����vlog����

Preroll: PolicyCast explores research-based policy solutions to the big and often complex problems we’re facing in our society and our world. This podcast is a production of the Kennedy School of Government at Harvard University.

Intro (Siri Chilazi): Trainings in particular are interesting because they’ve been around the longest, since the 1960s actually, and they’ve been studied since then too. So by now we have 400 or 500 studies, most of them very high quality, that have looked at the concrete outcomes of these trainings. And as much as they sometimes raise awareness, they might inspire people to act such that people report upon leaving the training: “Yes, I’m now more motivated to tackle unfairness at work and I learned so many things and the facilitator was amazing.” But then when we follow those folks up, any amount of time later to see whether the decisions that they’re making are actually any different. We just don’t see evidence of concrete behavior change.

Intro (Iris Bohnet): So, in many ways our approach is part of a new thinking that is not just relevant for making work fair, but more generally on—how do we get any information that we share to have impact on people’s behavior? And there are a number of things that we threw really at this problem, but the two may be most important, at least for me, were that we wanted this intervention to be very targeted. So the training shouldn’t be a generic training generally about, the value of diversity or creating empathy, although these are all important things. But they don’t have any direct impact or any direct connection to the work that I’m going to have to do tomorrow. So it had to be targeted and focused on particular people with particular responsibilities.

Intro: (Ralph Ranalli): Hi. It’s Ralph Ranalli. Welcome back to PolicyCast. Fairness is a thorny issue these days. In fact, the politics of the moment can make it a difficult subject to even talk about. Then there’s the problem of how do you actually achieve it in the workplace? Study after study has shown that the traditional, well-intentioned ways of doing things—trainings, leadership development programs, networking events, speaker events—simply haven’t delivered the results in terms of equity and diversity that they were intended to. Exactly why they haven’t worked has been the subject of exhaustive study by Harvard Kennedy School Professor Iris Bohnet and Senior Researcher Siri Chilazi. Both are affiliated with the Kennedy School’s Women and Public Policy Program, where Bohnet, a behavioral scientist, is a co-director. They say that most training programs focus on changing individual behavior, while their research and experiments have shown that changing systems instead—especially in a way that targets key moments in an organization’s operations where bias is most likely to occur—is much more effective. They join me today to talk about the data-driven and research-backed policies they say can make workplaces more fair, more cohesive, and more productive, techniques they also discuss in their new book: “Make Work Fair.”

Ralph Ranalli: So Iris, Siri, welcome to PolicyCast.  

Iris Bohnet: Thanks for having us.  

Siri Chilazi: Great to be here.  

Ralph Ranalli: So congratulations on the book, but, more appropriately,  congratulations on a body of work that the book represents, which is about making the workplace more fair. And more fair in a way that is actually effective, versus the ways that I think we’ve been doing it out in the workplace up to now, which have proven to be widely  ineffective. But can we start with just a little background on how the two of you both came to do this work? What was your route that you took to wanting to study it and take this deep dive into an important subject?  

Iris Bohnet: So I’ve been interested in the workplace and in fairness at work for quite some time, and I wrote an earlier book in 2016 that focused on what works and started with that discussion that we needed to bring more rigor and evidence to the question of fairness at work. And then I met Siri a few years later and, Siri was like, “This is a great book. And there’s more to be said.” And so in many ways it’s Siri’s fault that we wrote another book.  

But many things happened between 2016 and 2025. On the one hand, we have so much more evidence now than we did in 2016. So some of the hypotheses I advanced in 2016, we didn’t really have answers to, and we now do. And we have worked, and Siri has led some of that work, with companies with public sector agencies and NGOs, on their workplaces. And so we could tell more stories. And then finally, I think Siri also convinced me that we should actually share more stories. And that’s a bit of a leap of faith maybe for an economist like me, that we shouldn’t just dwell on the data. Of course, the data’s very important, but the stories travel much better than data. Siri, what do you think?  

Siri Chilazi: It’s been such an immense pleasure getting to work with you Iris for the last nearly a decade. I think that was one of the inflection points in my career, frankly, is I didn’t expect to become an academic researcher at all. I knew that I was very passionate about advancing gender equity at work. But I thought I’d do it from the corporate sector. And I kind of stumbled into this intersection of research and practice by accident. But I’m immensely grateful, that I did. And even though we couldn’t have planned the timing of when the book would come out, you know you start writing two and a half years earlier. You never know what’s gonna be going on in the world around the time the book comes out. We do feel like it’s an exceptionally timely addition to the conversation.  

Ralph Ranalli: Right. I definitely want to get to that timeliness and the atmospherics of the current moment in a little bit. But I wanted to start with this data that you have and with the research and to talk about the things that aren’t working first and what you found about the workplace training programs there are currently around diversity, equity, and inclusion. How they haven’t achieved their desired effect in producing equal opportunity. What is some of the evidence of that? How have you been able to quantify that?  

Siri Chilazi: At the heart of that insight is one of the core arguments that we advance in the book, which is, making work fair is not a program, but it’s a way of doing things. And a lot of the most common solutions that we have implemented with the absolute best of intentions over the last five to 10 years, and in some cases even longer, have been programs like leadership development programs, like networking events, like speaker events, like trainings, which you mentioned. Trainings in particular are interesting because they’ve been around the longest, since the 1960s actually, and they’ve been studied since then too. So by now we have 400 or 500 studies, most of them very high quality, that have looked at the concrete outcomes of these trainings. And as much as they sometimes raise awareness, they might inspire people to act such that people report upon leaving the training: “Yes, I’m now more motivated to tackle unfairness at work and I learned so many things and the facilitator was amazing.” But then when we follow those folks up, any amount of time later to see whether the decisions that they’re making are actually any different. We just don’t see evidence of concrete behavior change.

So when we say trainings don’t work, that’s what we mean. They haven’t been shown to be effective to change the actual outcomes that we would care about measuring and tracking when we’re talking about fairness. So moving from these types of programmatic, off to the side interventions to changing the things that we’re already doing on a daily basis, like running meetings, giving colleagues feedback, hiring, promoting others. That’s a much more evidence-based and effective strategy to yield those real, measurable results.

Iris Bohnet: And even some of the things that we have been doing, we can make better. So Siri and I, and a couple of others recently came together and gave trainings more thought. Because as Siri said, trainings are still a favorite tool of many, many organizations, and particularly in the United States, but also more broadly. And so we thought that maybe we could make them better. And I don’t know, are you convinced by our evidence theory?  

Ralph Ranalli: How did that work out?  

Siri Chilazi: I’m, I’m very convinced by the kernel of our evidence, now I want to see more companies replicating our experiment. But Iris you should tell the listeners what it was all about.

Iris Bohnet: So, in many ways our approach is part of a new thinking that is not just relevant for making work fair, but more generally on—how do we get any information that we share to have impact on people’s behavior? And there are a number of things that we threw really at this problem, but the two may be most important, at least for me, were that we wanted this intervention to be very targeted. So the training shouldn’t be a generic training generally about, the value of diversity or creating empathy, although these are all important things. But they don’t have any direct impact or any direct connection to the work that I’m going to have to do tomorrow. So it had to be targeted and focused on particular people with particular responsibilities. That was the first insight.  

And then the second one was that the training had to be timely. So, normally, these trainings happen during onboarding or always in September as a repeat, but they really aren’t particularly timely, particularly relevant again, for anything that I’m gonna have to do tomorrow. And that were the two guiding principles. But again, we threw a bucket load of behavioral science at this training really. And we ran the experiment with a large multinational engineering and telecommunications company that had about, or still has actually, about 100,000 employees active in more than 100 countries.

And when I say experiment, that’s literally what we did. So it was a randomized controlled trial where we gave some people the treatment. So some people got the training intervention and other people didn’t. So who did we focus on? We focused on hiring managers. Again, that could, that could have been any other group. But hiring is a very specific task that many people have to engage with. And then we also focused at a particular moment in that hiring task. And that is when these managers have to move people from the long list to the short list. So that’s after the first stage of screening that often is done by an algorithm these days, where you might be presented with a long list, let’s say, of 200 people, and then the manager has to choose, let’s say the 20 or 30 that they actually want to invite for an interview or want to advance to the next stage of the evaluation process. So it was very targeted and was very timely.  

And the intervention, the training was literally just a seven-minute video. It was delivered by senior manager from the company, which might also have mattered, and it focused up on a number of aspects of hiring. But maybe the two most important ones are that it reminded the hiring managers of the norms, of the values that this company upholds, including diversity and inclusion. And secondly, it reminded the managers of what a good hiring practice looks like. And a good hiring practice focuses on skills, and ideally on skills complementarity—what these new people might add to the team that I already have.  

In any case, it worked. So it worked on the dependent variable that this company itself identified. So this company was concerned that, given it was an engineering company, given that 80% of managers were men, they were likely overlooking women in the talent pool. And they might also be overlooking non-nationals, which is a bit of an uncommon term, but given that this was a multinational company, they were concerned that they might not benefit from the global talent pool. And the Swiss might be hiring the Swiss and the Mexicans might be hiring the Mexicans. So it worked and helped us hire more of these underrepresented groups, particularly women non-nationals.

Ralph Ranalli: Yeah. So that sounds a lot like taking that sort of personal focused training and focusing it almost down to the systemic level, which is the big...  

Iris Bohnet: mm-hmm.  

Ralph Ranalli: ...one of the big takeaways of your research, which is that it’s not about people. It’s not about fixing ourselves. It’s about fixing systems and creating systems and processes that bake fairness into an organization. You say in the book that you have a three-part process for doing that. Can you just break down that three-part process for us?  

Iris Bohnet: Yeah, happy to do that. So we think that you have to make fairness count. And when we say counting, we mean literally counting. As in we need to understand our numbers. We need to understand our data. We need to understand whether we pay people equitably, for example. So, counting is very important. It includes accountability, it includes transparency. So, measure and then make your data speak is kind of the summary of that first part. The second part—Siri, why don’t I turn it over to you?  

Siri Chilazi: Sure, happy to. And actually, the study that you just explained is a great example of part two, which is making fairness stick. How do we build it into all of our existing processes so that fairness isn’t an extra thing off to the side, something additional that we have to remember, but something that’s just built into the process by default. Such as delivering this short training video to managers as part of hiring. So that’s making fairness stick. Then Iris, back to you for the last part.  

Iris Bohnet: Yeah, the last part in many ways might be the hardest one, because it focuses on making fairness normal. So this is about social norms. This goes beyond the formal practices and procedures that we are very passionate about. But we also realize that of course, many of our experiences don’t only happen during hiring or performance appraisal processes, but also in meetings, during lunch breaks, when trying to find support for the work that we are doing. So that last part is actually very, very important.  

Ralph Ranalli: So one of the questions I have is: When you’re talking about rooting out bias and you’re talking about creating fairness, how do you deal with the fact that bias and fairness... the definitions of those terms are often different according to different people. How do you get an organization to adopt an organization-wide definition of fairness that everyone has buy in to, so that these processes can work towards that goal?  

Siri Chilazi: The short answer is by letting the data speak because the data don’t lie, right? They always tell the tale. So if we are seeing, let’s say, that women spend a longer time at a given rank or in a given role on average before advancing to the next level compared to men, that might cause us to ask questions about what’s happening at this role, what’s happening in the promotion process? What’s happening when people are given different projects to work on while at they’re at this level that’s causing this disparity? Because there are no systematic differences in ability between women and men. So the data will ultimately point us to those places where we’re seeing gaps, and that becomes then an opportunity for us to look at the processes and systems and say, is there potentially any unfairness built in here?

Conducting regular pay audits is another example. The data will show you, if you are paying people unfairly, not based on relevant criteria, like people having more experience or having gained better performance evaluations or so forth. So I actually think the better approach is to focus less on the definitions and focus more on data and closing the observed gaps, which can short circuit the need to come up with complicated definitions and getting everybody on board with them.

Iris Bohnet: Although Siri, I think we agree that fairness for us fundamentally is about creating equal opportunity.  

Siri Chilazi: Absolutely.  

Iris Bohnet: So again, I’m totally with you that there are many, many, many different definitions of fairness and probably all have a kernel of truth.  

Siri Chilazi: Mm-hmm.  

Iris Bohnet: But we want to make it very simple in that sense, right? We’ve got to be very straightforward. And a study that I know both of us love, that we talk about in the book, is a meta-analysis that looks at different audit studies that have tried to understand the likelihood that somebody is invited to an interview. And what these studies do is they respond to a given job ad and send to an employer two identical resumes, with only one aspect in the resumes differing. And that could be a name, that could be a club that a person might have been part of that says something about their religious identity, that could be something about their age, their marital status, parental status, other types of information. And we have about 300 of these studies that have now been conducted.

And probably many of our listeners were part of one of them, because literally thousands and thousands of organizations have been in those samples. And still today we find that generally people who have traditionally been underrepresented are still less likely to be invited to an interview. So that’s for us, a measure, like a gap, as Siri said. A gap that is not explained, or explainable by the observable characteristic. And that is a signal of unequal opportunity.  

Ralph Ranalli: So, there’s agreement on equal opportunity as a metric. But I’m not sure that in the current environment, where we have, the Trump administration, for example, withholding or threatening with to withhold funds from organizations that don’t eliminate their DEI programs. How do you make the case that systemic rooting out of gender inequity in the workplace is not just sort of a DEI program or some woke program that just falls under that definition that seems to be under such scrutiny right now.  

Iris Bohnet: So, as you said before, we are very data driven and try not to be ideological whenever we can. And just let’s measure, let’s measure what’s happening. Let’s measure whether you do have equal opportunity at your school, whether you do have equal opportunity at work. Let’s look at the pay differences that Siri has mentioned before. And, I think we are very open to what we find and if we find something is working and we’re surprised by that, we’re like: Wow, we need to think more about that. Like the diversity training study that we did ourselves. I think we entered it with a lot of skepticism, given the evidence that we have and we’re excited to learn more. So, as researchers, that I think might differentiate us from some others—that we became researchers because we are deeply curious about the world and try to understand how it works, and we try to make it better.  

Siri Chilazi: And as researchers, we’re also convinced by the existing evidence that fairness actually is the smart thing to do for all of us collectively. Another study that we love looked at the evolution of GDP per capita here in the United States from the 1960s to today- or actually 2019- to be very precise and found that 40% of the increase in U.S. GDP per capita over that nearly 60 years is attributable to the fact that the restrictions of the 1960s are no longer in place at work today. So women can participate fully in the workforce, so can people of color. People with disabilities are certainly much more able to participate now than they were more than 60 years ago.

So it’s not only the right thing to do morally, as we both very strongly feel. It’s also smart thing to do from the standpoint of our collective prosperity for society. And we also mention in the book and are very open about the fact that gender equity is a two-way street or perhaps a multi-way street. Our research tends to focus on the workplace, and in scenarios where women are currently lagging behind men. But there’s many other places, like in the educational sector, in health, in a lot of administration and literacy related occupations- like Richard Reeves has explained- where men are lagging behind. And that is also a huge problem for our society. Boys, for example, benefit tremendously from having male role models, the younger they are in school. So, it’s work for us all to partake in for the collective benefit of us all.  

Ralph Ranalli: Do you think that message is getting through to the C-suite offices? If I’m an HR manager, right, and I read your book, I’m all in. But do you think that message is getting through to the CFOs? One of the things I read, which I thought was very interesting, was your recommendation for flexible work arrangements as a way to bake fairness into the workplace. But yet we’re seeing CEOs ordering everyone, post-COVID, back into the workplace. Everybody has to be at their desk again. Despite the fact that you did some research about the efficacy of flex work. So can you talk a little bit about that research and maybe if there are any CFOs listening, how they should be paying attention to that and that it would benefit their bottom line.  

Siri Chilazi: I think this is a good example of what Iris said a moment ago about being ideological versus driven by the evidence. My first question to any CEO or CFO who says, “I know my people are most productive in the office,” would be: “Wonderful! Based on what data? Based on what metrics? How do you know? Because to your point, Ralph, there is quite a bit of evidence now, even from well before COVID and then certainly generated in the last five years, that shows that remote work and hybrid work does increase productivity in a lot of cases. Not always. Increases work satisfaction, life satisfaction, improves retention, especially for parents and people with disabilities.

So it can have all these wonderful benefits. That said, there’s not a one size fits all model for flexible work because some jobs just need to be done at least partially in person. So any kind of one size fits all policy for an organization of thousands is bound to generate some backlash or dissatisfaction or jealousy or whatever other negative emotions might come up. So one of the things that companies probably need to do is get smarter and more targeted about how they design these policies and roll them out, not unlike what they’re already doing in other areas of their business, right? We accept that when people have different job descriptions and different roles, they get paid different amounts, and we don’t think that it’s unfair that a finance manager gets paid more than a janitor. So I would also argue that it’s maybe not unfair that if a finance manager’s job is all done on the computer and can lend itself to remote work, that that person gets to work remotely a little bit more than a person who is in a role like a janitor that requires in-person presence.  

But I do want to add one more thing. And that’s a more general observation, which is that extreme opinions on any issue these days tend to be really magnified in the conversation. So if there’s a few high profile CEOs who come out publicly and strongly against remote work, it can often create this inaccurate perception that remote work is dead. No one supports it anymore. And that’s why collecting more large scale data systematically is so important because that reveals the true state of affairs on the ground.  

Iris Bohnet: And in fact, Siri, that made me think of a wonderful talk that we had in the Women in Public Policy Program and the Radcliffe Institute Seminar Series that meets every Monday. We just had one of the experts on working from home, Nicholas Bloom. He’s a professor of economics at Stanford and he has, in fact, presented us with that data. And it’s interesting how working from home has kind of stabilized in the United States. So the average is around two days a week. And that’s been quite stable since COVID. So, clearly, we had a peak, we were lower before COVID, then we had a peak during COVID where many, many more people work from home. And then now it’s kind of decreased again, but not to the level where we were before COVID and it’s kind of stabilizing in the last few years.

So I think that’s, again, very important that we don’t fall prey to what we call the availability bias in behavioral economics, where the most available information is those that come from very loud voices, and that, of course are picked up by the news. I think that’s very important. And the other thing, Siri, that I just wanted to stress, because I’m now so thinking of Nick’s work—again, I’m suggesting every CFO looks at his evidence because it’s really quite remarkable how he and his team have been collecting evidence, experimental evidence, in the U.S., in China, in India, you know, and really many different countries. But anyway, for the U.S.—the disability findings...

Siri Chilazi: Mm-hmm.

Iris Bohnet: I did not know that before his presentation that it was in particular people with disability, who have benefited from increasing work from home.

Ralph Ranalli: Can you talk a little bit more about some of those research success stories that you’ve found, about specific places and specific instances where adopting this organizational, structural approach to fairness has really proven itself without a doubt the way to go.  

Iris Bohnet: Maybe I’m going to start actually with working from home and then we can go more broadly into hiring, promotion, performance appraisals. I mean, we’ve done a number of studies in different areas, but on working from home, another study looked at India, where in India, one of the biggest challenges still is the low workforce participation rate of women. And it’s kind of surprising because it’s a growing economy and many more jobs are being created, and many different people have tried different types of interventions, and it appears as if the social norms are pretty strong and keeping highly qualified women out.

And so colleagues of Nick’s also did a randomized control trial in India trying to understand whether giving women in particular more opportunities to work from home might increase the likelihood that they accept the job offer and- which I thought was very interesting- might have long-term impacts where they’re more likely then to work on a job that might require them to come into the office more. That’s exactly what he found. And so I do think we have to try out many more things to better understand what works and what doesn’t work. Mm-hmm.  

Ralph Ranalli: Yeah, that was interesting because it seems like it was an approach that took into account the cultural norms. And instead of trying to bash headlong through the cultural norm, it sort of worked with it, because women work in the home. And then is that such a big step to working from home and then you take the step after that. So, is there a lesson there in terms of working with location-specific, cultural-specific factors, and factoring them into this? Are there any other instances where that’s happened?  

Iris Bohnet: I mean, absolutely. Now when I say absolutely, I mean we have to do both. So we definitely have to think hard about how to make our interventions compatible with the types of norms that people uphold. And I’m now thinking of a very, very challenging practice that is not something that Siri and I talk about in our book, but that speaks to this question. And that is female genital mutilation, where some really interesting work has been done around the question of why is it happening and could one find solutions to the why that don’t include that particular practice. And that’s, funny enough, another talk that I just heard by Eliana La Ferrara and Lucia Corno on the faculty at the Kennedy School and trying to change these, these practices by basically meeting where people are at and just ever so slightly changing a little bit what the input factors are to these initiation rites that seem to be so important. So I think that’s important.

But secondly, I also want to stress that we should work on the social norms, and we should try to think about how they come about. And I know this is a concept that Siri likes, this is pluralistic ignorance.  

Siri Chilazi: Mm-hmm.  

Iris Bohnet: And I think that’s an important one.  

Siri Chilazi: It’s an important one. Pluralistic ignorance refers to the phenomenon whereby our perceptions of what other people think are often miscalibrated. So I’ll give you a couple of examples. At Santander Bank in the UK, men working at the bank were asked how many of them were in favor of flexible work. And 99% said, “Yes, love flexible work. It’s amazing. I would love to do it.” But then they were asked, okay, how many of your male colleagues do you think are supportive? Well, they thought only two thirds. So that’s an example of the gap. And as you can imagine, what we think others think is actually, in many cases, a bigger influence on what we end up doing, than our own deeply held beliefs. So what in fact was found in the case of Santander Bank is when some of these men were informed that actually, 99% of your colleagues are also supportive. This is a great thing. Everybody loves it. They became much more likely to take parental leave- not because they changed their deeply held opinions about parental leave, but because their calculus had changed about how it would be received among their peers if they actually took more leave.

And there’s another great recent example of this here in the United States. A paper published late last year, late in 2024, found that the support for various diversity and inclusion efforts is actually much higher among a variety of segments of the American public than what people think. So depending on the exact population, in the 80s- percent of people say: “Yes, I believe we have to work to create a world that’s more diverse and inclusive. But when they were asked, “Okay, how many other people do you think are supportive of this?” The average hovered around 50%. So this is one important way to change norms, is to just accurately inform people about how others truly feel about an issue so that we can update our miscalibrated beliefs.  

Ralph Ranalli: Yeah, again, it goes back to the data and the actual data and communicating that data.

Siri Chilazi: Yes.  

Ralph Ranalli: So going back to the research on successful organizations that you found, because you’ve said that there are the obvious things to look at, like how do you promote, how do you hire, how do you do performance appraisals? But there are other places where you can bake fairness in, like how you organize meetings or how you organize work. Can you talk a little bit about what you found in terms of what a successfully fair organization looks like in terms of how it handles those day-to-day details of how its operation works.

Siri Chilazi: One of the other studies that we love comes from Novartis, the global pharmaceutical company that’s based in Switzerland. They were very interested in increasing employees’ sense of inclusion and their sense of psychological safety, which is their willingness to take interpersonal risks in the workplace, such as by speaking up against a senior leader or expressing an unpopular opinion.  

And these are important concepts, but it can also sound a little bit fluffy, right? Like inclusion, how do we even start tackling something that’s this amorphous? So we love the approach Novartis took, which is they honed in on a specific behavior, a specific feature of work that happens regularly, and that’s one-on-one meetings between managers and their direct reports. And they wondered if we could make these meetings somehow better for employees, would that change their feelings of inclusion and psychological safety. So they tested a variety of guidelines to managers on how to conduct these meetings. And it turns out there was one guideline that was by far the most effective at both increasing inclusion and psychological safety, but, interestingly, also increasing the frequency of the meeting, which was an unexpected side benefit. And this guideline to managers was to speak less, listen more, and give space in the meeting for their direct report to surface their needs and aspirations and then be supportive of those things.  

So that’s an example of making a really small and simple tweak to something that people are already doing. So this is not piling anything new and extra on your plate. You’re already gonna be having these one-on-one meetings. Just conduct them a little bit better so that, you get more out of them, but also you start to create the kind of culture that you’re trying to create.  

Iris Bohnet: And I think Siri, in the example that you just gave, a bit of a theme is emerging and that is that we have to focus on moments that matter.

Siri Chilazi: Mm-hmm.  

Iris Bohnet: And that could be those meetings, but that could also be the training that we discussed before that happens just before the hiring managers move people from the long list of the short list. Right. We have to get much better at thinking about those moments where people make these consequential decisions. And maybe just to add one other example that is beyond formal procedures that we think is really crucial and that you alluded to before, Siri, is this question of: What type of support do people get?  

So a few years ago we worked with a law firm in the United States and one of the leading law firms—global but headquartered in this country. And they were puzzling over the fact that they had a very diverse entry class of first year associates—given the fact that now law schools, of course, are much more diverse than they were 50 years ago- and still ended up with a pretty homogenous group of partners at the top. And so we looked at some of the data, we looked at some of the processes and actually came up with something quite simple. Because we saw that the patterns started in the very first year, where first year associates weren’t given the same types of opportunities. Namely, what happened in this law firm was that partners chose the first year associates they wanted to work with. And the data, again, told the tale, showing that partners basically replicated themselves and if partners were white and male and American, that’s what that support looked like. And even to the degree where partners preferred people who went to the same law school that they went to.  

So how do you break something like that, right? Again, you are identifying a moment. That is a moment that matters, that support that you get, because as we all know, that could then lead to a vicious cycle where you end up with non-important work to start with and you just never get exposed to the shiny assignments, to the important clients, to the important speaking engagements, et cetera. So, the law firm decided to centralize work allocation and has now one little office, partners who rotate, who give out work basically and match partners with first year associates during that first year.  

Ralph Ranalli: So we’ve reached the time when we need to talk about policy. So I’m a CEO, CFO, and I have decided to give your methodology a try. What would the top two or three things that you would both recommend that I do first—policies that I would adopt in order to start down this path of making my organization more fair and hopefully have a better organizational culture and have a better functioning workplace.  

Iris Bohnet: So I’ll let you take the lead here, but I wanted to say just one thing to make it very clear. We don’t think that our advice is restricted to the private sector. Because we, you know, we just talked about co CFOs. We’re thinking and have worked with mayors.  

Ralph Ranalli: Oh sure.  

Iris Bohnet: And we have worked with leaders of agencies, we’ve worked with NGOs. Because an organization is an organization.  

Ralph Ranalli: Right. Absolutely. And different organizations have different cultures and they have different priorities. Yes. Like a private sector company, a financial services firm is more focused probably on the bottom line, whereas an NGO is probably more focused on service delivery and...  

Siri Chilazi: yes.

Ralph Ranalli: ...less so on making money. So, I were the head of an organization, just an organization, what would your advice be? How do I start?  

Siri Chilazi: And I loved how you asked the question initially, Ralph, because you said, what would you do in a tailored way, which gets to the heart of the first part of our answer, which is there’s no one size fits all for everybody. So be like a doctor, right? When a patient walks into your office, you don’t start handing out pills at random. You run a bunch of tests. You ask a lot of detailed questions to understand the patient’s symptoms, and then based on all your knowledge, all the data at hand, you develop a diagnosis of what you think is ailing the patient. And only after that do you then prescribe the medicine that you think will be most effective. And then you follow that up with a follow up appointment four weeks later so that you can determine if the medicine is working as intended, and if anything about the treatment regimen needs to be tweaked.  

So I, as an organizational leader, would start the same way, looking at my data, letting the data speak, letting the data reveal the places where we have gaps, where we have pain points, where we have opportunities to improve. And then designing targeted interventions to fix those gaps if possible, implementing those interventions as experiments, just like we did with the diversity training video to actually test: Does this work the way we hoped it would? And then if it does wonderful, you might roll it out to the whole company. And rinse and repeat the process. Iris, what would you add?

Iris Bohnet: Obviously I would always start with data as well. And then I’m gonna put in a good word for behavioral science, which might come across as very biased, but I don’t just mean behavioral science narrowly defined that in academic term. But I do think secondly, organizations could benefit from subject matter expertise. So they could, you know, benefit from some people who, like Siri and I, come in a little bit as the doctors who do the diagnosis, because we already know a bit better what questions to ask, where to look. For example, what I just described before, sometimes it’s called performance support bias. So. It’s not new to us. And so we kind of knew that that’s something that we should definitely check.  

So maybe the second thing is to think a bit about: Where would that expert base be? That probably includes some people who have some understanding of how good organizations function, how productive processes and practices are designed. It probably includes some data scientists who can help us analyze the data statisticians. I’m not trying to pinpoint any one particular expertise, but just in saying you need people who can diagnose the problem and you need people who then can provide some of the data that helps you understand- and importantly measure- what’s happening. So I think that would be my second act.  

Siri Chilazi: Yeah. And I would just follow that up with while that kind of comprehensive organizational approach that Iris and I have described is ideal, and if you’re high enough in the organization, you might have the power to make it happen. Also, if you’re lower down in the hierarchy, right, an individual contributor, maybe the manager of a small team, maybe even this junior summer intern. There is a system that you do control. There is something in your own work that you can tackle to make that system more fair. In fact, our book begins with the example of Astrid Linder, a Swedish engineer who is the research director for Sweden’s equivalent of the National Highway Transportation Safety Administration here in the U.S. And she realized years ago, given that her job was to keep all people in cars as safe as possible, that it wasn’t happening because the data showed that women, even though they drive a little bit less than men, are significantly more likely around the world to be injured in car crashes. Why is that? It’s because cars are safety tested almost exclusively on dummies that are modeled on the male body, and there are some important biological physical differences between women and men, including joint laxity, bone density, and others, that then made women less safe. So Astrid Linder in her role didn’t have the power to change regulations worldwide. She didn’t have the power to force car companies to safety test cars differently. But what she could do and did was develop the world’s first crash test dummy modeled on a woman’s body, and she sparked a movement where many other organizations have now followed in her footsteps to develop different females dummies of their own, including ones that model a pregnant body. And she has also inspired and collaborated with car companies, including Volvo, who have started voluntarily adopting these new crash test dummies in their tests. And now we’re seeing regulators around the world, including here in the us, jumping on board the train as well. So even just one person with limited power by making their own work fair can spark a bigger revolution.  

Iris Bohnet: So maybe the advice, and Siri I don’t know whether you’d subscribe to that, is that maybe leaders should collect the data, should have expertise to help them design interventions and evaluate interventions. And, they might want to inspire a fairness hackathon.  

Siri Chilazi: Yes.  

Iris Bohnet: Something along those lines. Right?  

Siri Chilazi: I love that.  

Iris Bohnet: Right. Well, because as you say, in our work we have come across so many people at all levels and all walks of life that have made work a bit more fair in their domain of influence.

Ralph Ranalli: Well. Data-driven policy for positive change is our reason for being here at PolicyCast. So I can’t tell you what a pleasure this has been talking to the both of you. And I want to wish you the best of luck as you go there in the world and try make it a fairer place.  

Iris Bohnet: Thank you for having us.  

Siri Chilazi: Thank you. It’s been a pleasure.

Outro: (Ralph Ranalli): Thanks for listening. If you enjoyed this episode, circle back with PolicyCast next week, when my guests will be renowned scientist and presidential advisor Professor John Holdren and Jennifer Spence, the director of the Harvard Kennedy School Arctic Program. We’ll talk about how a region that is vital to regulating the earth’s climate and is home to unique ecosystems and human cultures is under increasingly under unprecedented economic, geopolitical, and environmental pressure—including U.S. President Donald Trump’s stated intention of grabbing Greenland away from Denmark. And please to check out some of our other Kennedy School podcasts, like Justice Matters from the Carr-Ryan Center for Human Rights Policy. In the most recent episode, co-host Aminta Ossom discusses the climate crisis and human rights with Sam Bookman, a scholar of climate change law and a post-doctoral fellow at Harvard’s Project on the Foundations of Private Law.And if you are a regular listener or are thinking of becoming one, follow us on Apple Podcasts or your favorite podcast app, and please leave us a review while you’re there. So until next week, remember to speak bravely, and listen generously.