Otis Dudley Duncan. Classic Articles

Population Investigation Committee
The Measurement of Population Distribution Author(s): Otis Dudley Duncan Reviewed work(s): Source: Population Studies, Vol. 11, No. 1 (Jul., 1957), pp. 27-45 Published by: Taylor & Francis, Ltd. on behalf of the Population Investigation Committee Stable URL: http://www.jstor.org/stable/2172508 . Accessed: 15/12/2012 00:02
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Taylor & Francis, Ltd. and Population Investigation Committee are collaborating with JSTOR to digitize, preserve and extend access to Population Studies.
http://www.jstor.org
This content downloaded on Sat, 15 Dec 2012 00:02:22 AM All use subject to JSTOR Terms and Conditions
The Measurement PopulationDistribution of

By
OTIS DUDLEY DUNCAN
of of distribution problems are The analysis explanation patterns population and humanecologist,and locationeconomist. geographer, for the demographer, out for however, takenresponsibility working a has None of thesespecialists, in method for dealing with the subject. Demographers, comprehensive is have given it littlesystematic attention, althoughthe literature particular, on distribution various in materials population richin elementary descriptive of a of regions. This paperattempts summary themajortechniques describing some unresolved problems indicating and measuring distribution, population wellbe thefocusoffurther research. that ofmethod might classificationmeasures population of is of distribution; The following a tentative of to is overlapping someof it does notpurport be exhaustive there evident and thecategories. A. Spatialmeasures of sub-divisions by (I) Numberand density inhabitants geographic of (z) Measures concentration
measures B. Categorical and metropolitan-non-metropolitan Rural-urban classification (i) size distribution (2) Community to or sites by (3) Concentration proximity centres to designated the illustrations Within compassof thispaperit is possibleto give onlybrief theseheadings. under
NUMBERS AND DENSITY BY GEOGRAPHIC SUB-DIVISIONS
of (3) Measures spacing measures (4) Centrographic potential (S) Population
in distribution The basic information employed most studiesof population of is thecensus enumeration population geographic of sub-divisions a country by unit. It will appear below that summary or other territorial measuresof distribution may depend heavilyon the areal units by which population are of enumerations tabulated. Such units-geographic sub-divisions the total is over whichdistribution described-areof threemain types: (i) territory of of units,but not political units,(2) unitsconsisting combinations political and (3) units speciallydelineated statistical for themselves politicalentities, to conforming politicalboundaries. International purposesnot necessarily of and of studies distribution typically employ countries combinations countries is as units, there interest wellin themoredetailed but as analysis thatcombines with comparisons.' intra-country international
1 Glenn T. Trewartha," A Case for PopulationGeography", Annalsof theAssociation American of Abbott Payson Usher, " The Historyof Population and xLIII (June, 195 3), pp. 71-97; Geographers, in xx Review, (January, in 1930), pp. II-I 32, reprinted Demographic Settlement Eurasia ", Geographical Analysis, edited by JosephJ. Spenglerand Otis Dudley Duncan (Glencoe, Ill.: The Free Press, I956) pp. 3-25.
27
z8
OTIS
DUDLEY
DUNCAN
carnumbers have developedseveralmethodsof portraying Geographers symbols designate to special such tographically as thedotmapor mapsemploying purposes, for whileappropriate particular sizes. Each ofthese, placesofspecified the to difficulties trying depictwithequal fidelity distribution when encounters of and settlement the large nucleations population.' in regionsof dispersed that method ofcourse, onlyrelatively with is, The basicdifficulty anycartographic with can conclusions be demonstrated a map. imprecise of portrayal numbersby A step beyondthe mere listingor cartographic to of is sub-divisions the computation the ratioof population area, geographic of calculation density. In view of the wide use and frequent i.e., population literature thatthe demographic contains little it figures, is surprising density treatment of discussion theconcept. A well-rounded methodological fundamental to a sufficient paper. Here it is perhaps of the subjectwould require separate that that are unresolved problems arisein theuse ofdensity, indicate there serious the obtained distribution. First, results or population statistics maps to study of will depend,to a significant degree,on the system areal unitsfor which are inasmuch there no wayto assigna uniquemeaning as is densities calculated, in of to thenotion density thevicinity a givenpoint. Second, constructing of in on is in or mapsthere difficulty deciding a suitable density dasymetric isopleth intervals. Each of the proposedsolutionsto this problemset of density " " it whether leans towarda " mathematical or a " functional criterion-is to density in unsatisfactory one way or another. Third,any attempt refine a than " gross" area encounters figures basing them on " net" rather by ". in considerable indeterminacy the notionof " net-ness Most presentations are clearnorconvincing. of " net" densities neither and between demographers is It wouldappearthatthere needforcooperation for in density.Whereas techniques studying statistical geographers developing a in the of have regarded study density primarily,problem as, thegeographers should be in a positionto clarify demographers presentation,2 cartographic of measurements. theanalytical and purposes limitations density
MEASURES OF CONCENTRATION
has of the in density to do with Ordinarily, majorinterest studies population or rather with theoverall average than over just thevariation density a territory of is withthatof this of Formally, problem identical territory. density theentire
1 John W. Alexander," An Isarithmic-Dot Population Map "
1943), pp. 43I-432.
xix Economic Geography, (October,
2 Alexander," An Isarithmic-Dot Population Map ", op. cit.: John W. Alexanderand George A. ", Maps of the United States: Techniques and Patterns Geograpbical Zahorchak, " Population-Density H. xxxiii (July,1943),pp. 457-466; JamesA. Barnesand Arthur Robinson," A New Method Review, of xxx (January, 1940), pp. Review, for the Representation Dispersed Rural Population", Geographical 134-1 37; PrestonE. James, The GeographicStudyof Population",Chapter in American " 4 Geography: Inventory Prospect, and editedby PrestonE. Jamesand ClarenceF. Jones(Syracuse: SyracuseUniversity Map of Farm Populationin the United States", Press, 1954); Eugene Mather, " A Linear-Distance xxxiv (September,1944), pp. 173-I80; John K. of Geographers, Annals of the Association American " Wright, A Method of Mapping Densitiesof Population,withCape Cod as an Example ", Geographical xxvi (January,1936), pp. 103-I0i. See also, Philip M. Hauser, Otis Dudley Duncan, and Review, Research Report AFPTRCReport, of BeverlyDavis Duncan, Methods UrbanAnalysis: A Summary TN-56-i (San Antonio: Air Force Personnel& TrainingResearchCentre,1956), pp. 67-69.
THE
MEASUREMENT
OF
POPULATION
DISTRIBUTION
29
", of studyingthe " unevenness" or concentration population. " Concentration language having a similar it may be noted,is a termlike so manyin the scientific to Latin form,thathas two specificmeanings: (i) as referring a stateor degree at of unevennessof population distribution a given point in time,(z) as referring to the process of increase over time in the degree of unevenness (the reverse "). Abstracting change, decrease of unevenness, is called " deconcentration any fromthe spatialpattern population distribution, measureof concentration of meaning of the term) seeks simplyto make operationalthe notion (in the first of the " degree of unevenness". A device forgraphicpresentation and two index numbersassociated therewith (although many other have been most widely used for measuringconcentration measuresare conceivable,some of which have been seriouslyproposed). These are (I) the Lorenz curve and Gini's " concentrationratio ", both originally suggested for measuring inequality of income or wealth' and adapted to the measurementof population concentration; and (z) what will be called here ", simplythe " index of concentration or more generallythe " index of dissimilarity ",A. Figure IA illustratesthe principleon which the Lorenz curve is constructed. Area units are arrayed in order of decreasing density. Then,
,oo
f f
Density Interval
/ ~~~(Pop.Ml 2)
90.
8
so
45- 89.9
ox
III
-
/:18-
44.9
Under 18
? 60 -1
QL
a0-
E
20
----
..............
`7
20
40
60
80
t00
CUm *b
POP
Cumulative Per Cent of Population
in FIGURE I. (A) Lorenz Curve for Measuring Population Concentration the United States, in Relation to DensityIntervals (County Basis): 1950. (B) Index of Concentration, in Relation to Lorenz Curve. A, in of Dwight B. Yntema," Measuresof theInequality thePersonalDistribution Wealthor Income", xxviii (December,1933),pp. 423-433; MaryJeanBowman, Association, Statistical of American Journal the in Economic " A Graphical Analysis of Personal Income Distribution the United States", American 1945), pp. 607-628. xxxv Review, (September,
c
30
OTIS
DUDLEY
DUNCAN
the one intointervals, computes them or units separately grouping these treating of and of cumulative percentage population of areawiththeaddition each unit of percentages area (Y,) of cumulated and plotsthe several (or interval units), of cumulated percentages population againstthe corresponding (X,). Such a all had the in throughout, thecasewhere units equal curvewouldfollow diagonal and would coincide with the X-axis if all densities(even distribution), point. at were concentrated one mathematical populationin the territory extremesof complete evenness (no Variation betweenthese hypothetical is to and by concentration indicated the degree which concentration) complete the from diagonal. departs thecurve ratio, The concentration CR, is givenbytheformula:
IO1000
CR=
2
i=I
Xi,1Yt- E X,Y l,
i-I
and cumulative X percentages, k is thenumber where and Y are therespective have beengrouped). Geometrically, if of of arealunits(or intervals units, they the between cutveand the thisformula expresses area on the graphcontained areabelowthediagonal. The Lorenz as of thediagonal a proportion theentire as of ratiohave been studied, measures population curveand concentration considers most onlythecase who,however, by distribution, intensively Wright, and restriction one which are all in which units ofequal areasize,an unnecessary discusses data. Wright to measures actual of 1 application these virtually precludes otherthan CR, that dependon the shape of the also some indexnumbers, the it from diagonal. to Lorenzcurveas wellas thedegree which deviates is the of A, The indexof concentration, algebraically simply maximum the vertical it distance setof k valuesof (Xi - Y,). Geometrically,is themaximum in as method to the from diagonal thecurve, illustrated FigureiB. An alternative to student employA as the is perhaps first of computation givenby Hoover,2 are concentration extensively.Ifxi andy, theuncumulated anindex population of then and of percentages population arearespectively,
A=-E~Xl-y,l
2 Z
Ik
the between two percentage disdifferences the or simply sum of thepositive tributions. before intervals units intodensity of the computing Incidentally, effect grouping lower value than that obtainedin CR or (usually) A is to give a slightly that data. It is evident from intuitively, the Lorenz ungrouped computation a curve encompasses slightly largerarea when drawnwith manyshortline thanwhendrawnwithonlya fewlong ones. When A is computed segments data onlyif as fromgroupeddata,the same value is obtained forungrouped is of density theterritory. of thelimit oneoftheclassintervals themeanoroverall
1 2
xxvii (December,I937),
" Annalsofthe Association American Geograpbers of JohnK. Wright, Some Measuresof Distributions,
I77-2II. ",
1941), I (November, History,
of Redistribution Population,i850-I940 Edgar M. Hoover, Jr.," Interstate

I99-205.
Of Journal Economic
THE
MEASUREMENT
OF
POPULATION
DISTRIBUTION
31
For mostpurposes, however, half-dozen a intervals so are usuallysufficient or forcomputing CR. Fromthegeometric relationships CR and A to theLorenzcurve populaof of tionconcentration,can be deduced it that Thus if A =o 5 (so%) thenCR is at leasto 5 and can be no greater than0 75. OftenCR approximates averageof theseextreme the values. For example, inFig. i, withA =o0* zA - A 2=o 83o,andtheaverage o * 587, is 7o8,as compared withthe actual value, CR=o 73o. For comparisons situations widely of of either aboutas wellas theother, they varying concentration, indexserves though rankclosely the wouldnot,of course, similar situations exactly samein all cases. in Inasmuch A canbe computed as without arraying units order density, areal of it is usually simpler the measure compute. to in of It can be shownthatbothCR and A can be expressed terms indexes of the dispersion the densities the areal units(the meandifference the of of and mean deviation,respectively).Hence a measureof concentration be can abouttheoverall as of regarded a measure thedispersion unitdensities of density. This beingthe case, the earlier remarks about the mathematical indeterminacy ofthedensity of concept applyas wellas to measures concentration. some of the problems interpreting Perhapsthe best way to appreciate of of is to numbers measures concentration examine index under alternative computed of States, conditions.Table i showsindexes concentration (A's) fortheUnited the of into I900 to 1950, forvarious systems dividing country sub-areas.These characteristics the index of of data serve to bring out some important concentration: The value of theindexis, in general, related the to (i) directly intowhichtheterritorysub-divided, inversely is of or number arealunits related is statement thefollowing: to the averagesize of the units. A morerigorous for can thantheindex An indexcomputed a givensetofarealunits be no larger for sub-divisions the first of set. Thus, the computed a set whichcomprises on Table i. Indexes Population of Concentrafion, computed theBasis of Counties, and States: 9poo q9o. to Economic Sub-regions, Geographic Divisions, theUnited for
Year
I950 ... I 940 ... ... ... ... ... ... ...
A < CR ? 2 A-
2.
County basis
58 9
S 59 538 52 8
Economic Subregionbasis
50 I 49 2
Geographicdivision basis
39.2 40 5
... I930 ... ... I920 I9I0 ... ... I900 ... * Not computed.
... ... ...
... ., ... ..
59*I
49*I 48 3 48 7 5X6
40? 8 4I.2 4I16 43 6
Source: United StatesCensus publications.
mustbe at leastas greatas the indexbased on subindexbased on counties of and if is regions (combinations counties), willbe greater there anyunevenness the of distribution countieswithinsub-regions.(z) Consequently, index by
32
OTIS
DUDLEY
DUNCAN
provides no unique answer to the question of what degree of population a concentrationcharacterizes territory. Any index value must be considered relativeto the systemof sub-areas on which it is based. (3) Moreover, the index does not give a unique answerto the question of whetherthe unevenness or (thesecond meaningof "concentration", is of distribution increasing decreasing noted above), as is evident fromthe fact that the series in Table i change in directions. contrary indexes are subject to these sources Although comparisonsof concentration of ambiguity,it does not follow that they are empiricallymeaningless. For example, subject to qualification upon more intensive study, the changes The decreasing in exhibited Table i mightbe given the followinginterpretation. the indexesbased on geographicdivisionsreflect spread of populaconcentration of area of the United States,a continuation the historical tion over the continental fromeast to west and the openingup of regionsto intensive of pattern settlement occupancy. The increasing indexes based on counties indicate urban and metropolitanconcentrationof population, and the arrestingof the increase of local deconcentration population within metropolitan afterI940 may reflect slightchanges shown by the indexes for sub-regions areas. The comparatively and a may represent balancingof the two tendenciesof regionaldeconcentration since many sub-regionscontain both metropolitan concentration, metropolitan the areas. If,in fact, changesin population and extensive non-metropolitan centres have followed the complex patternjust over the last half-century distribution it suggested, would, of course,be impossibleforanysingleseriesof indexnumbers to describe those changes adequately. The contraryresults obtained with a indexes may then reflect basic ambiguityinherentin any concept alternative of the thatdoes not specify system areal unitsto which it refers, of concentration of thana defectin the operationaldefinition the measureof concentration. rather
MEASURES OF SPACING
that is closely related to density An approach to the analysisof distribution and concentrationmeasures is the measurementof the spacing of population units. Theoretical contributionsto this problem are due primarilyto plant of ecologistsworkingon the spacing of members speciesin the plantcommunity' and economistswho have developed hypothetical systems and to the geographers of economicareas on theassumptionof even populationdistribution.2However, of the most directapplicationof the idea of spacingto the representation populawas developed by Barnes and Robinson3 in theirtechnique of tion distribution the " linear distance map ". This type of map is recommendedby the authors for displaying variations in density for relativelydispersed populations. It = of involves a transformation densityby the formula: Average distance *i i / of VDensity. In their illustration the technique,densityis taken as number to of farmhouses square mile,and the average distanceof a farmhouse its six per
1 Philip J. Clark and Francis C. Evans, " Distance to Nearest Neighbor as a Measure of Spatial xxxv (October, 1954), pp. 445-453. in Relationships Populations", Ecology, 2 August Losch, TheEconomics Location Press, I954). (New Haven: Yale University of 3 " A New MethodfortheRepresentation Dispersed RuralPopulation", op. cit. of
THE
MEASUREMENT
OF
POPULATION
DISTRIBUTION
33
nearest neighbours is deduced on the assumption that farmhousesare evenly distributed. A correction of the formulais given by Mather,' replacing the constant I II by I .07.2 In both these papers it is indicated that values of average distancecomputed by formulaagree ratherwell with observed average even in areas where the distribution farm of distances separating farmhouses, houses appears somewhatirregular. on the primarily impressionistic Evidently, lineardistancemap commendsitself of it depictedis a simplefunction density can contrigrounds. Since the quantity in of inherent the conceptof density. bute nothingto the clarification ambiguities " But it is perhaps true that a linear measure is more " understandable than a to ratioof populationto area, which is somewhatdifficult visualize. between Whereasthe techniqueof lineardistancemaps employstherelationship presentation, spacing and densitymainlyas a device for improvingcartographic of otherstudieshave been concernedwith measuresand patterns spacingin their of of own right. Clarkand Evans state: " The pattern distribution a population of of plants or of animals is a fundamentalcharacteristic that population, but to difficult describe in precise and meaningful it is a featurethat is extremely effort devisingappropriate to terms".3 Plantecologistshave devoted considerable tests of the randomness of a distributionin space. Recent studies continue but endeavour as well to specifydegree and patternof departures this interest, fromrandomness. Some of the resultsof this work will be indicatedbriefly. For a given universe of territory containingn units (individuals or defined groups of individuals)of population, let ri be the lineardistance of the ith unit of to its nearestneighbouringunit,irrespective direction. The r, are measured for the entirepopulation of units, or a random sample thereof. Let p be the densityof population units,area being measuredon the same scale as is used in it the determining r5. If r is the mean of the r5, can be shown that-rE=0-5 P of large random distribution density p. is the expected mean in an infinitely The observed mean, 1A) varies below this,to a theoreticallower limit of zero, become more " clumped " or " aggregated". In the limiting as distributions case each unit is contiguous to at least one otherunit. The observed mean may vary above
rE
to a limit of
uniform .0746/Vp, which occurs for a perfectly
in distribution which each unit is equidistantfrom 6 other units. The ratio, R, of actual to expected distance,or rAIrE, may thus vary fromzero to 2 *149I, test with a value of unityoccurringfor a random distribution. A significance of forthe departure R fromunityis available.4 In studies of human populationsindividualswould, of course, be found in a of pattern aggregationto at least the household level. With the methodreferred to, it would be possible to make an exact test of the assumptionof Barnes and Robinson and Mather that farm households in the United States tend to be
3 " Distance to Nearest Neighbor as a Measure of Spatial Relationships in Populations", op. it. P. 445.
1 " A Linear-Distance Map of FarmPopulationin the UnitedStates", op. cit. 2 A more precise value of this constantis I0746.
4 Ibid.
34
OTIS
DUDLEY
DUNCAN
No wouldbe considerable in spaceduniformrly. doubtthere regional variation has a withmeasures spacing of thevalue of R. The writer experimentedlittle of townsand citiesin two regionswheretherewas reasonto believethata wouldevidence to itself.Valuesof R approximating tendency uniform spacing different unity, from werefoundforsettlements Iowa in I.4, and significantly as of two-thirds Indiana,taking unitsthe urbanized and in the northern areas areas. No doubtin other urbanized and urban placesoutside regions tendency a of rather thanuniformity spacing, of towardaggregation communities, would be discovered thistechnique. Whilethisexercise by produced onlyverygross on measurements a smallscalemap,it suggests results, owingto theuse ofrough is a toward the of that tendency uniformitynotnecessarilyfunction thecloseness for withi8 milesin of spacing, rA was aroundI 3 milesin Indianaas compared Iowa. in of One can imagine described patterns distribution whichthetechnique i$ " it to " in value. Foronething, fails distinguishclumped patterns which oflittle from in of each unithas but one close neighbour patterns whichthe number and someelaborations is and units clump larger variable. Clark Evansindicate per not for that ofthetechnique wouldbe helpful some,butprobably all,suchcases. as of thatthemeasurement spacing, appliedto human It seemslikely population for tendencies toward will units, provemoreuseful detecting uniformity (which, theoretical are than for disas Losch indicates, of considerable importance) of criminating amongpatterns aggregation.
CENTROGRAPHIC MEASURES
one Owingto itspopularization theU.S. BureauoftheCensus, ofthebest by " or of " of known centres population is themeanpoint, centroid, centre gravity, if i.e., " thepointupon whichthe UnitedStateswould balance, it werea rigid were distributed thereon witheach plane without weightand the population and to exert influence a an individual on beingassumedto have equal weight thatpoint"-1 A measureof from to central pointproportional his distance has ;3 and criticized aroundthe centreof gravity been proposed2 dispersion if to studies. butitseems havebeenusedvery little, at all,in empirical in of have Two othermeasures central tendency arealdistributions received attention.The " medianpoint" is defined the intersection as considerable of two orthogonal lines,each of whichsplitsan area into two partswithequal is of The of numbers inhabitants. location this indeterminate point to somedegree of as withtherotation theaxesusedin itscalculation.Lines inasmuch it varies to to into parallel theseaxes maybe calculated dividethe distribution fourths are in each direction. Their pointsof intersection knownas " quartilides ". of and the of that describes degree pattern dispersion thepopulation setofpoints
1 U.S. Bureau of the Census, I950 Census Population, Vol. ii, Part I (Washington: Government of Printing Office, I953), p.9. 2 D. WeltyLefever," MeasuringGeographic Concentration by Means of the StandardDeviational xxxii (July,I926), 88-94. Journal Sociology, Ellipse ", American of 3 P. H. Furfey, Note on Lefever's ' StandardDeviational Ellipse' ", American " of Journal Sociology, xxxIII (July,I927), pp. 94-98.
The principle can be extended to " decilides", " centilides", etc., yielding a
THE
MEASUREMENT
OF
POPULATION
DISTRIBUTION
35
" of centre 1 A concept that the" median is different overtheterritory. somewhat aggregate travel. This is the point pointof minimum or moredescriptively, is from whichthe sumof the linearradialdeviations a minimum.It has been locationin his " hypothesis median of discussedby Quinn as an optimum of aggregate the are for location Methods available calculating point minimum .f"2 travel thecentre alongrestricted rather thanby shortest to is routes travel where airline distance.3 by of is It shouldbe notedthatthe locationof the centre gravity affected a whereastherecan be changein the positionof any unit of the population, the movement withoutaffecting medianpoint. Hart suggests considerable most useful renderthe centreof gravity that theirmathematical properties the of over forstudying areal shifts a distribution time,the medianpointfor for distributions the same time,and the mediancentre at different comparing locational for services. optima centralized investigating
POPULATION POTENTIAL
the of is on that As was indicated, centre gravity computed the assumption to his distance thecentral exerts influence an from eachindividual proportional in " point. The term" influence here,of course,is construed the sense of with social influence But investigators concerned as mathematical weight. is an have pointedout thatinfluence moreprobably inverse affected space by Stewart's comment thecentreon of function distance. Professor thana direct thispointforcefully: concept4 expresses of-gravity
of is centre gravity principally The factis thatthispopulation bureaucratic hocus-pocus. " on of to a It is nonsense compute centre a basissuchthat theleverage a dozenpersons at could conceivablycounter-balance in California, the long end of the teeter-totter, has fictitious fulcrum anyspecific teeter-totter's that sociological whatsoever theimpossible a meaning. Thereis on thecontrary wealthof evidencethatpeople exertmoreinfluence milesaway exert about people a hundred close at handthanfaraway,and thata thousand milesaway. as thesameinfluence fivethousand people fivehundred
IOO
". persons in New York, the short end of the teeter-totter
. There is no evidence
at " to If the" influence of each individual a pointis considered be inversely of from thenthe totalpotential population at to it, proportional his distance of of in a point, is thesumof thereciprocals thedistances all individuals the L0, of the is the from point. In practice, course, computation madeby population smallareaareequidistant within suitably a from that assuming all theindividuals Lo, whence, at = Potential Lo=
1
iD1'
D,'
xxx Geography, (January, ", JohnFraserHart, " CentralTendencyin Areal Distributions Economic Method and pp. 48-59; E. E. Sviatlovskyand Walter Crosby Eells, " The Centrographical I937), pp. 240-254. Review, Analysis Geographical xxvii (April, ", Regional viii 2 JamesA. Quinn, " The Hypothesis of Median Location ", American Review, (April, Sociological
I954), 1943), pp. I48-I56.
3
xIII Travel Metron, (I937), pp. 78-81. ", Minimum Aggregate

4
of Routes upon the Centerof D. E. Scates and L. M. van Nortwich," The Influence Restrictive October 7, Ig to Taken froma letter the New YorkTimes, 9.
36
OTIS
DUDLEY
DUNCAN
where Pi are thepopulations then areasintowhicha territory divided is the of and theDi are therespective distances theseareasfrom (usually of Lo measured from geographic the centre fromthe approximate or centre gravity the of of in for of population, each area). After computing potential a number points such as Lo, it is possibleto obtainvalues for otherpointsby interpolation; or one may construct isometric maps showinglineswhichare loci of points having equal potentials.' It maybe observed thenotion potential a point inprinciple that of at is perfectly precise, whereas conceptof density a pointis meaningless thatof the at and in of of density thevicinity a point ambiguous. Whilethedensity anyportion is of a territory dependsonly on the number inhabiting thatportion, potential at anypointdepends thedistribution population on of overtheentire territory. The two concepts related, are in however, that configurationequipotential the of one linesis determined thepattern variation density by of of overa territory; could say the same,of course, withrespect theinterrelations thevarious to of other measures distribution havebeendescribed. of that
<~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
_ ~~~~~~~~~~~~~~~3
FIGURE 2.
Isolines of Population Potential,United States:
I950.
Figure a is a map showingisolinesof population potential the United for States. The principal determinant the configuration the equipotential of of lines is the massiveconcentration metropolitan of population the Middle in Atlantic States, thecontours elongated theWestbythehighly but are to urbanized
1 For details see John Q. Stewart," Empirical MathematicalRules Concerningthe Distribution in xxxvii (July,I947), pp. 46I-485; reprinted Review, and Equilibriumof Population", Geographical editedby Spenglerand Duncan, op. cit.,pp. 344-37I, Demographic Analysis,
THE
MEASUREMENT
OF
POPULATION
DISTRIBUTION
37
zone stretching along the GreatLakesto Chicago. New York City, wherethe " peak potential occurs,maybe regarded the " centre of the country as from the standpoint demographic of influence, granted the assumptions which on thismapis constructed. Figurez maybe compared withFigure3, whichshowsisolines aggregate of traveldistance the same underlying for population distribution. This index
250DSATS
>
\-
s5
FIGURE 3.
Isolines of Aggregate Travel Distance, United States:
I950.
a reaches mimimum, the "mtedian i.e., centre occurs, a pointlyingwithin " at the lowestcontour shown. At thatpointaggregate traveldistance around is 92 billionmiles; thisis the totaltravelrequired move everyindividual in to thecountry that to point shortest by airline distance.The travel person per would be somewhat over600 miles. To thewriter's knowledge has notbeen observed it previously potential that and aggregate travel distance closelyrelated are concepts. The value of either at a point, can be expressed the formula, P,D7, wherem= -I forthe by Z Lo, potential computation m=I forcalculating and distance. aggregate linear travel The two shouldbe regarded alternative notmutually as but concepts; exclusive anditmay wellbe that valuesofmother than?: I will be found suitable certain for problems. 1
1 Theodore R. Anderson, "Potential Models and the Spatial Distributionof Population", Papers It is a curiosity Science the and Proceedings, Regional II thatpopulation Association, (I956), pp. I75-I8z. near the point of potentialis emphasizedby an investigator (Stewart)workingat Princeton University of whilethe" hypothesis medianlocation" was setforth an investigator maximum potential, by (Quinn) to in at the University Cincinnati, close proximity the point of minimum of aggregatetravel. It is left to speculateon thepossible"demographic of determination" demographic forstudents Wissenssoaiologie of concepts.
38
OTIS
DUDLEY
DUNCAN
CLASSIFICATION RESIDENTIAL consideredthus far may be considered" purely The measuresof distribution to spatial" measures, i.e., theycan be computedwithoutreference any qualitative of on or classificatory information theareal sub-divisions the universeof territory. are In the other main class of devices for describingdistribution found those of that depend on a prior classification areas as to size, location, rural-urban in or becomes character, the like. The measure of spatial distribution, effect, the frequency distribution the population with of a summary statistic describing scheme. respectto such a classification is used in describingdistribution the division classification The most familiar of the population into rural and urban residents. It would require a sizeable of the paper even to summarizesketchily variationsin definition thesecategories used criteriaof urban among various censuses.' Probably the most frequently of classification communities political or statusare size and legal or administrative sub-divisions. Thus the major componentof the urban population in the U.S. places exceedinga Census has always consistedof incorporated(legal criterion) (size criterion). To this core have been added specifiednumberof inhabitants certain other populations specially designated as urban; for example, in the of places of z,500 or more (boundaries I950 Census, inhabitants unincorporated of of which were delimitedaccording to geographicalcriteria)and inhabitants as portionsof urbanizedareas not classifiable urban exceptin termsof proximity of a administrato largerpopulationcentres. The importance modifying primarily tiveconceptof urbanpopulationis indicatedbythesize ofthepopulationincluded criteria: in I950, when 64 -o?/% the U.S. population of underthe supplementary of identified unincorporated places of this while 6- 60% were residents separately size or partsof urbanizedareas otherthan incorporated places of z, 500 or more. the However defined, proportionof the population residingin urban territory is undoubtedlythe most widelyused measureof population distribution, owing in to itsreadyavailability manycensuses. The ubiquityof themeasurein regional, national,and temporalcomparisonsperhaps accounts for the tendencyof many to demographers treatit as an aspect of population " composition" ratherthan ". distribution The categories," urban" and " rural", are so broad that sub-categoriesof each need to be introducedinto any refinedanalysis of distribution. In the " United Statesthe breakdownof ruralpopulationinto " farm" and " non-farm used for the last three decades; but it has become residentshas been widely " clear that the rapid growthin the " rural-non-farm sector of the increasingly as " rural expansion,but moreprobablyrepresents " populationis not interpretable "suburbanization " of the urban population. Changes in the meaning of are definedcategoriesof populationdistribution thusto be expected qualitatively of distribution when the spatialpattern undergoeschange.
1 A detailedaccount of the complicated of classifications the United States is in history rural-urban Classification the United States, in given by Leon E. Truesdell," The Developmentof the Urban-Rural SeriesP-z3, No. I (Washington: U.S. Bureau of the Census, I874 tO I949 ", Current Population Reports, 1949).
was urban,53
44%
in resided incorporated or placesof 2, 500 inhabitants more,
THE
MEASUREMENT
OF
POPULATION
DISTRIBUTION
39
Cutting across, though not independent of, the rural-urbanclassification, status. Although,again, operational of is the classification areas by metropolitan of vary, they generallyinclude rules for assigning definitions the classification centrecertainterritory contiguous to it and thoughtto share to a metropolitan is centres inevitacharacter. The designationof metropolitan in its metropolitan statususually being of the matter, criteria metropolitan bly a somewhat arbitrary stated in terms of size and spatial separation from centres of larger size. Conceivably, functionalcriteriafor metropolitanstatus could be developed,' but their routine use in statisticalcompilations would present considerable difficulty.The following figures(estimatedpopulation in millions for I956)2 illustrate the relation between urban-ruraland metropolitanclassifications: Total ...plt ... Metropolitan ... Urban ... ... Rural ... ... ... ... ... ... ... ...
...
Non-Metropolitan ... ... ... Urban..... ... Rural ...
... ... ...
I64 3 96*z *4 79 17
68*I
43
24. S 43 .6
Such a tabulationbrings out a facet of population distribution patternsthat is classification: obscured by a simple rural-urbanor even a community-size of population around major centresthat is characteristic the clustering tributary of highlyurbanizedcountries.
COMMUNITY SIZE
common practiceis to sub-divideurban population (and rural, Anotherfairly forthat matter)by size of community. Demographic differences community by as size withinthe broad urban categorymay oftenbe as important gross ruralof urban differences.3As in the definition urban places, the community-size of will be, in considerablepart,a function proceduresfor compiling distribution the data. A paramount problem is whether to treat each politically distinct or urban place as a separateunit in the distribution, to regarda clusterof such of places as a single unit, as in the United States Census definition " urbanized in in area " The distributions Table z show what a wide difference commumtyis size distribution produced by changingfromone to the other of these bases. distribution A community-size may be summarizedwith a varietyof statistics. that has been proposed for income distributions One such statistic mightwell be adapted to this purpose. The " equatorial communitysize " is calculated that divides the population in half. for by interpolating the size of community Thus, the data in Table z show that one-halfof the U.S. population in I950 8,8oo inhabitantsor more, on an urban place lived in places of approximately
1 Rupert B. Vance and Sara Smith, " MetropolitanDominance and Integration Chapter 6 in ", The Urban South, editedby RupertB. Vance and Nicholas J.Demerath(Chapel Hill: University North of CarolinaPress, '954). 2 " Civilian Population of the United States,by Type of Residence,March 1956 and April 1950 ", Current Population Reports, SeriesP-20, No. 7I (Washington: U.S. Bureau of theCensus,1956). 3 Otis Dudley Duncan and AlbertJ. Reiss, Jr.,Social Characteristics Urban and Rural Communities, of 19f o (New York: John Wiley & Sons, 1956).
40
OTIS
DUDLEY
DUNCAN
on Compiled by Table z. Population Distribution Site of Community, States: 197o. Two Bases, the United for
community United States,total Urban. x,ooo,oooor more
500,ooo to I,000,000 250,000 tO 500,000
I00,000
Size of
Urban place basis Number of places ... i8,548

4,74I
% of total population
IOO-O
Urbanizedarea basis Number of places I7,2I7

3,410
12
% of total population
IOO 0
..
...
.3
...
... ...
...
50,ooo toIOO,000 25,000 to 50,000 IO,000 to 25,000 5,ooo to IO,OOO 2,500 to s,0oo ... Under 2,500t ... Other urbant ...
to5o,ooo
... ...
...
... ...
...
... ... ...
...
... ... ...
778 I,I76
I,846 457
I26 252
65
23
5 I3
0 614II.5
6i
6*3 5 9 7 9 54
49
0*4
5 -5 5.8
24 70
I3
640 0 251I
4.3
i,605
-
547 908
172
59
5.8 7* 2 3 9
5.8
2X9 5 5
4.I 3 7
... ... ... ... ... ... ... Under i,ooo (incorp.) ... 290... ... Otherrural? Rural
...3 I,OOO tO 2,500
9,649
,807 4,158
36- 0
2.7 29 o29
4 3
13,807 4,158
9,649
36 0 4 3 2.7
o
* Includes 38 urbanizedareasand 2i urbanplaces. places insideurbanizedareas. t Incorporated t Unincorporated partsof urbanizedareas. ? Includesunincorporated places of under i,ooo, as well as otherruralpopulationnot in nucleated settlements. Tables sa and 5b (Washington: Vol. i, Number Inhabitants, of Source: 1950 Census Population, of Government I952). Office, Printing
area basis.' 28,300 or moreon theurbanized basis,or in citiesof approximately betweenthese two resultsemphasizes of again the relativity The difference are to of by measures distribution the system whichthe statistics compiled. withthe mediansize of size The equatorial community is not to be confused of for couldalso be computed a distribution townsand The latter community. in like a cities size,butnotfor distribution that Table a inwhicha considerable by is classified size of community of and by proportion thepopulation not actually as centres. The equatorial does size couldnot be described livingin nucleated of that except, course, not dependon the lowerlimitof the size classification, to mustbe low enoughforhalfthepopulation fallabove it. On the thislimit withthe lowerlimitof the size otherhand,the mediansize will varydirectly as close to it, inasmuch small of distribution citiesand towns,usuallyfalling outnumber largeones. greatly communities that withthe observation the size have Severalinvestigators been intrigued a from of distribution citiesand townsexhibits kindof regularity placeto place is in and overtime. For anylargeterritory whichurbanization at all advanced foundthatsmallplaces outnumber largeones, and the it is almostuniformly skewed. Auerbach2 credited is is withthe size distribution in general highly
on Interpolating the logarithmof communitysize. LIV Mitteilungen, ", Felix Auerbach, " Das Gesetz der Bevolkerungskonzentration Petermanns (19I3), pp- 74-76,
2
THE
MEASUREMENT
OF
POPULATION
DISTRIBUTION
41
of distribution formulm Other earlytreatments city-size were apparently producedwithoutknowledgeof Auerbach'sand Lotka's work. Goodrich2 of abovea givensizelimit the that number cities tends be equal to twice noted to of that abovea limit twice of thenumber cities is which equivalent size,a formula with rule a to the " rank-size ". Saibantes experimented fitting Paretocurve an distribution presented analysis concentration and of to the city-size indexes or analogousto Gini's indexesof incomeinequality concentration. Gibrat4 the distribution with a log-normal suggestedfitting city-size equation,and related an thereto. proposed indexofconcentration " Both the " rank-size and the Pareto-curve inapproachhave been further in drewattention theparallel to between vestigated thelasttwodecades. Singer6 and distribution Pareto'slaw of incomedistribution presented thecity-size and of evidence of goodness-of-fit the Pareto curveto a numberof city-size and Allen6 reviewed considerably distributions. extended work. recently Singer's rule whoevidently The " rank-size " wasmuch publicized Zipf7 by beganhiswork studies. Mathematical well and reference Auerbach's Lotka'sprior to without as of to contributions theliterature the " rank-size " weremadeby as empirical rule dealtwith the generalization and Stewart8 Hammer.9These investigators of
Alfred Lotka, Elements Physical J. of Biology (Baltimore: Williams& WilkinsCo., I925), pp. 306-307. betweenPopulationand the City Plan ", in The Ernest P. Goodrich. " The Statistical Relationship editedby ErnestW. Burgess(Chicago: University Chicago Press,I926). Urban of Communiy, vii della Popolazione ", Metron, (I928), pp. 53-993Mario Saibante " La Concentrazione 4 R. Gibrat,Les Inegalitis Economiques (Paris: Librairiedu Recueil Sirey,I93I). 5 H. W. Singer," The ' Courbe des Populations : A Parallel to Pareto's Law ", Economic Journal, XLVI (June,1936), pp. 254-263. 6 G. R. Allen, " The ' Courbe des Populations,': A FurtherAnalysis ", Bulktinof the Oxford xvi Institute Statistics, (May-June, pp. I79-I89. of I95)' Universit. and George KingsleyZipf, NationalUnity Disunity (Bloomington,Ind.: The PrincipiaPress,Inc., and HumanBehavior thePrinaiple Least Effort of (Cambridge,Mass.: Addison-WesleyPress, I941);
1
2
formula".
the of to of first presentation a formula summarize distribution citiesby size. in wereranked orderof number thatwhenthecitiesof Germany He observed of rankand itssize tended be moreor less to of inhabitants product a city's the valueofthis he that constant. Moreover, suggested theaverage product, expressed of an as a proportion the national affords indexof population population, concentration suitablefor comparing degreesof urbanization among countries. of rule the If the " rank-size " were to hold precisely, product rankand size of wouldbe equal to thesize of thelargest city. An average theproducts may, " as " be therefore, regarded an " adjusted or " estimated sizeof thelargest city. the coefficient comesdownto taking " adjusted" Hence,Auerbach's suggested of sizeofthelargest as a proportion thenational city population. of that In a briefdiscussion Auerbach's work,Lotka1indicated forthe IOO " citiesin the UnitedStatesthe " law of urbanconcentrationwas more largest precisely (rank) 0.93 x size=constant. Lotka stated, however: " It may be is how muchsignificance to be attached thisempirical left open question an to
Inc.,1949).
Rules Concerningthe Distributionand Equilibriumof Population 8 " Empirical Mathematical op. cit. 9 Carl Hammer, " Rank Correlation Cities and Refinement mimeographedreport,Bureau of of ", Applied Social Research,Columbia University, I. 195
42
OTIS
DUDLEY
DUNCAN
otherthan therulein whichtherankof a cityis allowedto have an exponent population and the formula estimating number aggregate for and unity, presented curve. of size limitsfromthe parameters the fitted of citieswithinspecified of persistence is recent contribution Madden'sdiscussion thetemporal Another patterns in of as " of the " rank-size relationship evidence stability the growth listedhere, incompletely investigations of cities.' Aside fromthe empirical rule of on havecommented thesignificance the" rank-size " of a number writers and distribution spacetheory population of of from standpoint thegeneral the merits of little that in It is curious one finds theliterature discussion therelative the to approaches summarizing city-size " of the " rank-size and Pareto-curve may be the between two formula: relationship The mathematical distribution. as indicated follows: and of of Let x= size of cityor town(number inhabitants) y=y(x) = number curve logyis for (rank).3 Thentheformula thePareto placesofsizex or larger deterlog A - a log x, ory= Ax-, oryxa-=A, whereA and a are constants is " " minedfromthedata. The generalized rank-size formula log x= log B determined from B - i log y,or x =By-, or xyP B, where and ,8are constants the data. These equations implythat ,B=i/ctor c=ilfl, and A=BI1P or the if by B= Al/at. However, A and othavebeendetermined minimizing sum inin of thesquaresof theresiduals logy whileB and ,Bhave beendetermined in the by dependently minimizing sum of the squaresof the residuals log x, holdprecisely. wouldnot,in general, relationships theforegoing has curves beencarried " " The bulkoftheworkon fitting rank-size or Pareto of have been out somewhatinformally.In the majority cases, constants of without any statedcriterion fit,or the data have graphically determined to on approximation a straight paperand their been plotted double-log simply little line has been noted by visual inspection. One findsin the literature curvesto cityfor fitting and techniques criteria of discussion mathematical to attention goodnessof fit. little size distributions, correlatively, formal and, " " of calculations " errors and " averageerrors in Singerand Allen present of citiesin varioussize groupsfromthe equationof the computing number are basedon suchcalculations invalidated curve. However, comparisons thefitted and case to case in the number size of intervals. from to a degree, variation by such test has knowledge, a formal of goodnessof fit, In no case,to thewriter's of thatin the majority caseswherethefit as x2, beenpresented. One suspects test mathematical would detect has been judged acceptably good, a stringent model. One consequence the from mathematical departures statistically significant on unduly the fitof of the " rank-size"approachhas been to focusattention verbal in numerous Thus, at theformula thevery upperendofthedistribution. that the "rank-size the ". writerhas encountered statement the discussions
economy.2
1 Carl H. Madden, " On Some Indications in of Stability theGrowthof Citiesin the UnitedStates", iv Change, (April, 1956), pp. 236-252. and Development Cultural Economic of op. 2 For example, Losch, TheEconomics Location, cit.; RutledgeVining," A Description Certain of in Change, (January, and 1955), Development Cultural ", SpatialAspectsof an Economic System Economic (New York: JohnWiley & Sons, Inc., i956). andSpace-Economy pp. 147-195; WalterIsard, Location 3 Singer and Allen let y (x) = number of places larger than x.
THE
MEASUREMENT
OF
POPULATION
DISTRIBUTION
43
thatits largest on ruledoes not hold forsuchand sucha country the grounds have is cityis muchtoo largeor too small. The netimpression thatsceptics of fitting city-size the possibility a simpleformula adequately oftenrejected of on distributions thebasisofan examination onlya smallpartof theevidence of formula of mayhave been guilty failing whereas proponents one or another evidence. as to attention negative to positive to giveas much of liberal criterion whatconstitutes wellbe inclined adopta fairly to One might numerous independent to fit an acceptable ofa curve thedataifon sucha criterion and if instances poor fit of fit of instances acceptable were to be discovered than on ad hocgroundsdiffering rather for systematically could be explained involvesat leastthree of each instance. The generality anyproposedformula or of considerations:the numberor proportion countries regionswherea timeperiods the in ofa good fit various and the is good fit obtained, occurrence of and over of of stability theform thesizedistribution time, therange community all virtually investigators a is sizesforwhich good fit obtained. On thelastpoint, thelogs) can be expected hold onlyfor to formula (in thatthelinear recognize however, mayvaryspatially minimum.This minimum, sizes above a certain the fall it maysometimes within rangeof available moreover, and temporally; or identifiable, it maybe lowerthanthe size data and thusbe approximately statements unjustified and aretabulated.Bothincautious forwhichfrequencies are Allen' thismatter to be foundin the literature;however, on speculation size for a numberof of estimates the minimum carefulempirical presents of formulas over space countries.On the generality " rank-size"and related thatthe fundof availabledata constitutes a it mustbe recognized and time, the comparability and periods. Moreover, of countries quite biased sample of in methods recording data by oftheavailable is greatly impaired variation census collection instances presented of size. The largest is community and tabulating withdatareadily in available secondary worked mainly by Allenwho,however, the where Pareto " rank-size or that " sources. It is evident casesareto be found thoughnot all, of whichcan be " explained fits formula quite poorly,many, one in to away" by reference deficiencies thedata. Intuitively, feelsthatthere certain abnormal periods " " " kindsof " countries and, perhaps, are certain to in whichsuch a formula would not be expected fitwell. However,until is it for rationalization the formula available, is difficult an adequatetheoretical thesehunches. to organize of of A careful significance theParetoor the " rankappraisal thetheoretical two extremes: between size" rulewould probably midway assignit a position on the one hand,a merely curiosity, and, on the other,a " law " empirical theoretical scheme verified and an under from accepted deduced fully rigorously " conditions. In otherwords,such a rule has " plausibility as well specified " between of this support. A " plausible connection as a modicum empirical of of and organization economic typeof regularity his doctrine thehierarchical Vining was set forth Ldsch2and restated Hoover.3 Moreover, by regions by
2 Edgar M. Hoover, " The Conceptof a System Cities: A Comment of on RutledgeVining'sPaper ", iII I955), pp. I96-I98. Change, (January, and Development Cultural Economic
1 "' The ' Courbe des Populations', A Further Analysis", op. cit. 2 TheEconomics Location, op. cit.,pp. 431-438. of
44
OTIS
DUDLEY
DUNCAN
model based on the of the possibility developinga relevant has suggested " that some processof development leads in the limitto stable assumption none of ".1 of distributions the sortthatwe have described As yet,however, of pursuedfarenoughto yielda theory community-size theseideas has been are to and that what conditions both necessary sufficient distribution predicts " there riseto theParetoor " rank-size form. Underthecircumstances, is give to efforts discover empirical the a limits indicated needforconcurrent definmitely in to termsthe of these rulesand attempts formulate theoretical of validity of distribution to aspects population distribution other of relationship thecity-size the and ecological organization.Pendingthese achievements, Pareto curve means of summarizing community-size may be accepted as a convenient circumstances. distributions undercertain
OTHER CATEGORICAL MEASURES
can of of An indefinitely number measures distribution be generated large basis or on areal byclassifying units somequalitative quantitative and tabulating in of falling each class. For example, and proportion inhabitants the number the of McKenzie,tabulated population counties Hawleyand Bogue, following fifty approximately milesof the seaboardand GreatLakes in the lyingwithin 2 within thiszone was concentration toward increasing UnitedStates. A trend of it populationby classifying Hagood studiedthe distribution rural-farm of cityin each county.3 A number investigators to according size of largest to distribution withpopulation from according distance have been concerned majorurbancentres. of not to it Ordinarily is desirable examine onlytheproportions thepopulation but also the of scheme, in falling the severalcategories such a classification of land area occupiedby the population each category. It is then aggregate their and densities to study to pattern.Severalinvestigators, possible compute from to centres.4 have studied according distance forexample, density gradients and ratio the to It is alsopossible generalize Lorenzcurve concentration technique rather thanby,density foruse in thisconnection.Unitareasmaybe ordered, that curve described), anyvariable is ofinterest.The resulting by (as previously to is no longerconstrained be concaveupwardtowardthe diagonal,but may followan irregular path,even going above thediagonalin whole or in part. The concentration ratio, computedby the formulaalreadystated,is then to with of by as interpreted a measure concentration respect thevariable which valuesif halfor moreof the area areasare ordered. It will take on negative
", " A Descriptionof CertainSpatialAspectsof an Economic System op. cit.,p. I 85. Amos H. Hawley and Donald J. Bogue, " Recent Shiftsin Population: The Drift toward the XXiv 1942), 143-I48. Statistics, (AugUst, of ", Review Economic I930-40 District, Metropolitan 3 Margaret JarmanHagood, " Rural Population Characteristics Chapter Iz in Carl C. Taylor, ", A. States(New York: Alfred Knopf, I949). etal., RuralLifein theUnited 4 Colin Clark, " Urban Population Densities", Journal theRoyalStatistical cxIv (No. 4, Society, of America: Deconcentration Shape of Metropolitan pp. 490-496; Amos H. Hawley, The Changing I95I), of sinceI920 (Glencoe, Ill.: The Free Press, I956); Donald J. Bogue, The Structure the Metropolitan of Community Arbor: University Michigan,I949). (Ann
1
2
observed from I900 to 1930, with a decrease occurringbetween I930 and I940.
THE
MEASUREMENT
OF
POPULATION
DISTRIBUTION
45
enclosed by the curve and the diagonal is above the diagonal. If, for example, the ratio may be interareas are orderedby distancefromcentres, concentration ".1 pretedas an " index of centralization resultfromcombinations distributions More elaborateproceduresforstudying of two or more of the techniquesdescribed. Bogue's unusuallycomprehensive investigationinvolved the computation of densities (or " habitationratios") for urban and rural population separatelyaccording to distance to nearest to size of centre, geographicregion,and proximity highways metropolitan centre, and to non-metropolitan urban centres.2 In this study the focus of attention by was on spatial gradientsin density,as affected the other attributesunder also. consideration; but it mightwell have included measuresof concentration The limitsto the possibilitiesof this kind are practical,ratherthan conceptual, in nature. The ingenuityof futurestudents of population distributionwill of doubtlessproduce a variety new techniquesof analyzingdistributions. In conclusion, it should be recognized that the study of population and connections with other branches of distributionhas wide ramifications classes of the populationare compared, of different demography. If distributions it is possible to develop useful measures of segregationand areal association.3 has but also populationdynamics an important Not onlypopulationcomposition, of the studyof migration of course, One important phase is, distributive aspect. of the investigation its impact on patterns population distribution. The same of fertility and mortality, while, on the other hand, can be said of differential are importantclues for the study of differentials turned up in researchon the variationin levels of the vital processes fromplace to place. But the purpose aspects of thispaper is not so muchto emphasizethesewell known distributional of general demography,as to suggest the advisabilityof making population itselfthe subjectof concertedresearchefforts. distribution
Hauser,Duncan, and Duncan, Methods Urban of Analysis, cit. op. TheStructure he Metropolitan of op. Community, cit. 3 Otis Dudley Duncan and Beverly Duncan, " A MethodologicalAnalysisof Segregation Indexes", American xx " Residential Sociological Review, (April,I95 5), pp. 2IO-2I7; Distribution and Occupational LX Stratification American ", Journal Sociology, (March,I95 5), pp. 493-503 . of
2
A Methodological Analysis of Segregation Indexes Author(s): Otis Dudley Duncan and Beverly Duncan Reviewed work(s): Source: American Sociological Review, Vol. 20, No. 2 (Apr., 1955), pp. 210-217 Published by: American Sociological Association Stable URL: http://www.jstor.org/stable/2088328 . Accessed: 15/12/2012 00:08
.
American Sociological Association is collaborating with JSTOR to digitize, preserve and extend access to American Sociological Review.
210
AMERICAN SOCIOLOGICALREVIEW explained in terms of increased capital outlay to improve facilities in order to maintain organizational prestige. (10) With a decline in the membership of an organization, there is no immediate or actual decline in income. This phenomenon is due primarily to the greater efficiency of the organizationin collecting dues and carrying out financial drives. (11) Material property will increase over a period of time, and this increase is closely related to the expenditures for staff and upkeep. Unless the material property is withdrawn from use or permitted to deteriorate, these service expenditurescannot be reduced below a certain level.
frequency of face-to-face interaction, preceded by the increasing structural differentiation and the separation of various functional activities. Furthermore,the association is confronted with the problem of enforcing the pertinent features of its programthrough a relatively expensive outlay. Thus with an increase of controls there is a corresponding increase of staff and administrative expenditures. This phenomenon of increased controls, efforts to improve communication,and the use of additional professional help can be found emerging not only when the membership declines, but also when the membership increases at a rapid rate. (9) Increase of expenditures can also be
A METHODOLOGICAL ANALYSIS OF SEGREGATION INDEXES *

OTIS DUDLEY DUNCAN AND BEVERLY DUNCAN
University of Chicago
in THERE have been proposed the literature several alternative indexes of the degree of residential segregation of the nonwhite population of a city.' This paper shows that all of these can be regarded as functions of a single geometrical construct, the "segregation curve." From this there are developed several important implications:
* Revision of paper read at a meeting of the Midwest Sociological Society, April, 1954. The clerical assistance of Florence Sugeno, Richard W. Redick and Robert Glassburg is gratefully acknowledged, as is the financial assistance of the Social Science Research Committee of the University of Chicago. This research was supported in part by the U. S. Air Force under Contract Number AF 33 (038)-25630, monitored by the Human Resources Research Institute. Permission is granted for reproduction, translation, publication and disposal in whole and in part by or for the U. S. Government. 1Donald O. Cowgill and Mary S. Cowgill, "An Index of Segregation Based on Block Statistics," American Sociological Review, 16 (December, 1951), pp. 825-831; Julius A. Jahn, "The Measurement of Ecological Segregation: Derivation of an Index Based on the Criterion of Reproducibility," American Sociological Review, 15 (February, 1950), pp. 100-104; Julius A. Jahn, Calvin F. Schmid, and Clarence Schrag, "The Measurement of Ecological Segregation," American Sociological Review, 12 (June, 1947), pp. 293-303.
(1) The proposed indexes of segregation have a number of hitherto unnoticed interrelationships which can be mathematically demonstrated. (2) Some of them have mathematical propertiesof which their proponents were unaware, and which lead to difficulties of interpretation. (3) As a consequence, the status of the empirical work already done with segregation indexes is questionable, and their validity for further research is undetermined. This paper consists of a summary of the mathematical analysis made of segregation indexes and of a documentation of the conclusions listed. The problem of validating segregation indexes is viewed as one of some importance, not only in its own right, but also as an illustration of the difficulties in finding an adequate rationale for much sociological research using index numbers.
THE SEGREGATION CURVE
Consider the k census tracts of a city. The tract contains Ni nonwhites and Wi To.Summing whites, totalling to Ni + W
ith
k k k
over i,
W, and YTi 1 T. For each tract compute the nonwhite proi

-
N,
ANALYSIS OF SEGREGATION INDEXES portion, qj Nl/Ti, and array the tracts in ranks 1 to k in order of magnitude of qj. With this ordering compute tract by tract the cumulative proportions of nonwhites and whites, letting the cumulative proportion of nonwhites through the ith tract be Xi and the cumulative proportion of whites be (N1 + N2)/N, Y2 (W1 Yi, e.g., X2 + W2)7W. The segregation curve is the function Yi f(Xi), as graphed in Figure 1. The observed segregation curve, together with the nonwhite proportion in the entire city, q =N/T, contains all the information involved in the calculation of any of the segregation indexes suggested in the literature. As is suggested below, research on segregation is not likely to progress far with the research problem limited to the study of this information alone.
-
211
The "Nonwhite Section Index," 5 here denoted D, for dissimilarity or displacements is the maximum vertical distance between the diagonal and the curve in Figure 1, i.e., the maximumof the k differences (Xi -I). Alternatively, suppose there are s tracts for - Y,. If xi Xwhich qi q; then D = and yi are the uncumulated proportions of the city's nonwhites and whites, i.e., xi
k
DEFINITION
OF INDEXES
We here define, with reference to the segregation curve, the several indexes proposed in the literature, omitting the proofs of the equivalence of our definitions and those originally given. In all cases these proofs involve only elementary algebra and geometry. The "Gini Index," Gi,2 is the area between the segregation curve and the diagonal of Figure 1, expressed as a proportion of the total area under the diagonal. It can also be defined as the "mean cost rating" of the cost-utility curve with Y- cost and X utility;3 or as the weighted mean difference with repetition,4 of the tract nonwhite proportions, qi, divided by the mean difference, 2pq, of the binomial variable of color, for the total city population, scoring each white person unity and each nonwhite zero (where 1 - q). The simplest forq = N/T, p
k k
mula for computing Gi is :Xi 1Yi 1, 1 1 keeping the tracts in the order established for constructing the segregation curve.
2
Ni/N, yi -_ Wi/W, then D - Y2 1xt - y4 Furthermore,D is the weighted mean deviation from q of the tract proportions, qt, divided by the mean deviation, 2pq, for the total population. It may be interpreted as the proportion of nonwhites who would have to change their tract of residence to make q for all i (hence the term, displaceq= ment). Our interest in the Cowgills' index 7 is confined to the mathematical form of the index, without regard to the important but logically distinct issue of the appropriatesize of area units. The general form of the index is the ratio of the number of areas occupied exclusively by whites to the maximum number of areas which could be so occupied. To obtain a relationshipto the segregationcurve we have considered a slight further generalization: the ratio of the number of persons living in exclusively white areas to the total whites in the city. The generalized Cowgill Index (Co) is then the length of that segment of the curve, if any, which coincides with the vertical drawn from (1,0) to (1,1) (see Figure 2). It may be noted that the foregoing indexes can be described as measuring directly the degree of departure of the segregation curve from the diagonal, which is the norm of even distribution. Other such indexes of "unevenness"could doubtless be suggested.8 The remainingindexes proposed in the literature can be related to the segregation curve only by explicitly introducing the city non5 Jahn, Schmid, and Schrag, op. cit., Index #4; Josephine J. Williams, "Another Commentary on So-Called Segregation Indices," American Sociological Review, 13 (June, 1948), pp. 298-303. 6 Donald J. Bogue, The Structure of the Metropolitan Community, Ann Arbor: University of Michigan, 1949; Edgar M. Hoover, Jr., "Interstate Redistribution of Population, 1850-1940," Journal of Economic History, 1 (November, 1941), pp. 199-205. 7 Cowgill and Cowgill, op. cit.
Jahn, Schmid, and Schrag, op. cit., Index #3.
3 Otis Dudley Duncan, "Urbanization and Re-
tail Specialization," Social Forces, 30 (March, 1952), pp. 267-271; Otis Dudley Duncan, Lloyd E. Ohlin, Albert J. Reiss, Jr., and Howard R. Stanton, "Formal Devices for Making Selection Decisions," American Journal of Sociology, 58 (May, 1953), pp. 573-584. 4 Maurice G. Kendall, The Advanced Theory of Statistics, London: Griffin and Co., 3d ed., 1947, Vol. I, Ch. 2.
212
I
AMERICAN SOCIOLOGICALREVIEW
I
Figure I
Figure
2(
Co
0~~~~~~~~~
0
00 0L.
Cm
rp
owie
I Figure 3
a Figure 4
q/
~ ~ ~ q~~~~~~~~~~~~~~~~~~~~~~p ~
'
q.~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~0
0 0 X 0 O Rep
Figure- 5
Figure
a: .22 .06 .01
'0
0 FIGURE
1. "Section" index in relation to segregation curve (curve for Macon, Ga., 1940 D = FIGURE 2. Generalized Cowgill index (Syracuse, N. Y., Co = .42). FIGURE 3. "Ghetto" index (Louisville, Ky., Gh = .60). FIGURE 4. "Reproducibility" index (Birmingham, Ala., Rep. = .62). FIGURE 5. Williams' model of the segregation curve, with D = .56. bX), for selected values of a. FIGURE 6. Hyperbola model, Y = aX/(l
.47).
ANALYSIS OF SEGREGATION INDEXES white proportion, q. We assume throughout that q .5. The "Nonwhite Ghetto Index," Gk,9 is found graphically by plotting the line Y = q(l - X)/p. The index value is then (Xv - Y.), denoting by (Xg, Y.) the point where this line intersects the segregation curve (see Fig. 3). The "Reproducibility Index," Rep,'0 is formally identical with the index of efficiency used in prediction work." To obtain Rep graphically construct the line parallel to Y = qX/p which is "tangent" to the segregation curve, i.e., which intersects but one point of the curve or which coincides with that segment (if any) which has a slope q/p. Then the value of Rep is the X-intercept of the auxiliary line (see Figure 4). The correlation ratio of the binomial variable, color, on tract is by definition the square root of the variance between tract proportions divided by the total variance of the population; i.e.,
213
by Jahn et al.,'6 seems somewhat superfluous. It also seems undesirable to restrict the use of eta to the case of tracts of equal size, as the latter authors do. Unlike the other indexes, eta involves a squared term, and no simple geometric relationship of eta to the segregation curve has been found. That a relationship exists is, however, indicated below.
INTERRELATIONS OF INDEXES
eta
_
Tpq
P In the case of a binomial variable, eta is identical with the mean square contingency coefficient, phi,'2 and is equal within a very close approximationto the intraclass correlation.'3 It is, therefore,a well known statistic, appearing, e.g., in Robinson's formula for ecological correlation as a measure of "clustering by area." 14 The "revised index of isolation" recently suggested by Bell 1' is identical with the square of eta. This term, as well as the term "segregation score" used
8 Leo A. Goodman, "On Urbanization Indices," Social Forces, 31 (May, 1953), pp. 360-362. 9 Jahn, Schmid, and Schrag, op. cit., Index #1; Williams, op. cit., p. 301. 10 Jahn, op. cit. 11 Lloyd E. Ohlin and Otis Dudley Duncan, "The Efficiency of Prediction in Criminology,"
AmericanJournal of Sociology, 54 (March, 1949),

pp. 441-451. 12 Williams, op. cit. 13 Leslie Kish, "On the Differentiation of Ecological Units," Ph.D. dissertation, University of Michigan, 1952. 14W. S. Robinson, "Ecological Correlations and the Behavior of Individuals," American Sociological Review, 15 (June, 1950), pp. 351-357. 15Wendell Bell, "A Probability Model for the Measurement of Ecological Segregation," Social Forces, 32 (May, 1954), pp. 357-364.
Previous work has made it clear that the foregoing indexes are not independent. Jahn et al.17 found moderate to high empirical correlationsamong four of them. Hornseth 18 demonstrated that the same four could all be expressed in formulas of one general type. Williams found relationships among them for a segregation curve of a specified type.'9 In fact, there are definite relationships among the several indexes which hold irrespective of the form of the segregationcurve, and which can be derived formally without reference to data. For example, the minimum value of Gi is D, and the maximum is 2D -D2. Some of the other relationships which have been found are the following: (1) qD/(1 - pD)?Gh-?D (2) 1 - p(1 - D)/q?Rep--D, unless the term on the left is negative, in which case Rep - 0. (3) GhkzetaV\1Gh (4) 1 - 2p(1 - Gk)?Rep-?Gk/(p + qGh) (5) Co is the minimum value of Gi and D; qCo/ ( - pCo) is the minimum of Gh; and if Co>1 - q/p, then Rep??1 - p (1 - Co)/q. (Some values of eta and Gh reportedby Jahn et al. which are inconsistent with the third relationship are to be attributed to their use of a formula for eta not weighted for tract size, or to computational errors.) From the above it is clear that there is almost necessarily a high correlationbetween D and Gi, as well as between eta and Gh. On the other hand the correlation between
#2. Jahn, Schmid,and Schrag,op. cit. 18 RichardA. Hornseth,"A Note on 'The Measurementof EcologicalSegregation'by Julius Jahn, Calvin F. Schmid,and ClarenceSchrag,"American SociologicalReview, 12 (October, 1947), pp. 603604. 19 Williams,op. cit., p. 302.
17
16 Jahn, Schmid,and Schrag,op. cit., Index
214
AMERICAN SOCIOLOGICALREVIEW
TABLE 1. SELECTED MEASURES OF CLOSENESS OF FIT OF ACTUAL SEGREGATION INDEX VALUES TO VALUES CALCULATED FROM THE HYPERBOLA, FOR 60 TRACTED CITIES:
D and Gh need not be high if there is considerable variation in q. These mathematical relationships, therefore, appear to account satisfactorily for the empirical intercorrelations of the index values reported by Jahn et al.
A MODEL FOR THE SEGREGATION CURVE
1940
Mean
Root-
Segregation Index D Gi Gh Rep Eta Co

1
Arithmetic Error 0.0 -0.006 -0.003 0.028 -0.016 0.078
Mean Absolute Error 0.0 0.015 0.044 0.065 0.039 0.078
MeanSquare Error 0.0 0.019 0.066 0.102 0.054 0.124
Correnation
Index:1
Williams' model of the segregation curve, referred to above, is represented geometrically by the line parallel to the diagonal of the graph which includes the points (D,O) and (1, 1 -D) (see Figure 5). For such a curve Williams showed that Gi - 2D -D2 Gk D, and eta +-\D. It can also be shown that Co -D and Rep = D. The Williams model, although a useful analytical construct, does not serve well to describe empirical segregation curves. We have worked with an alternative model which assumes that the segregation curve has the form of a hyperbola, Y = aX/ (1 - bX), where a and b are both non1. Figure 6 shows negative, and a + b selected curves of this form. It can be seen that the smaller the value of a, the greater is the degree of segregation, as measured by Gi or D. For a segregation curve of the form described, it can be shown that the segregation indexes have the following formulas: 1 + 2a(b + logea)/b2. (1) Gi (1 - Va)/(1 + V/a) (2) D (3) Co O when (4) Rep = (Vap -\Vq)2/bq, 0, when q/p<a. q/p?a, and Rep a 1 4abpq -a2(5) G Gh =1 2bpq No exact formula for eta has been found, but in view of the relationship between eta and Gh stated in the preceding section, the following is suggested as a close approximation: (6) eta - (Gh + V\Gh)/2, where the value of Gh is taken from the formula just given. To fit the hyperbola to census tract data for a city, D was calculated from the data, and the parameters of the curve were com-
1.0 .980 .945 .908 .958 ...

corresponding
Correlation index =
(2 d2) /k *Var (I),
where the d's are differences between
actual and calculated values of a segregation index, and Var (I) is the variance of the actual values. The correlation indexes for this problem are slightly lower than the Pearsonian correlations between actual and observed values.
puted from the formulasb = 1 - a, and a - (1 - D)2/(1 + D)2, the latter being
the solution for a of the second formula in this section. A hyperbola was fitted to the census tract data for each of the 60 tracted cities of the
United States in 1940. Then, predicted values of each of the indexes were computed, entering these parameters and the corresponding observed q's in the foregoing formulas. The results summarized in Table 1 make it clear that most of the segregation indexes can be predicted rather closely, given D, q, and the assumption that the segregation curve conforms to the hyperbola model. The largest errors occur for Co, since the hyperbola model requires this index to be zero. Even so, the model does not sacrifice a great deal of information. There are 18 cities whose Co index is actually zero, and another 13 have values of Co under .05. An index of this type is evidently of little use in comparing cities on the basis of census tract data. Two important conclusions may be drawn from the experience with the hyperbola model. First, there appears to be a characteristic form for the segregation curves of most large American cities, despite the considerable variation among them in degree of segregation. Second, for this universe of cities, there is little information in any of the indexes beyond that contained in the index, D, and the city nonwhite proportion, q. Each of the other indexes can be obtained to a close approximation given D, q, and the assumption of the hyperbolic form of the segregation curve. This conclusion might require modification if area units other than
ANALYSIS OF SEGREGATION INDEXES tracts were employed; and it has not been checked for any date except 1940.
EMPIRICAL CONSEQUENCES OF AN INDEX OF THE CHOICE
215
A number of criteria have been offered for the choice of an index formula,20with no consensus on the matter having been reached. In our judgment it has been insufficiently emphasizedthat the empirical results obtained with an index may be strongly affected by its mathematical properties. For example, Jahn et al. report several correlations between Gh and other variables,2l of which three are large enough to be statistically significant and possibly of theoretical importance. WVehave reworked these correlations with a somewhat different sample of 46 cities, calculating segregation indexes for the nonwhite, rather than the Negro, population. The correlation of Gh with the 1939-40 crude death rate from tuberculosis for these cities was .58. A much higher correlation, .83, was found between the city nonwhite proportion, q, and the tuberculosis death rate. Further, since q and Gh correlated .51, the partial correlation between Gh and the tuberculosisdeath rate, with q held constant, dropped to .32, which is on the borderline of statistical significance at the .05 level. Or, when the tuberculosis death rate was standardized for color, the correlation with Gh dropped to .29, which is of doubtful significance. In addition it was found that Gi correlated only - .02 and .06, respectively, with the crude and standardized tuberculosis death rates. It appears, therefore, that the tuberculosis death rate responds primarily to the proportion nonwhite, rather than to the degree of nonwhite segregation, and that the correlation obtained with Gh reflects primarily the correlation of Gh with q. The difference between the results obtained with Gh and Gi follows from the fact that Gi depends only on the segregation curve, while Gh involves not only the
20 Cowgill and Cowgill, op. cit.; Hornseth, op. cit.; Jahn, op. cit.; Jahn, Schmid, and Schrag, op. cit.; Julius A. Jahn, Calvin F. Schmid, and Clarence Schrag, "Rejoinder to Dr. Hornseth's Note on 'The Measurement of Ecological Segregation,"' American Sociological Review, 13 (April, 1948), pp. 216-217; Kish, op. cit.; and Williams, op. cit. 21 "The Measurement of Ecological Segregation," op. cit., pp. 302-303.
curve but also the nonwhite proportion, q. It was also found that the correlation between Gh and the per cent of overcrowded housing, .40, droppedto .02 with q partialled out, and that tihe correlation with Gi was a nonsignificant -.13, with q being a better predictor than either Gh or Gi, correlating .76 with per cent of overcrowding.Similarly the r of -.39, between Gh and Thorndike's "G" was clearly attributable to the high correlation, -.87, between q and "G," and even changed in sign to an r of .11 when q was partialled out. The correlation of Gi with "G" was a nonsignificant .18. These results indicate that the ability of a segregation index to predict other variables is an insufficient criterion of its worth. If we wished to predict the tuberculosis death rate, the percentage of overcrowding, or Thorndike's "G." we would use none of the segregation indexes, but instead, q, as the predictor. Yet the theoretically interesting question would remain, e.g., is the tuberculosis death rate associated with the segregation of nonwhites? To investigate this question we require a measure of segregation whose validity is established independently of its correlation with the death rate.
INADEQUACIES OF SEGREGATION INDEXES
The literature on segregation indexes contains references to a number of difficulties in their use and interpretation.Some of these require additional comment. All of the segregation indexes have in common the assumption that segregation can be measured without regard to the spatial patterns of white and nonwhite residence in a city. Yet it is common knowledge that in some cities-e.g. Chicago-the nonwhite population is predominantly clustered in a "Black Belt," whereas in other cities nonwhite occupancy takes the form of scattered "islands" or "pockets." Surely whatever variables of ecological organization and change are related to the degree of segregation must also be affected by the spatial pattern of segregation. We have found that in 1940, in 50 of the 51 nonsuburban tracted cities studied, the nonwhite population was more concentrated toward the ecological center of the city than the white population. (This study involved the use of an "index of centralization," which has not yet been described in pub-
216
AMERICAN SOCIOLOGICALREVIEW discussion of the problem of area unit has been given by Wright,23who has indicated the formidabledifficultiesin the way of finding segregation measures which are not relative to the system of area units used.
IMPLICATIONS FOR INDEX THEORY
wishedwork, but which is in some respects similar to the Gini segregation index.) Other aspects of the spatial pattern of segregation need to be studied as well. It seems unlikely that any single index of segregation will be found sufficient for the purposes of such research. In none of the literature on segregation indexes is there a suggestion about how to use them to study the process of segregation or change in the segregation pattern. As a first step in this direction we have experimented with an adaptation of the method of expected cases to determine (1) how much of the segregation of nonwhites can be attributed to differences between whites and nonwhites in income, occupational status, and rentals paid, and (2) whether changes in degree of segregation over a ten-year period are related to changes in these variables. It has been found, for example, that there is a markeddifferencebetween southern and northern cities in the influence on residential segregation of white-nonwhite differentials in labor force and occupationalstatus. When this variable is held constant, the Gini index of segregation is reduced by 12 to 22 per cent in most southern cities, but by only two to nine per cent in most of the cities of the North. Such preliminary findings indicate the advisability of taking account of socio-economic factors in analyzing differences in residential segregation. There is, further, a need for research to develop mathematical and empirical bases for anticipating the effects on measures of segregation of such changes as a markedincrease in the nonwhite proportion, an improvement of the nonwhite's relative socio-economic status, or an invasion-successionsequence. The problem of the appropriateareal unit for research on segregation has been forceAs fully stated by the Cowgills.22 they imply, it is easy to gerrymander tract boundaries to increase or decrease the apparent degree of segregation. However, the problem cannot be solved merely by reducing the size of areal units, e.g., to blocks. The objections made to the census tract basis apply also, mutatis mutandis, to blocks. For example, if all nonwhites resided on alleyways and all whites in street-front structures, then even a block index would fail to reveal the high degree of segregation. The most complete
22
In this paper we have not sought to formulate a comprehensiveset of criteria for determining the validity of a segregation index. Probably such an attempt would be premature in the present stage of empirical investigation and conceptualization of the phenomenonof segregation. In our judgment the criteria thus far suggested in the literature fall short of comprehensiveness,and not all of them are likely to be generally accepted. However, we feel that the work reported here is relevant to the problem of validating segregation indexes, and possibly suggestive for the general problem of index development in social research. Specifically, we have established the following: (1) There was a lack of clarity and consistency in the specifications for a segregation index originally proposed. Jahn et al. suggested that "a satisfactory measure of ecological segregation should . . . not be distorted by . . . the proportion of Negroes."24 Though it has never been made clear what would constitute "distortion" in this respect, it is apparent from our analysis that the nonwhite proportion, q, does enter into the formulas for such indexes as Gi, D, GC, and Rep, and that q is involved in different ways in the several formulas. For example, if the segregation curve remains constant, but q changes, then the values of Gk, Rep, and eta will be affected, but not those of Gi, D, and Co. As yet there is no criterion to determine which of these is the more desirable property for an index number. Lacking such a criterion it is perhaps doubtful whether a meaningful comparison can be made of the degrees of segregation of two cities with greatly different q's. (2) The empirical correlationsamong alternative indexes are clarified by determining the mathematical relationshipsholding among them-either in general, or under the assumption of a par23 John K. Wright, "Some Measures of Distribution," Annals of the Association of American Geographers, 27 (December, 1937), pp. 177-211. 24 "The Measurement of Ecological Segregation," op. cit., p. 294.
Cowgill and Cowgill, op. cit.
ANALYSIS OF SEGREGATION INDEXES ticular model of the segregation curve. Thus the assessment of segregation indexes is carried a step beyond the speculation of Jahn et al. that their four indexes comprise "two sets of independently discriminative measures," 25 and Hornseth's erroneous judgment that these four indexes "are for practical purposes identical measures."26 (3) The mathematical analysis of segregation index formulas discloses the areas of redundancy and ambiguity among them, i.e., it permits a conclusion as to the circumstancesin which two indexes will give interchangeableresults and in which they will give incompatible results. Hence it goes beyond the truism that the empirical results obtained with an index are in part a function of the mathematical form of the index, to indicate what specific property of the index is responsible for the kind of results obtained. Thus, for example, the finding that Southern cities are more segregated than Northern can be properly qualified if it is known that the index being used responds in a certain way to variations in q, and that Southern cities generally have higher nonwhite proportions than Northern. A difficult problem of validation is faced by the proponent of a segregation index formula. The concept of "segregation" in the literature of human ecology is complex and somewhat fuzzy, i.e., the concept involves a number of analytically distinguishable elements, none of which is yet capable of completely operational description. Yet it is a concept rich in theoretical suggestiveness and of unquestionable heuristic value. Clearly we would not wish to sacrifice the capital of theorization and observation already invested in the concept. Yet this is what is involved in the solution offered by naive operationism, in more or less arbitrarily matching some convenient numerical procedure with the verbal concept of segregation. The problem must be faced of considering a variety of possible selections of data and operations on these data in an effort to capture methodologically what is valuable in the work done with the concept prior to the formulation of an index. As we have suggested, it may be that no single index will be sufficient, because of the complexity of the notion of segregation,involving as it does considerations of spatial pattern,
25
217
unevenness of distribution, relative size of the segregated group, and homogeneity of sub-areas, among others. In short, we are emphasizing the distinction between the problems of (a) working from a limited set of data to a mathematically convenient summary index, and (b) working from a theoretically problematic situation to a rationale for selecting and manipulating data. Lazarsfeld and his co-workers27 have taken the lead in a much needed effort to codify the proceduresby which concepts are so specified that index construction may profitably be undertaken and to rationalize the decisions involved in formulating an index for research use. Sociologists will learn, as economists have, that there is no way to devise adequate indexes which avoids dealing with theoretical issues. Incidentally, one lesson to be learned from the relatively unproductive experience with segregation indexes to date is that similar problems are often dealt with under different headings. Most of the issues which have come up in the literature on segregation indexes since 1947 had already been encountered in the methodological work on measures of inequality, spatial distribution,and localization in geography and economics.28
27 Paul F. Lazarsfeld and Allen H. Barton, "Qualitative Measurement in the Social Sciences: Classification, Typologies, and Indices," in The Policy Sciences, edited by Daniel Lerner and Harold D. Lasswell, Stanford: Stanford University Press, 1951. 28 P. Sargant Florence, W. G. Fritz, and R. C. Gilles, "Measures of Industrial Distribution," Ch. 5 in National Resources Planning Board, Industrial Location and National Resources, Washington: Government Printing Office, 1943; Edgar M. Hoover, Jr., "The Measurement of Industrial Localization," Review of Economic Statistics, 18 (November, 1936), pp. 162-171; Wright, op. cit.; John K. Wright, "Certain Changes in Population Distribution in the United States," Geographical Review, 31 (July, 1941), pp. 488-490; Dwight B. Yntema, "Measures of the Inequality in the Personal Distribution of Wealth or Income," Journal of the American Statistical Association, 28 (December, 1933), pp. 423-433. See also Federal Housing Administration, The Structure and Growth of Residential Neighborhoods in American Cities, by Homer Hoyt, Washington: Government Printing Office, 1939, Ch. 5; P. Sargant Florence, Investment, Location, and Size of Plant, Cambridge: Cambridge University Press, 1948; and George C. Smith, Jr., "Lorenz Curve Analysis of Industrial" Decentralization," Journal of the American Statistical Association, 42 (December, 1947), pp. 591-596.
26
Ibid., p. 299. Hornseth, op. cit., p. 604.
Residential Distribution and Occupational Stratification Author(s): Otis Dudley Duncan and Beverly Duncan Reviewed work(s): Source: American Journal of Sociology, Vol. 60, No. 5, World Urbanism (Mar., 1955), pp. 493503 Published by: The University of Chicago Press Stable URL: http://www.jstor.org/stable/2772537 . Accessed: 15/12/2012 00:10
.
The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to American Journal of Sociology.
RESIDENTIAL DISTRIBUTION AND OCCUPATIONAL STRATIFICATION1

OTIS DUDLEY DUNCAN AND BEVERLY DUNCAN
ABSTRACT Ecological analysisis a promising approach to the studyof urban social stratification, differences in for the residential distributions occupationgroupsare foundto parallel the differences of amongthemin socioeconomicstatus and recruitment. occupationgroupsat the extremes the socioeconomic The of scale are the most segregated.Residential concentration low-rent in areas and residentialcentralization are inversely relatedto socioeconomic status. Inconsistencies the ranking occupationgroupsaccording residential in of to patternsoccurat pointswherethereis evidenceofstatusdisequilibrium.
and the Population Research and Training Center of the University Chicago, the clericalassistance of of David A. Lane and Gerald S. Newman, and the helpful suggestionsof Evelyn M. Kitagawa and Philip M. Hauser.
2 "The Urban Communityas a Spatial Pattern and a Moral Order,"in The Urban Community, ed. Ernest W. Burgess (Chicago: University Chicago of Press, 1926), p. 18.
The idea behindthispaper was forcibly ticular of methodological for set techniques stated-in fact,somewhatoverstated-by researchin comparativeurban ecology.4 RobertE. Park: "It is becausesocial rela- These techniques adaptable to a wide are tions are so frequently and so inevitably variety of problemsin urban ecological correlated with spatial relations;because structure, and objective economical permit physical distances so frequently are, or comparisons amongcommunities, thus and seem to be, the indexes social distances, overcome of of some of the indeterminacy a that statistics have any significance what- strictly cartographic approach.The techeverforsociology. And thisis true,finally, niquesare hereappliedto onlyone metrobecause it is only as social and psychical politan community, Chicago;however, comfactscan be reducedto, or correlated with, parativestudies, on conducted an exploraspatialfactsthat theycan be measured at torybasis,indicatetheir abilityto produce all."2 results. significant a This studyfinds close relationship beDATA AND METHOD tween spatialand social distances a metin ropolitan It thata syscommunity. suggests of The sources data forthisstudy, except tematicconsideration the spatial aspect as notedotherwise, of werethepublished volof stratification phenomena, thoughrela- ume of 1950censustractstatistics Chifor tively neglected students thesubject,3 cago and adjacentareas5(coextensive of by with shouldbe a primary focusofurbanstratifi- the Chicago Metropolitan District,as decation studies.Aside fromdemonstratinglineated 1940),and thecensus-tract in sumtherelevance humanecologyto the the- mary punch cards for this area obtained of ory of social organization, studyoffers from BureauoftheCensus.The ecologithe the further evidence thesuitability a par- cal analysis pertainsto employedmales for of yearsold and over,classified into I The authorswish to acknowledge the financial fourteen groups listedin support of the Social Science Research Committee theeightmajoroccupation
4 Otis Dudley Duncan and BeverlyDuncan, "A Methodological Analysis of SegregationIndexes," forthcomingin American Sociological Review; Donald J. Bogue, The Structure theMetropolitan of Community (Ann Arbor: Universityof Michigan, 1949), p. 72; Richard W. Redick, "A Study of Differential Rates of Population Growthand Patterns of Population Distributionin Central Cities in the United States: 1940-1950" (paper presented at the 1954 annual meetingof the AmericanSociological Society,Urbana, Illinois).
3 See, however, the discussionof "dwellingarea" by W. Lloyd Warneret al., Social Class in America (Chicago: Science Research Associates, 1949), pp. 151-54.
1950 United States Census of Population, BulletinP-D1O.
493
494
THE AMERICAN JOURNAL OF SOCIOLOGY
dis- occupation group the percentageof all groups thetablesbelow.The occupation in in and in regarded thisanalysis(farmers farm workers thatgroupresiding each area segment).The private household unit (tract or zone-sector laborers, managers, farm in- indexof dissimnilarity not reported) two occupaand occupation between workers, the of thousand theone tiongroupsis thenone-half sumof the cludeonlytwenty-one between males in the absolutevalues of the differences and a halfmillionemployed takenarea by distributions, the respective District. Metropolitan hypothetical A portion of the analysis is carried area. In the accompanying between through withthe censustractas the area exampletheindexof dissimilarity unit. There are 1,178 censustractsin the of District, which935 are in Metropolitan A B Diff. Area thecityof Chicagoand 243 in theadjacent rests on of area. The remainder theanalysis 1...... 5% 10% 15% 5 20 15 2...... delineated a schemeof zones and sectors, 15 40 25 3 ..... Tracts wereassignedto ratherarbitrarily. 30 15 45 4...... to of concentric thecenter the circular zones, Total. 100% 40% 100% withonecityat Stateand Madisonstreets, miles,twoup mile intervals to fourteen and mileintervals to twenty-eight miles, up occupationsA and B is 20 per cent (i.e., of morethan withresidual categories tracts as 40/2). This may be interpreted a measand the miles twenty-eight from citycenter 20 ure of displacement: per cent of the tractsin the adjacentarea too largeto be workersin occupationA would have to concategory classified zones.The latter by area move to a different in orderto make tains only 1.4 per cent of the employed identical with that of their distribution with males. Five sectorswere established, B. occupation radiallinesdrawn boundaries approximating is When the indexof dissimilarity comThe NorthShoresecthe from citycenter. puted betweenone occupationgroupand through such torrunsalongLake Michigan combined(i.e., total all otheroccupations as suburbs Skokie,Evanston,Lake Forest, males exceptthosein the given employed sectorexand Waukegan;the Northwest to occupationgroup),it is referred as an ParkRidgeandDes Plainesto tends through An index of segregation.7 equivalent and includes Arlington Heights;theWestsector the means of computing more convenient andBerwyn, thesuburbs Cicero, Park, of Oak indexis to computethe index segregation out running as faras Wheatonand Naper- of dissimilarity betweenthe givenoccupasectoris approximately ville;theSouthwest males (i.e., tion groupand total employed Blue Isbisected a linerunning through by "adjusting"theresultby all occupations), land,Harvey,and ChicagoHeightsto Park of the by dividing one minus proportion the Forest; and the South Shore sectorruns totalmale employed in included laborforce the along Lake Michiganthrough Indiana thatoccupationgroup. suburbs of East Chicago, Hammond, and The indexesof segregation dissimithe Gary,and East Gary.Combining zone laritywerecomputed botha tractbasis on and sector schemesyielded a set of 104 and a zone-sector basis to detersegment segments;that is, area units zone-sector of minethe effect the size of the area unit averagingabout ten times the size of a 6 For the use of the index of dissimilarity a as with considerable varicensustract, though of association" see National "coefficient geographic ation in area and population. and ResourcesPlanning The spatial "distance"between occupa- National ResourcesBoard, IndustrialLocation (Washington, D.C.: Governthe differ- mentPrinting tion groups,or more precisely Office, 1943), p. 118. is ence betweentheirareal distributions, 7 For discussionof the index of dissimilarity as To measured theindexofdissimilarity.6 a segregation by indexsee Duncan and Duncan, op. cit., one calculatesforeach and the literature therecited. computethisindex,
RESIDENCE
AND OCCUPATION
495
8 JeromeK. Myers, "Note on the Homogeneity of Census Tracts: A Methodological Problem in Urban Ecological Research," Social Forces,XXXII (May, 1954),364-66; JoelSmith,"A Method forthe -Classification Areas on the Basis of Demoof graphicallyHomogeneous Populations," American XIX (April,1954), 201-7. SociologicalReview,
on theresults. the of for 'While indexes tracts ordered distancefrom center the the by are uniformly to higherthan forzone-sector city,thatis, are classified according the segments, effect be disregarded this can for zonal scheme. A negative index of centhat the givenoccupapurposesof determining relativeposi- tralization signifies the or tions theoccupation of groups.The product- tion grouptendsto be "decentralized," away from momentcorrelation betweenthe two sets on the averagelocated farther of segregation indexesin Table 2 is .96. the citycenterthan all otheroccupations, The correlation betweenthe two sets of while a positive index is obtained for a occupation.9 dissimilarity indexes Table 3 is .98,with relatively in "centralized" the segment-based index (s) relatedto the tract-basedindex (t) by the regression OCCUPATION AND SOCIOECONOMIC STATUS equation,s = .8t - 1.3. These results indiof indicators the Selectednonecological cate thatforthekindofproblem dealtwith relativesocioeconomic statusof the major here the larger, and henceless homogene- occupationgroupsare shownin Table 1. ous, unit is as serviceable the smaller The professional as and managerialgroups one. This suggests thatsome of the recent clearly have the highest socioeconomic concern about census-tract homogeneity rank, while operatives, service workers, may be misplaced.8 and laborersare clearlylowest in socioThe indexoflow-rent concentrationob- economic is by status.The ranking socioecotainedby (1) classifying tractsinto inter- nomiclevelwouldprobably agreed by on be vals according themedian to rental mostsocialscientists. majoroccupation monthly The oftenant-occupied dwelling units;(2) com- groupscorrespond roughlywith the Alba putingthepercentage distribution rent Edwards scheme of "social-economic by intervals each occupation for groupand for groups."Edwards does not separatesales all occupationscombined; (3) cumulating workersand clericalworkersby "socialthe distributions, fromlow to high rent; economic group,"and the groupof service (4) calculating the quantity TXi-1Yi contains household, except private workers, TXiFYi_,whereXi is the cumulatedper- individualoccupationsvariouslyclassified centage of the given occupationthrough by Edwards as skilled,semiskilled, and the ith rentinterval, is the cumulated unskilled, Yi the predominantly lattertwo. percentageof all occupations combined, reof in A ranking terms medianincome and the summation over all rentinter- sultsin tworeversals rank.The 1949meis in vals; and, finally, "adjusting" the re- dian incomeof male managerial (5) in workers sult (as forthesegregation index)to obtain the Chicago Standard Metropolitan Area an indexequivalentto theone obtainedby was about $500 greater thanthatofprofescomparing the givelnoccupation group sionalworkers, substanbothwere although withall otheroccupations This tiallyabove thatforsales workers. mecombined. The indexvariesbetween100 and -100, with dian income for the craftsmen-foremen a positivevalues indicating tendency for groupwas about $500 higher thanthatfor residences thegivenoccupation of groupto clerical In workers. fact,themedianincome low be in areas of relatively rentand with for the craftsmen-foremen was only group relativeconcen- slightly negative values indicating wherebelowthatforsales workers, in tration high-rent areas. as the medianincomeforclericalworkers The indexof centralization computed was onlyslightly is above thatforoperatives. in the same fashion, exceptthat tractsare 9 The indexes of low-rentconcentration and of
identicalwith the index are centralization formally of urbanizationproposed in Otis Dudley Duncan, "Urbanization and Retail Specialization," Social Forces,XXX (March, 1952), 267-71. The formula of given here is a simplification the one presented there; and the area units and principleof ordering are, of course,different.
496
However,in median school years completed, professional workersclearly rank first, whilethereis littledifference the in for medians themanagerial, sales,and clerical groups. The mediandropssharply, over 2.5 years,forthe craftsmen-foremen group and declinesfurther each groupin the for orderof theinitiallisting. In theChicagoMetropolitan District the proportion nonwhites an occupation of in group appears to be closelyrelatedto its
obtained of rating salesworkers low prestige their by the NORC. Furthermore, data do prestigeratingsby sex. not differentiate area, the in Particularly a metropolitan male sales workergroup is more heavily as withsuchoccupations advertisweighted and real estate agentsand ing,insurance, and of salesrepresentativeswholesale manuthan is the case forfeconcerns facturing sales retail amongwhom malesalesworkers, are clerks thelargemajority.
TABLE I
SELECTED INDICATORS OF SOCIOECONOMIC STATUS OF THE MAJOR OCCUPATION GROUPS Median Median School Median corn Income Years Com1949 inin141t pleted4 Edwards' Socioeconomic
economic
Major OccupationGroup*
Group?
Nonwhite 1
Per Cent Cont
and kindred workers. $4,387 Professional, technical, except and proprietors, Managers, officials, farm ....... ... ............... 4,831 3, 698 Sales workers ....... .. ................. 3 7132 .............. Clericaland kindredworkers . 3, 648 workers. . and foremen, kindred Craftsmen, 3,115 Operativesand kindredworkers. 2, 635 Serviceworkers, exceptprivatehousehold.. 580 .2, Laborers,exceptfarmand mine
16+ 12.2 12.4 12.2f 9.5 8.9 8.8 8.4
1 2 3 4 5 5-6 6
2.7
l 7.4 4.9 12.4 23.0 27.4
f 2.8
2.2
* Does notinclude not and laborers, occupation farm workers, household private managers, and farmers farm reported. For males theexperienced in laborforce theChicagoStandard of Area,1950.Source:1950U.S. Metropolitan t Bulletin P-C13,Table 78. of Census Population, and West,1950.Source:1950 U.S. Census in malestwenty-five old and over, theNorth of years t Foremployed P-E No. 5B, Table 11. SpecialReport Population, States, for Statistics theUnited Occupation Comparative Source:Alba M. Edwards, equivalents. ? Approximate Office, 1943). Printing D.C.: Government 1870to 1940 (Washington, tracts in residing census 1950.Based on nonwhites District, malesin theChicago Metropolitan For 11 employed malesin the include 95.8percentofall nonwhite in nonwhite population 1950.Thesetracts 250 containing or more District. Metropolitan
to bases of The proportion very is The failure different ofranking status. socioeconomic by has results been discussed managerial,and giveidentical low in the professional, in on for writers stratification termsof "dishigher but sales groups, it is somewhat of affinity strata" and "status disequilibthanforthecraftsmen-foreworkers clerical the in The reversals rankbetween are proportions ob- rium.""l men group.Increasing and serviceworkers, servedforoperatives, 10 National Opinion Research Center, "Jobs in laborers, order. and Occupations: A Popular Evaluation," Opinion con- News,IX (September1, 1947), 3-13. is ranking in general The suggested 1"Cf. Pitirim A. Sorokin,Society,Culture,and Research formity theNationalOpinion with Personality (New York: Harper & Bros., 1947), toward pp. Center'sdata on popularattitudes of 289-94, on disaffinity strata. On status disoccupations, exceptthat sales occupations equilibrium cf. lmile Benoit-Smullyan,"Status, American occu- Status Types, and Status Interrelations," and appearto rankbelowclerical craft 154-61; inade- Sociological Review, IX (April, 1944), a Rural An pations in the NORC results.10 Harold F. Kaufman, DefiningPrestigein titleswithin Community("Sociometry Monograph," No. 10 of quate sampling occupational mayaccountin partforthe [New York: Beacon House, 19461). thesales group
RESIDENCE
AND OCCUPATION
497
subjectto professional managerial and are groups and the The lattergroups necessarily clericaland craftsworkers of are most fre- cross-pressures fromthe determinants quent.The upshotseemsto be thatno one residential the for selection; example, cleriranking be acceptedas sufficient all cal grouphas an income can to equivalent that for purposes.The examination residential of operatives but the educationallevel of of patternsdisclosesother instancesof dis- managerial workers. that equilibrium, whichare of interest To checkthehypothesis spatialdisboth in themselves as cluesto theinterpretationtances among occupationgroupsparallel and of thosealreadynoted. of the social distances, indexes dissimitheir major among distribution larity residential in
RESIDENTIAL PATTERNS
TABLE 2 Fouraspectsoftheresidential patterning ofoccupation groups considered. first INDEX OF RESIDENTIAL SEGREGATION OF EACH are The is the degreeof residential segregation of MAJOR OCCUPATION GROUP, FOR EMPLOYED eachmajoroccupation group with respect to MALES IN THE CHICAGO METROPOLITAN DIsall others, that is, the extentto whichan TRICT, 1950 occupation groupis separatedresidentially fromthe remainder the employed of labor By By force.The secondis the degreeof dissimiCensus Sector Major OccupationGroup* Tracts Segments larity residential in distribution amongmajor occupation thatis, theextent groups, to and technical, kindred which pairs of occupationgroups isolate Professional, 21 30 . workers..... .... themselves from another. one The third as- Managers, officials,and proprie20 .29 pect is the degreeof residential concentra- tors,exceptfarm 20 .29 tionofeach occupation groupin areas char- Sales workers 13 9 workers Clericaland kindred acterizedby relatively low rents.Finally, Craftsmen, and kindred foremen, 14 19 ............. workers ....... the degreeof centralization each major of 16 occupation group(i.e., the extent which Operativesand kindredworkers.. 22 to Service workers, except private an occupation group concentrated is 20 toward 24 ......... ..... household 29 the center themetropolitan of community) Laborers,exceptfarmand mine. . 35 is examined. each case the spatial patIn * Does private and farmmanagers, terning theresidences considered re- household not includefarmersand occupation reported. of is in not farm workers, laborers, lationto socioeconomic level. A clearrelationship theranking ma- occupation of are groups shownin Table 3. As of jor occupation groups by socioeconomic previously a of indicated, listing majorocculevelcan at statusand by degree residential by of segrega- pationgroups socioeconomic a tionis shown Table 2. Listedin theorder best onlyroughly in approximate social disgiventhere, indexes residential a the of segre- tance scale. Similarly, measureof discan a form U-shaped gation The pattern. highest similarity residentialdistribution in valuesareobserved theprofessionals and only approximate spatial distancebefor the the laborersand the lowestvalue for the tweengroups-theindexmeasures onlythe clericalworkers. distributions The degreeof residential dissimilarity theresidential of variesonly slightly set segregation to among the withrespect a particular ofareas and aspectsof professional, and sales groups; is insensitive otherimportant to managerial, of it for markedly theclerical the spatial patternsuch as proximity however, declines workers and then increasesregularly for areas of concentration. each successive the group. Nonetheless, data in Table 3 indicate This finding of thatresidmtial suggests seg- the essentialcorrespondence social and regationis greaterfor those occupation spatial distanceamongoccupationgroups. of statusthanfor If it is assumedthattheordering major groupswithclearlydefined with those groupswhose status is ambiguous. occupation corresponds increasgroups
498
ing social distance(e.g., thesocial distance group salesworkers; is,theresidenand that and sales workers of with is tialdissimilarity salesworkers craftsbetweenprofessional operatives,and laborersis greater thanthatbetween professional and men-foremen, thanthatof themanagerial greater and if it is assumed slightly managerial workers), in of difference terms their although that the index of residential dissimilaritygroup, less. approximates spatial distancebetween socioeconomic the levelis presumably of the The residentialdistribution clerical thetwogroups, expected would pattern to is be the following: at Starting any pointon workers moredissimilar the distribuand professional, manthe diagonal, the indexeswould increase tionofsales workers, than to that of the craftsreading or to theright(downor to the agerialworkers up left,in the case of the indexesbelow the men or the operatives.Hence, although
TABLE 3
INDEXES OF DISSIMILARITY IN RESIDENTIAL DISTRIBUTION AMONG MAJOR OCCUPATION GROUPS, FOR EMPLOYED MALES IN THE CHICAGO METROPOLITAN DISTRICT, 1950
(Above diagonal,by census tracts;below diagonal,by zone-sector segments)

MAJOR GRouP* OCCUPATIoN
MAJOR OCCUPATiON GRouP*
Prof., Tech., Kindred
grs., Offs., , Props.
Crafts- Oper- Service, LaborSales Clerical, men, atives, ers,exc. exc. Farm Wkrs. Kindred ForeKinPriv. and men dred Hshld. Mine
Professional, technical,kindred workers Managers, officials, and proprietors, ............ ........ exceptfarm Sales workers ........ ............ Clerical and kindredworkers.. ... kindred workers. Craftsmen, foremen, . kindredworkers..... . Operatives, Serviceworkers, exceptprivatehousehold......................... Laborers,exceptfarmand mine
.. 8 11 20 26 31 31 42
13 7 18 23 29 31 41
15 13 ....... 17 25 30 30 42
28 28 27 ....... 12 16 19 32
35 33 35 16 ....... 14 25 30
44 41 42 21 17 . ..... 19 21
41 40 38 24 35 26 ....... 24
54 52 54 38 35 25 28 ......
* Does notinclude not and farm household farmers farm and managers, private workers, laborers, occupation reported.
withprogrouped are workers often segments). clerical It basedon zone-sector diagonal, as and sales workers managerial, is clear that the expected pattern, though fessional, disin de- "white-collar," termsof residential essentially reproduced, not perfectly to they The scribes observed the pattern. exceptions tribution are moresimilar thecraftswhitethanto theother are fewand forthe most part can be ex- menand operatives such hypotheses collargroups. plained hypothetically; of inversions theexpected The remaining research. providecluesforadditional priexcept workers, involveservice pattern is The least dissimilarity observedbeof One-fifth theseare "janivate household. and workers, tweenprofessional managerial a and sextons." Presumably substantial and profes- tors and sales workers, managerial of live at their place proportion thejanitors the Furthermore, sional and sales workers. of work in apartmentbuildingshousing of dissimilarity each of these groupswith workersin the higherstatus occupation groupis of approxi- groups."2 is hypothesized thisspecial each otheroccupation that It of In three the mately samedegree. fact, the 12 Cf. Ray Gold, "Janitors versus Tenants: A patternconcern Status-Income Dilemma," American Journal of of inversions the expected the comparisonbetween the managerial Sociology, LVII (March, 1952), 486-93.
RESIDENCE
AND OCCUPATION
499
of accountsforthe tendency circumstance to to workers be less dissimilar the service on statusgroupsthan expected the higher At status.13 thesame basisofsocioeconomic of time the color composition the service direcacts presumably in theopposite group on segregation tion.In so faras residential across occupational basis of color,cutting comlines,existswithinthe metropolitan at occupationalstatusis rendered munity,
INDEXES
the indexeshad to be computed total for employed personsratherthan males. It is that the degreeof lowclear,nonetheless, rent concentrationinversely is related the to socioeconomicstatus of the occupation All groups. fourof thewhite-collar occupation groupshave negativeindexes,signifyingrelative concentration high-rent in areas, whereasall fourof the blue-collar groups havepositive indexes. Again, there is
TABLE 4
OF Low-RENT CONCENTRATION AND OF CENTRALIZATION FOR MAJOR OCCUPATION GROUPS, CHICAGO METROPOLITAN DISTRICT, 1950
INDEX OF CENTRALIZATION INDEX OF LOW-RENT CONCENTRATiox (ToTAL
(EMPLOYED MALES) . M etro_

__ _ _ __ _ _ __
MAJOROCCUPATIONGROuP*
Sector
_ _
__
EMPLOYED PERSONS)
District
politan
North Shore
Northwest
South-
South
west
Shore
technical,and kindred Professional, -32 workers ................. .... and Managers,officials, proprietors, .-30 exceptfarm . ........ ...... -25 Sales workers . 9 Clericaland kindredworkers Craftsmen,foremen,and kindred .11 workers 29 kindredworkers Operatives, Service workers, except private 7 ... ... ........ household .... 32 Laborers,exceptfarmand mine....
-14 -12 - 5 5 - 8 10 21 7
-15 -20 -15 7 6 21 16 9
-20 -16 -12 2 - 6 16 18 21
-29 -19 -12 1 -7 18 20 30
-20 -15 - 9 5 - 5 8 16 16
5 1 8 9 -26 - 4 36 - 1
* Does not include farmersand farm managers, private household workers,farm laborers, and occupation not reported.
as ineffective a determinant a relatively least partially sharpbreakbetween clerical the how- and theother three location.These factors, white-collar groups. The of residential grouphas a slightly ever,probablydo not whollyexplainthe managerial greater inlargestsingledeviationfromthe expected dex oflow-rent concentration theprothan indexofdissimilar- fessional the pattern, muchlarger group,despitethe higher income and craftsmen-foremen service levelof theformer. is evenmorestriking itybetween It than betweenclericaland service that the low-rent workers concentration craftsof workers. men-foremen is substantially thanfor higher of colunm Table 4 showsthein- clericalworkers, The first again the reverseof the of concentration theoccu- relative positionson income. It can be dexesoflow-rent Some cautionmustbe exer- shownthat in 1940 the combinedclerical pationgroups. sincethetabulathem, cisedin interpreting and salesgroup tended spenda larger to protheyare based did notdistintionon which portion its incomeforrentthandid the of and maleand female workers, between guish groupof craftsmen, and foremen, kindred noted in data, 13 This effect has been definitely workers. For example,for tenantfamilies not shownhere,forfemaleprivatehouseholdworkwith wage and salary incomes between of ers, about one-fourth whom"live in."
500
other theprofessional and in and $3,000 1939, without $2,000 as group, is also truein the headedby SouthShoresector. 63 income, percentofthefamilies The latter sector exhibpaid $40 permonth its a quite markeddeparture fromthe exor a clerical salesworker withonly38 per pectedpattern, that the onlydecentralin as or morerent, compared are whoseheadswerecraftsmen,ized occupations thosein theblue-collar centoffamilies category. Thereis a smallmeasureof conworkers.'4 or foremen, kindred for firmation the hypothesized concentration for in pattern, The indexof low-rent categorythe althoughpositive,is low that withinthe white-collar serviceworkers, groups. least centralized groups theprofessional are comparedto the otherblue-collar no and managerial, pattern and within blue-collar the to This exception the expected as the is doubthas thesameexplanation advanced category mostdecentralized thecraftsproportion men-foremen group. The high index servfor above; thatis,thata substantial high ice workersis doubtlessdue to the relalive workers incomparatively ofservice placeof tively high proportionof nonwhitesin with statusareasinconnection their this occupation, and the relativelycenemployment. of of The indexes centralization theoccu- tral location of the South Side "Black of are pationgroups givenin Table 4, bothfor Belt," a largeportion whichfallsin the The decentralization Districtas a whole and SouthShoresector. of the Metropolitan According to the otherblue-collar sectors. each ofthefive groupsis attributable within is zonalhypothesis, there an up- to the presenceof the Indiana industrial theBurgess of suburbs theperiphery theSouthShore status on in of wardgradient thesocioeconomic the from cen- sector.A similareffect some industrial of as thepopulation oneproceeds of end terto theperiphery the city.Hence one suburbsat the northern of the North cen- Shore sectoris observable the low cenin wouldexpectthe degreeof residential indexforlaborers thatsector. in of groupto be in- tralization tralization an occupation status. It is apparent thatexpectations basedon the relatedto its socioeconomic versely for mustbe qualified recogby The data providegeneralsupport this zonal hypothesis of proalthoughthereare some sig- nizingdistortions the zonalpattern hypothesis, industrial concentraThus, for the Metro- duced by peripheral nificant exceptions. of three thefour tions. Such concentrations as appear only in politanDistrict a whole, (indicating certain indexes negative are sectors, and, wheretheyare absent, white-collar and leads to a realistic exdecentralization), threeof the the zonal hypothesis relative indexesare positive(indi- pectation concerning pattern residenthe of fourblue-collar The excep- tial centralization socioeconomic status. by catingrelativecentralization). and tionalcasesareagaintheclerical craftsRESIDENTIAL SEPARATION AND DISSIMILARmen-foremen groups. ITY OF OCCUPATIONAL ORIGINS In threeof the fivesectors(Northwest, for that West, and Southwest),the hypothesized Therearegoodreasons supposing is indexes perfectly residential are patterns relatedto occupaof pattern centralization between tional For have except theinversion for mobility. example, reproduced, ecologists and noted a tendency for advances in socioworkers the craftsmen-foremen clerical For statusto be accompanied miwhich appearsin all sectors. the economic by group, deviation gration toward city's the Residenperiphery. NorthShoresectorthe principal low tial segregation doubtless of thebaris is one fromthe pattern the comparatively in of degreeof centralization serviceworkers riersto upwardmobility, so faras such is to by In and laborers. thissectorthe managerial mobility affected the opportunity and the than observe imitate wayoflifeofhigher moredecentralized groupis somewhat social strata.Amongthe findings reported 14 Data for the Chicago MetropolitanDistrict, above,at leastonemayhavean explanation 1940, fromTable 11, Families: Income and Rent, It that mobility. is surprising Population and Housing, 16thCensus of the United thatinvolves of the residential do patterns sales workers States:1940.
RESIDENCE
AND OCCUPATION
501
by thoseof is illuminated thedata in Table 5, which morethantheydo from not differ since shows indexesof dissimilarity and workers; professional managerial among the is groups with respect the to the incomeof sales workers well below majoroccupation of by theyranklowerin prestige, distribution each group majoroccupathat of either, is attainment substan- tiongroupof theemployed male's father.l6 educational and their workers. These indexes, less thanthatofprofessional therefore, pertainto differtially in But thereare data whichsuggestthat a encesamongthemajoroccupation groups or are of origin, recruitment. hyThe sizableproportion sales workers mov- background, occupational ingto a higher to is the level,or aspire pothesis be tested that,thegreater it betweena pair of occupation to sucha move,anticipating by following dissimilarity in group. groups occupational of theresidential pattern thehigher origins, greater the is dissimilarity residential in distribution. The OccupationalMobilitySurveyfound their
TABLE 5* INDEXES OF DISSIMILARITY IN DISTRIBUTION BY FATHER'S OCCUPATIONAMONGMAJOR
OCCUPATION GROUPS, FOR EMPLOYED MALES IN SIX CITIES IN THE UNITED OCCUPATION GROUPt MAJOR OCCUPATION MAJOR GROUPf Crafts- Opera- Service, Laborers, Clerical, Ine. Kindred Foremen Kindred Priv. ec me, te, Mine Mn Foremen Kindred TI~shld. STATES,
1950
Mgrs., Sales Wkrs. Ofs, Offs. Props.
and kindred technical, Professional,
20 16 27 ......................... workers exand Managers,officials, proprietors, . 11 ........ 28 . ......... cept farm 26 ........ Sales workers .. .. ..... ........... ....... . . . . workers Clericaland kindred . .......... ...... workers . ........ kindred foremen, Craftsmen, . kindredworkers ............... .............. . Operatives, Service workers, including private ........ ..... ........ household ................. ........
38 31 35 18 ........ . ....... ........
39 34 37 20 14 ....... ........
34 30 35 28 25 22 ........
46 42 47 39 31 23 20
* Source:Unpublished Table W-9.For description sampling enumeration proand data from of Survey, Occupational Mobility L. in 1954). LaborMobility Six Cities(New York: SocialScience Research Council, cedures Gladys Palmer, see not A and farmers farm and of household workers are t Does notinclude managers occupation reported. smallnumber private of laborers and withlaborers, except mine. workers, a smallnumber farm included withservice
Table W-56 ofthe Occu15 Based on unpublished pationalMobilitySurvey,takenin six citiesin 1951. For descriptionof the samplingand enumeration proceduressee Gladys L. Palmer, Labor Mobility in Six Cities (New York: Social Science Research Council,1954),chap. i and AppendixB.
in The pattern Table 5 is clearly that of like thatformales employed both 1940 and of of was with 1950there a movement 23 percentof ofTable 3. The indexes dissimilarity to in computed thezoneon as themenemployed sales workers 1940 respect residence, and sectorsegment proprietors, of intothegroup managers, basis,correlate withthe .91 by officials 1950. This is the largestsingle 6These indexes are based on the aggregated in movement themobility resultsof sample surveysin six cities in 1951. Alinteroccupational table, exceptthat 23 per cent of laborers though separate data are available for Chicago, moved into the group of operativesand these were not used here, because the sample was in too small to producereliablefrequencies most of workers."5 kindred the cells of the 8 X 9 table fromwhich the dismobility similarity aspectof occupational Another indexeswerecomputed.(In the intergeneof ration mobilitytable the classification fathers' and farm occupationsincludedthe group "farmers managers" as well as the eight major occupation groups listed in Table 5. This was desirable,since proportionof fathers-though very a significant fewofthe sons in thisurbansample-were farmers.)
502
lead to accurate, specific predictions indexesfor occupationalorigin.The cor- ground indexes of the patternof differences residential in relationis .94 for the residential withthe regression distribution. based on censustracts, index(t) on theindexof of the tract-based The ecological analysis has provided in dissimilarity occupational origin (u) support theproposition spafor that is strong being t = 1.2u - 1.8. The hypothesis tial distancesbetweenoccupationgroups definitely substantiated. thereby social distances, of In Table 5 all butoneoftheinversions are closelyrelatedto their in either terms conventional of inof measured on expected the assumption the pattern of statusor in terms of an unequivocalranking the occupation dicators socioeconomic in origins; that workers. of differences occupational the involve salesand service groups occupation groupsare workersare closer to professional the most segregated Sales of back- thoseat the extremes the socioeconomic to withrespect occupational workers of in and scale; thatconcentration residence lowworkers thanare themanagerial ground relatedto socioecogroups. rentareas is inversely each of theblue-collar from farther status;and thatcentralization resiof would nomic pattern a Actually, moreconsistent inversely relatedto sociosecond denceis likewise salesworkers by be produced ranking status.Theseresults in accord are In group. thisre- economic in place ofthemanagerial are with accepted ecological theory,provide origins spect the data on occupational the with the ecologicaldata supportfor it, and demonstrate relemore consistent research thetheory to of in vanceofecological status thanare thedata on socioeconomic of of Table 1. In terms theindexes dissimi- social stratification. These generalizations, however, perare workservice origins, in larity occupational to three white-collar haps no more significant the advanceers are closerto thefirst of thanare theinstances in blue-collar ment knowledge thanare anyof theother groups amongthe whichtheydo not hold and the additional in However, comparisons groups. advancedto accountfortheexclerical and blue-collar groups, service hypotheses measuresof socioranknextto last,orbetween ceptions.Conventional clearly workers statusdo notagreeperfectly to of economic as Again,thefactor and operatives laborers. is origins morecloselyrelated the rank order of the major occupation occupational thanare the indi- groups,nor do the several ecologicalinto residential separation case in pointoccurs the dexes.The prime at status. catorsofsocioeconomic scale, at the The last point deservesemphasis.Not middleof the socioeconomic on and junctureof white-collar only do the indexesof dissimilarity an conventional Clerical kindred and as blue-collar pattern occupations. area basis have thesame general have substantially moreeducation originbasis but workers those on an occupational and kindred occur than craftsmen,foremen, thatpattern from also thedeviations and the clericaloccupationsare and at thesamepoints inthesamedirection. workers, of in- usuallyconsidered greater than the prestige be This cannot said regarding several and status.If income thecraft related occupations. However, of dicators socioeconomic have higher managers craftsmen-foremen considerably separation, determined residential on and clerical incomes theaverage, and, amongmales, would outrankprofessionals, is identicalwith theirnonwhite proportion smaller.The workers would be virtually of of in operativesin theirseparationfromother pattern theindexes dissimilarityresiclearly places the cleriresidential dentialdistribution If groups. educationdetermined dif- cal groupcloserto the otherwhite-collar therewouldbe substantial separation, thanthecraftsmen-foremenand are, the for ferences between indexes professional groups of the clericalworkers' Neither indexof low-rent conworkers. and workers managerial is is thesehypotheses borneout by the data, centration less thanthatofthecraftsmen But in termsof residential in whereasdifferences occupationalback- and foremen.
RESIDENCE
AND OCCUPATION
503
C. W. Mills (New York: OxfordUniversity Press, 1946).
centralization clerical the group tends fall ventionallydesignatedas "socioeconomic to withthe lowerblue-collar and the status."Apparently, to attempts compound groups, craftsmen-foremen group with the other thesetwo can at best producea partially scale; at worst,theymay obscure white-collar In groups. general, wouldap- ordered it pear that"social status"or prestige more significant in differenceslife-style, consumpis important determining residential in and the as- tionpatterns, socialmobility. of sociationof clericalwithotherwhite-collar There is one important qualification groupsthan is income, although latter the resultsreported.Like census tracts, the sets up a powerful as groupsare not perfectly cross-pressure, evi- broad occupation denced by the comparatively The managerialgroup inhigh rent- homogeneous. income ratioofclerical of families. account cludesproprietors peanutstandsas well To for of executives, and night-club fully thefailure clerical to workers be as corporation are residentially decentralized like the other singers classified professional as workers white-collar groups, wouldhave to con- along with surgeons. one One would therefore siderwork-residence to a sharper differentiation Data on expect find much relationships. if work-residence patterns moredetailed occuseparation a 1951 Chi- ofresidential for cago sampleshow that clerical workers re- pational classifications were available. In semble craftsmen, the and kindred particular, pointsat whichcross-presforemen, locationdevelopshould workers the degreeof separation in much sureson residential morethanthey sales,managerial, pro- be moreclearly do identified. or fessional workers."7 Further research shouldseekother forces of producingresidentialsegregation. Perhaps the most suggestive Ethnic finding the studyis that dissimilarity occupa- categorizations other thanraceare doubtless in tionalorigins moreclosely is associated with relevantthoughdifficult studydirectly to in lackofdata. In general, patterns dethe dissimilarity residential distribution than for is any of the usual indicators socioeco- scribed of herewouldbe expected hold for to nomicstatus.This result also but deviations might can onlybe inter- females, significant in of But pretedspeculatively. one may suppose occur, partbecausetheresidence marthatpreferences aspirations and is determined moreby concerning riedfemales probably housing and residential patterns largely theirhusbands'occupationthan by their are formed childhood adolescent by and experi- own,and in part because the occupations encesin a milieu which father's of the occu- thatcomposeeach of themajoroccupation for groupsare different females from those aspect. pationis an important The discovery that"statusdisequilibria" formales (as mentioned above in regard to are reflected inconsistencies theorder- salesworkers). in in Bothraceand sexwouldbear ingof occupation groups of according their upon residential to patterns privatehouseresidential patterns a provides further rea- holdworkers, are predominantly who female son fordistinguishing "class" from"social and nonwhite. finalclass ofespecially A imstatus"elements'8 within complexcon- portant the is of factors the effect thelocation 17 Beverly of on Thereis evidence Duncan,"Factors Work-Residence workplaces residence. in Separation: Wage and Salary Workers, Chicago, thatresidences notdistributed are randomly 1951" (paperpresented the annualinstitute at of the Society Social Research, for Chicago, June5, withrespectto places of work.If location an difof workis controlled, even sharper 1953). 18 See "Class, Status, Party," in From of thanthat ferentiation residential patterns Max Weber:Essays in Sociology,ed. H. H. Gerth and described heremay be revealed.
UNIVERSITY OF CHICAGO

Otis Dudley Duncan. Classic Articles

Transféré par

Informations du document

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Otis Dudley Duncan. Classic Articles

Transféré par

Droits d'auteur :

Formats disponibles

Population Investigation Committee

The Measurement PopulationDistribution of

of (3) Measures spacing measures (4) Centrographic potential (S) Population

xix Economic Geography, (October,

Cumulative Per Cent of Population

1941), I (November, History,

of Redistribution Population,i850-I940 Edgar M. Hoover, Jr.," Interstate

... ... ...

40? 8 4I.2 4I16 43 6

Source: United StatesCensus publications.

uniform .0746/Vp, which occurs for a perfectly

". persons in New York, the short end of the teeter-totter

xIII Travel Metron, (I937), pp. 78-81. ", Minimum Aggregate

Isolines of Population Potential,United States:

Isolines of Aggregate Travel Distance, United States:

in resided incorporated or placesof 2, 500 inhabitants more,

Non-Metropolitan ... ... ... Urban..... ... Rural ...

... ... ...

Urban place basis Number of places ... i8,548

Urbanizedarea basis Number of places I7,2I7

A METHODOLOGICAL ANALYSIS OF SEGREGATION INDEXES *

W, and YTi 1 T. For each tract compute the nonwhite proi

Jahn, Schmid, and Schrag, op. cit., Index #3.

3 Otis Dudley Duncan, "Urbanization and Re-

a: .22 .06 .01

AmericanJournal of Sociology, 54 (March, 1949),

16 Jahn, Schmid,and Schrag,op. cit., Index

Segregation Index D Gi Gh Rep Eta Co

Arithmetic Error 0.0 -0.006 -0.003 0.028 -0.016 0.078

Mean Absolute Error 0.0 0.015 0.044 0.065 0.039 0.078

MeanSquare Error 0.0 0.019 0.066 0.102 0.054 0.124

1.0 .980 .945 .908 .958 ...

(2 d2) /k *Var (I),

where the d's are differences between

Cowgill and Cowgill, op. cit.

Ibid., p. 299. Hornseth, op. cit., p. 604.

RESIDENTIAL DISTRIBUTION AND OCCUPATIONAL STRATIFICATION1

1950 United States Census of Population, BulletinP-D1O.

THE AMERICAN JOURNAL OF SOCIOLOGY

THE AMERICAN JOURNAL OF SOCIOLOGY

Per Cent Cont

16+ 12.2 12.4 12.2f 9.5 8.9 8.8 8.4

l 7.4 4.9 12.4 23.0 27.4

THE AMERICAN JOURNAL OF SOCIOLOGY

(Above diagonal,by census tracts;below diagonal,by zone-sector segments)

MAJOR OCCUPATiON GRouP*

Prof., Tech., Kindred

grs., Offs., , Props.

(EMPLOYED MALES) . M etro_

-15 -20 -15 7 6 21 16 9

-20 -16 -12 2 - 6 16 18 21

-29 -19 -12 1 -7 18 20 30

THE AMERICAN JOURNAL OF SOCIOLOGY

Mgrs., Sales Wkrs. Ofs, Offs. Props.

and kindred technical, Professional,

38 31 35 18 ........ . ....... ........

THE AMERICAN JOURNAL OF SOCIOLOGY

C. W. Mills (New York: OxfordUniversity Press, 1946).

Vous aimerez peut-être aussi