part[…]Participation is completely voluntary." 7 The building structure allows for little communication between sewing departments and among work groups. Workers spend most of their time in their own group's working area on the production floor during work, and have little communication with other groups during and after work. Thus, we have few concerns about spillover of treatment to control groups. productivity, normal working hours, and overtime working hours. We took into account qualitative comments from departmental supervisors on the leadership style of each supervisor to fine-tune the group matches prior to randomization (see matching code below). When groups were paired within departments, we randomly assigned one group in each pair to the participatory meetings condition, and the other to the observer condition. We assigned the three groups that did not achieve a match within their departments to the observer condition, as desired by the factory (results are robust to the exclusion of these three unmatched groups; see the supplementary materials). The final sample consists of 31 treatment groups (N workers = 868) and 34 control groups (N workers = 881).
In Study 2, forty academic departments' administrative staff groups (out of 70 total) at a university in the United States agreed to participate in the study during the recruitment phase.
Each group was comprised of an academic manager (the supervisor) and staff members who directly report to the manager (the workers). Supervisors agreed to participate in the study, and we moved forward with their participation if a majority of their staff also consented to participate.
The median size of the administrative group was 6. The staff members' work is relatively independent, including job roles such as graduate and undergraduate administrators, finance managers, and event coordinators.
As in Study 1, we randomly assigned the 40 administrative groups to participate in a weekly morning participatory meeting (participatory meetings or treatment condition) or continue with their status-quo meetings (control condition). Supervisors and workers provided written informed consent before the experiment's commencement. To randomize, we used the same non-bipartite matching scheme in Study 1. Staff groups were matched based on their academic division (Natural Sciences, Social Sciences, Humanities, or Engineering), program category (full-degree programs or interdisciplinary centers), staff headcounts, faculty headcounts, and initial level of enthusiasm shown by their RSVP and attendance at our recruitment events.
When departments were paired, we randomly assigned one department in each pair to the participatory meetings condition, and the other to the control condition. Eight staff groups (N treatment = 5, N control = 3) dropped out of the study after randomization but before any data collection. 8  Here, we replace all the causal inferences in the results section from multiple regressions with between-group t-tests, treating group as units of analysis.

Attitudinal Changes toward Authority and Justice
General attitudes toward authority. The mean score of attitudes toward authority for the whole sample was 4.05 (SD = 0.37). This value accords with the "slightly agree" point of the scale, which indicates that, on average, workers tended to slightly agree with statements asserting complete obedience and respect for authority without question. However, workers in the participatory meetings condition reported significantly lower score in attitudes toward authority (M = 3.87, SD = 0.32) than workers in the observer condition (M = 4.23, SD = 0.33, t = 4.36, p < .001, d = 1.08). Participatory meetings changed participants' attitudes toward general authority such that treatment workers registered as less authoritarian on a traditional scale of authoritarianism.
These results were consistent when we analyzed each item within the index. One month after the end of the experiment, treatment workers who took part in these brief participatory meetings were significantly less likely to endorse obedience and respect for authority as the most important virtues children should learn (t = 2.86, p < .006, d = 0.71), less likely to agree everybody would be better off if people would talk less and work more (t = 5.04, p <.001, d = 1.26), and less likely to believe in supernatural power (t = 2.30, p = .020, d = 0.59).
Belief in a just world. For attitudes and perceptions in generalized justice, the mean score for the whole sample was 3.98 (SD = 0.23). This value is also just below the "slightly agree" point of the scale, which indicates that, on average, workers tend to slightly agree with statements asserting belief in a just world. Workers in the participatory meetings condition Robustness check without the three unmatched groups in the observer contidion.

Supplementary Note 5 -Robustness checks with 62 groups (intact pairs).
General attitudes toward authority. The mean score of attitudes toward authority ( = .52) for the whole sample was 4.03 (SD = 0.36). This value accords with the "slightly agree" point of the scale, which indicates that, on average, workers tended to slightly agree with statements asserting complete obedience and respect for authority without question. However, workers in the participatory meetings condition reported significantly lower scores in attitudes Joint Significance Test Because we tested the average treatment effects on multiple dependent variables, we conducted a joint significance test on the null hypothesis that the coefficients on average treatment effects from all the multiple regressions are jointly nonsignificant. As predicted, there was a jointly significant difference of the average treatment effects between workers in the participatory meetings condition and observer condition, F(1, 58) = 7.59, p < .001.
As a robustness check, the multivariate analysis of variance (MANOVA) was conducted to assess condition difference on all dependent variables recorded. The multivariate effect was significant by conditions, F(1, 63) = 7.89, p < .001. Thus, we conclude that participatory meetings significantly changed workers' attitudes compared with workers in the observer condition.
Supplementary Note 6 -Robustness checks including new workers.

Attitudinal Changes toward Authority and Justice
General attitudes toward authority. The mean score of attitudes toward authority ( = .52) for the whole sample was 4.00 (SD = 0.37). This value accords with the "slightly agree" point of the scale, which indicates that, on average, workers tended to slightly agree with statements asserting complete obedience and respect for authority without question. However,

Joint Significance Test
Because we tested the average treatment effects on multiple dependent variables, we conducted a joint significance test on the null hypothesis that the coefficients on average treatment effects from all the multiple regressions are jointly nonsignificant. As predicted, there was a jointly significant difference of the average treatment effects between workers in the participatory meetings condition and observer condition, F(1, 58) = 8.06, p < .001. As an additional robustness check, we conducted a multivariate analysis of variance (MANOVA) to assess treatment differences on all dependent variables recorded. The multivariate effect was significant by treatment, F(1, 63) = 8.45, p < .001, partial that participatory meetings significantly changed workers' attitudes compared with workers in the observer condition.

Supplementary Note 8 -Survey Materials
In Study 1, due to workers' time constraints, we conducted the survey during the 1-hour meal time and provided a free meal as incentive. On average, 4-5 groups (not segregated by treatment assignment) gathered for a mealtime survey session. The survey was described as "part of a research project that investigates worker experiences in the factory and in the larger society." Workers were assured that there were no right or wrong answers, that the survey was confidential, and specifically that researchers would not share individual answers with the factory management. To further ensure confidentiality, participants put their completed answer sheets in a sealed envelope and put their envelope in a transparent box with all other surveys. No identifiable survey data were shared with the factory, as agreed in advance of the study.
A considerable proportion of the factory worker population is illiterate, so a standard written survey was not be feasible. Prior in-depth cognitive interviews with an independent sample of factory workers guided the development of our survey instrument, which combined oral questioning in a group setting with workers checking boxes on individual answer sheets.
Researchers read each survey question aloud, and participants marked their responses on answer sheets (featuring boxes for "yes" or "no", or numbers from 1-6 on a Likert scale of agreement that was well-explained in advance). This procedure obviated the need to read or write Mandarin characters. Participants were encouraged to interrupt the researchers for question clarification, but were not allowed to discuss their answers or look around each other's answer sheets during the session. If people would communicate less and work more, everybody would be better off.
Every person should have complete faith in some supernatural power whose decisions he obeys without question.
Belief in a just world ( = .32) Although evil men may hold political power for a while, in the general course of history good wins out.
It is often impossible for a person to receive a fair trial in China.
By and large, people deserve what they get. got a chance, or would they try to be fair? Please show your response on this card, where 1 means that "people would try to take advantage of you," and 6 means that "people would try to be fair".

Conflict
In your mind, to what extent do the following social groups have conflict with each other?
The rich and the "ordinary" The capitalists and the working class If people would communicate less and work more, everybody would be better off.
Every person should have complete faith in some supernatural power whose decisions he obeys without question.
Belief in a just world ( = .21) Although evil men may hold political power for a while, in the general course of history good wins out.
It is often impossible for a person to receive a fair trial in the US. (R) By and large, people deserve what they get. Supplementary