18 becker tricks of the trade chap4

8/13/2019 18 Becker Tricks of the Trade Chap4

http://slidepdf.com/reader/full/18-becker-tricks-of-the-trade-chap4 1/21

hicago Guides

l t l n ~and Publishing

On Writing Editing and Publishing

]acquesBarzun

Getting into Print

WalterW Powell

Writingjor Social Scientists

Howard S Becker

Chieago GuidejorPreparing Eleetronie Manuseripts

Prepared bythe Staffof the Universityof Chicago Press

lides qftheField

lohnVan Maanen

Handbookqf iologieal fllustration

FrancisW Zweife

The Crcift qfTranslation

lohn Biguenet andRainerSchulte editors

Style

]oseph Williams

Mapping It Out

Mark onmonier

Indexing ooks

Nancy C. Mulvany

Writing Ethnographie Fieldnotes

RobertM. Emerson Rache Fretz and LindaL Shaw

Glossary qfTypesettingTerms

Richa rd Eckersley Richa rd Angstadt Charles Ellerston

RichardHende Naomi Pascal and AnitaWalkerScot t

The Crcift qfReseareh

Wayne C. Booth GregoryG. Colomb and]oseph Williams

ManualjorWriters qfTerm Papers Theses and DissertationsKate L Turabian

1 b

ricks o th r de

ow to Think aboutYour

ResearchWhileYou re oing It

Howard S ecker

The University of Chicago Press

Chicago and London



ONT NTS

Preface ix

TRI KS

2

IMAGERY

SAMPLING

67

4

ON PTS

9

5

LOGI

46

oda 2 5

References

Index 229



FOUR

Concepts Are Defined

We all work with concepts. All the time. We have no choice, as Herbert

Blumerpointedout in a critique ofwhat was called,when he wrote, o p

erationalism. He noted that you could no t have a science without con

cepts.Without concepts, you don t knowwhere to look,what to look for,

or how to recognize what you were looking for when you find it . Psy-

chologists, in their heyday when Blumer wrote, thought they could dowithout concepts, at least concepts defined in abstract theoretical terms.

They thought they could avoid such chronic troubles as arguments over

definitions by defining concepts simply, as what theymeasured by the op

erations they used to study the phenomenon they were investigating. In

the classic example, they said that intelligence, whose definition was

hotlydebated then as now, was what intelligence tests measured.

Sociologists equivocated in the same wayabout the concept of ttitude

Many researchers assumed that people had thoughts or dispositions or

ideas (or something)-summarizedas attitudes-inside them, waiting to

be released by the appropriate stimulus or situation. What an attitude was

wasn't dear. Scientists argued about the definition. But their inability to

define an atti tude didn't prevent them from inventing attitude measure-

ment,a procedure inwhich people's answers to a long list of questions pro

du ced a number that measured thei r a tt itude toward movies or

foreigners or schools or political parties.The scientistsmeasured the relia-

bility and validity of attitudes, and concocted statistics that described the

relations of attitudes to one another and to other facts about people. They

thought they could show that people differed with respect to attitudes

about this or that, and that those differences correlated with other differ-

ences in ways thatseemed meaningful.

Crit ics complained that there was no general understanding of this

thingthatwas being measured. Operationalists evadedthose complaints by

denyingthat theyhad saidanything about the actual content or meaning of

the measured attitudes: attitudes werejustwhat the tests measured, noth

ing more. No one believed that. If theyhad, there would have beenmuch

less research on attitudes or intelligence or the otherimportant ideas that

were defined operationally.Because,after all,no one reallycares about test

measurements in themselves-only about intelligence or racial attitudes

or propensities to violence or whatever the test is supposed to measure.

A favorite reply to attacks on attitude or intelligence tests was, You

110

CONCEPTS

don t want to call it intelligence? Fine Call it X. OK? You could de

flect this irritating and unsatisfactory riposte by actually referring to the

i tem in question as X. I see,you'veshown thatchildren of different racial

groups differ by ten points, on the average, on something called X. So

what? But, of course, no one cares about the differential scores of black

and white childrenon X.Without content, Xhasno relevanceto anyques-

tion of theory or policy. But people do care about differences in intelli-

gence because, if they really exist , they have serious poli tical and moralconsequences of a kindsomething that is just X could never have. By the

third time a critic called X what everyone involved knew was reallyintel-

ligence, the discussion would get more serious.

Thiscritique mayseemquaint and out of date, since few contemporary

social scientists would admit to being operationalists of the kind Blumer

criticized. But many contemporary researchers act as though they'd ac-

cepted a variant of that position. In this sense: they choose, as an indica

tor of the phenomenon they want to talk about , something tha t has an

imperfect, sometimes a highly imperfect, relation to the phenomenonit

self,and then treatthe indicator as though i twerethatphenomenon.They

askpeople their occupationand treat the answer as ameasure of socialdass,

bylocatingthe occupation named in a list ofjobs whose prestige has been

measured or placing it in a Census dassification of major occupational

groups.They may say thattheyare measuringwhat Kar Marx or MaxWe

ber orW LloydWarner or C.Wright Mills meant when theyspokeof so

cial dass but that's neither obvious or particularlybelievable. Peoplewho

make such measurements don t insist that a person's occupation is social

dass in the Marxian or Weberian sense, since they haven't demonstrated

anyrelationbetween the two empirically,but their analysesan d discussions

implicitlyassert the identity. Important as measurementmay be, it doesn't

do much for ou r understanding of the concepts we use.

Another way of defininga concept is to collect examples of things we

recognize as embodyingwhat theconcept refersto, and thenlook for what

the inevitablymessyand historicallycontingent ideasp eople routinely use

have in common. Somecommon sociological examples of such concep

tual work are skilI, crime, or profession.We try to formulate adefinition

that indudes all the things we think are alike and leaves out those that are

different. We are embarrassed if someone can show that something we

didn't think belonged inour collection in factfits the terms of the defini-

t ion. Thus, researchers tried to define a profession as a special kind of

111



FOUR

work, different from other occupations. What they wanted to include in

the aggregate their definit ion collected were such highly respected and

well-paid occupationsas medicineand law So theyframed theirdefinition

b listing the traits that characterized those occupations. (Freidson 1994

glVes a careful account of theseproblems,and offers realisticand useful solutions to them.)

Invariably, an industrious and clever crit ic would find an occupation

that all the definitional requirements (long years of training, a body of

esotenc knowledge, state licensing, and so on) but clearly didn't fit.

Plumbing used to b e good f or this b it of theoretical skullduggery.

~ l u m b e r s have the attributes included in standard definitions of a profes

SIOn: an e s ~ t ~ r i c body of knowledge (try fixing your own drains), long

years of trammg, state licensing, and the rest. But everyone knows that

plumbing is no t aprofession.Th e seerningparadox arises because the items

in the collection the definition is framed to coverhave been chosen on the

basis of.anunacknowledged variable: the social prestige of the occupation.

If prestIge correlated perfecdy with the other criteria, there would be no

problem.Bu t it doesn t.

Such problemsarise inmanyareas of sociological work.The theoretical

trick that helps solve the problem is to recognize thatwhat goes into the

collection the definit ion has to cover governs the kind of definition we

come up with. And collecting the examples is thekind of sampling prob

lem considered in chapter 3.So we look for answers to such questions as:

How do we make up those collections? What do we typically leave out?

whatharm does it do to be se1ective in Our choice of examples? Defi

llltIOnalproblems arise exacdybecausewe havechosen these collectionsin

ways thatignored the injunction of chapter 3 to include the widestpossi

ble variety of cases of a phenomenon in our sampie. Here are twofurther

e x a I ~ p l e s where the harm is more substantial, 01' at least more easily seen,

than m the case of profession (which is at least on the surface, mainly a

conceptual embarrassment, though the policy implications of the definition of that tenn areq uite serious, as Freidson [1994: 149-216] shows).

k ll

S ~ c ~ o l o g i s t s econornists, and other social scientists re1y implicidy 01' ex

phCIdy, on the idea of skill. They argue that differences in pay for in

stance, result from the scarcity of real skills, so thatpeople who have rare

CONCEPTS

skills get paid more.What would make a skill scarce? One thing would be

the differential distribution of natural talent to exercise the skill. People

who are tone deafwould find itdifficult to learn to playhundreds of songs

by ear, as I h ad to do t oho ld a job playing the piano in taverns. Some peo

pie can manipulate numbers easily and might be especially good a t ac

counting, keeping books, 01' managing people s money. Some are very

skilled with aneedle andcansew 01' knit 01' crochet excellendy. Some have

awaywith people, knowhow to ease theirfears 01' make them feelat home.Some have learned decisiveness and are good a t i t; theycanmake up the ir

minds in a difficultsituationwhile the rest of us stand around sucking our

thumbs.

Another contributor to the scarcity of a skill might be howlongyou

have to work 01' howmuch you have to pay to acquire it. People, on this

theory, wouldn't invest so much t ime and energy they rnightinvest else

where if i tweren t going topayoff. So thenumber of peoplewillingto ac

quire the skill will go down if the rewardsfor sellingit are low. If everyone

acts in this econornicallyrationalway the number of people in each occu

pation will reach an equil ibriumat a price users willpay for the skill and

practitionerswill accept.

We can certainlymake a long list of skillspeople have had over the cen

turies. It s clear,when we inspectsucha list, thatnot all skillsare equally re

warded. Skill alone does not produce big rewards. You need a skill that

someone else,who canand willpayfor it,wants. If you have avery rareskill

thatvery rich people want badly, you will be rewardedhandsome1y. If, for

instance,you are one of the few people who canrepairdamagedart works

owned by wealthy peoplewho prize them highly, you will be paidwellto

exercise that skill. If you have a skill many other people have-if you are

one of the millions who can quickly be taught to cook hamburgers at a

fast-food franchise, a group of whom there are more members than any

one needs-you will be paid the lowest legal wage (01' less if the bosses

think they won't get caught). But even a very rare skill won't do you anygood unlesspeople wealthyenough topayfor it at the rateyou d likereally

wantand need it. My ability to play hundreds of tuneswasn tworthmuch,

because theonly people who wanted itwere band leadersand tavernown

erswho could, if i tgot too expensive, managewith pianists who knew alot

less than I did.

The demand for skills varies historically. Temporary conjunctions of

circumstances can raise the value of skills ordinarily no t worth much.

113



FOUR

H ~ b s b , ~ w m 1 9 ~ 4 described the unlikely victory of a group of un

skIlled laborers m the great London gas strike of 1896. London, at that

time, was litlargelywith natural gas,manufacturedby cokingcoal-that is

byheating coal in large furnaces so that the gas it containedwould be

leased to be captured andpiped to households andfactories. Running the

furnaces-shoveling the coal in and keeping it burning-was unskilledla-

bor. Anyone could do it. It had never required anyspecial training, other

than what you got on the job. So, when the laborers who did this work

went on strike,conventionalwisdom and economic theoryalike said it wasunlikelythey could win.

But they did win the strike, and got a handsome settlement from their

employers, who were as greedy as capitalists are supposed to be. How did

the workers win? Hobsbawm shows that these unskilled laborers actually

hadsomevery important skills and thatan unusual conjuncture of circum-

stances at the timeof the strike hadmade those skillsmore valuable to the

employers than theyordinarily were. Put the questionthis way:why didn t

the employersjustgo out andhiresome other unskilled men to shovelcoal

into ,the ~ u r n a c e s Why didn t they just wait the strike out, manipulating

public opmlOn to make their stubborn employees look responsible for the

disconlforthouseholderswere sufferingand thus bringthem to heel?There were several reasons why employers didn t take these obvious

steps.The sellers of gas were facing new competition in the form of elec-

tricity. Still a novelty, electricitywas potentiallY just as good away tolight

your houseand, if astrikewent on fora while, customersmight be tempted

to experiment with the new form of energy.The longer the strikewent

on,the more customers the purveyors of gas would loseto electricity.

Further, the employers couldn t replace these unskilled laborers as easily

as you might have thought. To be sure, what they did required no great

schooling.But the machines theytended, while not higWy technical and

thus not requiring, say engineering knowledge to run, were o ld and

~ r o t c h e t y . The gas manufacturers had been coasting, collecting theirprof-

Its and notmaintaining the machineryanymore than absolutelynecessary.

So the machines worked but, like all oldmachines, had to be coaxed.You

had to know when to give thefurnace a good kick, and where to kick it.

These might no t be skills in the conventional sense, but if the men who

shoveled the coal didn t have them the furnaces didn t work. The bosses

couldhire other unskilledworkers but,lackingthat special knowledge, the

newmen couldn tdo thejob.

114

CONCEPTS

That combination of circumstancesgave these unskilledlaborers some

skills thatwere at least temporarilyvaluable, and they used their advantage

skillfully to win higher wages.Th e important lessonfor us is that theiden

tical abilitymay be skilled or unskilled, depending on circumstances.The

meaning of the concept of skill depends on which cases you have in mind

when youdefine it.

So skill, if youwant to raise yourwages bywithholding it,must be a skill

that someonewith moneywants. Suppose you have the skills, and they are

scarce and peoplewant them, but those potentialpurchasers of your ser-

viceswould rather not payyou as much as your skillsmightbe worth on the

open market. This, I take it, is the point of research and work on what is

called comparable worth. Here s the problem: many people think

women have been historically, and still are, discriminated againstin thela-

bor market. A great variety of statistical studies show that employers pay

women less than men any time they can get awaywith it. And who can

blame them? Capitalism, as Marx said, is a tough system and employers

who pay more than they need to for components of their products will

soon be driven out ofbusiness by shrewdermanufacturerswho can seil the

same product cheaper.

The gas worker example sheds some light on this problem. Suppose the

law finally forbids out-and-out discrimination on the basis of gender;

womenmust be paid what men doing the same jo b are paid.Women will

still make less. Now why? Because the distribution of men and women

ac ross occupations is skewed. No women play major league baseball and

very few nurses are men, and ball players make a lot more than nurses. A

disproportionate number of schoolteachers arewomen; a disproportion-

ate number of corporate executives are men. If you pay all nurses, what

ever their gender, the same, and pay all executives, men o r women, the

same, but pay nurses less than executives,womenwill end up making less

on the average becausemore of them are injobs that don t pay as weil.

How canthatinequitybe remedied?Some reformers have attacked theway pay scales are set (it is primarily governmental agencies that are vul

nerable to such attacks), noting that salaries are setwith reference to the

skills allegedly required to do the work, but that skills important in

women s occupations (that is, occupations most ofwhose members are

women) are either ignored o r not valued higWy in such evaluations. If

technical skills arevalued more highly than the skillsnecessary to dealwith

complex social situations, and thejobs women are more likely to have-

115



FOUR

l ike nursing and teaching require fewer technical skiils and more hu-

relations skills, then women will be paid less even though they are

Just as higWy skilled, although in different areas.

Of course, proponents of the status quo will argue that it can t beshown

that are commensurable. But that, of course, is thepoint . If they

aren t, ts because we haven tagreed onhow to measureskiil.And if that s

true,. then how do we know that men s skiils are worth more? And it sjust

thatJudgment that is embodied in the verywage scalesbeing attacked.

. I:ve been a long timegetting to the conceptualpoint, because the point

lies thekind of examples I ve given, not in abstract talk.The point is that

conceptspresuppose thatyou have inspected the fuil range of things they

coverwhen you formulate and define them. Now we can see one of the

reasons for my earlier emphasis on methods of sampling thatproduce ex-

amples of that range. If you leave some phenomena out because of con-

ventional prejudice or forany of the other reasons I discussed there,your

concepts willbe flawed. Generalizations ofwhich those conceptsare com-

ponents wiil contain a lot of noise, random variation thatisn t random at

ail, but rather the resultof systematic social biases in the selection of cases

you usedto define your concepts.

rime

The same reasoning applies to the weil known phenomenon of white-

coilar crime.Why did Edwin Sutherland find it necessary to devote his

presidential address to the American Sociological Association (1940) to

the subject of white-coilar crime? Because he wanted to accuse his col

leagues of aconceptual error that had asirnilarbasis in inadequatesampling

b:lsed on conventional, sociaily approved prejudice. Crirninologicaljour-

nalsand books,at thetime Sutherland delivered his blast, were filled with

theories about crimeand researchon crime.Whatwas crime, this thing ail

the theories and research were about? Activity that violated the crirninal

law That seemed fair enough. The mountains of research that had been

done showed that crime was highly correlated with poverty, with broken

hornes, and ail the other conventionalindices ofwhatwas then called so-

cial pathology. Sutherland asked a simple question: how can tha tbe t rue

when there are crimes being committed by very weil to do people who

do no t exhibit the conventionalsigns of socialpathology,and by the largest

CONCEPTS

and most respected corporations in the country, which sirnilarly did not

come from broken hornes?

Th e answerto thatwassimple enough. No one, no conventional crim-

inologists certainly, thought the crimes weil to do people and corpora-

tions comrnittedwere, in somefundamental way really crimes. Besides,

the culprits involved were seldom convicted of criminal violations, be-

cause these cases were oftensettled as civilsuits. lf therewere no criminal

convictions, how could there be any criminals?Th e governmentwas typi

caily more interested in getting the bad guys to stop their mailfrauds and

securityswindles and forcing them to pay off those who had been cheated

than in sending anyone to jail. ut thatwas no t a natural consequence of

the nature of the crimes,whichcouldjustas weilhavebeen prosecutedun -

der criminal statutes, and occasionaily were. resulted from judgments

madeby prosecutors,who exercised the discretion the law gave them as to

whetherto pursuecriminalor civilremedies.

Prosecutorshad other reasonsfor not pushingfor criminalconvictions.

As Katz s later (1979) research showed, white-coilar crime and crimes of

the more conventional kind differ in anotherimportant way ordinary

crime, there s no question that a crime has been committed. Someone has

been robbed or assaulted. The question is: who did it? white-coilarcrimes, on the other hand, there s no question about who did it. The big

grocery chain l abel meat tha t weighed 14 ounces as weighing one

pound.The question is not who did itbut rather is i t a crimeor not? Such

a thing, after ail, might have happened because a scalewas faulty and the

company didn t know about i t, or because a crooked butcher was skim

ming some of the profit for hirnself, or forany of a number of reasons that

would show that the companylacked criminal intent. So, for both sets of

reasons, white-coilar criminals are convicted of crimes far less often than

common criminals.

Sutherland s impeccable reasoning was that if you decided not to in-

dude the crimes rich people and corporations committed when you cal

culated your correlations, you gua rant eed t he result t ha t c rime was

correlated with povertyan d its accompaniments.No t because it reailywas,

but because you were using a flawed concept, one that pretended to con-

tain all members of a given dass, but actuaily left out a large number of

those members on the uninspected grounds of social prestige.You didn t

have an empirical finding, you had a definitional artifact.

7



FOUR

Defending against Sutherland, conventional criminologists argued, es

sentially, that everyone knew that those rich people and corporations

weren t really criminals. That is, if you accepted the conventional no

tion of what a criminalwas-a tough guy with amaskwhojumpedout of

the bushes, stucka gunin yourribs and tookyour money, a guywho made

a careerof crime, liveda life of crime, shared the culture of crimewith oth

ers like him (and these criminals were, in conventional thought, male, of

course)-then itwasclear thatthe nice people who wore suits andties and

took yourmoney in broaddaylightover a deskin a fancy office,and the or

ganizations in whose buildings those officeswere situated, didn t look like

thatat all.Theymight takeyour money, but no twith a gun; in fact, the way

theydidit youmight not even knowyouhad been robbed unlesssomeone

pointedit out to you.

Sutherland arrived at his understanding of white-collarcrime by using

a trick basedon a common feature of organizationallife. As I suggested in

the discussion of sampling, organizations typically tell lies about them

selves. If that s too harsh, we mightjust say that they like to put theirbest

footforward, and prefer not to mention things thatwouldmake them look

bad, especially when those events and activities can plausibly be inter

preted as random deviations or characterflaws attributable to individuals,things thatare in anycasebeyondwhat anyonecould reasonablyexpect the

organization to guard against. It s the general explanation police depart

ments give when any of their officers get caught misbehaving: There s a

fewbad apples in every barrel. Thisexplanation is designedto counterany

suggestion that would accept the more sociological hypothesis that the

barrel makes the apples rotten-that is, that the department s organization

and culturemight lead officers who would otherwise be law-abiding into

bad ways.

Socialscientists willbe led astray if theyaccept the lies organizations tell

about themselves. If, instead, they look for places where the stories told

don t holdup,for theevents andactivitiesthose speakingfor the organiza

tion ignore, coverup, or explain away, theywillfind a wealth of things to

include in the body of material from which they construct their defini

tions. Sutherland s trick was simple. He looked for facts corporations

might not put in their annual reports: the civil suits against them and the

settlements they had made of such claims; and the violations of criminal

law sociologists did not count because corporations hadmanagedto avoid

criminal prosecution,instead settling them as matters of civillaw.

CONCEPTS

When youfind events andfacts thatare not accounted for in the stories

conventionally told about a class of organizations,you have usually found

a new element or variable thatneeds to be incorporatedinto the defini

tion of the phenomenon under study. A more general version of Suther

land s trick produces the labeling theory of deviance (see for example

Becker 1963). In thisway: the conventionalstoryabout deviance is that the

organizations responsible for dealingwith it actually do deal with it effec

tively.They may not preventit from happening-police departments may

not be able to controleveryrogue cop-but once it is known to have oc

curred they find it and punish it. Corporations may not be able to prevent

employees from cheating customers, but they track down and punish the

cheaters. And so on.

Butwhenyou discover that not all deviations aretrackeddown, andthat

the selection of which ones to track down is not random, you have good

reason to think that you have found another element in the puzzle

namely, a step in the process of detection and punishment that consists of

not detectingsome people ornot punishingsome thathavebeen detected.

You thusknowthat deviance includes both a possible infraction of alaw

or rule, and a process of acting in some fashion againstwhoevermight be

thought to have committed the infraction. When Sutherland saw thatsome who committed crimes were not treated the way others were, he

knewhe was onto something.

Keep inmind that what Sutherlandsawwasno tmuch of a secret.Every

organization enforces the rules it is responsible for in a partial and discre

tionary way. Sutherland s originality consisted in making that discretion

the subject of study. 1 11return to this separationof rule-breakingfrom the

perception and punishment of rule-breaking in chapter 5,when we look

into the uses of combinatoriallogicfor socialresearch.)

All these examples showthat concepts thatdon t cover the full range of

cases to which they allegedly apply are flawed. Generalizations that in

clude flawed concepts as terms in the explanatory equation will not explain everything they claim to apply to, as explanations of crime based

on juvenile delinquents activities could not explain the crimes of large

corporations. Including the full range of cases forces us to revise our gen

eralizations, make them more complex and more interesting.Then,con

taining less noise and less unexplainedvariance, they will explain more of

what theyare supposed to explain.

The trick here, to repeat, is recognizingthat the definitions of concepts

9



FOUR

rest on what the examples they are based on have in common. However

abstract (or theoretical ) the resulting definition is it bears the marks of

that often uninspected selection of cases. That's why I've insisted on the

necessity of strivingfor imagery thatenlarges our ideas about what might

be present in the worldwe study. If our imagery is based on a biased sam-

pIewe will have trouble. If we systematically look for excluded cases, our

work willimprove.

Defining Concepts: Some Tricks

To review our results to this point:we define concepts as opposed to dis-

covering their true nature), and our definitions are shaped by the collec-

tion of cases we have on hand with which to think about the problem.

Suppose we have gathered a good collection of cases andwant to proceed

with creatinga useful concept. How do you do that? It's true thatit takes

some imagination andsome free associating and some consultingof what

others have said in the past, butyou can do all thatand still notknowhow

to createa concept.What do you actuallydo?

Socialscientists ask themselves this question when they begin to gather

data without havingmuch sense of what the problem theyarestudyingac-tually iso That happensmore oftenthan wewould like to admit.It happens,

for instance, when we agree to studya practical problem, a problem de

fined by its importance to the people involved in it. (Since so much re

search is funded because the problems are pract icai ly and pol it ical ly

important, this situation is common.) Are black students getting a fair

shake in education? , however anyof those terms is defined, is not aques-

tion framed in sociologicalterms. That'sno t to say it isn't importantor in

teresting, but rather that when we study i t we will have to turn i t into a

sociological question before we have anything distinctive to say about it.

Bu t we don't knowwhat thatquestion will be, no t yet.We onlyknowit af-

ter we see what kinds of organizations, institutions, and processes are in

volved in the production of that problem (what kind of a machine is

operating to make things happen thatway), and only our research will tell

us that.

So we find ourselveswith apile of data, trying to figure outwhat it could

be about, sociologically. Studentswho find themselves in this fix oftensay

they want to narrow their problem down, a ritual phrase some teacher

taught them to say to ward offgetting in over their heads. For students, but

120

CONCEPTS

no t only students, thatalso means finding away to say something thatwill

be defensible against all attacks; if they make the problem narrow

enough they canfind out all about i t, nail i t down, andnone of the v g ~ eenemiesthey sense around them canget them. (I've discussed thosefears 111

Becker 1986b.)

Students learning to do fieldwork commonly suffer from this disease.

They finally get theirnerve upto interviewsomeoneandthendon'tknow

what to ask.When they observesome socialsituation,they aren'tsure what

constitutes their data, which of the thingsthey see andhear theyare sup

posed to write down.That's becausethey don't know their problem

is what they're studying. They know they have to do lt , so they put ~ n y -thingdown. Or so it seems.As a result, their notes are scattered, essennally

incoherent; their interviews wander because they don't give the p ~ o p l ethey are talking to any systematic guidance about what they would like to

know.

But there is some orderto what theyhavedone, becauseyou can't make

the simplestdecisions unless you have someidea as to whatyou are doing.

The students' imagery of people and places and situations like the

they'reexamining hasledthemto do whateverthey did,ask what they d1d

attend to what they did, ignorewhat they did. They now have to find out

what they had in mind that led them to do all that.The problem is to un

cover the imagery that got them into this fix.

My trick here is aversion of anold pador game. In gan:e s ~ l e o n esays for instance, NineWagner. The objectof the to I ~ a g l e the

question that is the answerto. In this case, the questlOn that eliClts ~ h a tswer is Wh o wrote thatpiece? Mozart? And the answer I took hbernes

with the spelling) is Nein Wagner So, trying to figure out whatyou are

doing, you say to yourself, Th e dataI have here are the answer to a q u ~ s -tion.What question could I possiblybe asking to whichwhat I have

tendownin my notes is a reasonable answer? Iaskstudents to rereadthe1r

notes with this in mind, to pretend that they dideverything they didpur

posefully and have succeeded in doingjustwhat they set out to do. Now

theywillfind out what they did.

The exercisegenerally makes studentsunhappy.They seethat, ~ h ~ e v e rvague idea they had in mind when they began their work, they d1dn t.get

anywhere neardoing it. Unspoken assumptions and unacknowledged1m-

agery-about the problem, but more likely about what they can reason-

ably expect in the way of cooperation from people--have led them to

121



FOUR

investigate topics they didn t have in mind and didn t care about, usually

very minor and superficial matters whose virtue was that they came to

mindduring alull inthe conversation. Th e students wanted to know about

patterns of social organization but, under the pressure of performing s

knowledgeable researchers when they knew they weren t any such thing,

they asked the peoplethey interviewedand participatedwith about trivia.

They want to know about umest among the factory workers they are ob

serving, bu t theyhave only talked to them about thefood in the company

cafeteria or last night s football game on television. And they know that s

no t it. They didn t do what they should have done to find ou t what they

wanted to know.

I tell them no t to be unhappy. No w they know what theywere actually

investigating, what the ir f irst a ttempts actua lly asked about , and they

know that what theylearnedwasn t what they wanted to know. Knowing

tha t, they can change direc tion, reformula te the ir quest ions , and have

somethingdifferentto pu t in their notes.Theirdataare nowmore likely to

beabout what they want to be investigating.And, if it appears theymayno t

beableto see something they think it s important to see ot ask about some

thing they think it s important to ask about, they can consider alternate

ways to getat what interests them.

Their reformulated questions constitute the beginnings of conceptual

construction. They see what they aren t interested in and don t want to

know about.They usually don t find thisvery thrilling and think they have

just wasted their time on a wrong lead. But they haven t. They can onlys y

that X doesn t interest them by havingsome notion of what would interest

them. Naming the objectofinterest is the beginning of conceptualization.

I ve made i t sound s though this trick could only be done by sociolo

gists whoworkwith qualitative data, unfetteredby researchdesigns,able to

keep changing their minds s they do theirresearch. In fact, the introduc

tion of microcomputers into everyday sociologicallife hasfreed quantita

tive sociologists from their dependence on mainframe computers, fromthe long waits those machines inserted between getting an idea, thinking

how to testi t on your data, and actually getting the results. Freed from the

mainframe, quantitative analysis is much more interactive. People r u n o f f

factor analyses that once took ayear ofhand calculation inthe time it takes

to refill their coffee cup. Th e cost of calculation having been lowered so

dramatically,researchers can do analysesjust for the hell of it, to see if there

is anything to a hunch Ragin and Becker 1988). And that i n t ur n means

122

CONCEPTS

that quanti tat ive researchers too can inspect the answers they have to see

what questions they imply. Th e same trickswill work forthem.

Let the Cas e ifin the Concept

This is a slightly different way of exploiting the recognition thatconcepts

are defined. Sociologists, concerned to generalize, want to establish that

what theyhavestudied is not theonly one of its kind.What good woulditbe to getsure knowledge about something when you couldn t apply that

knowledge anywhere else? Thisconcern is enshrined in the well-known

distinction between idiographic and nomothetic sciences. Students espe

cially, I think,want to pu t theircase(the thingthey studied) into some con

ceptualcategory, for the very good reason that if they can do that then all

thejustificationsfor why you shouldstudysuch thingsareready-madeand

easily available to them.

Bu t there s a problem with that. It s n o t d e a r tha tyou can s y anything

very useful if you focus only on what is common to your case and other

cases withwhich it shares membership in some dass. Th e more seriously

you take the case, the harder you t ry to und erstand it fully so that there s

nothing about i t that you have to hide or ignore, the harder it is to see it s

being justlike any other case it might superficially resemble.

Consider this s a choice betweenletting the conceptual category de

fine the case and letting the case define the category.We let the category

define the case by sayingthat what we have studied is a case of x let s say

of bureaucracy or modernization or organization or any of the other

common concept s we use t o understand the social world. Doing that

leads us (not necessarily, bu t often enough in practice) to think thatevery

thing that is importantabout the case is contained inwhat we know about

the category.Analytically, then, wejust have to inspect the caseto see that

it has all the attributes a member of that category is supposed to have and

thus is one of the things described by thatconcept.We check, s y to seethat ou r case has all the features Ma x Weber said a bureaucracy should

have. Gu r analysis is complete when we show that it does have all (or

most) of those things,and have explainedwhyit doesn thave the onesthat

aren t there.We ignore those elements of the case whose presence or ab

sence the categorydescriptionignores.This strategyhelpsus develop the

ory by adding cases t o the collection of examples of the type, and

variations to ideas and principles othershave developed to explain them.

123



FOUR

This is something like thenormal science work of articulation described

by Kuhn (1970, 27-34).

The more the world, as exemplifiedin our case, includes just wh at ou r

concept includes and no more, thebetterour analysisworks. But the world

is hardly everjust as we imagined it. In fact, sucha rare similarityprobably

occurs onlyunder somevery special circumstances. It occurs,for instance,

whenwe havetailoredou r conceptto fita particularinstance.If I construct

a theory of revolution by generalizing from the American or Russian

Revolution, then my theory wil l f it thecase I based i ton. The world and

our concept resemble each other,too,when we have enough control over

the world to m ke it exactlyfitour categories. La.tourexplains that science

works, which is to say thatitspredictions are verifiedin practice, because

scientists change the world unti l i t is just like the setting in which they

made their discoveries (1987,249-50). Louis Pasteur could protect cows

fromanthraxby vaccinatingthem onlywhenhe couldpersuadefarmers to

replicate the essential features of his laboratory on their farms. He says

Factsand machinesare like trains, electricity, packagesof computerbytes

or frozen vegetables; they can go everywhere as long as the t rack a long

which they travel is not interruptedin the slightest (1987,250). It is ex

tremely difficult to lay the tracks on which social science can travel. Too

many other people have conflicting ideas about how the social world

should be arranged to let us arrange it so that our theories will work. So

suchtracks are best laid in computer simulationsand sometimesin labora

tory experiments.Unlike Pasteur,socialscientists can rarely persuade any

one to turn their real (not simulated) hornes or communities into the

tracks on which our theoriesmightrun.

So the strategyof lettingthe concept define the case accomplishesa lot,

but at a price: we don t see and investigate those aspects of our case that

weren t in the descriptionof the categorywe started with.The things we

leave out, however,come back to bother usoWhetherwe include them in

our investigationor not,they are sti ll thereand continue to operate in thesi tuation we re studying, almost surely influencing the phenomena we

want to understand. It makessense to include themin ou r analysis even if

our concept doesn t make room for them.Which is the argument for the

alternative strategy: letting the case define the category.As i n t he earlier

example, take theAmerican revolution as the modeland define a category

thathas all the attributes (everysingle one,becausewedon t knowwhat to

leave out) of that case.Anythingwe find ou t about the casebecomesa cru-

124

CONCEPTS

cial part of the concept.What does that accomplish? Can we ever create

any generalizationsworking that way?

Letting the case define the concept lets you define d i ~ e n s i o n s :o u

might see varying in other cases.Youdiscovert hat the e x e c u t l ~ e s of s v l ~ sand loan associations sometimes stealmoney by manipulatmg banking

regulations whose complexity makes it difficult for~ r o s e ~ u t o r s to decide

whetherwhat they indisputably did is a crime.That Idenufies an aspectof

crime you would no t see in cases of assault,where no one doubts that

hittingsomeone with a clubis acrime.Th e generalizationthat resultsfrom

your study is thatthe clarityor ambiguityof an action s criminality, the

things that affect that , are something to include in all future studles of

crime. In a way the result of working like this is notmore answers, but

more questions.

Generalizing Bernie Beck s rick

I snuck in amove in the above analysis,when I saidits outcomewas a new

aspect of crime to be includedin future research-the ~ l r i t y or ambigu

ity of an action s criminality. explainwhat s involvedm that movenow.

Sociologists often know no intermediate stops between the raw facts of

the case they studied and the largest ,most general categories of social

I . Th they may describe the findings of their research on, sayana YS1S. us, . . .

drinking alcohol, and jump from t ha t t o talk about .ldenutle.s self-

conceptions or some other highly abstract aspect of sOClal orgamzatlOn or

interaction.As a rule, our research does no t have anythingvery new to say

about self-conceptions or identity. Researchers usually use such general

ideas to orient theirwork,to suggestan overallapproach and avery general

set of questionstheymight ask.Th e ideas serve as what Lewontin refers to

as informing and organizingmetaphors whose role is t o b r i n ~ order i· (1994 509) What the researcherswho use them dlscovernto con uSlon , .

will probablyno t lead to anyreformulationof those generalideasor ques

tions. At worst, the researcher announces triumphantly that what

studied was indeed a case of the development of identity or the adapuve

characterof social organization.That kind of result isn t useful to anyone.

It doesn t add much to whateverwarrant the very general theories it is at

tached to already have. And the general theories don t add much to the

specific studies.Th e advice theyoffer is too g e ~ e r lWhat is useful is the description of somethmg more general than the

125



FOUR

p a r ~ i c ~ l a r f a c t ~ we discovered, but less general than notions of identity or

sO lal mteractlOn. Something in between, something like what Robert

~ r t o n alerted us to as theories of the middle range. I moved fi om the

s a v ~ n g ~ l ~ a n .convictions to the idea of the darity or ambiguity of an

actIOns cnmmahty, but I didn texplain how I did that.When I teach field

work, I often make that kind ofjump in discussing the possible extensions

of a student s.findings. This is the aspect of what I do thatmost often pro

vokesthe feelmgthatsome kindofmagic trickis beingperformed, thattheway I getfromA toB isn t somethinga person can learn to imitate.

During the twenty-five years I taught at Northwestern, my office was

always.next door to that ofBernardBeck, one of sociology s great teachers

and thinkers, whose qualitiesare less well-known than they ought to be. I

l e a r ~ e d more from him thanI will ever repay,a lot of itfrom eavesdropping

on hIS conversations with graduate students about theirwork in progress.

N what I ~ e a r d has beenmore useful to me thanhis trickfor getting

to thls mtermedlate level of thinking about a research result. Since he has

neve.rpublished this t ~ i c k whichhas the elegance of simplicity, I m taking

thehberty ofborrowmg Itfrom him.

Beck says to a s t ~ d e n t who has gatheredsome data and now is tryingto

und,erstand what hIS or her dissertation research is ab out, Tell me what

you vefound out, but without usingany of the identifying characteristics

of the actual case. I ll use my own dissertation, a study of the careers of

schoolteachers in Chicago, as an example (the resul ts are reported in

Becker 1970, 137-77).Had I been a studentaskingBeckfor help figuring

out what generalization my research could produce, he probably would

first haveaskedme what I had actuallyfound out about Chicago teachers.

I couldhave offered this condusion:

These t e a c ~ e r s make the ir careers by moving from school to

s ~ h o o l wlthm the C h i ~ a g o s c ~ o o l system, rather than trying to

nse to 1 ~ I ? h e r b e t t e ~ pald posItIOns, or movingto other systems in

other ltIes and thelrmoves betweenpositions in the school system can be ~ n e r s t o o as t rying to find a school in which the

people theymteracted with-students, parents, principals, other

teachers-would act more or less theway the teachers expectedthem t

Had I toldBeck all that, he would,usingthistrick,havesaid to me, Tell

me wha t you r research is about, but now you are forbi dden to use t he

words teacher, school, pupil, principal, or Chicago. To answersuch a

126

CONCEPTS

question, Iwould have to choose words more generalthan the specifics of

my case, butnot so general as to lose the specificity of what I found. If I

started talking about identity or rational choice or similar high-level

abstractions, I would lose what I hadlearned about careermov ements re

sulting from choices betweenmore and less comfortable work situations.

So I rnight have answered tha t my study showed how people in bureau

cratic systems choose between potential positions by assessing the way all

the other participants will treat them and choosing places where the balancewill be best, givenwhatever theyare tryingto maxirnize.

That s how I made the movefrom the fact thatbanking executives steal

to the statementI made about the darity or ambiguityof an action s crim

inality. Irestated the assertion that th e executivesof savings and loan asso

ciations sometimes steal money by manipulating banking regulations

whose complexity makes it difficult for prosecutors to decide whether

what theyindisputably did is a crime without usingany of the specifics. I

didn t say executives or savings and loans or any of the other specifics.

I saidwhat dass each of thosebelonged to and so ended up talking about

the ambiguity of an action s crirninality, a dimension that could be useful

in the study of any crirninal activity.And I could take another step and talk

about something less specific than criminallaw-mles in general-and

that would let me introduce such interesting cases as whether the ball the

pitcher throws is a ball or a strike, the mIes for deciding that being as

ambiguous as any in the criminallaw.

You could argue that, after all,baseball and banking don t have much in

common. Right.Every time we makesucha comparisonandfind sucha

similarity, we will also immediately find such a difference.Both the sirni

larity and the difference give us general categories to think about and

use inour analyses. The similarity says by way of generalizing, Every set

of mIes is c lear to some degree and ambiguous to another degree. Th e

difference says by way of a different kind of generalization, Within the

organizations (like baseballand banking) in which mIes are made and en

forced, there are other thingsgoing on,such thatthosemIeswillvaryalong

adimension mnning from clarity to ambiguity. Making such comparisons

reveals further complexities in the creation and applicationof mIes, com

plexities that can be attended to in future research.

The immediate consequence of that result is thateverystudycan make

a theoretical contribution, by contributing something new that needs to

be thoughtabout as adimensionof thatdass of phenomena.Th e only time

127



FOUR

thatwouldn tbe truewould bewhen th tw .e 0 cases studledw .d . .every respect-something SO nl k I ere entlCahn

u e y as not to be worth worrying about.

Concepts Are Generalizations

Here s a different approach to the same oint Al .them and speculate about th d d P . though we thmk about. em an efine theIdeas, or speculations or matters of d fi m, concepts are not just. I e mÜIOn In fact co .Ica generalizations which d b . , ncepts are empIr-

, nee to e tested an d fi dempiricalresearch results-th t fk re me on the basis of Wl aIS, 0 nowledge of the world

e commonly have difficulty I .phenomena: theysort offit b t app ym

g

lconcepts to real cases of social

, u notexactyTh bfine phenomena by 0 b . at s ecause we seldom de-

. ne unam IgUOUS criterion Wl d .

trunk, It S anelephant, and that s th t If . e on t say If It has a

basis of price that s a mark t If a ,olr peopleexchangegoods on the, e . we t aked tha t

sure whether a case was or wasn t f h . way, we would know for

(one 0 t e thmgs .

That s somethingof ane we were 111terested in.xaggeratIOn Wewould st ll h all

associated with decidin h ave t he p roblems. g w at a trunk or an e h

pnce is.) xc ange on the basis of

Concepts thatinterest us, however, usuall .

Weber didn t d fi b Yhave multIple cntena. Maxe ne ureaucracy by one criterion H .

characteristic features th . . e gave a long lIst of. e eXIstence of written fil . b

reers, decisions made by r I d es, JO s defined as ca-

du es, an so on Smularly s . I . .

efine culture withmult I . . . . , OCla SCIentIsts usuallyIp e cntena It co f h

handed down from SIStS 0 s ared understandingsone generatIOn to the next· f h

that embody the basic val f 0 co erent propositionsues 0 asocIety, and so on.

In theworld we live in however heutes requiredfor them to b b nomena seldomhaveallthe attrib-

e, unam 19uously b fmultiple criteria.Anor . . h mem ers 0 a class definedby

gamzatIOn as wntten fil dstrict rules but has n h 1 es, an makes decisions by

, 0 career pat s for f uncti . Inot? An organization has all o ~ n e s s It a bureaucracy, or

b

on paper, the attnbutes Wl b b

ureaucracy, but is the k d f e er attn uted t o a 0 orgamzatIOn which h has this incident, reported b d suc t gs happ.en

y or on and hIS coll .public saccess toinformation tha t I II eagues a study of the

city county,and state offices in Ill wa.s ega y s u p ~ o s e d to be availablefrom

tion laws: mOlS under vanousfreedom of informa-

When a professor from the Center for U .western University sought . Affam at North-

some votmg data 111 Chicago, for ex-

128

CONCEPTS

ample, he was clearly and repeatedly told, in person, by a c lerk

with an Irish surname, thatthose data, whilelegally public,were

not available. While he was arguing to the contrary one day, anItalian surnamed clerk glanced at the professor s name on the

written request, and interrupted to say: Masotti.You Italian?

Dr. Masotti said, Si, and spoke briefly in Italian to the clerk,

who then called to another fellow Italian who labored for 30minutes to produce a complete set of the initially unavailable

data. (Gordon et al. 1979,301)

Even if it has files and rules and all the otherWeberian criteria, is that a

bureaucracy?

A first reasonthese quarrels overdefinitions are important is that the de-

scriptive titles that embody these concepts are seldom neutral, but rather

areterms of praise or blame. Culture, for instance, is almostalways a good

thing ( bureaucracy, as in the above example, is almostalways bad).So we

care, beyond technical theoretical considerations,whether we can say that

a group has culture or not.Wedo no twishto rewardwith the approbation

signaledby the honorific title somebunch thatdoesn t deserve it. Suppose

agroup s membersshare understandings,an elementI mentionedabove as

often included in definitions of culture, but invent those understandings

on the spot,insteadofhandingthem down fromgeneration to generation.

Is that culture, or not? Some socialscientists will no t want to givea bad

group that does such things (for instance,a delinquent gang) the honor of

having real culture ; theywant to save sucha goodword for praiseworthy

organizations (Kornhauser 1978). (An interesting problem arises here

when historians discover that what seemed to bejust such handed-down

traditions embodying primordialvalues, etc.,were actuallyinvented not so

long ago, the way they have discovered that Scottishculture, as embodied

in the traditions of the ancient clans and their customary tartans, was in

ventedby woolen merchantswith excessstock on hand.)

Another problemcan be put more technically. Suppose you have x cri

teriafor an object , and you call objectsthat have all x criteria What do

you callthe objects thathavex 1 or x 2orx nofth criteria?Thesim

pIe solution is to call them not and ignore all the differences among

them-that is, treat them as though theonly thing that is important about

them is what they aren t.But that is often unsatisfactory becausehardly any

of the objects we study have all the criteria;instead they havevaryingmix

tures of them-whatWittgenstein called family resemblances. The bu

reaucracieswe study are similar,but theyaren t identicalthe waymolecules

129



FOUR

of copper are.We can,of course, give every combination of possibilitiesa

name. In fact, we seldom do that, because these devices quicklygeneratea

very large number of possibilities we aren t prepared to handle theoreti

cally or practically. (Methods for handlingthat complexity exist,and I will

discuss themin chapter5.)

So concepts like bureaucraey are really, s we ordinarilyuse them, gen

eralizations that s y Look, these x criteriaactually do go together,more

orless, all the time,

enoughso that we can

pretendthatthey are ll there

in

everyObject 0 even though almost all in factjust have most,no t all, of

them. That makes aproblembecausemany of your cases don t act syour

theory s ys theywill, precisely because theyare missing an important at

tribute that is responsiblefor that aspect of the behaviorof

We can often finesse these difficulties, because the number of cases is

small or because the objectswe collected don t lackattributes that are im

portant for the problem we are pursuing. But when we can t, we should

recognize that our concept was notjust an idea but an empirical gener

alization thatsaid thatall thosecriteria alwayswent together.

A good example from the world of practical affairs has to do with the

concept of living somewhere. When the 1960 Census failed to count a

large number of young blackmales, the politicalconsequences forced sta

tisticians and survey researchers to take the problem seriously.The practi

cal questionconfronting the research committee considering this problem

was how to conduct the next Census so s to count the people who had

been missed the last time (parsons 1972,57-77). The U.S. Census must

count people where they live, for purposes of political representation, so

the questionbecamea double one: how can we find them w r t y liv so

that they will fill outour forms, and what does it mean to live somewhere

(because ifwe understand what itmeans to livesomewherewe will know

how to reach them)?

The expert committees discussions revealed a profound ambiguity in

the notion of living somewhere.What does i t mean to live somewhere?Foreverycriterion proposed,you couldimagine aperfectlyreasonableex

ception.Youlivewhereyou sleep: ifI m on vacation inMexico do Ilive in

Mexico? It s whereyou usually sleep: I m a travelingsalesman, I don t usu

ally sleep n y w h ~ r e in particular. It s where you getyour mail: many peo

pIe gettheirmailat General Delivery or the City LightsBook Store in San

Francisco, but they don t live in those places. It s where you can always be

reached: for me, at the moment, that s the SociologyDepartment at the

130

CONCEPTS

University ofWashington, but I certainly don t live there. It s where you

keep your clothes, it swhere For most people, most of the time, all those places are the same place.

They usually sleep inthe place they get theirmail,which is also where they

have their clothes and can most easily be reached. But for most people

sometimes andforsome people all the time, these aredifferentplaces:they

keep their ~ l o t s one place and sleep in another. For them the concept is

just not adequate and, if we want to take them into account, we have tobreak the concept down into its component indicators and treat each

separately. In otherwords, we have to realize that the e ~ p i r ~ l g e ~ e r l l z -tion embodied in the concept is not true: all those cntena don t go to -

gether all the time. .

You canmake thisfailure of the indices of aconcept to stlck together s

we dlike them to the jumping-offpoint for expanding and complicating

your theory of the world. Marisa Alicea (1989) did that in her study of re

turn migrants to Puerto Rico-people who, havingmovedto NewYork

or Chicagofrom SanJuan or Ponce, then gobackto the island. Sheshowed

that, in fact, they move back and forth between their two homes f re

quently. Thus, it smisleading to think of them smigrants and farmorere

alistic and useful to see them s people who have, s she says, dual home

bases. Taking that result seriously means that another fact built into

the concept of living somewhere -that people can only live in one

place-has to be seen s simply another possibility that mayor may not be

true in a given case. .

I have sometimes upsetlistenerswith such examples,which seemto en-

tail an extremeconstructivismthat makes it impossible to do anyresearch

at all.They are especially upset if I follow the living somewhere example

with amention of how Harold Garfinkel (1967) confounded demogra

phersby describing the case ofAgnes,a transsexualwho had changedgen

ders sociallyand then physically, and then asking howthe ~ n s u s couldbe

sureit had correctly classified someone smale or female.Dld you have to

take down everyone s pants i n orde r t o b e sure of the classification, he

wanted to know? If youcouldn t use evenso simple anidea sliving some

where or being male or female, how could you observe or count any-

thing? Alicea s research shows that seeing the concept s an emplncal general-

ization helpsyou to avoid analyticerrors.We conventionallythink ~ h ~ t ~ -grants live in only one place ata time and, when theymove, stop livmgm

131



FOUR

theplace they used tolive in andgo to live somewhereelse.Weil,of course,

they do go somewhere else. But they actuaIly have some sort of home

~ h ~ t sort, of course, is the researchable question thatmakes itworth get-

tmg mto these complications) in two places, both the mainland U.S. and

theirhometownin Puerto Rico.Youcan t assumethat living in thesecond

placemeans exactlywhat meantwhen theylivedwhere theyusedto live

before theymigrated.Before moving, theymighthave thought of Home

as the onlyhome theyhad.But having acquiredHorne ,theymight d e c i d ~that theyneedn t give up thefirstone,and thenmight 1 ~ o v e back andforth

between the two the waypeople with alittle more ma neygo to theirsum

mer c o t t a ~ e s ~ v e r y year. The pathos of the story is thatthese people may

not have, m eItherplace, some of the nice things a real home gives you,

such as a secure economic base or an affectional base of peoplewhoknow

you andlove you. (Buthavingtwo hornes isn t necessarily adeprivationei

ther. CarolStack s researchshowshow poor children who can r un awa

andlive for a while with a neighbor or relative two doors down the

can profit fromhavingmultiplehomes.)

The trick of seeingconcepts as empiricalgeneralizationshelps solve the

~ r o b l e m s createdby an unthought-through insistence that all the proper

t1es of a concept always go together.Uncoupling them, andtreating them

as c a p a ~ l e of varyingindependently, turns a technicalproblem into an op

portullltyfor theoreticalgrowth and articulation.

Concepts Are Re1ational

I oncetaughta dass caIled Classics of SocialResearch. One of the books

we readfor the dasswasJaneMercer s abeling theMentally etarded 1973 ,

a study of the waythe label mentallyretarded was applied in the River

side, California schoo1s. This study proves, as weil as anyone but an ideo

loguewould want it proved, that borderline retardation as opposed to the

: real . retardation that is accompanied by obvious physical handicaps, etc.) a d1sease Mexican andBlack kids get when they go to school, and are

cured ofwhen they leave schoo1.

I was moved in dass one day to givea lecture on theidea that all terms

describing people are relational-that is, that they only have meaning

when they are considered as part of a system of terms. This is no t a new

idea. I think I first saw pu t thatway bya Marxist historian (perhaps E. Thompson or Eric Hobsbawm) who said thatdass was a relational term:

132

CONCEPTS

termslike middle dass or working dass only have meaning in relation

to one another or to upper dass; and the meaning is the character of the

relationship. Working dass means that you work for people who are

members of the owner dass.

That seems obvious enough. But it s one of those obvious things that

people acknowledge and then ignore. How do they ignore it.? By imagin

ing that a dass, by havinga characteristic culture or way of h.fe, w o u ~ d be

what it is no matterwhat systemof relations it was embedded m.That sno tto say that there aren t dass cultures, but rather to insist thatsuch c L 1 l t ~ r e sresult from some group of people being related to some other group m a

way that creates, at least in part, the conditions in which their distinctive

way oflife develops. .

A similar meaninghasbeen attachedto the idea of a countrybemg un-

derdeveloped. In this case it was done by the simple d e v ~ c e of.treating

underdevelop as averb, to underdevelop, which made 1t b V ~ U S that

there were some other countries or organizations that were makmg that

underdevelopmentbe whatitwas. In this case,there are obviouslytwo sep-

arate things: to be underdeveloped onlyhas meaning in relation to other

places thatare developed, and the distribution of development .as atrait

is created by the deliberate actions of some of those other o r g a l l l Z a t l O ~ sI took this up in dass when one of the students, a clinical psycholog1St

who found Mercer s condusions hardto accept, insisted thatmentalretar-

dation was, afterall, real, not just amatter of definition or relations. At least,

she said, there are somecases in which children are profoundly retarded. I

startedmy reply by asking the studentswhether they thought I was taIl

short. (Ifyou measured me, I would be about 5 10 , which, these days,

not particularly taIl, but not short either.) They looked confused and

waved their hands as if to say that I was medium. I insisted on an answer

and of course, theycouldn t giveit. I saidthat I used to be a shortermem-, d h

ber of the faculty, when one coIleague who mea sured 6 9 an anot er

one who was 6 6 were around,bu t thatI hadgotten tallersince theyleft. I

asked a visitingJapanese student if i t wasn t true t ha t I would be tall in

Japan. She laughed uncomfortably and finally saidyes. I said thatwhen I

was in high school Iwould have been a reasonableheight to playbasketball

but no t anymore, and went on to point ou t thatheight was about as real a

fact as youcouldhope to know about anyone-certainly as real, say, as re

tardationor intelligence.

The trick here is to place any term thatseemsto describea trait of a per-

133



FOUR

son or group in the context of the system of relations it belongs to. That

shows you thatthe trait is notjustthe physical fact of whatever-it-is, but

rather an interpretation of that fact, a giving of meaning to i t, that de

pends on what else it is connected to. The first thingit is connected to is

other traits,which havesimilarly been givenmeaning, so that theyconsti

tutea system of possibilities.The gradedseries that runsfrom profoundly

retarded to retarded to normal to gifted to genius is a good ex

ample.

But, the analysis can go on,what else is this system connected to?Why

do these distinctionsseem natural to a no more than ordinarily reason

able person?Whydo theyseem reasonable enough and importantenough

to ac ton? I pointed out that I myself was profoundly retarded -in the

area of drawing. I could neverdrawa tree or adogtheway the good draw

ers in my dass could.As a result, I had always felt ashamed.This disability

had affected my life in nontrivial ways. Another student owned up to be

ing profoundly retarded in the area of music, so unable to carry a tune

that she had always been told to just mouth the words whenher grammar

school dass sang in assemblies.

Why were these statements ironie, no t serious? Because, obviously,

these disabilities don t make any difference. No thing reallybad happens

to if you can t draw or carry a tune. may be unpleasant and mildly

shammg.You maywish you could do these simple thingswith as litde trou

ble as others.Butourworld is not so organized as to require us tobe abletosing or draw.

Ou rworld, however, is soorganizedthatpeople mustbe ableto dosome

things that retarded people can t do easily orwell or atall. To get along, at

leastat alevel somepeople andinstitutions define as minimal, you have to

be able toreada litde, do alitde arithmetic, catch on to what sgoing on

andpickup various kinds of ideas and skillswithin a certainlength of time,

readmaps, tell time, und erstand directions, and so on. Otherwise, you are slow.

LewisDexter (1964),writingabout The Politics ofStupidity, pointed

out that all those skills result from ou r ancestors and contemporaries hav

ing builtand maintained aworld thatmakes thoseskillsmore or less neces

sary.You could build another kind of world where a similarnecessity for

physicalgrace and dexteritywould be built into itsphysicalappurtenances.

In sucha world, it might be necessary, in order to open a door, to perform

some rather complex physical movement awkward people would have

134

CONCEPTS

trouble with; some very awkward peoplewouldn t be ableto openit at all.

We might call these people gawkies and have special entrances to places

builtfor them, perhaps give them special remedial dasses inthe hope of re

daiming them for a productive life, althoughwe might have to conclude

sadly that their genetic endowmentmade it impossible.

So there is a great difference between a physical trait and its social im

portance. We all have all sorts of traits, only a few of which are socially

marked as important because of theway theyare embeddedin a system ofrelations. Theybecomeimportantwhen the organization of physical and

socialarrangements makes them necessary. Takeheight. If you are above

or below a certainrange ofheightour physicalarrangementsmake it awk-

- ward. If you re short yourfeet won t reach the floor when you sit in stan

dard chairs; if you re tall you ll bump your head on doors if you aren t

careful. Our social arrangements are somewhat more forgiving; but still,

very tall women andvery short men are exposed to troubles finding part

ners the rest of us don t have.

All this has a historical dimension. Several centuries ago, people s aver

age height was less than it is now-so doorways built in the fifteenth and

sixteenth centuries, unless they have been rebuilt, will catch carelesscon

temporary people andbump their heads.Or take the skill of doing simple

arithmetic. Anyone, these days, who cannot do addition,subtraction, and

other simple arithmetic operations is certainly slow maybe even re

tarded. But thoseskills werenot always required.PatriciaCline Cohen s alculating People (1982) showed that it wasn t until weil into the nine

teenth centurythat the ordinaryAmerican reallyneeded such skills; before

that storekeepers and derks might needthem,butnot the average person.

She calls theseskills numeracy, in analogyto literacy. The termempha

sizes that it is because these are socially valued skills now, built into our

everyday operations, thatwe see them as such important human abilities;

at an earlier time such skills might have been interesting cultural orna

ments just as singingand playing the flute were, but certainly not impor

tant.

Skills and traits not only become more important, they also become less

important. Diana Korzenik s book rawn to t (1985) describes the

changes,back andforth, in the importance of skill in drawing in American

society. In the middle andlatenineteenth centurysome importantpeople

decided thatthe reasonthe United Stateswas falling behind in industrial

ization was that Americans did not kriow how to draw. Much invention

135



FOUR

and adaptation of machinery took place on the floor of industrial shops,

where workmen dreamed up improvements and inventionsbased on their

detailed experience of the operations involved. For tha t to happen eff i

Clently, w o r k ~ n n had t o be able to draw plans from which the necessary

partsand eqUlpment couldbe built. But Americanworkmenhad not been

trained in mechanical drawing and were no t as good a t i t as were, for in

stance, Germanworkmen. Steps weretaken: amovement to have remedial

classes for a d ~ l t s so thatworktnencould acquirethis necessary skilI;a push

to have drawmgtaughtmore systematically in the elementary schools.But

that empha.sison drawing was relatively short-lived; other developments

madedrawmg no t so important after all, which meant that in the 1930sI

could go through elementary school andbe thought a bright studenteven

though I couldn tdraw (and had, in addition, tertible penmanship,which

would have been a severe handicap in the pre-typewriter era).

gets to saywhich traits are importantenough to be madethe basis

for senous and fateful distinctions? Sometimesit is our immediate associ

~ t e s will d e ~ i d e themselveswhethermy inabili ty to draw or your

mab ty to do anthmetlC or her inabilityto carrya tune are serious enough

to warrant special negative treatment, or whethermy ability to remember

be . r e ~ d y to play one thousandpopular songs on thepiano or your abil

Ity mlltate Cary Grant or Groucho Marx or Judy Garland warrant

speClal rewards. Sometimes, and this is where Mercer s results are so im

portant, the decision is put in the hands of specializedprofessionals who

possess special esoteric methods for making these determinations. One

of ~ e r c e . r s truly shocking findings is that gross racial and ethnic dispro

P O ~ t l O n s labeled retardation do no t appearwhen teachers recommend

chl1dren m their classrooms for intelligence testing-the children recom

mended display the same proportion of Mexicans, Blacks, and Anglos as

the generalschooJ population.No, the gross overrepresentationofMexi

cans.appears o y when intelligence tests are given, when the decision to

classlfy a chl1d as retarded is madeby someonewho has . f. . no expenence 0

the child m thereall ife of the classroomand cannot interpret the bare test

scores m thelight of other knowledge of the chi ld So th i al. . e pro eSSlOn lza-

t ~ o n of these d e c i ~ i o ~ s through the development of occupational special-

~ l e s monopohes, anotherimportanthistorical variable affectinghow

mdlVldual trait s come to be embodied in a set of social relations that

make them Important.

Politicsan d power similarlyaffect how systems of relations make some

136

CONCEPTS

traits important. If a negative trait is being assigned to people, powerful

people can often prevent that from happening to them or theirs. If some

thing good is being passed out, they will do theirbestto see thatthey and

theirs get i t. In the 1980s, the U.S. Congress (presumably trying to give

middle-classpeople something to balance the specialresources being allo

cated to the educat ion of poorer, so-called underprivileged children)

authorized a program for gifted and talented children. I supposethat the

distinction mirrors, on the positive side, the distinction between pro

foundly and mildly retarded.

This programcreated a problem for teachers of visual art in the public

schools: how do you choose the children who are gifted or talented and

thus deserve extra training and opportunities? Even though middle-class

parents are,by and large,more interested in otherkinds of skills and t a l e n ~ sthan they are in visual art, still, if i t s there to get,theywant it.Theywant It

to the degree that the people who decide who gets such special treatment

need ascientificallydefensible way ofmakingthe choices involved.Which

is how I ended up ata conference thatwaslabeled as beingabout creativ

ity in the arts but turned ou t in fact to be about Can o u devise. a test of

Someabilitysuchthat I cantellparents thatchildren got mto the Glftedand

TalentedProgram on the basis of this test score and please leave me alone,

I can tdo anything about it if your child s score was low?

So the teachers problem became a testers problem.What do youmea

sure to assess ability in visual art?This was a serious problem because it is

muchmore difficultto agree on a criterion in artthanit is in arithmetic or

reading.There is, however, one thing that everyone knows is important

forvisual art ,and that happens to be the thing I can t do:drawing. Unfor

tunately, i t isn t obvious that the abili ty to draw, even supposing th.at i t

might be relatively easy to test, is closely related to, say s u c ~ e s s a l ~ a lartist, anymorethan such conceptual abilities as ability to vlsuahze spatlai

relationships or color sense or you name it. Furthermore, it s obvious that

if you use a criterion such as success as an artist youmightwa.nt to includesuch socialand businessskills as hustling. Still further, some vlsual arts, no

tably photography,don't require anydrawingability atall, soany test based

on drawingwould necessarilymake some gross errors.

What s thepoint of this lengthy digression about gifted and talented ?

That thepower ofmiddle-class parents can affect the way this systemof re

lations is set up andthus make i tmore or less important, and more or.less

available to people of different kinds.Bu t their powermay no t be sufficlent

7



FOUR

to overcome the power of the entrenched professionals into whose handsthese determinations have fallen.

A second point to this example is that there are atleasttwo kinds of sys

tems of relations involved. In one, the reputationally desirable position is

in themiddle, at themean of whatever is beingmeasured, likeheight.This

recalls EverettHughes's suggestion, discussed earlier, that we inspect devi

ations from the average in two directions, looking both for people who

have more of whateverit is andpeople who have less. In his example, one

doesn'twant to deviate trom themodalway of organizingsexual relations,

eitherbybeing worse than others (in ways thatproduce labelslike rake

or slut ) or by being better (being, say a goody-goody ). In other re

lational systems, however, reputations and their results for one's life get

better thefartheryou go in one direction,and worse thefarther you go

in the other direction. Intelligence is likethat, as are other traits like artisticability.

Tosummarizethissetof tricks:Pu t terms into thefull setof relations they

imply as tall implies short and gif ted implies n ot gifted ).Then look

at the way thatset of relations is now organizedand has been organizedat

other times and inotherplaces as in understandingthatno t knowing arith

metic has a differentmeaning and differentconsequences thanit did 150yearsago).And, finally, seehow thingscameto be organized thewaythey

are hereand now, andwhat connections to other socialarrangements sustain that setof relations.

Th e Wittgenstein Trick

I've owned a copy of LudwigWittgenstein's Philosophical nvestigations for

years, but I readit the way EverettHughes told me to readthe sociological

writings of GeorgSimmel: not to geta full understandingof whatthe au

thor might have meant, but rather as away of generating ideas I could use

in my own researchand thinking. One ofWittgenstein's ideas has become

a standard part of my repertoire. Because it was provoked by a passage in

the nvestigations I think of it as the Wittgensteintrick.

Discussing the philosophical problems of intention and will i n one of

the numbered paragraphs thatmake up thebook,Wittgensteinmakes this

remark: Let us not forget this: when I raise my arm,'myarm goes up. And

the problem arises: what is left over if I subtract the fact that myarm goes

up from thefact thatI raise my arm? (Wittgenstein 1973, §621).That's the

138

CONCEPTS

essence of the trick: if I takeawayfrom an eventor object X some quality

what is left?

, This trickhelps us stripawaywhat is accidentally and contingently

of anidea fromwhat is atits core, helps us separate what's central to our Im

age of a phenomenon from the particular example it e m b ~ d d e d in, as

Wittgensteinisolates the core of our intuitive imageof mtentlOn by sepa

ratingthe contingent physicalaction fromit. Here's an example.I was once

part of a panel organizedto talk about modern art. One of theother pan

elists had become a serious, big-money coilector of contemporary art

three years earlier.When itwashis turn to speak,he talked ~ n o w l e d g e a b l yandat length about his coilection, which of course conslsted of a large

number of paintings, sculptures, and other objects. As I listened to him, I

thought, I have a house fuil of paintings and other objects,just he ~ o e s ,but I don t have a coilection.Why not? So I did theWittgenstem t n c ~ . I

askedmyself: What is left overifwe subtract from the idea of a coilectlOn

the fact that this coilectorhas alargenumberof paintingsand other art

jects in his house? I turned to my data-the talk the c o i l e c t o ~ was glV

ing-for the answer. He immediately gave me part of the so?utlOn to my

problem: his coilection, as opposed to my mere mass of obJects, had, he

said,a direction. Itwas not just an airnless assortment of stuff, the resultof

whim and caprice; less pejoratively, it did not represent the u n ~ u t o r e d ap-

lication of his own taste. Rather, it resulted fromand embodledknowl

:dge and trainedsensibility (his own and thatofhis advisers), thus had

a concrete and explicit aim and structure. Likewise, his coilectlOn had a

future. was headed somewhere. would be the object of repeated

evaluationby knowledgeable experts. It was part of a world artistic ac

tivity and progress,itsvery accumulationan actof substancem.that V0rld.

My stuff, in contrast,was just that: stuffIhad bought b ~ c a u s e I.hked It, stuff

I had tradedmy photographs for; accumulatin.g itwasJusta p r ~ ~ a t e ~ h a thad no significance to anyone but me nu.ne. word , ust

portant here, signifying as i t oftendoes m phllosophlcal talk merely or

no more than. ) . .

In fact, as the coilector talked, I realized that having the obJects m the

house (or the office or anyplacehe actuailylived or w o r k ~ d w ~ s no t reaily

necessary to his having a coilection.Accumulatingthe obJects m one place

is not necessary to the idea of a coilection.Why not? If you are a d ~ a l ~ r spe

cializing in new,trendyart (the kind the coilector c o i l e ~ t e d yo.u mSlst.be

fore you seila piece (thedealerwho was the thirdpanelistexplamed thlS to

139



FOUR

me) that the purchasermake the work available for loan to museums for

exhibitions. If you, an artdealer,are trying to build anartist s reputation, it

does neither you 01 the artist any good to havean importantpiece sitting

in someone s living roomin the Midwest, no matter howmuch yousold it

for. The piece mustbe where i tcan beseenby importantpeople (that is,

peoplewho are important actors in the world in which such paintings are

exhibited, bought, and sold) and thus contribute to the development of a

career.Many museums have shows thatare part of thisprocess,and the pur-

chaser of a work mustmake the purchase available for them. In fact, I had

beenin Amsterdam a few months earlier, and had seen, in a show ofwork

by NewYork artists at the StedelijkMuseum, manypiecesby the artists the

dealer on the panelrepresented,some of them in the collector s collection.

Truly with it collectors thusmight not see sizable fractions of their c01

lections for long periods of time. In fact, of course, some people s collec

tions, 01 portions thereof, are often on more 01 less permanent loan to

museums (whichhope to be left those works in the lender s will).

Using theWittgensteintrick, then,what is leftwhen you takeaway from

collection the idea that you have a lot of art stuff in the house?What

seemed to be left (in this situation at least, but I t hink i twou ld bea com-

mon view of the problem) was the idea of the collector as apersonwho hasthe financial and cultural resources (the latter what Pierre Bourdieu has

called culturalcapital ) to chooseand acquire objects thatrepresentwhat

will eventually turn out to have been major trends in modern art. In his

talk, the collectorsaid somethinglike this: T he idea is to find outhow to

getthe bestwork of anartistwho will be historicallysignificant,worksthat

will turn out to be a major part of art history.Your reward is to have your

judgment approvedby history. On thisview,where the objects are is irrel

evant, and havingobjects in itselfdoesn t makeyou a collector.The objects

aremerely the visible symbols of the decisive action the collectorhas taken

by staking bigmoney and a reputation for sagacity and sensibility on the

choice of art works, and it s that act ion that is crucial to understandingwhat a collection s (Which is why somemembers of theart world dispute

the characterization of]oseph Hirshhorn, for whom a majorart museum

inWashington, D. C. is named, as a greatcollector.Can you,they complain,

be a greatcollectorif, as he is saidto have often done, youjustwalk into an

artist s studio and after a quick look around buy everything in it?Where s

the sagacity and sensibility in that? This, of course, is an art world com-

plaint, not asociologicaljudgment.)And it s notjust theaction the collec-

140

CONCEPTS

tor takes that s important for understanding the idea of a c o ~ e c t i o n : ob -

viously; it s also the action the rest of the world takes by m ~ m what the

collector accumulated significant in art history 01 not. I ve drawn on

Raymonde Moulin s analyses [1967, 1992] of French interna-

tionalart markets far some of these ideas. An attentlve readerWIll see, too,

that this trick is anotherway of describingwhatyou vestudiedwithoutus

ing any of the specifics,which is whattheBeck trick does.)

En1arging a Concept s Reach

The Wittgensteintrick,then, lets us isolate the generic features of a s e ~ i e sof cases we think have something in common, the features ou t of w ~ l hwe can construct the generalization that is a concept. Once we 1 ~ 0 -latedsuch a genericfeature of some socialrelation 01 process and glVen lt a

name and thus createda concept, we can look forthe same phenomenon

in p l ~ s other than where we found it.The study of prison cultures fur-

nishes a nice example. .

Students of prisons (e.g., Sykes 1958) had demonstrated that the m-

mates of men s prisons developed an elaborate c u l t u ~ e . Inmates ~ r e a t e d. ents that took overmany of thefuncuons of keepmg 01 -onv1Ct governm . .

derin thejoint; they developedinformal but orderly markets m c ~ g a r e t t e s ,drugs, tailor-made prison uniformsfor the snappily ~ r ~ s s e d convlct, and:

. ty fpersonalservices they organizedsexual ctlVlty;t hey enforcedane

strict code of c,:onviet behavior emphasizing the necessity of.neverglvmg

informationabout other prisoners to prison guards and OffiClalS. .

Analysts of prison culture attributed these inventions to the depnva-

tions of prison life: deprived of autonomy, prisoners carved out a g o v e r ~ -mental structure that got some autonomy back for t h e ~ l and a C O ~ V I C tcode (ofwhich the prohibition on snitching on other pnsoners to .pnson

staffwas a major component) that preserved that a u t o ~ o ~ y . ; . d e p ~ l v e d of

drugs, sharp clothes,and other goods theywereu ~ e d

to m clvillanh ~ e ,

theyorganized markets to provide those things; depnved of sex,.they ~ m p r o -vised a system of predatory prison-specific homosexual r e l a u o n s ~ l p s t ~ a tdid not threaten their self-conceptions as macho men. The soclOloglcal

generalization, a specificationof amore general set of ~ d e a s that goes back

toWilliam Graham Sumner, was that prisoners collecuvely d e . v e l o ~ a cul-

ture thatsolves the problems created by the deprivations of pnson life.

So far, so good.With this theory in mind,Ward and Kassebaum (1965)

141



FOUR

studied a women s prison. They found none of the things the theory of

prison culture hadled them to expect.Quite the opposite. Even the offi

ci lsof the prison complained about the lack of aconvictcode: thewomen

were forever snitchingon each other in awaythatmade a lot of troublefor

them and thus for the prison staff. No real underground market existed.

Sexlife was not organized in the predatory style of the men s prison; in

stead, the women developed pseudo-families, with butches acting s the

husbands and fathers of acollection ofwives and daughters. (See also Gial

lombardo1966.)

Did these differences-the absence of any of thethingspredictedby the

available theory of prison life-invalidate the generalization that the dep

rivations of prisonlifeleadto thecreation of aprison culture?And didthat

in turn mean that no gentralizations about prisons were possible? No t at

all. They meant that thegeneralizations are not about how llprisons are

just the same, but abouta process, the same no matterwhere it occurs, in

which variations in conditions create variationsin results (which is actually

amuch dassier form of generalization anyway).

In thiscase,the theorywasn twrong, but youhadto putin the rightval

ues of the variables, so to speak, to see how it was right.You couldstill s y

that the deprivations of prisonlifeled to the creation of prison culture, but

that was true only if you understood thatthesedeprivationswere different

for men and women.Women were not deprived of autonomy because, s

they explained to the researchers, they had never been autonomous; they

had always lived underthe protection of and been subjectto the authority

of aman: afather, husband,or lover.Whatprison deprived them ofwas ex

actly thatkind of protection.So, rather thandevelopa convictgovernment

to replace the autonomy they didn t miss, they developed a systemof ho

mosexual relationships i n which one woman stood in s the masculine

protector.

New women prisonerswere especiallyafraidbecause, due to variations

in the gender distributions of crime, men s prisons have a lot of profes

sional criminals serving time for robbery, burglary, and other less violent

crimes, whilemost women prisonersare in for drugs and prostitution, and

for the typically amateur crime of passion -that is murder. Since there

are thus more murderers in them, women s prisons sound like very dan

gerousplaces to be,even to the murdererswho know that they themselves

aren t dangerous (theyjust wanted to kill that one person who done them

wrong). So even the murderers are looking for someone to take care of

142

CONCEPTS

them. Similarly, women s prisonstypicallyallow inmatesto buy things they

want, like cosmetics and dothes, so there is no need for an underground

market.

In short, women prisoners are deprived of different things than men,

both because theirliveson the outsideand, therefore,their needson thein

side,differ,and because the prison is run differentlyfor them.Their culture

responds to that difference.The generalization is still tme, even though the

results are quite different.

The generallesson here, the trick to be appliedelsewhere, is not to mis

take a specific instance of something for the entire dass of phenomena it

belongs to. Deprivation probably leads to the collective development of

cultural practicesdesigned to relieveit in all sorts of settings,butwhatcon

stitutes deprivationmay vary considerably.

We are mostl ikely to confuse part of a dassfor the whole in this way

when t he dass has a well-known name that applies to an equally well

known set of instances. That swhy people who study education almost

always studyschools. That s where education takes place,isn t it? Everyone

knows that.Education, conventionally defined, consists ofknowledgeable

peopleteachingpeoplewho are less knowledgeable, and typically, not sur

prisingly,lesspowerf ul and less well-placed (children or immigrants,for in

stance),and doingit in schools.That s what education s

If, however, we think of education and learning s generic social

processes, there s no reason to think thatthose processes take place only in

schools. We might try to redefine the subject matter s people learning

things, wherever and however that activity happens and whoever does it.

Then we could indude in our collection of cases the way thieves teach

other the latest techniques of their trade, or the way young people teach

other to use drugs or engage in sexual activity.But that s just cheap irony,

because everyoneknows that thoseactivitiesaren t education, atleast no t

what anyreasonablelaypersonmeans by that.Education means schools.

But there s no reasonto assume thatlearning takes place in schools at all,even though that is the story schools tell about themselves and the story

well-socialized members of ou r society believe, or at least pretendto be

l ieve so that they won t appear to be nuts.You canstudy, s an example of

learning, how young people learn to use marijuana . You may find, s

Schaps and Sanders did in 1970 (andit mightbe different at another time),

that young women typically learn from their boyfriends, while the

boyfriends learn from each other.Byignoringthe conventional instances

143



FOUR

that define the concept,you have enlargedits reach.You have discovered

new people who do t he job of teaching andnew relationships in which it

is done.

It s quite l ikely that the process bywhich boyfriends teach their girl

friends to smokedope has a lot in commonwith other activities in which

knowledge, skill, and ideas are passed on. It might, for instance, resemble

the sys tem described by Gagnon and Simon (1973), i n which young

women teach their boyfriends to engage in romance, which they have

been practicing in sol itude for qui te a whi le , whi le the boyfr iends are

teaching them to engage in sex, which t y have similarlybeen practicing

in solitude. If the process works, and each learns what the other knows,

they canmanage to fall in love in themoreor less standard way

These processes of peer te<i ching andmutuallearningmay, in turn, have

their counterparts inside schools and other so-called educational institu

tions.Personalcomputerusersoften teach each otherhow to use theirma

chines, despite or perhapsbecause of themore conventional standardized

instruction availablehere and there. Students in conventional educational

institutions have repeatedlybeenshown (e.g., Becker, Geer, and Hughes

[1968] 1994) to teach eachotherhow to dealwith the constraints,requi re

ments, and opportunities those places embody:howmuch of the assignedwork you really have to do, for example.

To take another variation on the standardmodel of education, some

kinds of teaching and learning are, unlike the elementary and secondary

education that form the archetypal instances that define the concept, to

tally voluntary: lessons in pianoplaying,tennis, and Frencharealllikethat.

They takeplace in profit-making establishments,are oftenif no t always in

dividual, and have no fixed term.The students get no credits and no de

grees. Theyjust take lessons until theyfeel theyaren tgetting anything out

of it anymore. The distribution of power between student and teacher is

so different from the stereotypical school that thisis bound to be a some

what differentgeneric type. (Seethe discussionin Becker 1986a, 173-90.)

An excellent, perhaps the best, way to enlarge the reach of a concept is

to forge t the name entirely and concentrateon the kind of collective ac

tivity that is takingplace. A good example of this strategy is Erving Goff

man s analysis (1961) of what places that had the genericfeatures of total

institutions hadincommon with respect to the waytheirinmates(bethey

nuns, sailors at sea,or mentalpatients) had to l iveand the kinds of adjust

mentstheymade tol iving thatway Or his analysisof the characteristicso-

144

CONCEPTS

cialforms thatgrew up around people who had stigmas of various.kinds

(Goffman 1963).The brilliance of these analyses was to show that , m .the

generic sense hehad in mind, everyone had some sort.of s t i g ~ a no t

1 h bllnd or missing a limb and every mstl tut lOn was, meop e w 0 were . .

some respects,a total institution.Exchangmg the c o n v e ~ t l n a ~ contents ofa concept for a senseof itsmeaning as a form of collective action enlarges

itsreach and ou r knowledge.

It s t ime in the next chapter to consider some more formal ways .of

working with concepts, ways that use the devices of serious formalloglc.

145

18 becker tricks of the trade chap4

Documents