Compartments and Consequences
How We Structure Diversity, Equity, and Inclusion Content May Be Undermining Its Purpose
I need to make this clear at the outset: I’m not sharing my sources on this one. At least not all of them.
That’s an opening line that deserves explanation: I’ve spent some time with a collection of professional guidance documents on leadership and mentorship in veterinary medicine. It’s a fairly robust collection the kind of materials our organizations produce to help us become better teachers, sponsors, and colleagues. I think they’re good documents, written by people who care deeply about the profession. I also think they have a structural problem worth examining.
But I’m not publishing the bibliography. I’m not linking to the PDFs or websites for the documents I evaluate this time. I ran a line-by-line analysis, coded for content1, tested my methodology, and I’m asking you to take my word for it.
The simple reason is that I don’t want to hand anyone a target list. The critique I’m making is about how we talk about diversity and mentorship, specifically, about the unintended consequences of treating them as parallel tracks rather than integrated principles. That’s a structural observation, not an indictment of anyone’s values or intentions. That these documents on leadership and mentorship all contain sections on diversity, equity, inclusion, and belonging speaks to the effectively unanimous belief that DEIB endeavors are important and valuable to the profession of veterinary medicine.
But I’m also more than skeptical enough to understand that nuance doesn’t always survive first contact with the internet. I’d sooner leave the authors and their specific works out of it than watch a good-faith analysis become ammunition for bad-faith actors.
So you’ll have to trust me on the numbers. I didn’t make them up. I’ll show you the methodology. But the sources stay private.
One more thing before we begin. I know my voice can be intense and words often land with sledgehammer subtlety when a lighter touch might be more effective. I’ve worried that the messenger might overwhelm the message here, that because a guy whose intensity runs high and engine runs hot said it, the substance would get lost in the noise2. But I’d rather risk being misheard than stay quiet about something I think matters to how we train the next generation of veterinarians.
Here’s what I found.
You Can Add Up the Parts, but You Won’t Have the Sum
You can build the coolest stuff with Claude Code, and instead of doing that I used it to build software that would process and analyze the content of a bundle of leadership and mentorship documents I collected from various sources concerned with veterinary medicine.
I had the software go line-by-line in these documents, and analyze the content of each sentence. For sake of this essay, I’ve broken down the analysis into three categories. “Core mechanic” language, sentences that impart fundamental knowledge and insight on matters of leadership and/or mentorship to the reader. DEIB language, sentences that contained keywords and phrases regarding DEIB topics (as defined by the program itself). And overlapping language, sentences that including both core mechanics and DEIB language.
The results across all documents:
Core Mechanics: 42%
DEIB Language: 7%
Overlapping Language: 2%
I’ll again note that the documents were titularly specific to leadership and mentorship, all of the documents included sections on DEIB. This is not an indictment of the documents, but rather, I’ve come to believe, in how they are arranged.
I don’t know if 42% of sentences dedicated to core mechanics is appropriate. I don’t know if 7% of sentences pertaining to DEIB language, especially when the topic of the documents was leadership and/or mentorship, is appropriate. But the finding that grabbed my attention is the decided lack of overlap. “Separate but equal” is a rightfully fraught phrase, yet it’s one that slips into the mind on reviewing these findings.
When we separate, section, segment, or compartmentalize, the DEIB initiatives begin to feel less like core values and more like an extra credit assignment. Something “nice to have” rather than vital to the health and robustness of the profession.
You Can Strike Up the March, There is No Drum
Six times as much content dedicated to core mechanics than to DEIB, and hardly any overlap between the two. And yet, reading them, the DEIB sections are conspicuous and ubiquitous. They are utterly unmissable in reviewing any of the documents.
Why do we have the perception of overemphasis or sense of over-representation of these topics when they are demonstrably not overemphasized or over-represented?
It seems that there are a number of factors at play here.
First, the DEIB content tends to be clustered, concentrated in its own section rather than woven throughout. This may be a deliberate choice of a single author (“here is the section on DEIB”) or an organizational decision (“the expert on DEIB in this group of authors will write the section on DEIB”).
When the content does appear, it often carries a different tone. The mechanics are procedural: here’s how to set goals, here’s how to give feedback, here’s how to structure a meeting. The DEIB language is more frequently prescriptive: you must recognize, you are required to ensure, it is essential to acknowledge. Neither style is appropriate or inappropriate in either circumstance, but the contrast of language is noticeable.3
And I think there’s an irony worth naming, one that perhaps helps to make the DEIB sections stand out to the reader. Many of these documents explicitly caution against paternalism in mentorship, that is the tendency to impose one’s views, to assume you know what’s best for someone else, or to substitute direction for collaboration. And yet the DEIB sections often have exactly that register.
Again, not inappropriately. The guidelines on leadership and mentorship amount to, more or less, good advice. The sections on DEIB often reflect the standards set forth in federal employment law or the Civil Rights Act. An author conveying those messages codified by legislation and caselaw indisputably has more weight behind their words than someone relating sociological, psychological, or organizational research.
I think that these differences are highlighted by the structure of these documents. When DEIB is deliberately separated from core content, you create two different rhetorical modes. The DEIB language accumulates more intensity because its case is made in isolation, without the procedural scaffolding to anchor it.
The result is a document that says diversity matters while structurally indicating diversity is separate.
There is a Crack in Everything
I don’t think I’ve done anything to indicate the notion, but I want to be explicit: I am not advocating against diversity initiatives. I am arguing that the structure of how we present them in certain contexts could be undermining their effectiveness of acceptance, implementation, and impact.
The research on diverse teams is more nuanced than many of its champions or critics tend to acknowledge. A 2019 meta-analysis on team creativity and innovation found that what is known as “deep-level diversity” - differences in values, perspectives, problem-solving approaches, etc. - is positively associated with team creativity and innovation.
But surface-level diversity alone? Those characteristics that are, ahem, skin deep? The demographic characteristics that are immediately visible or easily observable, like race, gender, age, ethnicity, or physical appearance are not associated with a significant improvement in team creativity and innovation.
The research shows that the benefits of diversity are not automatically imparted because we have assembled a team possessing widely disparate Social Security Numbers, mismatched chromosomes, or varying amounts of melanin. The benefits of diversity emerge when teams are working on complex, interdependent tasks and members actively engage with one another’s different ways of thinking.
This distinction matters. We are not assembling teams to sort widgets. Veterinary medicine is a complex and wicked system. We diagnose disease that doesn’t announce itself in patients who cannot speak for themselves. We manage cases where the textbook all but shrugs at us, we navigate client relationships that require an ever-increasing degree of sophisticated communication and emotional intelligence. These are precisely the conditions under which cognitive diversity, the sort that comes from genuinely different life experiences and perspectives, becomes an asset.
Further, the value has to be cultivated. A 2016 meta-analysis of over 260 diversity training studies found that integrated approaches, those where diversity principles are woven throughout a curriculum rather than sectioned off, produced significantly stronger effects on both behavior and attitudes than did standalone modules. And it wasn’t close. For behavioral learning, integrated training nearly doubled the impact.
There’s a warning embedded in more recent work too: a 2025 review examined the unintended consequences of diversity initiatives - the authors called it “backfire” effects. When diversity programs are implemented in ways that signal special treatment, the programs can inadvertently increase perceptions that target groups lack competence. The initiatives meant to help end up reinforcing the stereotypes they were designed to combat.
And, predictably, the relationship between initiative strength and effectiveness isn’t linear; it’s an inverted U. Push too hard, overemphasize the message, and you reinforce stereotypes. Pull back, do too little, and you’ve done nothing to change perception or behavior.
The through-line for all of this is integration. Diversity principles work best when they’re embedded in the fabric of how we lead, teach, and mentor. They work less well when they’re announced as a separate agenda item.
Compartmentalization then doesn’t just feel like an afterthought, according to a pile of peer-reviewed research, it makes DEIB initiatives function like one.
Which brings us back to the leadership and mentorship documents collected in veterinary medicine.
Ring the Bells that Still Can Ring, Forget Your Perfect Offering
So what do we do about it? I’ve offered a problem list, a diagnosis, what’s the treatment plan? What would it look like to weave DEIB principles into the fabric of mentorship guidance rather than sectioning them off?
Consider the difference between these two framings:
Compartmentalized:
Chapter 4: Setting Goals with Your Mentee
Chapter 7: Considerations for Mentoring Across Difference
Integrated:
“When setting goals with your mentee, recognize that their aspirations may have been shaped by experiences you haven’t had. A first-generation veterinarian may not know what questions to ask about partnership tracks. A mentee from an underrepresented background may have learned to keep ambitions modest as a protective strategy. Curiosity serves far better than assumption. Your job isn’t to assume what they want, it’s to facilitate opportunities for them to discover it.”
The first approach teaches goal-setting, then separately teaches you to think about identity. The second approach teaches goal-setting in a way that already accounts for difference. The DEIB insight isn’t an add-on, it’s not extra credit, it’s part of the core skill.
Or consider feedback, a core element of both leadership and mentorship. A compartmentalized document might have a section on delivering constructive criticism, and then a separate section reminding you to be aware of cultural differences in communication styles. An integrated approach would note, right there, in the feedback section, that directness lands differently depending on someone’s background.4 That a mentee who grew up in a household, culture, or country where questioning authority was discouraged may need explicit permission to push back. That silence doesn’t always mean agreement.5
This isn’t necessarily about changing the amount of DEIB content, I’ve stated outright that isn’t what I researched or studied. It’s about changing the means of delivery, placing material - material we’ve already agreed to be valuable - differently, in places where it, according to research, will have more of the desired impact. The same insights, distributed differently, stop feeling like a lecture and start feeling like practical strategy that serves the practice of veterinary medicine.
Years ago, I wrote a snarky criticism of a set of leadership guidelines. I reread it leading up to this piece, and one line still rings true: “we might seek mentorship from people who aren’t like us, who don’t share our worldview, but with whom we connect through our shared pursuit of medicine.”
That’s integration. That’s not “diversity is important” as a standalone assertion, with no more weight than a bumper sticker or a hashtag. That’s diversity built into the definition of what good leadership and good mentorship looks like. It reframes difference not as an obstacle to navigate but as a resource to pursue.
The documents I reviewed aren’t wrong to include DEIB content, but I they’ve missed an opportunity in quarantining it.
That’s How the Light Gets In
These guidance documents that I evaluated are not neutral. They don’t just describe best practices, they shape those practices. Often for years.
When a professional organization publishes guidelines on leadership or mentorship, that organization is making a claim about what leadership or mentorship is, about what matters, about to whom we ought pay attention, and what doing it right looks like. Those messages of those documents trickle down until they become part of the water in which we swim. They set the terms for a generation of veterinary professionals.
We take this seriously in clinical medicine. When we write guidelines on managing hyperthyroidism or kidney disease, we bring considerable rigor and we also insist on a framework that takes us from understanding, to diagnosis, to treatment, to follow-up. It’s a structure set forth and shaped by guidelines and clinical insights that’s been shaped by literal centuries of research and education.
I want that same rigor and thoroughness brought to how we write about all aspects of mentorship and leadership. Not because the current approach to DEIB content is good, bad, wrong, right, excessive, or lacking, but because the current structural isolation may be undermining its own goals. Compartmentalization sends a signal, whether intended or not: here’s the diversity bit, it’s not a part of the core content.
I’m entirely confident that no author on any of the source materials intended that message at all. But that is what the structure indicates, and there’s research to prove it.
I don’t have a clean prescription here. I’m not going to pretend rewriting dozens of these documents is simple, cheap, or easy. Integration is not just a matter of moving some paragraphs. Rewriting these documents, writing new documents, will require rethinking how we conceptualize the relationship between professional skills and cultural awareness. We need not to treat them as parallel tracks but a single track, simply one viewed from different angles.
The research is extensive and clear: integration works better than compartmentalization. The perception data from my modest analysis suggests that a reader easily picks up on the structural separation, even if they can’t articulate why. Hell, it’s taken me 2500 words to get close to making the point.
The reason why is that the stakes are high, shaping the next and the next and the next generation of how veterinarians learns to teach, sponsor, and support one another warrant the effort.
We can do better by saying it differently. By weaving together rather than bolting it on. We deserve guidance documents that practice what they preach.
And, much more importantly, so do all those who come after us.
Claude Code is an insanely useful tool.
Yes, I’m aware of my own demographic and the potential for more than mere irony of writing critically about professional guidance documents in a field where men are the minority. I’m not claiming persecution, and I’m not quite naive enough to think that gender dynamics are as a simple as statistics and headcounts. But that’s a footnote, not the argument.
It is noticeable, I believe, even to someone who didn’t use the world’s most advanced commercially available natural language coding model to build an analytical engine for a fistful of leadership and mentorship documents in veterinary medicine.
I, for one, still benefit from being reminded of that occasionally.
During one of my wife’s first visits to a family dinner at my folks’ house, my sisters were yelling at each other in the kitchen. My wife was concerned. “Is everything alright” she asked. I was confused. “Aren’t they fighting?” she pressed. No, they were just yelling at each other. And sometimes swearing. It was learning moment for both of us.