The question
Every claim about work borrows from somewhere. Someone who says that people slack off when no one can see their share is borrowing from an experiment that, in GAEKJU's summary, had people shout and clap; someone who calls a team's polite agreement a mask is borrowing a term from a study of a Malaysian village, again as GAEKJU summarises it. The finding arrives without its place, and the decision to use it here, in this office, is made without anyone noticing that a decision was made.
This study reads a small body of public records in which that decision is written down. GAEKJU is a studio of the RISCON family that asks where people's flaws at work are left in words, and whether those flaws really cost the work. Its library holds 496 questions of three kinds. Of the second kind GAEKJU says on its library page, "Outside cases cite published studies"; the other two kinds, record readings and synthetic scenes written for comparison, are not read here. This study reads only the outside cases. Of its 51 outside cases, 16 ask the domain question, "Is it the same across domains?". Each of the sixteen takes a published study, summarises it, and sets it beside another workplace. Each then gives three readings of what the pairing means, leans to one, states a dated judgment, names what would show that judgment wrong, and proposes what to change.
The research question is narrow. How does such a case carry a finding from the place where it was published to a workplace it never studied, and what limits does it write into the carrying? Who, in the case's own terms, would decide whether the carrying holds, and what would reopen it?
The question is about the form of the records, not about the studies they cite or the people in them. An outside case is GAEKJU's summary, in Korean, of someone else's research. This study does not check the summaries against the studies, and it says nothing about a study beyond what the case says of it. Nor does it repeat the counts that many cases take from GAEKJU's own record of one organisation, a separate kind of question that this study does not read. What it reads is the reasoning that joins a study to a second place.
Do findings travel?
The question belongs to a long argument about external validity: whether a result found in one setting holds in another. The argument has sharpened in three directions, and each gives a lens for the cases.
The first is philosophical and practical at once. Nancy Cartwright, in Knowing What We Are Talking About: Why Evidence Doesn't Always Travel (2013), asks when a well-established result that something produced an outcome in a study setting, "there", is evidence that it will produce that outcome in a new setting, "here". Her answer insists that the two be kept firmly apart, and asks for evidence of two further facts: that the cause can play the same role widely enough to cover both places, and that the support factors it needs in order to work are present here. Angus Deaton and Nancy Cartwright, in Understanding and Misunderstanding Randomized Controlled Trials (2018), press the same point against the trust placed in trials: at best a trial gives an unbiased estimate for the sample it selected, and extending the result to other groups, or to any individual, needs a justification of its own. Their argument adds a turn that matters below: demanding "external validity" of a single study expects too much of it, and prior knowledge should be built upon rather than set aside.
The second direction is formal. Judea Pearl and Elias Bareinboim, in External Validity: From Do-Calculus to Transportability Across Populations (2014), treat the transfer of a causal effect from a study population to a target population as a problem that can be decided on paper before any data are seen. Their selection diagrams write down what is known about the differences and the commonalities between the two populations; the procedures then say whether the effect can be carried and, if it can, which findings from each population are needed. This study reads their result modestly: a transfer is licensed by explicit knowledge of the differences and commonalities of both places, and the procedures name which findings from both are needed.
The third direction is empirical and concerns psychology and the study of organisations. Joseph Henrich, Steven J. Heine and Ara Norenzayan, in The Weirdest People in the World? (2010), review results across the behavioural sciences and find substantial variability in experimental results across populations, with WEIRD subjects frequent outliers and among the least representative populations; there are, they argue, no obvious grounds for calling a behavioural phenomenon universal from one subpopulation. Tal Yarkoni, in The Generalizability Crisis (2022), shows a subtler gap: the verbal claim a researcher makes often reaches across stimuli, tasks and research sites that the statistics rarely model, so that sweeping words rest on quantities that do not support them. On the question of the laboratory, Craig A. Anderson, James J. Lindsay and Brad J. Bushman, in Research in the Psychological Laboratory: Truth or Triviality? (1999), compared laboratory and field effect sizes across domains that include social loafing and found the correspondence considerable. Gregory Mitchell, in Revisiting Truth or Triviality: The External Validity of Research in the Psychological Laboratory (2012), extended the comparison to 217 lab-field pairs from 82 meta-analyses and found that the answer depends on the subfield, the topic and the size of the effect: results from industrial-organisational psychology carried to the field most reliably, while effects from social psychology laboratories most often changed sign. Within the study of organisations, Gary Johns, in The Essential Impact of Context on Organizational Behavior (2006), argues that researchers underrate context, the situational opportunities and constraints that affect both the occurrence and the meaning of behaviour, and Denise M. Rousseau and Yitzhak Fried, in Location, Location, Location: Contextualizing Organizational Research (2001), write an editorial encouraging more contextualisation in methods and reporting.
Do concepts travel?
A second argument runs beside the first. It is less about effects than about words. David Collier and James E. Mahon, in Conceptual "Stretching" Revisited: Adapting Categories in Comparative Analysis (1993), start from a plain observation: when scholars extend their models to new cases, they must adapt their categories to fit the new context. Sartori's work on conceptual "traveling" and "stretching", as Collier and Mahon present it, guides this task, but it assumes classical categories with clear boundaries and defining properties shared by every member. Categories built on family resemblance, or radiating from a central case, do not behave that way, and a strict classical test can make a scholar abandon a category too early or change it in the wrong way.
The two arguments meet in the cases this study reads. Some outside cases carry an effect: a study found, as GAEKJU summarises it, that people put in less effort in a group, and the case asks whether the loss appears elsewhere. Others carry a concept: a study recorded, in the case's words, the everyday resistance of the weak, and the case asks whether a short "yes" said to a superior is that resistance. The first kind of carrying is a question for Cartwright and for Pearl and Bareinboim. The second is a question for Collier and Mahon. GAEKJU's cases do both, often in the same paragraph.
What this study adds
The works read here discuss methods and categories: what a design licenses, how populations differ, how a term is adapted. This study does not say they are silent on the moment a finding is put to use in a judgment about a particular workplace; it reads that moment as it stands in a written record. GAEKJU's outside cases offer an object for it: a public, dated record of sixteen transfers, each in the same form, each saying how far it goes and what would undo it.
This study makes three claims about that object. First, the cases carry a finding only as far as a likeness they name, and most of them state in the same breath what the study did not measure; in Cartwright's terms, they keep "there" and "here" apart in writing. Second, a single concept carried to different questions lands on different readings, so that the cases adapt a category case by case rather than stretching it once; this study reads it as close to the adaptation of categories that Collier and Mahon describe, seen in a record rather than in a theory. Third, the tests that would reopen the cases' judgments are placed almost entirely in the workplace the finding was carried to, which this study reads as close to the lesson that a transfer needs knowledge of the target, but no case says by when the test is to be made and only one says who would make it. The form names the evidence that would decide the matter and leaves the decision without an owner.
Chapter 1 sets out the data and the method, then reads the cases in two sections: how far each carries its study, and where each places the test of the carrying. An appendix gives the cases word for word in this study's English, and the conclusion returns to the debate.