Reactivity

How does a magazine ranking remake more than a hundred schools in its own image? Two sociologists, Wendy Espeland and Michael Sauder, interviewed 136 deans, admissions directors and career-services heads at American law schools, tracking what happened inside each school after U.S. News (the most influential college-ranking magazine in the United States) published its annual list. 136 interviews means the conclusions were not run off a database; they were asked, sentence by sentence, of the people it happened to. What they saw was not "rankings reflect how good a school is" but schools remaking themselves to the ranking's taste: cutting need-based aid to "buy" high-scoring students, hiring staff to chase down lost graduates, copying last year's standings when rating their peers. They named the phenomenon reactivity: measurement does not only record the world, it reshapes the thing being measured. It is also the name of this whole red branch. N00 says measurement is intervention; this is the first full field evidence for that sentence.

Reactivity started life as a slur in methodology. The methodologist Campbell defined a "reactive measure" in 1957 as one that "changes the phenomenon it is trying to measure," and an older generation of researchers spent its energy stamping the problem out. Espeland and Sauder flipped the contaminant into the object of study: if the measured always respond to the measure, then instead of pretending you can sterilise the instrument, ask by what mechanism the response happens and what it produces. The flip also exposes a contradiction inside measurement itself. We want it neutral, accurate and undisturbing to its object, and at the same time we use it as an accountability tool meant to change that object's behaviour. You cannot have both, and that contradiction is exactly where public measurement becomes political. (The full etymological case, and their three reasons for keeping the word, are in the sources at the end.)

After the list comes out, the first thing to deform inside a law school is money. Schools cut faculty salaries and left vacant posts unfilled, then turned the savings into merit scholarships worth three or four hundred thousand dollars a year, built purely to move the number, all to lift the LSAT (the law school admission test) median by a point or two. An admissions director at a fourth-tier school said one year's budget for these scholarships exceeded his entire financial-aid budget for the previous three years combined. That figure shows the flow of money being reversed outright: away from the students who need help most, toward the students who raise the number most. Then there are the glossy brochures mailed to voters. Peer reputation scores carry 40% of the ranking weight, and schools spend tens of thousands to over a million dollars a year printing and posting them. The authors asked one faculty member to collect this mail; in under a year the pile stood nearly three feet high. Deans almost all admit they throw it out on arrival, yet they mail it every year, because not mailing is riskier. Everyone knows it does nothing, no one dares stop: that is institutionalised waste.

Next, jobs get redefined. Career offices shift from "help students find work" to "get the employment number up": one four-person team spent close to six weeks a year tracking down graduates it had lost contact with, and a proposal to hire private investigators was dropped only because it cost too much. Last comes gaming. In 1995 U.S. News named 29 of 177 law schools whose reported LSAT medians were higher than the numbers they gave the ABA (the American Bar Association, the official accreditor of law schools); Detroit Mercy and Alabama were off by 4 points. Same students, two numbers, and which one you report depends on who is reading: one school, two sets of books. After the public shaming, the count fell to 13 the following year. Shaming worked, but only cut it in half, because the incentive to fake is structural and does not disappear along with a few named faces.

Espeland and Sauder's 2007 paper in the American Journal of Sociology reduces these deformations to two mechanisms and three kinds of effect. The three effects are the stories above: resources get reallocated, work gets redefined, and people game the measure. Mechanism one is the self-fulfilling prophecy: the ranking changes what applicants, employers and donors expect, expectations turn into real flows of applications and gifts, and a gap in the standings that may have started as pure noise gets hardened into a real gap. The paper traces four channels for this. First, external audiences: raw scores across law schools are packed tightly together, but ordering them blows statistically insignificant differences up into real consequences. Cross-admit data show that when school A ranks consistently a little above school B, students admitted to both flip from two-thirds choosing B to 80% to 90% choosing A. Two schools with nearly identical scores, and the one ranked slightly higher takes almost every shared admit: that is what "noise being hardened into fact" looks like. Second, and most hidden: the deans who fill in peer reputation scores do not know most of the schools, so they score off last year's ranking. One interviewee's exact words were "honestly, I'm rating with the rankings, it's a self-fulfilling prophecy." Stake's 2006 empirical work confirms that past rankings are the strongest predictor of current reputation scores. A survey meant to measure reputation independently gets swallowed by the ranking: the ranking issues its own certificate. Third, universities allocate new money to programs by their standing, and the money increasingly flows toward programs that are already ahead; back and forth, the ranking reproduces the stratification it was supposed to be measuring. Fourth, and most complete: schools actively remake themselves into what the ranking measures. One school whose mission was access was forced to give places meant for minority students with low LSATs and high GPAs to high-scoring applicants: the closer admitted students hew to the LSAT ruler, the higher the ranking's "validity" appears to be.

Mechanism two is commensuration: pressing schools with wildly different missions (one training Native American lawyers, one running nine law journals) into a single ruler, so they come out as number 1 and number 99. The dimensions thrown away do not become unimportant, they become invisible. That move has a lesson of its own, R03. The two mechanisms run on different clocks: commensuration changes the form of the information and takes effect instantly, while the self-fulfilling prophecy has to wait for expectations and behaviour to align step by step. They intensify each other too: precisely because a ranking is a radical simplification, it spreads fast, routes around professional gatekeepers and quickly wins outside believers; the more people discuss it and cite it, the more mechanism one's self-fulfilling prophecy gets fed.

The trap in the sandbox has a formal name: arms race. As long as your rival is spending on gaming, you dare not stop; once everyone spends, holding your place requires manipulation, and real educational quality falls behind. The ranking is still there, the standings are still perfectly legible, but what it measures has become "who is better at manipulating it." One dean in the Espeland and Sauder interviews put it exactly: this is "a self-fulfilling nightmare."

In November 2022 the theory got a live test. Yale Law School dean Heather Gerken announced the school was withdrawing from the U.S. News rankings, calling them "profoundly at odds with the mission of this profession"; Harvard followed the same day, and Berkeley, Georgetown, Columbia and Stanford pulled out within days. The top schools staged a collective boycott. Did the ranking die? No. U.S. News switched to public data and kept ranking the schools that had left, while heavily revising its methodology (cutting the reputation weight, adding employment and debt measures). Once the method changed, the standings swung violently, and in April 2026 Stanford knocked Yale off the top spot it had held for 36 years (Reuters, 2026-04-07). This is exactly Sauder and Espeland's judgment in their 2009 paper: resistance is not the opposite of power, it is part of how power runs. Withhold your data and the ruler measures you with whatever data it can get. The full breakdown of the withdrawal wave is in R13 The rankings strike back.

The theory has limits, and the sharpest criticism comes from inside the field. Healy's 2017 review names three. First, reactivity risks being a tautology: calling observed organisational change "reactivity" describes what is changing but does not adequately explain why numbers and rankings in particular (rather than doctrines, rituals or slogans, which also coordinate behaviour) have this power. Second, the argument slides between "the specific power of rankings" and "the general effects of quantification"; Healy thinks the genuinely distinctive mechanism is quite narrow, a near-monopoly public ranking that forces every unit into a mutual order. That judgment has a control group behind it. Sauder and Espeland themselves compared law schools with business schools in 2006: the former faced a single dominant U.S. News, the latter faced at least five lists with different methods, from BusinessWeek, U.S. News, the FT, the Economist and the WSJ. Multiple rankings dilute each other, schools can pick the flattering one to tell a story about, and discretion and loose coupling come back. Monopoly, zero-sum stakes and wide circulation are the volume knobs on how intense reactivity gets, not optional background. Third, a methodological limit: all the core evidence comes from self-reported interviews in a single field, and the authors themselves concede that coding frequencies are "only a rough indicator of importance." Causal identification is weak. 136 interviews can tell you what the mechanism looks like; they cannot measure how strong it is.

Brankovic, Ringel and Werron corrected the picture from the other end in 2018: Espeland and Sauder treat competition as a pre-existing condition that rankings act upon, whereas in their view it is the periodic publication of rankings itself that turns implicit, local status differences into explicit, global competition. Competition is the product of rankings, not their premise. Later lessons in this branch take these apart in detail: the full lineage of the self-fulfilling prophecy and its mirror image in R02, the analytic framework for commensuration in R03, and the Foucault-style answer to why rankings cannot be buffered away in R04.

One open question: the top schools walked out together and U.S. News could still rank them by force using public data, which shows that "the measured refuse to cooperate" is not enough to end reactivity. So what actually ends it? Audiences no longer using the measure, or a substitute ranking taking over? The D1 report lists this as the field's first open question.

The one-line takeaway: a ranking is not a mirror, it is a machine for remaking things; whatever it measures, the world slowly grows into.