The Safety of Work

Ep.98 What can we learn from the Harwood experiments?

Episode Summary

Can an 80-year-old research experiment provide any valuable insight into today’s work environments?

Episode Notes

In 1939, Alfred Marrow, the managing director of the Harwood Manufacturing Corporation factory in Virginia, invited Kurt Lewin (a German-American psychologist, known as one of the modern pioneers of social, organizational, and applied psychology in the U.S.

to come to the textile factory to discuss significant problems with productivity and turnover of employees. The Harwood study is considered the first experiment of group decision-making and self-management in industry and the first example of applied organizational psychology. The Harwood Experiment was part of Lewin's continuing exploration of participatory action research.


In this episode David and Drew discuss the main areas covered by this research: 

  1. Group decision-making
  2. Self-management
  3. Leadership training
  4. Changing people’s thoughts about stereotypes
  5. Overcoming resistance to change


It turns out that yes, Lewin identified many areas of the work environment that could be improved and changed with the participation of management and members of the workforce communicating with each other about their needs and wants.This was novel stuff in 1939, but proved to be extremely insightful and organizations now utilize many of this experiment’s tenets 80 years later. 


Discussion Points:



“The experiments themselves were a series of applied research studies done in a single manufacturing facility in the U.S., starting in 1939.” - David

“Lewin’s principal for these studies was…’no research without action, and no action without research,’ and that’s where the idea of action research came from…each study is going to lead to a change in the plant.” - Drew

“It became clear that the same job was done very differently by different people.” - David

“This is just a lesson we need to learn over and over and over again in our organizations, which is that you don’t get very far by telling your workers what to do without listening to them.” - Drew

“With 80 years of hindsight it's really hard to untangle the different explanations for what was actually going on here.” - Drew

“Their theory was that when you include workers in the design of new methods…it increases their confidence…it works by making them feel like they’re experts…they feel more confident in the change.” - Drew




The Practical Theorist: Life and Work of Kurt Lewin by Alfred Marrow

The Safety of Work Podcast

The Safety of Work on LinkedIn


Episode Transcription

Drew: You're listening to The Safety of Work podcast episode 98. Today, we're asking the question, what can we learn from the Harwood experiments? Let's get started.

Hey, everybody. My name is Drew Rae. I'm here with David Provan. We're from the Safety Science Innovation Lab at Griffith University. Before we ask what we can learn from the Harwood experiments, David, we got to ask, what the heck are the Harwood experiments and why are we talking about them?

David: Slightly different introduction, Drew. If I go back to my undergraduate psychology degree in the 90s, we learned a lot about Kurt Lewin or we're taught a lot about Kurt Lewin and his ideas around change management and his general field theory of how groups of people interact with each other.

It was really popular in psychology (I think) in the 90s, to learn a lot about it. We don't hear a lot about it today, but a close colleague of mine, Ralph Shreeve, hasn't come through the safety science academic world. He's come through organizational development, individual development, and change leadership schools. Kurt Lewin's work is still very heavily referenced in that field. I thought it was a great opportunity just to bring it into the Safety of Work podcast and have a bit of fun with the experiments.

I didn't actually answer your question. The experiments themselves were a series of applied research studies done in a single manufacturing facility in the US starting in about 1939. This organization and this site had what they considered to be a major problem with productivity and employee turnover. The managing director of the company also had a very strong academic bias and went on to complete a PhD or had just recently completed his PhD. He invited the university to come and help him solve his problems.

Drew: This is sounding very similar to the Hawthorne experiments.

David: I think there are a few, and this would have come after that. There is some of this real, it seemed to happen in the first half of the last century where the universities and industry were—I don't know, Drew—maybe in some ways more tightly coupled. The industry didn’t look to the universities to help them solve their problems. The Hawthorne experiment was one of the famous pieces of work as well.

Drew: Is there anything that you'd like to talk about around these studies before we get right into the paper itself? We've talked a bit on the podcast before about organizational science. If you’re coming into safety, there are a few different directions you can come in from engineering, you can come in with psychology, and then there are several different approaches to management. Is there anything to say particularly about this organizational science approach to management?

David: I don't think I've got anything specific to say, Drew. I really liked some of the work in the organizational studies field because it is specifically trying to understand how organizations function. Organizations are a special group dynamic, because there are formal and informal structures at play. I think the field that's specifically trying to understand organizations directly is a very useful field for managing safety inside the companies that Ellison is working.

Drew: We're not actually going to study a paper. We're not going to talk directly about a paper that emerged from these studies. Like a lot of this organizational research, it's hard to find one particular paper that the authors themselves actually produced that gives you a good view of what happened.

As I understand it, most of what we know about the studies really comes from people talking about the life of Kurt Lewin, his contributions, and how he went about things rather than directly from his published work, even though the work itself was so influential.

David: I think because this work was being done on behalf of the organization directly to solve organizational problems. That was always the primary focus. There is no single published piece of work that covers all of these experiments.

You're right, Drew. It appears as though in 1969, Marrow, who was the managing director of the company for over four decades, (I guess) published a book. He published a book on the life and work of Kurt Lewin, because they became very close friends. There was a chapter in that book that basically, from the managing director of the plant's perspective, told the story of all of these experiments.

I didn't have any copy of that book. Actually, I've just gone on and found a secondhand copy. I'll have it shortly, but what we did do was find a paper. I'll introduce it now, Drew. The paper is titled Kurt Lewin and the Harwood Studies: The Foundation of Organizational Development.

The author was Bernard Burnes. At the time, he was at the University of Manchester, but he's now a professor in organizational change at the University of Stirling. The article was published in the Journal of Applied Behavioral Science in 2007. 

Bernard Burnes, I did a bit of looking, because I hadn't come across him before. He's published a number of books on change management and strategic change management. He seems to have a popular book titled simply Managing Change, and it's now in its seventh edition. That seems very widely sold.

Drew: In terms of this article, we can trust that Burnes is familiar with the field and knows what the contributions are that Lewin made to the field. You and I have both read around this topic and a number of different biographical stuff about the study. My first instinct was to look up criticisms of the Lewin Harwood studies just to see what was out there.

David: Let me introduce who Kurt Lewin was and then let's go straight into those. There are a couple of limitations of what we're going to talk about, and then we'll talk about some of the key experiments and what we can practically learn from it.

Kurt Lewin was a German Jew. He emigrated to the US in 1933, at the time, to escape the rise of fascism. He always seemed to be a humanitarian, those who have written about it, and he believed that the human condition could only be improved by resolving the social conflicts like racial, religious, industrial conflicts.

His view that he held, all the way through his life and career was that resolving this social conflict was achieved through the facilitation of group learning processes that involve democratic participation. So that individuals could understand, restructure their perceptions of the world around them, and contribute to decisions that impacted on their lives. His focus of work was never the individual, but always the group.

Drew: He started off as an academic. He got called in to do this work by Marrow, starting off with one just short trip that then basically evolved into almost a lifetime of work. More than a lifetime because the study has continued for several generations of scholars after Lewin's contribution. All it have seemed to say was that it kicked off when Marrow, who ran the Harwood Manufacturing Plant called Lewin out to have a visit and give some advice.

The two guys had already met before. Lewin, when he was doing his own PhD, had talked to Marrow. Marrow was obviously a fairly scholarly type plant manager. They were good friends. Lewin had this opportunity to contribute to some of the problems that the plant was experiencing.

It should be fairly easy to imagine we're talking 1930s rural Virginia. They've opened up a new plant, presumably, because local tax incentives or whatever makes you want to open a plant in the middle of rural Virginia. But they can't get hold of any expert workers because in a new plant no one's got experience in operating in this manufacturing.

So you have 300+ workers, but very few of them have actually done this type of factory style production work before. Most of the workers were young women. First job, low productivity compared to what management was expecting of them.

David: Just a little bit of extra context. You're right. Out of any of the major industrial locations in the US at the time. Just to add some context to the challenge that they faced, this was also a time when World War II was starting in 1939, so that changed the labor market dramatically in the United States, and also, again, for what it actually meant.

When they got all of these trainees who had never worked in these factories before, after 12 weeks of training, they would only produce about 50% of the volume that apprentices would produce in industrial centers in the rest of the US. There was a dramatic productivity challenge which initiated all of this work. 

At the same time, because there wasn't a long history of employment in these types of industries, the employee turnover was significant. Many of the recruits would leave before they'd even finished their 12-week training period.

Drew: It's not just that the workers are new. The management of the plant thought they knew what productivity they could expect out of new workers. They weren't even getting that. The workers were coming in being below what they expected and being washed out before they finished the training.

David: I don't even know how to describe this study, whether it's action research, whether it's applied research. How would you characterize this set of experiments?

Drew: I think what we would call them today would be quasi experiments. One of the reasons why we don't know what to call them is that the method didn't exist. That's Lewin's big contribution. He's actually inventing this way of doing research.

He's got a lot of the form and style of an experiment, but it's not being done in lab conditions. It's not being done in carefully controlled conditions. It's being done out in the wild in an active factory with real workers trying to directly address real problems. Your dependent variable in the experiment is the productivity at the factory.

David: You're right, Drew. We didn't have so much of a language in academia around this industrial research, although it did come after things like the Hawthorne experiments and things like that. The protocols that they designed for this will be considered relatively robust contemporary research methods for this type of work.

They, typically in all of these experiments which we’ll talk about had experimental and control groups. They came up with some objective measures of performance. They maintained verbatim transcripts of some of the meetings and data for transparency and analysis. They were trying to take as much of the laboratory experimental design into their field work.

Drew: It's funny, David. I started off when I was first reading this. Just feeling really critical towards the methods they're applying, I was thinking, these are not fair tests. Their control groups are heavily loaded to fail and their main experimental groups are heavily loaded to succeed. Then I had to just stop and check myself and realize, hold on, they had control groups?

There's so much stuff that calls itself research, even today, that doesn't include any controlled comparison. These are the people who are pioneering this methodology. They really are trying not just to improve the factory, but to improve the factory in a way that has built in tests of its own credibility and built in tests of their ideas, so that they can claim some generalizable results afterwards. They can take it to other people and say, look at what we did. This is why it worked. Here's the general principle we can take into other factories and other work.

David: I think it's a really good point. I intend to say it a few times during this podcast. The context of this is 80 years ago. When we're talking about this, try and think through this podcast, not necessarily with your own lens now in 2022, but your 1939 lens that you will be putting over what we're going to talk about.

Drew: This is my favorite indicator of the time. When we're doing research today, we use so many euphemisms. We talk about your safety climate, organizational culture, and measures of work satisfaction. These people are measuring things like worker aggression towards management. Was this simpler time with simpler language?

David: Yeah, and in many ways, simpler in research designs. I think we'll talk about that at the end. They just got in there and ran some experiments. It's cool what they did.

Drew: Should we jump in with some of the actual experiments?

David: Yeah, we're going to talk about five group decision making. These had titles. I think the last thing I'd say before we jump in is we're going to talk about five different experiments. Burnes who authored this paper said, we really need to think of these as one long study, because it was done in the same organization.

The second study was done after the first study, the third after the second. Because they were trying to solve operational problems, each study built on what they learned and what had happened at the previous one. But also, the previous study had changed the starting point for future studies. We remember these are five studies conducted between 1939 and 1947 in the same single manufacturing plant.

Drew: One very last thing before we go into the studies themselves. Lewin's own principle for these studies—this is I think a direct quote—"No research without action, no action without research." That's where the idea of action research came from. The purpose is that each study is going to lead to a change in the plant. Each change in the plant is going to be something that we are studying to work out its effectiveness.

It's a constantly evolving cycle of making a change in a way that you can evaluate it and lead to the next change. Theorize a bit, lead to the next change, theorize a bit, lead to the next change.

David: Drew, let's dive in and talk about the five experiments. The first one is about group decisions. At the time that this research kicked off, Lewin had received the invitation from Marrow to come to this plant in Virginia and help with these operational challenges that he had.

Lewin had been researching the effects of autocratic and democratic atmospheres on groups of children in school settings. Looking at the atmosphere, this was the term at the time for culture in organizations. I think the first vice president of safety in DuPont in the 1920s talked about the safety atmosphere inside DuPont. We're talking about autocratic or democratic atmospheres or cultures in groups of children and looked at the impact on children's learning and found that the democratic environments created better outcomes for children.

They argued less. They were more productive. They were friendlier to each other as opposed to when the teachers were telling them what to do. That was Lewin's starting point. We thought, okay, well, let's maybe think about how this works in adults.

How this first experiment got designed is we'll get a group of our company's most productive operators, we'll meet with them several times a week. These are brief 30-minute informal meetings. We'll ask this group to discuss what their barriers are to increasing production further.

They started discussing what their individual working methods are. This is the first time that these operators started sharing their working methods with each other. In doing so, it became clear that the same job was done very differently by different people.

The group got to talk about why this was so, what the advantages and disadvantages were of their individual approaches and the approaches of others. And then they identified what changes the company's management could make to improve productivity. The company basically accepted all of those ideas that they came up with that the company could do to fix it.

Drew, the idea of getting a group of people together, and talking about work as done and what's going well, not going well, and what the company can do about it, 80-year-old idea?

Drew: At least, but certainly it was novel for this particular company at that time. Possibly this indicates that this is just a lesson we need to learn over, and over, and over, and over again, in our organizations, which is that you don't get very far by telling your workers what to do without listening to them in return.

David: Sorry, I was a bit too subtle, Drew. That was my joke about the newness of learning teams in safety.

Drew: Yes.

David: I'll be more direct next time. I got these operators together, how do you do your job? We do it like this, we do it like this, so I do it like this. Okay, what's good about the way you do it? What's good about the way I do it? If the company changes certain things, what would make us better?

The company accepted all of those changes. The group then voted on whether they were going to increase their daily output and to what level they should set it out. The management basically just said to the workers, okay, well, if we fix these things, do you want to increase your output? It's important at this time, all of the workers were paid a piece rate. They got paid based on how much they produced.

They gave the workers the option, do you want to increase the output? Over a period of five days, they voted to increase the output from 75 units per hour to 87. When they got that achieved, management asked them if they wanted to increase it any further. They said, yup, let's increase it to 90.

Over a five-month period, the performance of this experimental group compared with the other groups in the factory went from 75 units per hour up to 90. No other group in the factory showed any significant increase in output, albeit nothing was being done in any of those other groups different to normal operations.

The conclusion of that study was that Lewin maintained that the ability of the group to make its own democratic decisions was the key factor. Democratic, remember at the time where we're literally (I think) you're talking about people voting, giving the group the opportunity to vote. Having that process where the group made its own decision was the key factor rather than the discussions that took place.

To test this hypothesis, one of the subsequent researchers in Lewin's team went on to compare the effectiveness of group discussions with group decisions. What he did was hold meetings with two other groups of workers.

In both of those other cases, what they did was only discussed ways to increase productivity, but they didn't vote on whether they should do it. The result was only a slight improvement in their productivity. They weren't actually setting their own goals. The conclusion is not just the group discussion that occurred, a lot of it was down to that involvement in the goal setting process.

Drew: With 80 years of hindsight, it's really hard to untangle the different explanations for what was actually going on here. I don't want to be too critical. Just to be clear, there has been a lot of criticism of this study because there are lots of different ways you can interpret when you select a group of the most productive workers to do this study on versus other workers. This isn't really democratic because management's whole mission is to increase productivity here. The workers are basically voting amongst themselves to agree with what management is asking them to do. All sorts of little nitpicks.

But I think what's really important here is that the conclusion that was taken out of this study at the time was this ongoing idea that we still have in organizational studies, that worker participation in decision-making, and actually giving workers some power over their both conditions and their goals, is very effective in aligning workers with management expectations.

It seems like a bit of a paradox. But if you give people more choice, they're more likely to actually willingly go along with what you are aiming for, as opposed to just dictating what the conditions are, in which case, they're more likely to rebel.

David: I think even more so now than then with that, because at this time, people were paid piece rate. The management had the definite goal to increase production, but workers benefited when production increased as well. Whereas now, today, in many organizations where that particular payment structure doesn't exist at least in the employment setting, there's even less incentive for workers to align with management's productivity goals.

Drew: Yes. What we really don't know is whether given a free choice, those workers would have gotten rid of the piecework payment altogether and ask for a regular salary.

David: Yeah, that would have been a different type of control group, Drew, is there anything else you want to say? We'll do some practical takeaways at the end about all these five experiments, but is there anything more you want to say about that before we move to the next?

Drew: Just it's hard to emphasize the novelty of creating these separate control groups. The idea of the worker consultation and the measuring piece rates, that's exactly what was done in the Hawthorne studies. At the Hawthorne studies, we had no concept of comparing these two other groups, whereas this was set up with deliberate control groups.

The other thing that was novel was just this idea of thinking about the conditions around the workers rather than what management actions. The idea that we would solve the problem, not by changing exactly what management did, but by changing the environment and the group dynamics of the workers, was also very novel.

David: We mentioned the Hawthorne experiments a few times. We're assuming listeners know what it is, but not enough space in this episode, but you'll be able to search for it. If there's a bit of interest in some of these earlier experiments, we can always do a future episode on that study itself.

Drew: Yes. The purpose of this episode, just think of the Hawthorne studies as this, but done much worse.

David: Turn the lights up, turn the lights down, and see what happens. 

Let's move on. The second experiment is titled Self-Management. This is how the individual behavior is a product of the group, not more so than a product of the individual. That is this broad idea that Lewin started work on what he called Field Theory in 1917.

He proposed that group behavior is shaped by this really "intricate field" of symbolic interactions and forces that not only affect group structures, but modify the individual behavior. This was central in his approach throughout his whole career in analyzing individual group and societal behavior. He said, any changes in group behavior stems from changes to the forces within this field.

He went on to talk about force field analysis with his plan, change process, about unfreezing, moving, and refreezing organization. This is the idea that there's this complex dynamic that exists around a group that influences the behavior of the group. Anything introductory that you want to add before we talk about the experiment?

Drew: I think what he's really talking about is almost like the idea about the way you're running a marathon. In a marathon, each individual person's effort is their own effort. You don't get the same drag you get when cyclists are racing. The pace you're able to go at is still subtly affected by all the other people around you. It's much easier to run at the pace of the group than to run faster or slower than the pace of the group.

That's the idea that he's testing out here. If management sets the pace versus letting the workers have some way to collaboratively set the pace, then even though each worker is working individually, still that field around you is what influences your individual behavior as part of that group.

David: What this experiment involves is they got a group of workers. They tried to lower management's control over how fast they worked and increase the workers' control. They gave workers these hourly pace cards. Workers could decide every hour at what pace they wanted to work. That then obviously added up to their daily work rate.

Workers, like I said, were on piece rate, so they had a lot of motivation to do more work. When they were given this opportunity to set their hourly pace themselves, their output increased from 67 units average per hour to 82 units and stabilized at that new level, and the control group remained unchanged.

This is interesting because if a work group has the capacity to go from 67 to 82 per hour, then there was this idea that leadership should just tell the other group to just do 82 units per hour, because clearly it can be done. But this experiment, one of the conclusions of this (as I said) is that telling workers that this is the pace we want you to work at doesn't lead to that individual or that group adopting that goal as their own.

Drew: Part of Lewin's theory was that there were forces that pushed production in the other way. You've got the obvious incentive that you get paid more when you produce more. No one is going to work absolutely, absolutely flat out constantly. 

One of the things that you do to work out what's like a reasonable amount of work to be doing at a reasonable pace, is you see what everyone else around you is doing. If you're working faster than someone else that might tend to bring you a little bit slower because you see their pace as reasonable productivity. Or you might even have a really rational fear that if we start producing 80 units an hour, then management is going to adjust the minimum target. The minimum target is now going to be 80 units an hour, and we're not going to be able to meet that.

Often, there are these subtle perverse incentives as groups interact that can bring things down. What he was trying to do was introduce a group mechanism that could subtly push the group in the other direction, almost like a group nudge to work faster, to encourage their best interest towards that incentive of producing more and getting paid more.

David: I was talking to a group of underground plant operators. They said, do you know what our reward is for a month of record production? That amount becomes the next month's target. It was very cynical. They actually didn't want to push too hard because then, you're right, Drew, they're constantly moving. These workers obviously knew that if tomorrow we wanted to work a little bit slower, that was within our control, We can push a bit harder when we feel that we're able to, and then we know that that's not going to create an expectation.

One of the conclusions, which I thought was important, was that they said that this experiment is not independent of the style of personal leadership in the day-to-day running of the plant in terms of these work group supervisors. You can imagine even setting up these studies. 

You've got an operations leader, a shift supervisor or something who's facilitating this process with the work group, like telling them, you can have these pacing cards, how do you want to work? What support do you need? When are we going to talk?

You've still got a lot of coordination and facilitation happening by line management. What the research is going to say is we can't separate out this study from the influence of the actual leader, the style of personal leadership of these teams. Anything you want to add, Drew?

Drew: Just that we're sure we're talking about 80 years old because so far we've been through learning teams. We're now into frontline leadership skills.

David: Yeah, that's the next experiment. Like we said, this progressed so I said, oh, gee, we've really noticed that when we're doing this thing that it takes a certain type of leadership to facilitate these democratic work groups. What we should think about doing is doing a leadership training experiment.

They showed that this autocratic approach to management that they referenced Taylor's work, it didn't result in increased efficiency but decreased efficiency. There's been a lot of studies since Taylor's work on scientific management that was referenced in an article and referenced in this paper by Rose in 1988, that showed that a lot of this autocratic style of leadership didn't improve productivity. Perhaps, Marrow was aware of all of that work that had happened in the 3 or 3½ decades since Taylor's work and why it specifically went to Lewin. We can never know.

I guess they decided, look, we want to train our supervisors and run an experiment around that. This is the quote, “At Harwood, supervisors tended to be appointed on technical competence and not on interpersonal skills.” I hear that every second week in organizations today.

“Their view was that organizations are cooperative systems, the supervisor needs to actually facilitate that cooperation. If supervisors don't possess the interpersonal skills necessary to gain the cooperation of their subordinates, their peers, and their supervisors, then an organization can't be successful.” Drew, please jump in.

Drew: I just wanted to jump in with how much I love the way they actually did this training. This is a direct quote from the PhD student, the subordinate of Lewin, who was running these training sessions. He starts off the training session by saying, "What we'll try to do is make it not a lecture, not a class, but a clinic where we'll bring in the problems that are bothering us for discussion."

Their way to train leaders was themselves to listen to those leaders, and get them to come in and come along talking about your problems, get used to our management style that involves bringing your problems to us, and talking about them and workshopping them.

David: This was called out in the conclusions of this paper about the contributions to organizational development, because roleplaying was not a thing. This type of training was not a thing in organizations. This just didn't happen.

This was done between 1944 and 1945. We'll do six training sessions. We're going to bring supervisors in. We're going to ask them these three questions. What's the most frequent problem that you meet? Not problems with machines or the product, the personal problems that bother you. What's your most difficult problem? And what's the most distasteful problem that you meet? I'm not even sure this was confined to the workplace setting. I think these might have even been, what are your problems in life?

What they're actually doing is sharing them amongst their peers. They use these to roleplay exercises. It was all designed around gaining insights into their own and other people's behaviors, or that seeing the world through other people's eyes in developing empathy, compassion, and understanding. I think it would have been fascinating to think about doing this in the 1940s, this type of thing with leaders.

Drew: This is like super modern coaching. They do this roleplay scenario and then they're told to go away in between the sessions, try out the solutions that we've workshopped, and come back to the next session and tell us how it went. You could see this happening in organizations today in training frontline leaders.

David: Absolutely. By the sixth meeting, the supervisors appeared to have found the program effective in developing the interpersonal skills they reported. We don't have all the data, but they reported that they found it easier to deal with their superiors and peers, and that their skills in getting cooperation of subordinates has significantly improved as has their subordinates' productivity.

We don't have all the measures and all of the data, but they're basically saying that facilitating a process where people get to explore their challenges and of themselves in others solve problems, try out new approaches, reflect on that, improves their leadership, and ultimately their team's productivity. Pretty cool.

Drew: Yeah. Learning teams, frontline leadership. I think our next one is diversity training.

David: We're in diversity training now. The fourth study is about changing stereotypes. We talked a lot about bias when we're talking about diversity and inclusion. If we're going to think about stereotypes when we talk about here is a bias or bias in terms of diversity and inclusion, basically what happened is the world's still at war, World War III or just in post war, so the plant has real difficulties recruiting.

There are lots of young females, but this plant had a policy very similar to other plants around the country of not hiring women older than 30. I got to do some research and find out why that is the case, Drew, but we don't hire women over 30.

Drew: They're just considered to be unreliable, David.

David: Apparently, yeah, and not as hard working as younger women. Apparently, if you're over 30 and female, you're not a hard-enough worker to come and work in a plant. Again, it's 80 years ago. The answer that they thought was, and this is quite an issue, why do we have that policy? Let's get rid of that policy, let's challenge it, and let's recruit older women.

They got huge resistance from management at all levels. The researchers, it seems, suggested this when they started seeing these problems. The management at all levels of the business said, no, they're less reliable, they're less productive than younger females, we're not going to do that.

The researchers suggested, look, let's examine the cost of employing older workers. They agreed on four criteria to evaluate this. They said, look, let's see. Because if we can get older workers and we don't lose too much productivity, maybe it's still a good option for us.

They agreed on four criteria and said, right, okay, we're interested in productivity. The speed of learning new jobs, the level of sickness and absence, and the turnover rate. They're pretty cool.

They're pretty direct measures whether this is a good idea or not. How fast do they work? How fast do they learn their new jobs? How often are they away? And how frequently do they quit?

Drew: Basically, every excuse someone's giving for not recruiting an older worker is one of those four things.

David: It's one of those four things. Basically, when they collected the data, I assume we don't know the method, but they assume they hired a group of older women. They started tracking that group of older women against the over 30s and the under 30s in two separate data sets. They found on all of these four measures, older workers outperformed the younger ones.

Then it gets a little bit more interesting. Plant management got impressed by these results because they can see dollar signs in front of their eyes, so they agreed to start recruiting older workers. They said, great, let's do it. However, all of the female supervisors, all of the middle and frontline management remained steadfastly against the idea. Regardless of the evidence going and showing them all the evidence, they still believe that older workers were slower and more prone to absence.

Drew: What they said was, yeah, it's just that the particular older workers that we've got, they're the good ones. They're the exceptions. In general, all the workers still have all of these problems.

David: We did do an earlier podcast on beliefs and changing beliefs. We talked about vaccination, pro- and anti-vaccination. We talked about data not being that compelling for individuals. What was maybe a little bit more compelling was storytelling.

Here's what these researchers did. Because they were more focused on the group than the individual, they said, look. It's clear that we can't overcome this stereotype on an individual basis. Maybe we have to deal with this as a group logic or something in the field and the dynamics around the group.

They held a number of meetings with these supervisors. What they did is they presented the findings regarding older workers and they presented a range of people's views and discussed them. They said, here's this view, let's discuss it. The view that they're less productive, here's the data. By being allowed to compare prejudices freely with the data that's been gathered, the supervisors seem to gain new insights, and they became less hostile to the idea of hiring older workers.

Apparently, it was quite a slow process of change, but the supervisors did agree as a group to make a serious effort to recruit older workers. Apparently, over the next two decades, because these researchers were on site for 40 years, like you said earlier, generations of research, basically, these ideas had completely changed over 20 years.

Drew: I think it's important to point out that it wasn't just that over that very long period of time that attitudes across the whole country changed. This plant then for that entire period of time was ahead of the rest of the US in terms of hiring older workers. They're exceeding productivity for that same reason.

David: Yeah, I think they said they exceeded the standards of productivity of most plants in the nation over that whole period. It was that point of actually creating an opportunity for people to hear different views and perspectives, see some data, see the world through multiple lenses, and in the process, maybe let go a little bit of the strength of the lens through which they're viewing the world themselves, just by exploring alternative ideas and views.

Drew: David, I don't know about you, but that's something that I constantly struggle with personally, but it's obviously empirically true. Letting someone express something that they're feeling or thinking can actually diffuse and weaken that feeling rather than strengthen it. Particularly when we feel strongly about ourselves, we tend to do the exact opposite. We want to turn what they're saying into heresy, or something that they should be canceled over, or something that they need to be corrected on.

It's not that they're right. They're obviously wrong. They're obviously prejudiced. This is almost bigotry against older workers. But still, giving them space to express that in a safe environment when they can talk about discussing it with the data, is more effective in changing it than simply trying to shut it down and force them to accept the data.

David: Also, this overarching idea that it's about group dynamics. An individual seeing that, oh, actually, not everyone in this group thinks exactly the same way as me. Maybe I've got to moderate my position to fall closer into line with the position of the group, and then voting on what we'll do, so let's vote. Are we going to employ older people? If more than 50% vote, then even those people who have very outlying views, will part of belonging to a group is moderating your behavior to the norms of the group.

Drew: I don't have the studies in front of me, so take this with a little bit of a grain of salt. I think there's some evidence that the way the process works is people who have those more reactionary views, if they've been allowed to express those views when it comes to actually the group vote, they'll sometimes change their vote and go the other way.

I've been heard that I don't like the gays. I don't want the older workers working here. I think this is all wrong. Should we get rid of the discriminatory policy? Yeah, okay. I'll go along with the group and vote for it. All they needed was to be heard, and then they're willing to do the thing that we want them to do.

David: I don't know that research either, but I do play a role now that I'm outside organizations. I do play a role as just basically someone that heads of safety and organizations can vent to when they want to air frustrations about their organization with someone outside of their organization. I think just having the opportunity to say what they really want to say and then get back to work is helpful on its own.

Drew, let's talk about the fifth and the last experiment, and then we'll wrap up with some practical takeaways. This last one was titled Overcoming Resistance to Change. This really laid the groundwork for Kurt Lewin's theory of planned change, which even after he spoke about it, it's 40 years later when Kotter talked about the seven- or eight-step change process, which is used by all the big management consultants. It really came down to this early work of Lewin.

Basically, where it stemmed from is the majority of grievances that were presented at Harwood seemed to stem from a changed situation. Basically, production, workers, methods. The production work and the job tasks change several times a year, a new technology, a new process, new materials, new product. This was a big problem.

Workers complained bitterly about being transferred off one job, because we need you to go now and work on this plant, and then now we need you to go and work on this job. Because they went from jobs that they knew well and they could do with a high degree of productivity, and therefore, make money in the piece rate to a job that they didn't know, they were probably slower, they didn't enjoy as much, and they made less money, so morale would plunge.

They even talked about workers being reduced to tears, experienced machinists went on to another job for a period of time, then when they went back to their previous job, they never got to their previous levels of output. What the company wanted the researchers to help them with was, was there a way of introducing change that didn't lead to a decrease in productivity?

Despite a number of attempts that they tried to do change in different ways, in the year that Lewin died in 1947, his researchers basically experimented with a democratic participative approach to change, basically because everything that they've learned about participative decision-making to do with production, they thought, well, how can we apply this participative democratic approach to a change process? Anything you want to add to the introduction?

Drew: The first thing I found really interesting that I hadn't known before we started preparing for this podcast was, this work is where the whole idea of resistance to change comes from. Prior to this work, no one had really conceptualized it in those terms, that when you're making a change, they will automatically have this resistance that you then need to diffuse or overcome before you can make the change. That whole thinking about change like that comes from their work at Harwood.

The second thing is this wraps together all of those other bits. The bit about needing to give people a voice and get them to be heard, the bit about getting them to actually make a group decision as ways of overcoming the change, all those previous studies lead up to this one. 

Along with the same concern I've always had in the background of this, none of this is really giving the workers a genuine choice, This is still all about really overcoming the group resistance in order to make the change that management wants.

In some sense, they've got the illusion of choice that they're being listened to. They're being allowed to change other minor things. They're being allowed to talk about how the change will be implemented, but they're not allowed to stop the change. But still, that's enough. Letting them have some control, some autonomy, some voice, all reduces the resistance to the central change that needs to be made.

David: I think you're right. I think there is what needs to be changed and how it's going to actually work. As we talk about this in a moment, management might have said, well, we actually need to introduce this new machine into the production process. But then, how much the workers were involved in deciding while one person is going to stand here, and they're going to do this part, and then they're going to do this, and then things are going to go over here.

I think you're right. I think the management was still trying to get it changed through. There were different levels of involvement in the workers of exactly what that would mean for the workers.

Lewin first introduced this notion of resistance to change and talked about it as being predominantly a systems concept. A resistance across the group, not necessarily an individual's resistance. Also, I talked about how it affected both managers and employees equally.

They were thinking, how can we identify and promote those forces that increase workers productivity and reduce those that acted to restrain it? We still talk now in theories of planned change about, what are the things that enable the change, and what are the things that are barriers to change? That all came from here. All the barriers to change, that all came from this work.

They've got four experimental groups. What they did is group one was the control group. Any change to their work process was undertaken in the same fashion as normal. The workers were called into the office, they were told that changes were going to take place, and they were given the opportunity to ask questions to clarify their understanding of what that meant for them. This was very typical. At the time in 1940s, US manufacturing, the management decided what workers did and workers were told what it meant for them.

Group two were given more detailed information. They're allowed to nominate representatives of the group to participate in designing the new jobs in setting new production rates. This is interesting. This study design might be the emergence of the workplace health and safety representative or the IR representative of the work group.

Drew: It's certainly the earliest I've heard of it.

David: Yeah. Basically, they get to nominate representatives to represent them, they get to work with the management on the setting, and then the rest of the group gets told what's coming out of that process. In groups three and four, all members participated in the design of the new jobs and the setting of new rates.

Basically, they got the whole work group to do not just the representatives of the work group but everyone in the work group. What they concluded or what they found was that groups three and four achieved an increase of about 15% on their pre-change productivity. What was the productivity before the change, and then what's the productivity after the change? It even got better. Even with the change process, it seems that their productivity improved.

They said it was proportional to their degree of participation. We don't know the results, but we assume that group two was maybe less than 15. You assume that group one went backwards slightly from what was happening prior to the experiment.

Drew: A little nugget from the explanation as to why the participation works. It's not just about they feel good about being included. Their theory was that when you include workers in the design of the new methods and consult them like that, it increases their confidence. The change undermines people's confidence, it makes them feel uncertain.

The consultation works by making people feel like they're experts. They feel more confident in themselves, more confident in the change. That then makes them go into that new change as happier people who feel better and more secure in themselves. They're less likely to be psychologically harmed by the change, which is what causes them to be. It's not just that they hate the change that makes them unproductive. They actually feel worse about themselves that makes them unproductive.

David: They also measured in what they called aggression expressed against management. I don't know how they measured this. I don't know whether it was actually a count of the physical assaults that took place on management. I don't know. It could well have been something like a direct measure like that, again, remembering the time it was. What are your thoughts on that as a measure?

Drew: All I can do is just think of my own workplace and think, okay, what actually happens when we have a change is people do feel uncertain, people do feel insecure, and people gripe about how much they hate management. It's actually a pretty good measure (I think) of change, but we wouldn't call it that anymore.

David: No, I'm just wondering if management sat down with the research and said, yeah, let's do this experiment. Let's try and make sure that we can maintain productivity when we make changes. Then someone pops up and goes, hey, by the way. Can you see if this also makes the workers punches in the face less or something? They've added that as a measurement to the study.

Drew: Now, I'm just picturing this plant with a little sign up the front, days since the last assault against management.

David: Oh, yeah. The management get rewarded for not being assaulted, so they just go unreported. All right, there are the five studies. Anything you want to add before we go into some practical takeaways?

Drew: Let's talk a little bit about the research itself. Lewin sets out really quite clearly what I see as the tension between academic research and industrial research. He's trying to look for social laws. He calls them if-so relationships, hypothetical conditions and hypothetical effects.

They don't tell you what actually exists locally at a particular place or a particular time, like at a particular factory. You still got to do that job of diagnosis for any given place. You can't just say, hey, industrial democracy works, let's do it. Hey, this principle of change management works, just do it. 

You got to know what the local problem that you're trying to fix. But if, in the process of trying to fix a local problem, you can find a general law, you can then take that law, test it out elsewhere, and have a general principle part of your toolbox that you can use in management.

I thought he expressed really well this different aim of local change versus research. The people in his group lived that ideal of shifting back and forth between industry and academia. Some of them went to work for the company for a while. Lots of them worked for the company for a while, then had big academic jobs elsewhere.

Really, what we want in that working relationship is people moving between the two of them, not thinking, oh, this is the academic's job, this is industry's job. Let's not talk to each other. Let's insult each other rather than working out each appropriate role is in solving problems.

David: Great, Drew. Maybe a few limitations. I'll kick us off and then it'd be good. A couple that were pointed out in this paper is that this was not an atypical manufacturing organization given the time. It was a greenfield site. It had a different workforce makeup in terms of new to industry, gender, and experience. It was in a non-traditional industrial hub in the United States, so a different location.

It was not a typical manufacturing organization at the time. There's also this view that the organization was conditioned to respond well to participative management approaches. Even before these experiments, their CEO, Marrow, had spent the previous almost five years interacting with Lewin, doing his own PhD, so there were ideas that the organization was already running quite a participative approach.

Throughout the course of the experiments, as one experiment built on the other, the organization may have been conditioned to increase productivity than more participation because that's what happened in the past. A bit like the Hawthorne effect.

Drew: I don't think those criticisms invalidate the results. The act of running these experiments in the organization is totally consistent with the philosophy that they're testing with the organization. If involving your workers in a decades-long program of experimenting to improve worker conditions, participation, and reactions to change, if workers respond positively to that, isn't that proof of the whole theory?

David: Yeah, it's what you're trying to prove. If you're trying to prove that over a 10-year period, you can go from being 50% behind industry to consistently in front of industry, then that’s case proved. Are we right to do some practical takeaways?

Drew: Yes, let’s.

David: Okay, let's do this. From the first study on group decisions, I've said, in your organization, if you're doing learning teams, which is similar to what they did in 1940, it's one thing to be asked to participate in a learning team. It's a completely different thing to be given a genuine contribution to decisions that result during that learning team.

What they learned in that study is that getting people just to be engaged in the discussion is nowhere near the same thing as getting people to be engaged in the decisions that result from that discussion.

Drew: Yeah. Actively having the group make a decision rather than put forward suggestions seems to be the key marker. They had measurable differences between those two strategies.

David: Just to see how novel this was at the time, I read a different paper in the preparation for this episode, where they said that this was the first known instance of managers and workers talking together inside the plant, where managers got a chance to tell workers about what their productivity goals, targets, objectives, and all the costs were, and workers got to tell management what their problems and challenges were. That was such a novel, dynamic discussion process that had to be facilitated by the researchers quite carefully. Anyway, well done.

Drew: Second one, David?

David: From the self-management study, this idea of target and goal setting, seems to be an intrinsic motivation process. People have to believe that the goals are important, they have to believe that they're achievable, and it seems from the pacing cards, there seems to be little point in setting other people's goals for them. You'll likely get the same result as not having any goals at all. His idea of if you are setting goals, objectives, and targets, in your business, facilitating a team based goal setting process is probably more likely to get to your goals.

Drew: There is an important contingency there, which is that these goals mattered to the people involved. This wasn't a performance appraisal, let me just set my own goals for the year. This is, we're setting a goal that matters to us, because it affects what we get in our bottom line.

David: Maybe that's one thing for productivity now. Like I mentioned, there are different employment processes now. I like to think that in safety, there's a benefit to the employees if their workplace is safer, and there's a benefit to the company.

I'm not talking about setting injury rate goals or maybe I am, I don't know. But I still think in relation to safety, you might be able to get that alignment of benefit to organization and to work group.

Drew: Yeah, particularly if there was a particular safety activity that everyone agreed was worth doing. Having worked together, what the targets for that activity, how often it will be performed, what's an acceptable rate of that activity, would seem to make a lot more sense than trying to set some arbitrary target and ask everyone to meet it.

David: Yeah, I like that. From the leadership training one, my first practical takeaway is leadership is more than technical competence with the work that you're leading. I did put a little ‘duh’ in there, but we still talk about it. Having the capability in a leader to explore the world from other people's perspectives and to facilitate collaborative problem solving processes seem to be really important for team performance.

Drew: David, can I add another one there that I think is also a sort of a ‘duh’? Leadership training is about more than putting up PowerPoint slides to tell people how to be leaders. It's about modeling through the training, the type of leadership that you want them to exhibit.

David: I love that, Drew. Yes, absolutely. Think of the roleplay for the experiment. From those changing stereotypes one, this idea that data alone may not be enough to change individual beliefs. But if you can facilitate a space for peers to discuss the data together with the diverse ideas and perspectives of the group around that issue, to let them gain exposure to these different perspectives, it may lessen their attachment to their single perspective and may allow the group to arrive at a new shared conclusion. In this case, they took a group from unanimously one position to voting in favor of a different position.

Drew: Yeah, over a long period of time, but still, it worked. The next one from the study is about overcoming resistance to change. If people are involved in the change decisions and given an opportunity to contribute in ways that build up their comfort and their confidence in their own expertise, they're less likely to resist the change, they're less likely to be hostile to management, and they're less likely to feel bad themselves and distressed as a result of the change, decreasing their productivity.

David: I just have a general one outside of these five because it relates to Lewin's field theory and a lot of these participative approaches in organizations. This idea of focusing on groups, not the individuals in organizations, so much of what we do in safety is about the individual. 

We actually talk a lot about the individual. This idea that it's the group dynamic that moderates the individual behavior, it may not be the individual that moderates the individual behavior in organizational settings. How do we think in our approach to safety at a group or team level, perhaps, as opposed to an individual? Do you have any thoughts on that or if you want to add any other practical takeaways?

Drew: No. I would rather move straight from that to the question for the week about what can we learn from the Harwood experiments, which seems to be a heck of a lot, but it's all stuff we keep forgetting again.

David: I liked the way that you're raised to that. Just for our listeners to know, every week, we run out trying to summarize the whole episode in one sentence. If you've noticed, we both try and raise to ask each other the question. Well done, Drew, you got in this week.

I think what we've learned from this is very simple, do less telling and more collaboration. Sounds simple, but just think through all your safety processes, practices, and approaches, like how can this be more participative and less autocratic?

The second one that I just throw in there (I think) is that applied research is not as hard as we might think. When you read through these five studies and what they did, they just went and did it. They thought, what's the problem that we're facing? How might we think about doing it differently? How might we set up an experimental group and a control group? How might we measure it and evaluate it, and then run it? this really motivated me a little bit more in the context of, it shouldn't be that hard to do a lot of this type of work.

Drew: David, I'll quickly throw in one important thing (I think) from this study. This applied research wasn't just going out to a university and hiring them in as consultants. Most of this work was in the end done by people that the company hired themselves. They used people with research training as employees, sort of middle management change managers within the organization.

I think that's a model that we can still learn a lot from today. You don't have to go out to a university. You can just hire someone who's got that sort of thinking and attitude towards change improvement, and bring them into the organization as part of the team.

David: Yeah, great. All that can come to us at Griffith.

Drew: If anyone would like to form a 40-year collaboration, where you hire some of our PhD students and we continue to run studies with you building your organization over decades, I've still got 20 years of my career left.

David: Perfect. There you go. You don't get an offer like that every day, and I'll get involved in interesting stuff. 

That's it for this week. We hope you found this episode thought-provoking and ultimately useful in shaping the safety of work in your own organization. Join us on LinkedIn for discussion or send any comments, questions, or ideas for future episodes to