A Debate about Building Lawful and Effective Workplace Diversity Practices

[00:00:00] Welcome to the debate. So, imagine your company finally implements a perfectly structured, perfectly legal, identity-neutral hiring rubric. You would think bias would just, you know, disappear, right? But what if the data shows that the moment a company loudly declares itself a pure meritocracy, managers actually start giving male employees larger bonuses than equally performing women?

[00:00:22] Right. It's the ultimate paradox. You put up a banner saying the workplace is completely objective, and people immediately let their guard down. They allow their implicit biases to completely take over. And that paradox is exactly what organizations are slamming into right now. I mean, between 2023 and the end of 2026, the legal ground under workplace diversity, equity, and inclusion programs just completely ruptured. Oh, absolutely.

[00:00:45] We've seen this massive surge in executive orders like EO 14173 and Supreme Court rulings, most notably Ainsley, Ohio Department of Youth Services. And these have fundamentally altered the landscape. We are talking about major organizational anchors like IBM and Deloitte paying out tens of millions of dollars in False Claims Act settlements over practices that just three years ago were considered standard industry operating procedure. And organizations are terrified, frankly.

[00:01:38] Open access HR practices. So the central question we have to answer in this new 2026 enforcement era is this. Does this organizational shift toward purely structured open access HR practices represent a genuine evolution in achieving workplace fairness? Or does it risk creating false progress that merely shields organizations from legal liability while preserving all the underlying biases? Exactly the core tension.

[00:02:03] I argue that structured identity neutral processes are the scientifically and legally superior method for eliminating bias and achieving true fairness for every single employee. And I argue that relying solely on these neutral structural reforms risks a really dangerous complacency. It triggers that exact paradox of meritocracy we just mentioned, and it ultimately fails to address the specific systemic realities of underrepresented groups. Well, let's start with the science.

[00:02:29] Because for decades, corporate DEI programs have actually been at war with what industrial organizational psychology tells us actually works. That's a strong statement. But it's true. The current legal correction we are seeing forces companies into a model that finally aligns with the empirical data. Identity-restricted programs and, you know, demographic targets, they introduce noise into systems that should be based purely on capability. But do they? Just look at Paul Sackett's 2022 massive revision of the personnel selection literature.

[00:02:56] When they looked at decades of hiring data, they found that structured, criteria-based selection, specifically standardized interviews with anchored rubrics, has the absolute highest operational validity for predicting job performance. It sits at a .42. Wait, hold on. Before we gloss over that number, for people who don't spend their lives reading IO psychology journals, what does a .42 operational validity actually translate to in the real world of hiring? Fair point. So in the context of predicting human behavior, a 1.0 would be a perfect crystal ball.

[00:03:26] You would know exactly how good someone will be at the job. Unstructured conversational interviews, like the typical tell-me-about-yourself-sit-down, they hover at a fraction of that predictability. Right. They're highly subjective. They are basically a coin toss, infected by the interviewer's personal biases. So a .42 means that by defining the exact skills needed, asking every candidate the exact same questions, and grading them on a predetermined scale, you massively increase your ability to predict actual job performance. Okay, but... It outperforms job knowledge tests.

[00:03:55] It outperforms cognitive ability tests. And it does so while naturally ignoring demographic variables. By removing demographic targets, we are actually improving organizational performance. Look, the SACA data is robust. I will give you that. But you're putting entirely too much faith in the idea that a standardized rubric just magically eliminates human prejudice. We have to look, though, at the concept of false progress, right? From the 2019 study? Yes.

[00:04:22] Researcher Lisa Leslie identified this back in 2019. And false progress happens when there is a decoupling of policy adoption from policy implementation. Meaning they adopt the rule but don't actually follow it properly. Exactly. An organization adopts a strict rubric. The executives look at the process metrics, meaning they just check to see if the HR paperwork is compliant, and they declare victory. The metrics look fantastic on a dashboard.

[00:04:48] But the actual opportunity landscape for marginalized employees hasn't changed one bit. But if the rubric is actually used, the opportunity landscape does change. I mean, it forces the manager to evaluate the candidate's actual answers against a standard, not against their gut feeling. But you are assuming the manager is a rational robot executing code. Let's unpack the Castilla and Bernard 2010 research on that paradox of meritocracy we opened the show with.

[00:05:15] Why do managers give men higher bonuses than equally performing women the moment a company declares itself a meritocracy? It is a psychological mechanism called moral credentialing. Right. They feel they've already proven they are fair. Precisely. When people believe the system they are operating within is perfectly objective, they stop auditing their own decision making. They think, well, the system is fair. I am using the system. Therefore, my gut instinct must be fair. Their guard drops. I hear that, but...

[00:05:44] So dismantling identity-conscious programs out of legal fear and replacing them with a purely bureaucratic structure, it just leaves vulnerable employees exactly where they were decades ago, just hidden behind a wall of compliant paperwork. See, you're arguing against the concept of structure by pointing out that humans are flawed. But that is exactly why the structure is necessary in the first place. Take Meta, for example. Okay. They recently abandoned the diverse slate rule. You know, the requirement that a certain number of diverse candidates must be interviewed for a role.

[00:06:14] To some, that looked like a massive retreat from equity. But to me, replacing demographic arithmetic with rigorous job-analyzed structure is a massive upgrade. Think of structured HR processes like double-blind peer review in scientific publishing. Right, but you can't double-blind an in-person interview. You can see the person sitting right across from you. Well, you can't blind the visual, sure, but you blind the evaluation criteria. Peer review isn't about ignoring the author's identity out of malice.

[00:06:42] It's about systematically preventing irrelevant variables from polluting the evaluation of the data. Okay, I see the analogy. You anchor the rubrics, and you force interviewers to document the specific rationale for their scores. When you don't do this, you end up like the New York Times. Ah, you're referring to the 2026 EEOC lawsuit. Exactly. The EEOC sued them because a manager allegedly bypassed their standard structured process to add a favored external candidate to the final interview panel for a deputy editor role. Right.

[00:07:10] And that favored candidate ended up getting the job despite scoring objectively lower with the interview panel than the internal diverse candidates. When managers are allowed to bypass structure, bias wins. The structure is literally the only defense mechanism we have. I don't disagree that bypassing structure is a huge problem, but you are missing what happens when managers are forced into the structure against their will. Frank Dobbin and his colleagues studied over 800 U.S. workplaces across three decades. A 2015 study, right? Yes.

[00:07:38] Their 2015 findings highlight the concept of the iron cage. When bureaucratic reforms are implemented primarily to constrain managerial discretion, which is exactly what is happening right now under this DOJ enforcement wave, they provoke intense psychological reactance. Managers don't like being put in an iron cage. So what is the alternative? We should just let them hire whoever they want based on culture fit? We both know how that ends. No, no, of course not. But you have to recognize how they adapt.

[00:08:06] Dobbin found these control-oriented reforms can actually lead to declines in diversity. Because they fight the system? Yes, because managers are smart. They learn how to reverse-engineer the rubrics to justify hiring who they already wanted to hire. If a manager prefers a candidate who went to their alma mater or who reminds them of themselves, they would just write a glowing, perfectly rubric-compliant justification for that candidate's communication skills or strategic thinking. The structure doesn't eliminate the bias.

[00:08:33] It just becomes a highly sophisticated laundering system for it. But if a manager is reverse-engineering a rubric to launder their bias, that is a failure of organizational audit, not a failure of the rubric itself. Google's readout work guidance clearly demonstrates this. When pre-made questions and standardized rubrics are audited properly, candidates are actually 35% happier with the process. Sure, but... Because it forces interviewers to justify their scores with specific behavioral examples from the interview, not they generalities. You can't fake a behavioral example during an audit.

[00:09:03] But if your entire system relies on HR conducting a perfect forensic audit of every single manager's interview notes for every single hire just to prevent them from cheating, I mean, you have built a deeply flawed system. It's not about perfection. It's about deterrence. Even so, that vulnerability at the hiring stage completely explodes when we move past recruitment and look at ongoing employee development. Things like mentoring and sponsorship. Managerial discretion is exponentially higher when deciding who gets mentored. That's true.

[00:09:32] Historically, we corrected that systemic imbalance but targeted interventions. Now, companies are rushing to these completely generic open access models. Well, they are rushing to open access models because identity-restricted development programs are legally indefensible today. Look at the DOJ's $21.5 million settlement with Deloitte. It was incredibly specific. The Spring Warden Compass programs. Right. They targeted Deloitte's programs precisely because eligibility was restricted by race and sex.

[00:10:01] Or the EEOC suing Coca-Cola Beverages Northeast over a women-only networking event at a resort. Yes, the landscape is hostile. Following the Supreme Court's ruling in Muldrow v. City of St. Louis, an employee excluded from a program no longer has to prove it cost them a promotion. They just need to show some harm from being excluded. We simply cannot legally restrict access to career development based on demographics anymore. I understand the legal reality. But legal compliance is not the same thing as actual fairness. But in this case, I argue it leads to a better outcome.

[00:10:31] Look at the core mechanism of mentoring. The actual active ingredient in career development is the sharing of informal knowledge, getting visibility to leadership, and having advocacy. Right. When you open a mentoring program to everyone, you preserve that exact developmental mechanism. But you completely remove the discrimination flaw. Needs-based design helps everyone. A first-generation white professional from a working-class background or a fully remote worker, they lack access to that informal knowledge just as much as anyone else.

[00:11:00] Why should they be excluded from development? No one is saying they should be excluded. But converting a targeted intervention into a generic open-access free-for-all completely ignores how corporate power dynamics actually work. How so? Think about it like trying to fix a broken ladder for a specific group of employees by simply leaving the company toolbox open in the hallway for anyone to use. Who is going to grab the tools first? The people who need them? No, the people who already feel entitled to them. The people who already know how to build.

[00:11:28] You're assuming marginalized employees won't opt in. I'm pointing to the historical data. Alexander Kaliff and her colleagues analyzed decades of private sector data, and they found that targeted mentoring historically had notable, highly specific gains for black women. Yes, historically. If you take a vital, incredibly limited resource like executive sponsorship and you put it into a completely open-access lottery system, the employees who already know how to navigate the corporate political environment will overwhelm the system.

[00:11:57] I think that's a cynical view of open access. The Pew research data from light 2024 is damning here. Despite this massive corporate pivot to fairness for all, 18% of black workers still say their employer pays too little attention at DEI. That is a highly significant indicator. It tells us that these generic, sanitized open access models are leaving vulnerable employees feeling completely abandoned by their organizations. But you are ignoring Dobbin and Caleb's own later synthesis from 2016.

[00:12:25] They looked at what actually moves the needle for minority representation in management. And it wasn't the exclusionary nature of the programs that worked. It was the exposure. Exposure is important, yes. Voluntary training and mentoring programs led to significant increases in management representation precisely because they forced people to interact with colleagues they wouldn't otherwise meet. The exposure is the active ingredient.

[00:12:47] When you open a program to all and you use transparent selection criteria, you are removing the subjective executive favoritism that typically locks underrepresented groups out in the first place. Okay, but this brings us to the most dangerous part of this entire shift. Let's say we accept your premise that we have to use broad access rather than targeted interventions. How do we measure success? If organizations are terrified to track demographic outcomes to hold leaders accountable, what replaces it? Process-focused measurement replaces it.

[00:13:16] And honestly, it is vastly superior. When you impose demographic quotas on leaders, it inevitably drives individual discrimination. Because they hire to the metric. Exactly. It forces a middle manager to look at the demographic arithmetic of their team rather than the capability of the person sitting across the desk. That exact dynamic led to IBM's legally disastrous diversity modifier. Where executive pay was directly tied to hitting demographic hiring targets. Which the DOJ jumped all over. Right, because it essentially mandates illegal discrimination.

[00:13:46] Emilio Castilla's 2015 model shows us the alternative. Castilla's accountability model? Right. Castilla studied a firm that introduced rigorous accountability in the pay decisions without using demographic targets. They audited pay based purely on performance criteria. Before bonuses were finalized, managers had to explicitly explain and document any divergent decisions in writing. Okay, but without demographic targets, how did it affect minority pay? That's the best part. They didn't apply group-level demographic adjustments.

[00:14:13] But because they held leaders rigorously accountable to the process criteria, the pay gap shrank significantly for women and minorities. That is how you fix the system. You track adherence to the structure processes. In theory, yes. But let's look at what is actually happening right now in 2026. Major companies like McDonald's, Google, Target, they are retiring their demographic goals for senior leadership. Right. Complying with law. And they are replacing them with exactly what you are championing. Barrier analysis and tracking.

[00:14:43] Adherence to structured processes. But let's be intellectually honest about human behavior here. Tracking how many managers filled out a rubric correctly completely lacks the urgency of actual representation data. Why does it lack urgency if it fixes the underlying pay gap? Because it removes the mirror. If a company sees that its senior leadership is 95% homogenous, that creates an urgent, undeniable mandate for change. The board demands action. Or they just panic and hire purely for optics.

[00:15:10] But if a company drops that metric and instead just looks at a dashboard showing that 95% of its managers successfully filled out the hiring rubric on time, they pat themselves on the back. It creates a state of comfortable compliance. This is Leslie's false progress in its purest form. The HR metrics look flawless. The DOJ leaves you alone. But the organization is just as stratified as it was 30 years ago. I disagree completely. Demographic targets are the definition of false progress.

[00:15:39] When managers are pressured to hit a demographic number, they often hire just to hit the target without fundamentally changing the underlying culture, the inclusion mechanisms, or the support systems for those employees once they arrive. Well, retention is a separate... It becomes a revolving door. But when you measure the process, you know, when you track whether job postings actually reached a wide pool, when you audit pay for performance equity, you have fixing the machinery that generates the outcomes. You create a durable fairness capability that doesn't collapse every time the Supreme Court issues a new ruling.

[00:16:08] You're talking about fixing machinery, but you are ignoring the human cost of treating people like inputs in a legal compliance engine. I'm not ignoring the human cost. While organizations are busy scrubbing their policies to be perfectly legally neutral, they are rapidly dismantling the DEI infrastructure that historically supported psychological safety. When companies abruptly cancel diversity initiatives or rebrand them in the generic leadership training, they send a very clear chilling signal to their minority employees.

[00:16:36] It implicitly invalidates their lived realities. I don't think it invalidates. You cannot just slap a sanitized corporate narrative of fairness for everyone over a workplace culture that still harbors systemic bias and expect everyone to magically feel welcome. Okay, but I completely... I reject the premise that highly visible identity-restricted DEI programs were like universally fostering psychological safety. In fact, we have overwhelming evidence that they were frequently toxic to the workplace climate. Toxic in what way?

[00:17:05] So look at the EEOC lawsuit against Washington University. The university held a mandatory training session for an administrative team and utilized shared racial identity spaces, which literally meant putting participants into race-separated breakout rooms. Oh, I remember this case. A black senior program manager, she objected. She stated she would be the only BIPOC participant on her team in this setup, and she was like kind of forced into this separated space anyway.

[00:17:30] She described the entire experience as dehumanizing, and she was ultimately terminated after complaining about it. Like, that is the inevitable result when organizations design programs around rigid identity categories rather than human needs. Look, that is an egregious example of poor implementation by a specific facilitator. It doesn't invalidate the core concept of acknowledging differing lived experiences in the workplace. But it is not an outlier. It is symptomatic of an approach that the evidence shows fundamentally does not work.

[00:18:00] Edward Chang and his colleagues ran a massive field experiment in 2019 with over 3,000 employees. The behavior tracking study? Yes. They tracked actual behavioral outcomes after diversity training, who employees nominated for awards, who they invited to coffee, who they mentored. The online diversity training changed attitudes in the short term, but it completely failed to change actual behavior. Training alone rarely changes behavior. That's fair. And Lisa Nishi's 2013 research proves this from another angle.

[00:18:28] What actually reduces relationship conflict in diverse groups is not mandatory segregated training. It's a baseline climate of fair employment practices. It's voluntary, skills-based leadership behaviors. I agree that inclusive leadership behaviors are vital. Right. And when managers are trained on the mechanics of inclusion, like how to run a meeting where everyone gets a voice, or how to give specific, actionable feedback without relying on personality critiques,

[00:18:54] that builds psychological safety much more effectively than a mandatory lecture on privilege. I hear you. But by stripping away this specific language of diversity and equity just to appease this legal environment, we lose the vocabulary needed to address real, persistent biases. General training usually results in generic outcomes. Not necessarily. If we only teach managers general fairness, they remain entirely blind to the highly specific,

[00:19:20] nuanced ways that women, people of color, and other marginalized groups get sidelined in meetings or passed over for stretch assignments. General training might lead to generic outcomes, but structured accountability does not. The convergence of the 2026 legal landscape and empirical science is unequivocally a net positive. For decades, organizations relied on demographic arithmetic and identity-restricted programs that were legally dubious and, frankly, empirically weak. Well, they had their plays.

[00:19:46] But moving toward transparent, criteria-based accountability, like highly structured interviews, open access development, rigorous pay audits, creates a durable framework. It protects every individual equally, ensuring decisions are driven by job-related merit rather than subjective bias. The structural shifts are legally necessary right now to survive the DOJ and the EEOC enforcement wave. I grant you that. But they have to be implemented with deep intellectual honesty.

[00:20:11] If executives use these bureaucratic structures to simply declare the workplace magically meritocratic, they will reproduce the paradox of meritocracy. We have to audit the structures. But if we don't, the systemic barriers will just remain hidden behind perfectly filled-out rubrics, failing the very workers who need fairness the most. We certainly look at the efficacy of structure versus the risk of complacency differently. But it seems we have reached a vital point of convergence here.

[00:20:38] The era of superficial, unmeasured, and legally dubious DEI practices is decisively over. Yes, the performative era is absolutely dead. Any path forward now demands rigorous data, genuine organizational learning, and a focus on evidence-based practices rather than symbolic gestures. It is a profoundly complex landscape, especially with active legal flux, like the ongoing NATO-HE litigation that continues to shape how federal contractors operate. There is so much more to explore in Westover's research,

[00:21:06] and we invite you to dive into the source material to evaluate the optimal path for your own organizations. And as you navigate this terrain, think about that open toolbox in the hallway. You can't just leave the tools out and assume the house will build itself. And you can't build a starty house without a standardized blueprint. Thank you for joining us on The Debate.