Democracy and worker representation in the management of change: Lessons from Kurt Lewin and the Harwood studies

1939‒1943

Lewin is invited to Marion by Alfred Marrow in 1939, owner of Harwood and a student of Lewin. Marrow describes the difficulty that while wages are higher at Marion than elsewhere in the area, the factory is continuously dogged by low productivity. Lewin suggests reducing forces that cause worker strain, such as pressure from supervisers and that workers should be encouraged to think of their job targets as achievable. On Lewin’s suggestion, against local opposition, Marrow hires some highly skilled workers recently made redundant from a factory about 40 miles away and following this the output of the Harwood workers increases (Alden, 2012: 55; Marrow, 1969: 143). Lewin convinces Marrow to hire his student Alexander Bavelas to research how job targets might be made more achievable by enabling the direct involvement of employees. This fits Lewin’s wider aim to test the efficacy of democratic leadership over autocracy. One of the first experiments conducted by Bavelas involves workers deciding their output as a group, rather than being dictated to by management. Bavelas’ experiment runs from December 1942 until August 1943 and is proclaimed a great success by Lewin who argues that this indicates ‘a substantial permanent increase in production created in a short time by certain methods of “team decision”’ (Lewin, 1945: 308–309). Another experiment by Bavelas using ‘pacing cards’ that runs more or less concurrently with that on group decision is also judged successful.

1944‒1946

In 1944, Bavelas is replaced by John French Jr, another of Lewin’s students. French writes to Lewin for guidance on research design (French, 1944, in Highhouse, 2007: 337), then sets about replicating Bavelas’ studies. By February 1944, it transpires that Bavelas’ ‘permanent’ success in raising productivity has failed to materialize in the rest of the factory, following Lewin’s report that production at Marion has fallen quite steeply (Alden, 2012: 57; Marrow Papers, 1948). Lewin advocates opening up communications between management and workers and the cultivation of a more democratic style of management. French (1945b: 325) is also downbeat about the situation in 1944, referring to the ‘absentee owner’ (Marrow) and describing management as remote, authoritarian and uninterested in the views of operators. However, he also reports that in 1943 Bavelas eventually gained approval for a change in the structure of the organization involving the establishment of a labour management committee as a common forum for management and employees (French, 1945b: 329):

The dynamics of this method can be illustrated in the creation and functioning of a Labor-Management committee in a factory where there was no union or other group representing labor. Before the organization of this committee under the leadership of the plant psychologist, the management was not aware of the needs of the workers, nor inclined to do anything about those needs which were not obviously related to production. (French, 1945b: 329; emphasis added)

Harwood’s top management first rejected Bavelas’ proposal, ‘fearing that the committee might “stir up trouble” and bring them closer to unionization’ (Alden, 2012: 59–60; Marrow Papers, 1940). However, they eventually acceded to this so long as the committee refrained from discussing substantive matters such as wages or hours, which were deemed as the jurisdiction of management (Alden, 2012: 60). French notes that when he first headed the committee the representatives requested that the workforce be allowed to devote one minute each day to silent prayer for members of the military. He comments, ‘[n]aturally the initial reaction of management was not favourable to this loss of working time’. Management finally agreed to five minutes’ prayer one day a week. The committee started by considering issues of minor importance but eventually was consulted about more substantive issues such as wages, overtime hours and vacations (French, 1945b: 330). We return to discuss this committee later in relation to the unionization of the Marion factory.

In 1944, French conducts six training workshops to educate supervisors in participative techniques (French, 1945a). He also sets about replicating Bavelas’ striking findings from the group decision and pacing cards experiments. These are described as a resounding success, with ‘incredulous workers discovering that management is sincere in acting to cooperate with them on an equal footing’ (Lippit and Hendry, 1945: 316). French (1950) demurs, candidly admitting his failure to replicate Bavelas’ findings, obtaining ‘very different’ results that he puts down to the difficulty in learning a democratic leadership style, reporting that results improved along with his skill (French, 1950: 87). Marrow, impatient for results, recommends to French in July 1944 that he conduct group meetings with workers to help reduce absenteeism and relay propaganda messages to them over the loudspeaker system (Highhouse, 2007: 338).

The Marion factory has been lobbied by unions since its inception in 1939 (Marrow, 1966) and in March 1945 the United Mineworkers union achieves a vote of 47% (Marrow, 1966). However, in August 1946, ‘Pajamas and the Ego’ appears in Fortune magazine to describe Marion as an ‘industrial heaven’ where ‘labor unions seem to be unable to make headway’ (Fortune, 1946: 140). Gomberg (1966a) of the ILGWU accuses Marrow of authorship and ‘press agentry’ (1966b), a claim denied by Marrow (1966) and reiterated by Gomberg (1966c). Daniel Bell states that Marrow ignores Lewin’s advice to keep the experiments at Marion anonymous, deciding to trumpet the success of Bavelas’ group decision experiments in increasing output and escaping unionization. Bell continues that when this comes to the attention of the ILGWU, they send an organizer to the factory who distributes leaflets entitled, ‘To the Workers of Harwood Manufacturing, Do you know that you are being used as guinea pigs?’ (Bell, 1960: 463, note 61). Gomberg’s account, which tallies with Bell, is that following publication of the article the union dispatches an organizer to Marion, which, ‘was organized in a record short time’ (Gomberg, 1966a: 32). While this is disputed by Marrow (1966), the period between the worker vote for membership of the ILGWU and formal ratification of the agreement in Gomberg’s office on 31 December 1946, is just four months (Gomberg, 1966b). Henceforth, the union is formally authorized to represent the workers interests by negotiating piece-rates (Gomberg, 1966c).

1947

Kurt Lewin dies in February. The inexperienced Lester Coch is appointed as Personnel Manager in place of John French, who takes the role of ‘consultant’. French and Coch attend the first National Training Laboratory (NTL) workshop at Bethel in June and shortly afterwards they embark on the most significant experiment to date. There is disagreement as to the precise timing of the resistance to change experiment. Marrow (1966) places this ‘in the late Fall of 1947, nearly a year after the employees had chosen the International Ladies Garment Workers Union to represent them’ (see also Marrow, 1969: 150). This is most unlikely given that the timings provided by Coch and French suggest that the experiment ran for 167 days (Coch and French, 1948: 520, 522, 523), which would have carried them from the beginning of October till mid-March 1948. Bennis (1966) places the date in June 1947, ‘not until about a half-year’ after the signing of the union agreement at the end of December 1946. Sometime in late June is most likely, given that Coch and French (1948) state that the experiment was conducted sometime after they returned from the NTL workshop in early June. Why does Marrow persist in seeking to push the date to the late Fall of 1947? We discuss below that this may have been because he sought to distance the experiment from the industrial action that occurred at that time.

The overall climate at Marion in 1947 described by Coch and French (1948) is markedly different to that described by French (1945b). Where French had relatively recently described a remote and authoritarian regime, Coch and French (1948) state that labour relations were good from the day the plant opened; Marion is a ‘liberal and progressive’ factory that provides all kinds of employee services, which ‘has enjoyed good labor relations’ and places a ‘high value on fair and open dealing with the employees’, who are, ‘encouraged to take up any problems or grievances with the management at any time’ (1948: 513).

Four groups of workers are chosen for the experiment. Managers first modify the job and set a new piece-rate. The control group of 18 pressers are treated in the ‘normal’ manner by being called into the office and told that the job modification and change to the piece-rate are necessary because of competitive pressure, then given the opportunity to ask questions. Experimental Group 1 consists of 13 pajama folders, whose structure mirrors a representative democracy where members are given more detailed information and asked to nominate representatives to participate in designing the new jobs and in setting the new production rates. Experimental groups 2, 3 and 4 consisting of eight and seven pajama folders and two sewing machine operators respectively, participate fully in direct democracy, designing their new jobs and setting new production rates. In contrast to the control group, the need for cost reduction is vividly conveyed to members of the experimental groups who are told that the retail price of a finished garment is half it was in the previous year (Coch and French, 1948: 521).

Bennis (1966: 36) writes that the workers enjoyed ‘due process . . . and that the union completely supported the experimental program’. Gomberg denies this, stating that: ‘[t]he union not only did not cooperate in the subsequent experimentation, it was unaware of its existence’ (1966b: 35). Marrow confirms Gomberg’s assertion writing, ‘[t]here was no publicity and no announcement of the research. It was carried out as part of normal plant procedure’ (1966: 36), although he later states as ‘improbable’, Gomberg’s assertion that the union was not informed (1966: 37). Gomberg (1966c: 48) reiterates this claim. While the first draft of Coch and French (1948) mentions that the workers are represented by a strong union, this is excised from the final draft (Alden, 2012: 66), where the union is mentioned once in relation to a dispute over status.

The choice of the 18 hand pressers as the control group is significant, as this group is singled out for mention by Lewin who died some months before the resistance to change experiment. This group is described as exhibiting particularly strong ‘we-feeling’, restricting production to 50 units per day and scapegoating a member who produced too much. Lewin notes that when the group was disbanded this ‘nervous’ worker achieved rates as high as 92 units per day as an individual (Coch and French, 1948: 520; Lewin, 1947: 26–27). Coch and French (1948) report that almost from the start of the experiment the control group expressed hostility to the supervisor and methods engineer, restricting production and filing grievances over the piece-rate, ‘but when the rate was checked it was found to be a little loose’ (1948: 521). They report that 17%, or three out of the 18 members, left during the first 40 days (1948: 522). The group was broken up after 32 days and its members reassigned to new jobs scattered throughout the factory (1948: 523). Two and a half months later, the 13 pressers remaining of the original 18 (we are not told what happened to the two unaccounted for) were brought together for a second experiment, involving a pressing job of comparable difficulty to the first. Coch and French (1948: 524) report a dispute relating to ‘some anxiety over their seniority status . . . that is resolved in a meeting of their elected delegate, the union business agent and a management representative’. This is the only time the union is referred to in their paper. The authors report that this time the group was subjected to the ‘total participation’ condition and quickly achieved the desired efficiency (1948: 524).

From the ILGWU perspective, Gomberg recalls a similar sequence of events but with a more serious outcome. In the early days of unionization, he is contacted by Grace McWhorter the ILGWU representative at Marion who reports that in her first experience of processing rate grievances, Lester Coch describes her as having ‘the brain of a cockroach’, telling her he wants to deal with Gomberg direct (Gomberg, 1966c). Gomberg describes that, on arrival at Marion, Coch shows him charts of ‘Lewinian field vectors’, informing him that the aggressiveness of workers is manifest in their restricting production. Coch suggests two ways of solving this issue, either by raising the disputed piece-rate, or for Gomberg to cooperate with Coch. In Gomberg’s account (1966c: 48) he informs Coch ‘in no uncertain terms that we were not interested in his concept of cooperation in which we work together to achieve his exclusive objective. The rate dispute was settled’. Gomberg (1966c: 48) notes that this did not settle matters as ‘tensions grew until the stoppage of work in “Utopia”, which led to the meeting between Rolnick of my staff and Marrow in Marion, Virginia’. Gomberg admits that ILGWU officials did not at first understand the context of the experiments at Marion:

When the union officers, in the early days of unionization made the mistake of behaving like statesmen toward this kind of manipulation (which they did not understand completely) the factory elected an aggressive shop steward who damned both the union’s officers and the plant management. Dr. Marrow, the company president, came down to visit the plant with a staff member of the union’s engineering department and departed hastily when he saw the violence of the workers’ feelings. (Gomberg, 1966a: 34)

The imputation is of an unofficial stoppage led by a worker angered by the union’s lack of responsiveness to the ongoing experiments involving changes to the piece-rate. Marrow (1966: 37) provides a different explanation of what appears to be the same sequence of events:

Alas the facts are much less dramatic. In protest against a local union official, a small group of workers began an unauthorized walkout. Top-level union officials, distressed by the illegal wildcat strike, telephoned Harwood executives (including Marrow who was at the plant) to give assurances that the walkout would be ended immediately. It was an intra-union matter which the union was going to handle. The union did.

When stating that the union was not informed of the experiment, Marrow (1966: 36) omits to mention that the experimenters changed the piece-rate: ‘The aim was to see if technological changes in job methods could be introduced without the usual manifestations of hostility and loss of production’. Indeed, he makes no mention of the piece-rate at all, which is odd given that he had earlier acknowledged that grievances were raised in relation to the changed piece-rate (Marrow, 1957). In relation to a work stoppage, Marrow (1966: 37) denies that ‘observation was arrested at the most crucial part of the Coch-French experiment’s results’, ‘that the investigation was therefore incomplete’ and that, ‘it was unfortunate that the investigators did not stay at the plant location to see what happened after they ceased their observations’.

Michael Rose (1988: 172) states to the contrary that workers did strike over a union dispute at this time and that this involved a large number of workers, including members of changed experimental groups. Walter Bennis (1966: 36) accepts that there was a dispute during the resistance to change experiment but confines this to workers participating in the experiment: ‘we are talking about only 28 workers out of 600’. In fact, 46 workers participated in the experiment (experimental group 1: 13; experimental group 2: 8; experimental group 3: 7; control group: 18), an error amended by Marrow (1966: 36). At almost 8% of the workforce, this is far from an insignificant proportion, amounting to over 9% of all female employees. Contrary to Bennis’s argument, fallout from the experiment is likely to have had a considerable effect on other employees. At the very least, it is foreseeable that different justifications for the cut to the piece-rate provided to each of the experimental groups would have circulated widely throughout this relatively small plant, resulting in speculation, confusion and consternation among the workforce.

Epilogue

John French continued his association with Harwood and with Marrow for many years. However, Lester Coch left in 1948, the year that his famous study with French was published. Gomberg and Marrow agree on two things in the course of their exchange; firstly, neither mentions that Coch left the factory in 1948; and secondly, they agree on the excellent suitability of Coch’s replacement, Gil David. According to Marrow (1966: 37), David, who was appointed as personnel manager on French’s recommendation, ‘established close working relations with union officials and kept them informed of his projects, research programs and experiments’. Contrasting David with Coch, Gomberg (1966c: 48) describes relations with the latter as cordial, ‘his conduct and behaviour is exactly what was to be expected from any intelligent manager subject to collective bargaining constraints’. Gomberg (1966b: 35) raises no objection to replications of the resistance to change experiment at different factories of the Harwood group (French et al., 1958.) and later in Norway (French et al., 1960) because the plants were unionized and the unions were kept fully informed. These experiments received much less attention than Coch and French (1948), perhaps because they were less successful in proving their hypotheses.

One single source of legitimate authority?

Lewin wrote that the precise way in which democratic principles are implemented is contingent on the aims specified in local contexts such as the factory, business, community centre or school, each of which is governed by a specific objective, which in the context of business organization is specified by management (Lewin, 1944: 200; 1945, 1947). Just as the manager cannot ignore the reality of the market, the psychologist cannot ignore those with the power to set the overall goal of the organization (Lewin, 1945). John French states in relation to the experiments at Marion that ‘[t]he most important method – it is almost a principle – is that the psychologist must work explicitly within the framework of the major goals of management. He has neither the position nor the authority to impose more democratic forms of managing’ (French, 1945b: 325). Moreover, the experimenter is ‘necessarily constrained’ to demonstrate the greater efficacy of democratic methods in achieving management goals, to increase productivity, reduce costs and avoid labour trouble. (French, 1945b: 326; see also French, 1950: 91). Coch and French were themselves managers, and although workers who were selected for the full participation condition in the various experiments conducted at Marion were allowed to set their own goals, these were always aligned with the management goal for higher productivity through increased efficiency. There was no commensurate increase in compensation.

A unitarist versus plural view of organization?

In the context of the appeal for a return to Lewinian values to provide an ethical footing for organizational change based on employee participation (Burnes, 2009), it seems curious to argue that Lewin’s understanding of democracy might be antithetical to the interests of workers. Yet, Lewin (1945, 1947) does not justify democracy at Harwood on ethical grounds, but by arguing for its superior efficiency over autocracy and laissez-faire modes of organization. Here, Lewin ought to be understood within the context of his time. His justification of democratic leadership as more efficient than autocracy or laissez-faire chimes with the arguments of Mill, Schumpeter and Lipset, which antecede the Business Case for democracy (Johnson, 2006: 255). In the Handbook of Social Psychology, Morawski and Bayer describe how ‘[a]mongst historians there exists fair consensus on a reigning social psychology of this moment as one of an overriding sensibility of social engineering or a “psychotechnology” in the service of a “liberal technocratic” America’ (Morawski and Bayer, 2003: 232). Lewin’s discussion of the Harwood experiments is entirely technical, focusing on how to improve worker productivity (Lewin, 1945, 1947: 24–34). Indeed, the first of the six programme areas identified by Lewin’s group at MIT was to understand why ‘group enterprises [were] so inefficient or ineffective in getting things done’ (Marrow, 1969: 184).

The Lewinians’ deep commitment to prove the efficiency of direct democracy is evidenced by Bennis who seeks to justify its use in the Harwood studies: ‘we view democracy as applicable not because of some vague yearning for human rights but because under certain conditions it is a more “efficient” form of social organization’ (Bennis, 1966: 35). Lewin places strong emphasis on democratic leadership to confront the problem of we-feeling, or local loyalty to the status quo in a group. Democratic skill requires strong leadership: ‘The leader in a “tough democracy”, embarking on a programme of democratic re-education, must “have power and be able to hold onto power” and be “rather forceful”’ (Lewin, 1945: 304). Effective change cannot occur unless there is a ‘felt need’ by all those concerned, such that no one group should dominate and everyone should play a full and equal part in the process (Lewin, 1947). By means of open and honest discussion, the aim is to foster belongingness and group cohesion through consensus.

Some are concerned that the scope accorded to the leader by Lewin can result in the abuse of democratic participation as a means of engineering the outcomes desired by those in power (Billig, 2015; Graebner, 1986, 1987). One difficulty is that Lewin does not offer any idea as to how to differentiate strong democratic leadership from the ‘excessive strength’ that ‘might tip the balance towards improper pseudo-democratic, manipulative, or authoritarian methods’ (Graebner, 1987: 143). It is argued, in the absence of proper representation, that democratic methods can operate as a control technique for management. Billig (2015) summons attention to how apparently neutral phraseology of the kind used by Lewin and his associates, such as ‘would you like to? . . .’) conveys a ‘preference structure’ for agreement, such that agreement is invited and disagreement can be problematic (Pomerantz, 1984). Lewin himself advised in relation to the framing of facts that ‘[t]he reality is presented correctly, but [only] those aspects are brought to the fore which are linked with the psychological situation of the person in question and are favourable in bringing about permanent motivation’ (Lewin, 1948: 134, in Van Elteren, 1997: 347–348).

How were the facts framed to the workers at Marion? Were they provided with the knowledge to enable them to make a free decision based on adequate information? Was there really full and frank discussion in the spirit Miriam Lewin (1987) describes, where operators could question the provenance of the information provided by the researcher-managers? Workers were granted autonomy and free choice in determining their output, so long as this conformed to the overall goal of increasing productivity, with no necessary increase in compensation. In the Harwood experiments that were supposedly exercises in self-management, managers set clear limits as to the required minimum (Van Elteren, 1997: 346–347). Coch and French (1948) framed the problem for the experimental groups as dramatically as possible by asking them to compare two garments: ‘one was produced in 1946 and had sold for 100% more than its fellow in 1947’ (1948: 521). The workers then participated in designing their new jobs (1948: 522). There is no mention of questioning or discussion of the reasons offered for changing the job and the piece-rate, but rather the ‘facts’ were simply presented to the workers by the researcher/managers. Nor was there any indication that the researchers might change their opinion or modify their goals if the workers decided on a lower production target.

Imagine the situation of the worker who has recently voted for a union to represent her interest by negotiating piece-rates. Managers summon her work group to a meeting where they are informed that the job has been simplified and the piece-rate cut. It is little wonder that she will then file a grievance. Regarding the experimental groups, it is insufficient to state that their members were free to make a decision. This is because a free decision depends crucially on the manner by which information is framed and presented to the individual. It seems beyond dispute in the context of a decision to alter a job and its related piece-rate that a union official trained in work study techniques and with industry-wide knowledge of piece-rates and productivity will be much better equipped to discuss these issues and to raise appropriate questions with management than will a poorly educated line worker. For instance, they might seek to check the veracity of the information presented, such as the vivid example used by Coch and French (1948) where they stated that an identical garment produced in 1947 sold for half the price of that produced in 1946. A union representative would have found no evidence of a general fall in prices, never mind such a dramatic fall as described by Coch and French (1948). The Consumer Price Index rose by 17% between June 1946, just after the discontinuance of wartime price controls and March 1947, just before the Coch and French (1948) experiment (United States Department of Labor, 1949). Expenditure on clothing increased from $247 per head in 1941 to $385 in 1944 and stood at $437 by 1950 (United States Bureau of Labor Statistics, 2015: Tables 1–3). Employment in manufacturing was high with a low layoff rate and high quit rate reported across US manufacturing in 1946 against the background of a tightening labour market (US Bureau of Economic Analysis, 1946: 3). Average hourly earnings for manufacturing non-durables increased from 80 cents per hour in 1943, to over one dollar per hour in 1947 and one dollar 20 cents in 1948 (Goldstein, 1960). While sales of nightwear such as pyjamas may not have been as strong as other sectors (Atkinson, 1947), our union official would have had no shortage of data to enable them to reframe the facts presented to the workers by Coch and French (1948).

Selectively framing the facts in a manner conducive to managers as occurred in Coch and French (1948) is described as a trick, representing a possible misuse of democratic participation (Lawrence, 1969: 51; Van Elteren, 1997: 348; Van Strien, 1978: 293). This is likened to the ruse resorted to by parents who use ‘democratic’ methods with their children in conformance with their own pre-ordained decisions (Graebner, 1986: 151; Van Elteren, 1993). Billig (2015) draws a parallel between this technique and the class discussions used today by schoolteachers, which can hardly be called democracies because the teacher is in control and retains power even if this is exercised subtly. Miriam Lewin contests this analogy, arguing that Kurt Lewin espoused a vision of democratic leadership where ‘the leader clearly and openly puts forward his or her own beliefs and suggestions but with genuine respect for the opinions of others, and a willingness to modify goals or plans in the light of followers’ views’ (Lewin, 1987: 131). She distinguishes democracy from pseudo-democracy, where ‘discussion, criticism and arguing are permitted but the real goals are always set in advance by the administration and the students know that they have to discuss until they hit the “right thing”’ (Lewin and Lewin, 1941, in Lewin, 1987: 135). However, the researchers at Marion had the power to frame the agenda for discussion in line with their aims as managers. They were not prepared to permit criticism of their goals, nor indeed to change their goals, but rather limited their vision of democracy to discussion of the means to attaining them. In this context the analogy with the controlling parent or teacher seems apt and indeed a degree of coercion is evident in relation to the control group whose members were scattered throughout the factory by management following the first part of the experiment, which they would have construed as punishment (Gardiner, 1977).

Gomberg’s pluralist explanation of organizational democracy is very different from that espoused by Lewin and his associates. Gomberg argues that democracy is a means for distributing power to ensure a balance of interests so that no one interest can dominate the others and is best served in an institutional climate where no single institution can claim a full personal commitment from any individual (Gomberg, 1966a: 30). In the absence of union representation, the apparently full democratic participation of young, poorly educated and inexperienced workers at Marion, coached by educated experts in line with a management agenda amounted to involuntary manipulation (Gomberg, 1966a: 33). He is flummoxed that behavioural scientists converted to participative democracy can fail to understand what is in his understanding democracy’s most fundamental tenet, that democracy cannot exist where management is free to give and take without any countervailing force. Kariel (1956) makes the related point that Lewin’s procedure, which aimed to create group consensus works against traditional liberal-democratic institutions based on the reconciliation of different interests. Moreover, Lewin’s focus on efficiency fails to take technocratic power into account because there is no place for the necessary democratic external checks and balances placed on the autocrat from outside the organizational boundary (Gomberg, 1966a: 32). The appropriate representative in the context of Marion was the union, which was not consulted (Gomberg, 1966b).

Attitude to union representation

Bell’s (1960) contention that Marrow authored the Fortune (1946) article that claimed the experiments at Marion helped forestall unionization, must be set aside for lack of confirmation. However, Gomberg argues that the aim to forestall unionization might well have influenced management to engage in the studies in participative democracy, that in the USA management only became interested in ‘human relations’ and ‘behavioural science’ following the passage of the Noris LaGuardia Act in 1932, because they could no longer resort to using tactics such as ‘yellow-dog’ contracts, which had allowed companies to sack employees who agitated for unionization (Gomberg, 1966a: 34). While there is no mention of a union in the index to Marrow’s (1969) book on Lewin, Marrow earlier argued that unions effectively limit direct democratic participation by focusing narrowly on the discussion of grievances (Marrow, 1950 in Alden, 2012: 7). The hostility of top management at Marion towards unions is better documented in relation to their initial objection to Bavelas’ initiative for the formation of a labour-management committee, motivated at least in part, by fear that this would lead to outright unionization. The factory had been lobbied by three unions since it opened in 1939 (Marrow, 1966), raising the possibility that management’s eventual agreement to a workers’ council not only furthered the democratic agenda pursued by Bavelas and French, but also constituted their attempt to keep the unions out (Rose, 1988: 172).

What then of the researcher-managers? The tendency for psychologists in the USA to ignore unions has been remarkable (Stagner, 1981: 321). There is general agreement that psychologists at the time of the Harwood experiments prevailingly worked to further the interests of management against organized labour (Baritz, 1960; Gordon and Burt, 1981; Haslam, 2004; Huszczo et al., 1984; Shostack, 1964; Zickar and Gibby, 2007). In the wake of the Wagner Act, management turned to industrial psychologists to screen employees for potential agitators and to identify areas of organizations that offered potential for unionization in addition to focusing on key areas of concern to management such as productivity, turnover and job commitment (Zickar and Gibby, 2007: 66). A number of commentators refer to the ‘bitter’ battle that was ongoing at the time for union representation for industrial workers. One argument is that Lewin offered a practical solution to mediate between the antagonistic forces of management and unions (Ash, 1992: 205). Others argue that although not specifically developed as an anti-union strategy Lewin and his associates were either indifferent to unions or left the impression that unions would be unnecessary if management implemented his ideas (Van Elteren, 1997: 348; Wesner, 1995: 59). The latter argument seems most convincing, given that neither Lewin nor his associates mention a union except in the sketchiest of terms. A key tenet is their belief in the superior efficacy of their own version of direct democracy; indeed a key question for Coch and French (1948) was whether direct democracy is superior to representative democracy, for which they reported limited support (1948: 524).

Resistance to change, or defence of a property right?

Fox contends that a unitarist frame of reference acts powerfully to influence interpretation and action in line with a harmonious view of organization. Expressions of disagreement by workers to the aims of management may result in accusations of stupidity, or of not being good team players (Fox, 1966: 372). In the context of the Harwood studies, the concept of ‘resistance to change’ was framed from a managerial perspective (Dent and Goldberg, 1999), leading Rose (1988: 172) to comment that the title ‘announces its management bias with almost disarming frankness’. Given the sensitivity of piece-rates it is insufficient to argue that the dispute was solely about status, although this is what several commentators seize upon. On the other hand Lawrence argues that the researchers’ treatment of the workers in the control group ‘effectively’ communicated that, they were not the skilled and efficient operators that they thought they were, that they were doing the job inefficiently and some ‘outsider’ would tell them how to do the job right’ (Lawrence, 1969: 12).

As mentioned earlier the group of pressers who formed the control group were singled out for mention by Lewin (1947) and Coch and French (1948). A central concern for Lewin (1947) is how to prevent such groups of workers maintaining the status quo by restricting production and scapegoating those who fail to conform. Coch and French (1948) describe resistance to change in relation to the frustration experienced by changed workers immediately after the change and to we-feeling, where ‘a strong psychological subgroup with negative attitudes towards management will display the strongest resistance to change (1948: 519).

Alternatively, from a pluralist perspective, Coch and French (1948) ignore the interests of workers. In this account, the pressers who restricted production were acting to protect their property rights and those of the union (Gomberg, 1961: 120; see also, Mathewson, 1931). This practice, known as featherbedding, contrasts with stretchout, or the injustice that occurs when workers perform more work without additional pay (Gomberg, 1961: 128). In this understanding, without the checks and balances provided by union representation, Coch and French (1948) can be interpreted as a partisan exercise in ‘stretchout’ that effectively worked against the democratic interests of the employees.