Abstract
This paper addresses a prior question that most analyses of racial disparity in professional fields never ask: Why are unequal outcomes so consistently attributed to discrimination rather than to formation deficits, temporal lag, or structural constraint? The question is not rhetorical. It identifies a recurring pattern in institutional, academic, and public reasoning—one in which moral explanations for demographic disparity reliably displace temporal, structural, and formational explanations, even when the latter possess greater explanatory power. This paper introduces the concept of constraint misattribution cascades to describe the mechanism by which this displacement occurs, demonstrates how moral explanation crowds out temporal explanation through a series of identifiable cognitive and institutional steps, and provides a diagnostic checklist for recognizing attribution errors in disparity analysis. The paper is not an argument against the existence of discrimination. It is an argument against the inferential habit of treating discrimination as the default explanation for every observed gap, and it provides the conceptual vocabulary necessary for distinguishing between genuine discriminatory mechanisms and the misidentification of non-discriminatory constraints as evidence of bias.
I. The Problem of First Attribution
When a disparity is observed—when a high-skill field is found to contain a demographic distribution that deviates from the composition of the general population or from some other chosen baseline—something must be said about it. The disparity demands narration. It sits in a report, a study, an institutional self-assessment, a news article, and it asks, silently but insistently: Why?
The answer given to that question in the first instance—the first attribution—exercises a gravitational influence over every subsequent stage of analysis. If the first attribution is “discrimination,” then the analytical apparatus that follows will be organized around locating, measuring, and remediating discrimination. If the first attribution is “pipeline deficit,” the apparatus reorganizes around formation and access. If it is “temporal lag,” the apparatus orients toward historical sequence and generational development. The first attribution does not merely describe; it directs. It establishes the investigative trajectory, determines which data are sought and which are ignored, and—most consequentially—defines which conclusions are available before a single additional fact is examined.
This paper is concerned with the pattern by which the first attribution, across a remarkably wide range of fields and institutional contexts, defaults to discrimination. Not discrimination as a carefully identified mechanism with specific operational characteristics, but discrimination as a general atmospheric condition inferred from the disparity itself. The pattern is so widespread and so consistent that it cannot be explained by the particularities of any single field. It requires a structural account—an explanation of why the explanatory machinery of contemporary institutional life is calibrated to produce this specific output with such reliability.
The account offered here is not conspiratorial. It does not propose that actors within institutions are deliberately concealing non-discriminatory explanations in order to advance a racial narrative. What it proposes is something more mundane and, for that reason, more durable: that the cognitive, institutional, and social incentive structures surrounding disparity analysis are arranged in a way that makes discriminatory attribution the path of least resistance, and that this path-of-least-resistance dynamic produces systematic misattribution at scale.
II. Defining Constraint Misattribution
A constraint is any condition that limits the rate, volume, or demographic composition of entry into or advancement within a professional field. Constraints are ubiquitous. Every field has them. They include the cost of training, the geographic distribution of training institutions, the length of apprenticeship or formation required, the availability of mentorship, the economic viability of early-career compensation, the prerequisite knowledge or skill base required for entry, and the institutional structures through which credentials are obtained and evaluated.
Constraints are not, in themselves, discriminatory. A field that requires fifteen years of intensive preparation beginning in childhood is constrained by that requirement regardless of who is attempting to enter it. A field whose training infrastructure is concentrated in four cities is constrained by that concentration regardless of the racial composition of those cities’ populations. A field that pays poorly for the first decade of professional life is constrained by that economic reality regardless of which demographic groups find that reality more or less tolerable given their existing wealth and resource structures.
However—and this is the critical turn—constraints do not operate on all populations equally, because the preconditions for meeting constraints are not equally distributed. A field that requires early childhood formation in a particular skill will draw disproportionately from populations that have access to early childhood instruction in that skill. A field that requires the economic capacity to sustain years of low or no income during training will draw disproportionately from populations with existing wealth or family support structures capable of subsidizing that period. These differential impacts are real. They produce observable demographic skew. And they demand explanation.
Constraint misattribution occurs when the demographic skew produced by a non-discriminatory constraint is attributed to a discriminatory mechanism. The constraint is real. The skew it produces is real. But the explanation assigned to the skew is wrong—not morally wrong, but causally wrong. It identifies the wrong variable as operative. And because causal misidentification produces remedial misidentification, the interventions that follow from constraint misattribution are directed at a mechanism that is not, in fact, producing the observed outcome.
This is not an abstract problem. It has institutional consequences that are measurable and significant. When a field misattributes a pipeline constraint to a gatekeeping bias, it reorganizes its admissions, hiring, or promotion processes to counteract a bias that is not the operative cause of the disparity. The reorganization may or may not produce demographic change at the point of intervention—it often does, because direct demographic intervention at a specific gatekeeping stage can override whatever mechanism is actually operative—but it does not address the source of the disparity, which continues to operate upstream. The result is a field that has altered its internal processes, often at significant cost to its evaluative standards, without changing the underlying condition that produced the skew in the first place.
III. The Cascade Structure
Constraint misattribution does not typically occur as a single inferential error. It occurs as a cascade—a sequence of individually understandable steps, each of which makes the next more likely, and which together produce a self-reinforcing attribution pattern that becomes extremely difficult to interrupt once established. The cascade has identifiable stages, and understanding these stages is essential to diagnosing the pattern when it appears.
Stage One: Disparity Observation
The cascade begins with the observation of a demographic disparity. This observation is usually statistical: a field reports its demographic composition, or a study measures the representation of various groups at different levels of the field, or an institution conducts an internal audit. The observation is, at this stage, purely descriptive. It identifies a number—a percentage, a ratio, a gap—and presents it as a fact requiring explanation.
The observation stage is not, in itself, problematic. Demographic data are useful and sometimes necessary for understanding the composition of professional fields. What matters is what happens next: how the observation is framed, who frames it, and within what interpretive context the framing occurs.
Stage Two: Baseline Selection
The observed disparity must be measured against something. The question “Is this field underrepresenting Group X?” requires a denominator—an expected representation against which the actual representation is compared. The selection of this baseline is one of the most consequential and least examined decisions in all of disparity analysis.
The most commonly selected baseline is general population proportion. If Group X constitutes thirteen percent of the national population, then thirteen percent representation in any given field is treated as the expected norm, and any deviation below that figure is treated as a deficit requiring explanation. This baseline has the virtue of simplicity and the authority of democratic intuition: in a society committed to equal opportunity, should not every field roughly mirror the society it serves?
But general population proportion is, for most high-skill fields, the wrong baseline, and its wrongness is not subtle. It assumes that the only variable standing between population proportion and field representation is the field’s own selection process—that absent discrimination, demographic proportionality would naturally obtain. This assumption is false for any field that has prerequisites which are themselves unequally distributed. If a field requires a doctoral degree, and doctoral degree attainment varies across racial groups for reasons that include but are not limited to discrimination—reasons that also include differential access to undergraduate preparation, differential economic capacity to sustain doctoral training, differential geographic proximity to doctoral institutions, and differential cultural valuation of particular career paths—then the relevant baseline is not the general population but the population of doctoral degree holders. And that population is already skewed before the field’s own selection processes come into play.
The selection of general population as baseline is the first amplifier in the cascade, because it guarantees that the measured disparity will be larger than the disparity that is actually attributable to the field’s own practices. It inflates the signal before analysis begins.
Stage Three: Explanation Compression
Once a disparity has been observed and measured against a baseline that magnifies it, the explanation must be generated. This is the stage at which the most consequential compression occurs.
In principle, the explanatory space for a demographic disparity in a professional field is vast. It includes every factor identified in Section II of this paper—formation deficits, temporal lag, geographic concentration, economic barriers, prerequisite distribution, cultural factors, institutional structure—as well as discriminatory mechanisms at various levels. In practice, however, the explanatory space that is actually explored in most institutional and academic analyses is dramatically narrower.
The compression happens for several identifiable reasons. First, many of the non-discriminatory explanations are complex, multi-causal, and resistant to summary. They require detailed knowledge of the field’s formation pipeline, its economic structure, its historical development, and the specific mechanics of its prerequisite system. They do not fit neatly into a press release, a diversity report, or an op-ed. Discriminatory explanations, by contrast, are narratively compact. “The field is biased” is a complete sentence. “The field’s demographic composition reflects the interaction of a fifteen-year formation pipeline with differential access to early childhood instruction in a context where the geographic distribution of training institutions intersects with residential segregation patterns established by mid-twentieth-century housing policy, compounded by the economic non-viability of the field’s early-career compensation structure for individuals without pre-existing family wealth” is not a sentence that survives institutional communication.
Second, non-discriminatory explanations are distributionally uncomfortable. They point to causes that are diffuse, upstream, and not within the control of the institution conducting the analysis. An institution that concludes “our field’s demographic composition is primarily the result of constraints that operate before candidates ever reach us” has identified a problem it cannot solve by changing its own practices. An institution that concludes “our field’s demographic composition is primarily the result of our own biased practices” has identified a problem it can, in principle, address. Institutional actors prefer actionable findings, and discriminatory attribution is more institutionally actionable than constraint attribution, even when constraint attribution is more accurate.
Third—and this is the factor most resistant to frank discussion—non-discriminatory explanations carry social cost. An analyst who proposes that a disparity is the product of formation deficits rather than institutional bias is exposed to the accusation that he or she is “blaming the victim,” denying racism, or providing cover for discriminatory practices. The accusation does not need to be accurate to be professionally damaging. The asymmetry of social risk is stark: attributing a disparity to discrimination carries negligible professional risk in most institutional contexts, while attributing it to non-discriminatory causes carries significant professional risk. Rational actors respond to incentive structures, and this incentive structure reliably produces discriminatory attribution regardless of the underlying causal reality.
Stage Four: Moral Displacement of Temporal Explanation
This is the stage at which the cascade becomes self-reinforcing and extremely difficult to reverse. Once a discriminatory explanation has been offered and institutionally adopted, it acquires moral weight. To question it is no longer merely to offer an alternative causal account; it is to challenge a moral commitment. The explanation has been incorporated into the institution’s self-understanding, its public communications, its strategic planning, and often its compliance obligations. It has become, in a meaningful sense, identity rather than hypothesis.
At this stage, temporal explanations—explanations that locate the disparity’s origin in historical sequence, generational development, or the simple arithmetic of pipeline maturation—are not merely crowded out. They are morally displaced. To offer a temporal explanation is to be heard as saying “be patient,” and to say “be patient” is to be heard as saying “the current state of affairs is acceptable.” The temporal explanation is received not as a diagnostic claim but as a political one—an endorsement of delay, a counsel of complacency, a defense of whoever benefits from the status quo.
This moral displacement is deeply consequential, because temporal explanations are, for many observed disparities in high-skill fields, the most powerful explanatory category available. The arithmetic of pipeline development is unforgiving. If a population was excluded from a field’s formative institutions until a specific date, and the field requires a specific number of years of formation to reach professional competence, and a further number of years to reach the upper echelons, then the timeline for proportional representation is calculable. It is not a matter of opinion. It is a matter of addition. But this arithmetic is, in the current discursive environment, functionally unspeakable in many institutional contexts, because it has been pre-categorized as an expression of racial indifference rather than recognized as an exercise in causal reasoning.
Stage Five: Institutional Lock-In
The final stage of the cascade is institutional lock-in. Once a discriminatory explanation has been adopted, moralized, and embedded in institutional practice, it generates its own evidentiary ecosystem. Diversity officers are hired whose professional existence depends on the continued validity of the discriminatory explanation. Reports are commissioned that measure the field’s progress against benchmarks derived from the discriminatory attribution. Training programs are implemented that address the biases identified by the discriminatory framework. External evaluators assess the institution’s performance against standards that presuppose the discriminatory explanation.
None of these institutional developments are insincere. Most are staffed by people who genuinely believe in the explanation they are operationalizing. But sincerity does not ensure accuracy, and the institutional apparatus, once constructed, has no mechanism for revisiting its foundational attribution. The question “What if the disparity is not primarily caused by what we think it is caused by?” is not a question the apparatus is designed to ask, because asking it threatens the apparatus itself.
The cascade is now complete. An observation has become a baseline has become an explanation has become a moral commitment has become an institutional infrastructure. And at each stage, the possibility of alternative attribution has narrowed until, at the final stage, it has been effectively foreclosed.
IV. Moral Crowding: The Mechanism of Explanatory Displacement
The cascade described above depends, at its most critical juncture, on a specific cognitive and social phenomenon: the tendency of moral explanations to crowd out non-moral explanations in contexts where both are available. This phenomenon—which this paper terms moral crowding—operates through identifiable mechanisms that deserve independent examination.
The Salience Asymmetry
Moral explanations are more cognitively salient than structural or temporal explanations. This is not a deficiency of reasoning; it is a feature of human cognition. The human mind is, as extensive research in cognitive psychology has established, preferentially attuned to agent-based explanations—explanations that identify an actor, an intention, and a consequence (Heider, 1958; Malle, 2004). Structural explanations, which identify impersonal forces, systemic configurations, and emergent outcomes, are cognitively more demanding and less narratively satisfying. Temporal explanations, which locate causation in sequence and duration rather than in agency, are the least salient of all, because they offer no protagonist, no antagonist, and no resolution—only arithmetic and patience.
When a disparity is observed and multiple explanatory frameworks are available, the moral-agentive framework (“someone is doing this to someone”) will reliably be more salient, more memorable, and more communicable than the structural-temporal framework (“this is the result of a configuration that developed over decades for reasons that do not reduce to anyone’s intentions”). The salience asymmetry does not mean the moral explanation is wrong. It means the moral explanation will be selected more readily than its competitors regardless of its accuracy, because cognitive selection is driven by salience, not by truth value.
The Unfalsifiability Gradient
A second mechanism by which moral explanations crowd out non-moral ones is the unfalsifiability gradient—the degree to which an explanation is resistant to empirical disconfirmation. Temporal and structural explanations are, in principle, testable. If a temporal explanation predicts that a disparity should diminish at a specific rate as pipeline maturation proceeds, that prediction can be checked against data. If a structural explanation predicts that a disparity should be concentrated at specific stages of the professional pipeline where specific constraints operate, that prediction can be checked against the distribution of the disparity across stages.
Discriminatory explanations, as they are typically formulated in institutional and academic contexts, are significantly more resistant to disconfirmation. If a field’s demographic composition does not change after bias-reduction interventions, the explanation is not that bias was not the primary cause; the explanation is that the interventions were insufficient, or that bias is more deeply embedded than initially understood, or that “structural” racism operates independently of individual bias and therefore survives individual-level interventions. Each of these secondary explanations may be true in a given case. But the pattern—in which the discriminatory attribution is preserved regardless of the outcome of interventions designed to test it—is characteristic of explanatory frameworks that have become unfalsifiable. And unfalsifiable explanations, whatever their moral authority, are not diagnostic tools. They are commitments.
The Social Enforcement Mechanism
The third and most powerful mechanism of moral crowding is social enforcement. In most institutional contexts where disparity analysis occurs—universities, corporations, professional associations, government agencies—the social costs of offering a non-discriminatory explanation for a racial disparity are markedly higher than the social costs of offering a discriminatory one. This asymmetry is well-documented in the literature on preference falsification and spiral-of-silence dynamics (Kuran, 1995; Noelle-Neumann, 1974) and need not be belabored here, but its effects on the explanatory landscape must be acknowledged.
The social enforcement mechanism operates not by forbidding non-discriminatory explanations but by raising their cost. An analyst who offers a formation-based explanation for a disparity does not face formal censure; the analyst faces informal costs—reputational suspicion, exclusion from collaborative projects, characterization as racially insensitive, and the persistent imputation of concealed motives. These costs are sufficient to shift the distribution of publicly offered explanations toward the discriminatory end of the spectrum, not because analysts uniformly believe discriminatory explanations are correct, but because offering them is professionally safer. The result is an explanatory environment in which discriminatory attribution appears to enjoy overwhelming consensus, when in fact it enjoys overwhelming compliance—a distinction that is invisible from the outside but structurally decisive.
V. The Temporal Explanation and Its Suppression
Because temporal explanations are the category most consistently displaced by moral crowding, they deserve extended treatment. A temporal explanation for a demographic disparity in a high-skill field has the following general structure:
- The field requires a formation period of n years, beginning at age y, to produce a competent professional.
- The population in question was effectively excluded from the field’s formative institutions until date d, due to legal segregation, economic exclusion, or institutional policy.
- The first cohort capable of completing the full formation pathway therefore could not have entered the field’s professional ranks until date d + n.
- Advancement to the field’s upper echelons requires an additional m years of professional experience and accomplishment.
- Proportional representation at the field’s upper echelons therefore cannot be expected until date d + n + m, assuming no other constraints operate.
This structure is arithmetically straightforward. It makes predictions that are empirically testable. And in a remarkable number of high-skill fields, it accounts for a substantial portion of the observed disparity without any reference to ongoing discrimination. This is not because ongoing discrimination does not exist; it is because the temporal arithmetic is so powerful that it would produce significant underrepresentation even in a field with zero ongoing discrimination, simply as a consequence of the lag between access and maturation.
The suppression of temporal explanation is one of the most costly consequences of moral crowding, because temporal explanation is the category most capable of distinguishing between two fundamentally different situations: a field in which disparity is being maintained by ongoing discrimination, and a field in which disparity is diminishing along a trajectory consistent with pipeline maturation. These two situations look identical in any single cross-sectional snapshot—both show underrepresentation at a given moment—but they are causally and morally distinct, and they require different responses. Temporal explanation is the tool that distinguishes them. Its suppression is therefore not merely an intellectual loss; it is a diagnostic loss that degrades the quality of every institutional response built on the analysis from which it has been excluded.
VI. A Diagnostic Checklist for Attribution Errors
The foregoing analysis implies a set of diagnostic questions that should be asked whenever a demographic disparity in a high-skill field is attributed to discrimination. These questions are not designed to disprove discriminatory explanations; they are designed to test whether a discriminatory explanation has been earned through investigation or assumed through default.
1. Has the baseline been specified and justified? Against what expected representation is the disparity being measured? Is the baseline the general population, the pipeline population, the applicant population, or some other denominator? Has the choice of baseline been defended, or has it been silently adopted? Would a different defensible baseline produce a different magnitude of disparity?
2. Has the formation timeline been calculated? For the field in question, what is the length of the formation pipeline from initial entry to professional competence? From professional competence to upper-echelon status? When did the population in question gain effective access to the field’s formative institutions? Does the arithmetic of pipeline maturation account for some or all of the observed disparity?
3. Have structural constraints been identified and quantified? What are the non-discriminatory constraints that limit entry into or advancement within the field? Cost of training? Geographic distribution of training institutions? Economic viability of early-career compensation? Availability of prerequisite formation? Do these constraints operate differentially across racial groups for reasons that are independent of the field’s own practices?
4. Has the disparity been decomposed by pipeline stage? Is the observed underrepresentation concentrated at a specific stage of the professional pipeline—entry, mid-career, upper echelon—or is it uniform across stages? Concentration at a specific stage suggests a stage-specific cause; uniformity suggests an upstream cause. Has this decomposition been performed?
5. Has the trend been examined? Is the disparity stable, increasing, or decreasing over time? If decreasing, is the rate of decrease consistent with pipeline maturation arithmetic? If stable or increasing despite the passage of time since access was gained, what specific mechanism is maintaining or increasing it?
6. Has a specific discriminatory mechanism been identified? If the explanation offered is discrimination, what is the mechanism? Is it a specific practice, policy, evaluation criterion, or institutional structure? Can it be described with enough specificity to be empirically tested? Or is the discriminatory attribution an inference from the disparity itself—a circularity in which the outcome is both the evidence and the conclusion?
7. Has the explanation survived alternative causal testing? Has the discriminatory explanation been tested against non-discriminatory alternatives? Have formation, temporal, structural, and economic explanations been generated and checked against the data? Has the discriminatory explanation been shown to account for variance that the alternatives cannot, or has it simply been offered first and never challenged?
8. Is the institutional apparatus capable of revising its foundational attribution? Does the institution’s analytical and operational structure include a mechanism for revisiting the original attribution if new evidence suggests it is incorrect? Or has the attribution been locked into institutional infrastructure—staffing, reporting, compliance—in a way that makes revision structurally impossible?
These eight questions do not constitute an algorithm. They do not guarantee a correct attribution. But they impose a discipline on the attributive process that is, in most current institutional practice, entirely absent. Their consistent application would not eliminate discriminatory attribution; it would ensure that discriminatory attribution, when offered, has been tested against alternatives and found to be the explanation with the greatest warrant. That is all any diagnostic standard can do. It is considerably more than is being done now.
VII. Implications for the Series
This paper has introduced a conceptual vocabulary—constraint misattribution, the cascade structure, moral crowding, temporal displacement, the diagnostic checklist—that will be employed throughout the subsequent volumes of this series. Each volume will examine a specific high-skill field in which accusations of structural racism have achieved varying degrees of institutional authority. Each will apply the diagnostic checklist developed here, not as a mechanical exercise but as a disciplinary framework for ensuring that the field’s specific causal picture is examined with the precision it demands.
No volume will begin with the assumption that a field is or is not racist. Each will begin with the assumption that the field’s demographic composition has causes, that those causes are discoverable, and that discovering them accurately is both possible and necessary. The present paper provides the tools for that discovery. The volumes provide the occasions.
What this paper has demonstrated is not that discrimination is never the cause of demographic disparity in high-skill fields. It has demonstrated that the inferential pathway from observed disparity to discriminatory attribution is, in its current institutional form, systematically biased toward discriminatory conclusions by identifiable cognitive, social, and institutional mechanisms. These mechanisms do not operate because people are dishonest. They operate because the explanatory environment is structured in a way that makes discriminatory attribution easier, safer, and more institutionally functional than the alternatives—regardless of accuracy.
Correcting this bias does not require denying discrimination. It requires restoring the full explanatory field: making temporal, structural, and formational explanations available, testable, and socially permissible alongside discriminatory ones. It requires, in other words, making diagnosis possible again in a domain where diagnosis has been largely replaced by attribution.
That is the work this paper begins. The volumes complete it.
Notes
- The term constraint misattribution as used throughout this paper should be distinguished from the broader concept of misattribution studied in social psychology, which typically refers to errors in identifying the cause of one’s own emotional or physiological states (Schachter & Singer, 1962). The present usage is narrower and more specific: it refers to the misidentification of the cause of a demographic pattern at the institutional or field-wide level. The two phenomena share a family resemblance—both involve assigning a causal role to the wrong variable—but they operate at different scales and through different mechanisms.
- The concept of an explanatory hierarchy in which temporal causes are consulted before moral causes is not novel to this paper. It draws on a long tradition in the philosophy of science that distinguishes between proximate and distal causation and insists that causal claims be ordered by their temporal and logical priority (Aristotle, Physics II.3; Salmon, 1984). What is novel is the application of this principle to the specific domain of racial disparity analysis, where the ordering of causal claims has been, as this paper argues, systematically inverted.
- The baseline selection problem described in Stage Two of the cascade has been noted by several researchers in the quantitative analysis of disparate impact. Wax (2009) has argued that the use of general population baselines in employment discrimination analysis produces systematically inflated disparity estimates, while Sowell (2019) has demonstrated across multiple national contexts that demographic proportionality in professional fields is historically anomalous rather than normative, even in the absence of discrimination. The present paper does not adopt either author’s specific policy conclusions but treats their empirical observations as methodologically significant.
- The distinction between accessibility and mastery—introduced in the prolegomenon to this series and presupposed throughout this paper—has analogs in the educational literature on the difference between access and attainment. See Bowen and Bok (1998) for a detailed examination of this distinction in the context of selective university admissions, though the present paper does not share all of that study’s normative commitments.
- The concept of preference falsification invoked in the discussion of social enforcement mechanisms was developed most comprehensively by Kuran (1995), who demonstrated that public expressions of belief can deviate systematically from private belief when social costs attach to dissent. The application of this concept to the specific domain of disparity attribution is, to the author’s knowledge, original to this paper, though the underlying dynamic is well established in the social science literature.
- The unfalsifiability gradient described in Section IV is related to, but distinct from, Popper’s (1959) criterion of falsifiability as a demarcation between scientific and non-scientific claims. The present paper does not argue that discriminatory explanations are non-scientific. It argues that, as institutionally practiced, they frequently exhibit a resistance to disconfirmation that degrades their diagnostic utility, even when their underlying theoretical structure is capable of generating testable predictions.
- The diagnostic checklist in Section VI is designed as an analytical tool, not as a legal standard. Its questions are not equivalent to the burden-shifting framework established in McDonnell Douglas Corp. v. Green (1973) or its progeny, though there are structural similarities. The legal question of whether a specific practice constitutes actionable discrimination is governed by statutes, precedents, and evidentiary standards that are beyond the scope of this paper. The analytical question of whether a specific attribution is causally warranted is governed by the standards of causal reasoning outlined here.
- The phrase epistemic zoning law, used in the prolegomenon to describe the function of the series’ methodological framework, captures something important about the present paper’s ambitions. Zoning law does not tell anyone what to build; it tells them what kinds of structures are permissible in which locations. Similarly, this paper does not tell anyone what to conclude about any specific field’s demographic composition. It specifies what kinds of inferential structures are permissible in the analytical space this series occupies. Arguments may be built freely within these zones. They may not be built outside them.
References
Aristotle. (350 B.C.E.). Physics (R. P. Hardie & R. K. Gaye, Trans.). In J. Barnes (Ed.), The complete works of Aristotle (Vol. 1). Princeton University Press.
Bowen, W. G., & Bok, D. (1998). The shape of the river: Long-term consequences of considering race in college and university admissions. Princeton University Press.
Heider, F. (1958). The psychology of interpersonal relations. Wiley.
Kuran, T. (1995). Private truths, public lies: The social consequences of preference falsification. Harvard University Press.
Malle, B. F. (2004). How the mind explains behavior: Folk explanations, meaning, and social interaction. MIT Press.
McDonnell Douglas Corp. v. Green, 411 U.S. 792 (1973).
Noelle-Neumann, E. (1974). The spiral of silence: A theory of public opinion. Journal of Communication, 24(2), 43–51. https://doi.org/10.1111/j.1460-2466.1974.tb00367.x
Popper, K. R. (1959). The logic of scientific discovery. Hutchinson.
Salmon, W. C. (1984). Scientific explanation and the causal structure of the world. Princeton University Press.
Schachter, S., & Singer, J. (1962). Cognitive, social, and physiological determinants of emotional state. Psychological Review, 69(5), 379–399. https://doi.org/10.1037/h0046234
Sowell, T. (2019). Discrimination and disparities (Rev. ed.). Basic Books.
Wax, A. L. (2009). Disparate impact realism. William & Mary Law Review, 53(2), 621–693.
