Profiles, Probabilities, and Stereotypes.

• Author: Stein, Michael Ashley
• Position: Book Review

PROFILES, PROBABILITIES, AND STEREOTYPES. By Frederick Schauer. Cambridge: Harvard University Press. 2003. Pp. xiii, 359. $29.95.

INTRODUCTION

Published in 1949, Joseph Tussman and Jacobus tenBroek's article The Equal Protection of the Laws (1) has exerted longstanding influence on subsequent Fourteenth Amendment scholarship. (2) Insightfully, Tussman and tenBroek identified a paradox: although the very notion of equality jurisprudence is a "pledge of the protection of equal laws," laws themselves frequently classify individuals, and "the very idea of classification is that of inequality." (3) Notably, classification raises two sometimes concurrent varieties of inequality: over-inclusiveness and under-inclusiveness. Of these, over-inclusiveness is a more egregious equal protection violation due to its ability to "reach out to the innocent bystander, the hapless victim of circumstance or association." (4)

Despite this shortcoming of classification, Tussman and tenBroek objected to the process of classification only where the categories were either empirically unsustainable or based on legally proscribed characteristics. (5) The use of classification as a method of administrating policy was not itself opposed by the authors, both of whom were distinguished civil libertarians. (6) According to Frederick Schauer's Profiles, Probabilities, and Stereotypes ("Profiles"), (7) this broad and appropriate acceptance of classification is in stark contrast to current mores, where decisions based on categories and generalizations--what Tussman and tenBroek called classifications--are frequently denigrated as stereotyping, or, even worse, profiling. In response to this now-prevalent sensibility, Schauer defends the morality of using generalizations as a means of mediating modern-day life. He further argues that the use of classifications is inevitable and can also be desirable.

Part I of this Review sets forth Schauer's definitions, theses, and conclusions. Next, Part II critiques some of the assertions presented in Profiles. Finally, Part III extrapolates Schauer's analytical framework on generalizations to employment discrimination under the Americans with Disabilities Act ("ADA"), (8) an area not addressed in the book.

1. GENERALLY SPEAKING, GENERALITIES WORK

   Profiles defends the morality of decisionmaking based on generalizations against a contemporary inclination to equate such decisionmaking with the unseemly practices of stereotyping and profiling. Moreover, Schauer argues that determinations based on categories are both inevitable and useful.

   As a definitional matter, Profiles divides generalizations between spurious categories lacking statistical support and nonspurious categories that are empirically sustainable. The nonspurious category contains two further varieties: universal generalizations that are always true because of either definitional ("all bachelors are unmarried") or empirical ("all humans are less than nine feet tall") reasons, and those generalizations that are relatively truer for members of a particular group than they are in general ("bulldogs tend to have poorer hips than most other dogs," or "teenagers are relatively bad drivers in comparison to the overall driving population") (pp. 7-19). It is this last category, in which decisions are based on largely accurate proxies, where Schauer focuses most of his attention and that he defends on pragmatic grounds.

   Initially, Profiles supports generalizations as "an unavoidable feature of our decision-making existence" (p. 76). This is because we live in a complex society where people simply cannot conduct an in-depth and individualized analysis prior to every decision they make (pp. 75-78). Consequently, Schauer asserts that we utilize generalizations as heuristics on a daily basis, whether in determining that flying from Boston to New York is faster than taking the train or in believing that Ford cars and Hotpoint refrigerators are sound products (p. 76). Accordingly, even assessments we think are individualized are not always so (pp. 101-07). He further notes that sustainable generalizations form the bases for valid and rarely contested determinations in a number of areas (pp. 1-6). The Internal Revenue Service, to cite one example, effectively uses a "discriminant function" score to sort through millions of annual tax returns and identify certain criteria that, on the whole, are more likely to be associated with dishonest filing practices. These include the underreporting of income among cash-paid occupations like waitpersons and taxi drivers and the overly aggressive assertion of charitable and tax shelter deductions by physicians and lawyers (pp. 160-67). Even more prevalent are practices used by insurance actuaries who, every day, apply generalizations as to life expectancy, driving ability, neighborhood safety, and the extent of related risks to person and property (pp. 4-6).

   Nevertheless, because of a deep-seated Western cultural bias in favor of particularity, Schauer maintains that generalizations have developed a bad reputation and are therefore disparaged as stereotypes, or even worse, as profiling (pp. 1-3, 15-19). In making this point, Profiles illustrates the way generalizations have been treated in the past. For instance, the Eleatic Stranger from Plato's The Statesman recognizes the impracticality of tailoring a general rule to each individual. At the same time, however, he states that it would be immoral as well as a "disgrace" not to individualize that general rule whenever possible (pp. 27-41). (9) In a similar vein, Aristotle concluded that an equitable solution should always follow when legal generalizations produce unfair results (pp. 41-48). This concern with individualized justice as a necessary corrective to broadly based decisions motivated the practice of Equity in Western jurisprudence from its inception in Roman law through the role of the Chancery in England (pp. 48-54). Perhaps the most damning assessment of generalizations was made by William Blake, who averred that "to generalize is to be an idiot. To particularize is the alone distinction of merit." (10)

   To Schauer, this negative estimation of decisionmaking based on empirically sustainable classifications is unjustified for three main reasons. First, nonspurious generalizations are frequently as accurate as individualized analyses. Second, the use of statistically valid generalizations is more efficient than particularized assessments. Third, because of the accuracy and efficiency advantage to generalizations, decisions arising through a generalized process are widely perceived of as having greater uniformity and fairness.

   Profiles first asserts, as a general premise, that using sustainable generalizations often produces results at least as accurate as those arising from individualized analyses. This is chiefly because particularized assessments rely on individuals using idiosyncratic judgment to weight and exclude factors, and thus are often statistically less accurate relative to actuarial assessments (pp. 92-101). Consequently, Schauer asserts that in a number of contexts, including predicting criminal-recidivism rates, a formulaic look at relevant factors is a better predictor than an individual determination (pp. 92-101). Even Equity itself was arbitrary and unpredictable and so necessarily reverted to more rule-based processes (pp. 48-54). (11) Hence, according to Schauer, cultural or historic pining for individual decisionmaking as a source of greater justice may be misplaced.

   Secondly, Profiles avers that relying on statistically valid categories in the decisionmaking process is more efficient than utilizing particularized assessments. Schauer illustrates this argument by citing the example of uniform age-based exclusion (pp. 118-21). Some drivers who are over the maximum allowable driving age would be perfectly safe drivers, just as some individuals not yet entitled to vote or drink would do so responsibly (pp. 118-21). Although using categories will inevitably lead to both over-inclusiveness and under-inclusiveness, he avers that testing the particular abilities of each individual covered by these rules, such as eighty-year-old drivers or fifteen-year-old imbibers, would be prohibitively expensive. (12) The key to the justness of these rules, Schauer maintains, is not that they limn perfectly accurate categories, but rather that on the whole they are statistically sustainable and not arbitrary (pp.121-26).

   As a consequence of the greater accuracy and efficiency obtained through the use of classifications, Profiles argues that people perceive decisions based on generalizations as more uniform and fair. A paradigmatic example of uniformity is application of the Federal Sentencing Guidelines ("Guidelines") to determine the length of a criminal's jail time through a quantitative formula. (13) Although some judges have railed against the Guidelines for constraining their individual discretion, (14) Schauer asserts that the Guidelines may promote fewer mistakes exactly because of this limitation. Moreover, when general rules, especially legal rules, are applied evenly to everyone they create a perception of fairness. (15) In other words, uniform Guidelines ensure that those who come before a judge receive equal treatment and perceive that treatment as fair (pp. 260-61).

   Finally, because Profiles is not procrustean (even though it does affirm that there is something that might be said in favor of Procrustes), (16) two broad exceptions are made from a uniform application of generalizations. First, Profiles repeatedly contrasts its advocacy on behalf of empirically sustainable generalizations with its disapproval of spurious categories. For example, Schauer reveals how racial profiling is an egregious example of an unfounded categorization (pp. 175-98) by describing how O'Hare Airport workers subjected African-American women to unnecessary and...