Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at .
http://www.jstor.org/action/showPublisher?publisherCode=upitt. .
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
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.
University of Pittsburgh- Of the Commonwealth System of Higher Education is collaborating with JSTOR to
digitize, preserve and extend access to Ethnology.
http://www.jstor.org
Polynesian Relationships: Initial
Correlation and Factor Analyses of
Cultural Data1
Peter W. Hoon
DalhousieUniversity
Techniqueto delineategroupsamongsixteenNorthwestCaliforniatribes
basedupon250differentculturaland archaeological traits.The procedure
Driverand Schuessler usedis basicallysimilarto the procedures used by
the authorin this study.Initially,a rawdatamatrixwas formedwith the
sixteentribesin columnsand the 250ratingcharacteristics in rows.If a
tribein a columnpossesseda giventrait,a "I" was enteredat the inter-
sectionof the columnrepresenting the tribeand the row representing the
trait.If ehetraitwas absentfor the giventribe,a "o"was enteredat the
appropriate locationin the matrix.The presenceand absencedecisions
werebaseduponthe judgments of Driver,Schuessler,andinformants who
had intimateknowledgeof NorthwestCaliforniatribes.Earlierresearch
hadindicatedthatinformants wereableto agreeon thepresence or absence
of a ratingcharacteristicto a scientifically
acceptable degree.
Driverand Schuessler's secondstepin Q factoranalysiswas the inter-
correlationof the culturesby the standardSpearmanproduct-moment
method.The resultof thisstepwasa correlation matrixwhichshowedthe
correlationsbetweenall possiblepairsof cultures.This matrixbecamethe
inputdatafor the thirdstage,factoranalysis,whichgroupedtogetherthe
cultureswhich had relativelyhigh correlations with one another.These
groupsof cultures,or factors,showedethnographers whichculturescovaried
together(showedpresenceand absenceof principallythe samecultural
rating characteristics).Driver and Schuesslerconceptualized factorsas
compositesof severalcultureswhich shareda relativelyhigh degreeof
similarityand whichin turnsuggestedthe cultural prototypes withinthe
culturalareaurlderanalysis.
The currentinvestigationutilizesfactoranalysisin a somewhatdif-
ferentvein thanDriverand Schuessler (I957) did in theiroriginalwork.
In the currentinvestigation,
factoranalysisis usedto delineatebroadgroups
among twenty Polynesiancultures.Once the basic groupshave been
identifiedby factoranalysisindividualrelationships amongcultureswithin
a groupare examinedon the basisof theirintercorrelations. Both factor
and correlation resultsare used to develophypotheses of culturalcontact
and migrationamongthe Polynesian societiesunderstudy,whichin turn,
are comparedwith the findingsof otherPolynesianworkers.
FACTOR
ANALYSIS
OFTWENTY
POLYNESIAN
CULTURES
Method
TwentyPolynesiancultureswereassignedeither"o"or "I" (I - pre-
sent; o absent)for IO4culturalratingcharacteristicsfrom the Ethno-
grarphic (Murdock,
AMtlas I967). The IO4ratingcharacteristicswereselected
froma totalof 58I possibleratingcharacteristics
on the basisof criterisde-
velopedbyThurstone andDegan(I95I). If a particular ratiIlgcharacteristic
werepresentin two or morebut eighteenor lessof the twentyPolynesian
cultures,it was includedin the analysis.The rationalefor thisselectionof
rating characteristics
was to maximizethe capabilityof the ratingchar-
acteristicsto discriminateamong the twenty culturesby making the
POLYNESIAN RELATIONSHIPS 85
amountof covariance betweenculturesas largeas possible.A ratingchar-
acteristicwhichis presentin o-nlyone cultureor presentin all but one,
merelyintroduces uniqucvariancefor thatculture,anddoesnot showhow
the culturein questiondiffersfrom the other culturesincludedin the
analysis.
Followingratingcharacteristic factoranalysis
selection,a widelyavailable
computerprogrampackage(Dixon I969) generatedmeansand standard
deviationsfor each Polynesiancultureand phi-coefficient correlationsbe-
tweenall possiblepairsof cultures.Fruchter(I957) suggeststhat phi-coef-
if the variance
ficientsmaybe a biasedestimateof variableintercorrelations
of the variablesdiderssubstantially.Examination of the standarddevia-
tionsof the twentyculturesindicatedthatthey dideredby g per cent or
less.
Subsequent intercorrelation
to the generationof the phi-coefficient ma-
trix,the computerpackageextractedS1X factorson the basisof the recom-
mendations of Kaiser(I+O). Unitieswereinsertedas communality esti-
matesandthe programwasinstructed factorswhose
to extractsuccessively
eigenralueswere I or greater.This procedure producedsix factorswhich
in turn were rotatedby the varimaxprocedure(Table I). The purpose
of the varimaxrotationwas to producea simplestructure, definedas the
highestpossiblecorrelations (loadings)on the fewestpossiblenumberof
factors(HarmonI+8). Anceptually,the rotationprocessis analogousto
determiningthe best fit of six regressionlines to six clustersof points
TABLE 1
Factor Structureof Twenty PolynesianCulturesAfter Orthogonal
Rotation of Six Factors
I II III IV V VI
Uvea .79 -.04 .07 .17 .22 -.12
Tonga .74 .15 .21 .02 -.10 .09
Futuna .68 .04 .20 .15 .20 .17
Pukapuka -.08 .79 .13 .05 .28 -.03
Samoa .55 .58 .01 .01 -.01 -.13
Ellice .32 .55 -.02 . 45 -. 20 .03
Hawaii .17 -.08 .73 -.05 -.06 .11
Tahiti .30 .20 .68 .28 .05 .02
Raroia -.29 -.01 .56 .52 .03 -.01
Marquesas .06 .32 .52 -.10 .44 -. 25
Kapingamarangi .18 .31 .45 .17 .23 .01
Tokelau .07 .05 .09 .81 .05 -.01
Manihiki .15 .30 .13 .65 .40 .27
Ontong-Java .17 .01 .01 .60 . 40 -. 31
Niue .26 .11 .15 .57 .38 .33
Mangareva .18 -.12 .39 .40 .29 .30
Tikopia -.07 .23 .13 .16 .73 .08
Mangaia .34 -.08 -.06 .23 .67 .09
Maori -.15 .06 -.01 -.11 .09 .86
Easter .30 -.19 .12 .24 .01 .68
86 ETHNOLOGY
on factorII. Sincethe.correlations
betweenSamoaand Elliceand Samoa
and Pukapukaare virtuallythe samesize, a reasonable hypothesisis that
thetwoweresettledaboutthesametimefromSamoa.
ManyPolynesian scholarshavereportedthe similarities
betweenthe cul-
tureswhichappearon factorIII, andhavefrequentlyreferredto the geo
graphicalareain whichthey are locatedas the generalculturalareaof
EasternPolynesia(FornanderI+9; ChurchillI9II; SharpI957; Oliver
I96I; EmoryI+3; Buck I+4; Marshalland Giles I+8; Poulsens968;
GreenI97I). Polynesianlinguistshavealsonotedthe.similaritiesbetween
the cultureswhichappearon factorIII (ElbertI953; Dyen I+5; Pawley
I966, I967). What kind of migrationpatterndoes factoranalysisof the
culturaldatafromthe EthnographicStlas suggestwithinthe EasternPoly-
nesiangroup?Is there supportfor a singleE.asternPolynesianculture
servingasthemigration sourcefortheothers?
The intercorrelations
betweenthe EasternPolynesiangroup (Table4)
indicatesthat Tahiti has the highestcorrelation
with each of the other
EasternPolynesian cultures.Tahitiappearsto havebeenthe singlesettle-
TABLE 4
Intercorrelationsof Five CulturesWith PrincipalLoadingson Factor III
Hawaii Tahiti Raroia Marquesas Kapingamarangi
Hawaii
Tahiti .37
Raroia .22 .37
Marquesas .22 .36 . 19
Kapingamarangi . 23 .43 . 28 .32
Mangareva . 26 .43 . 27 .27 . 11
gO ETHNOLOGY
TABLE 5
Intercorrelationsof Five CulturesWith PrincipalLoadingson F^actorIV
Tokelau Manihiki Ontong-Java Niue
Tokelau
Manihiki 49
Ontong-Java .37 .50
Niue .50 .69 .34
Mangareva .36 .46 . 26 .54
contactsseemmostprobable:Tokelaucontactwitheitheror bothManihiki
andNiue; C)ntong-Javacontactwith Manihiki;andcontactbetweenNiue
and Mangareva.Thesecontacthypotheses suggestedby the culturaldata
wouldappearto be in conflictwith Bayard(I+6), who suggestedon the
9I
RELATIONSHIPS
POLYNESIAN
TABLE 6
CulturesShoxvingthe Highest CorrelationsWith Tikopiaand Mangaia
Niue Manihiki Ontong-Java Pukapuka Mangareva Marquesas
Tikopia .46 .44 .33 .33 .32 .32
Manihiki Ontong-Java Uvea Niue Tikopia Futuna
Mangaia .43 .41 .39 .36 .30 .29
TABLE 8
CorrelationsBetween CulturesWith PrincipalLoadingson Factors I and II
Pukapuka Samoa Ellice
Uvea .07 .41 . 25
Tonga .05 .45 .29
Futuna . 10 . 27 . 24
UVEAsu2
2 w FUTUNA 2 , 2 SAMOA
/ .42 .36 /245 .21 .39 .56 \
TONGA TAHITI TONGA TAHITI TONGA FUTUNA
.41 > .42,
1
*411 3
2/ MANIHIKI
X A ELLICEN 2 2z TOKELAU 1
/.49 .41 .35 .43 \ 49 . 44
TOKELAU TAHITI SAMOA MANIHIKI MANIHIKI RAROIA
19 .19 .31
FUTUNA \ 1X FUTUNA
1 / FUTUNA
.17 .45 \ /.45 .31 45 .34
TIKOPIA NIUE NIUE MANGAREVA NIUE MANIHIKI
< .46 .54 > .69 >
2 2
2/ MANIHIKIS
w PUKAPUKA 2/ KAPINGAMARANGI a1
43 .41 /.34 .23 .43 .41
MANGAIA TAHITI SAMOA ELLICE TAHITI MANIHIKI
.17 .35 > .41 >
1
X MANIHIKIX
2g TONGA 2, v TONGA
41 .50 .4S .05 g .45 . 29
TAHITI ONTONG
--JAVA SAMOA PUKAPUKA
.33 .34 > 35
1
EASTERN POLYNESIACRITICALTRIADS
supportshypotheses
The analysisof triadcorrelations drawnfromfactor
and singlecorrelationanalyseswith respectto the cultureson factorIII.
Together,thesethreeanalysespointto Tahitias the genesisculturefor the
remainingculturesloadedon factorIII. However,the correlationsbetween
\ TONGA ? , \ MANGAREVA
9t) LTHNOLOGY
KAPINGAMARANGI \
J<0Xx
ONTONC 0
r r MARQUESAs
MANGAIA \
24
EASTER1. .
-
I b:
MAOR
DATA
OFCROSS-CULTURAL
LIMITATIONS
The datauponwhichany kind of statisticalanalysisis madein cross-
culturalresearchcome from the insightsand observations of the field
ethnographers themselves,andso suchdatacan be no moreobjectivethan
theethnographers whogatheredit. Sincethedatain the SawyerandLevine
(I966) factoranalysisandthedatain thisstudywerederivedfromMurdock
andhis co-workers' analysisof manymonographs writtenby otherethno-
graphers,one obviouslimitationis the loss of accuracyinevitablewhen
aremadethrough"second
classifications hand"interpretation of monographs
writtenbyotherethnologists.
Despitethepossibilitiesthatthereareinaccuracies createdby ratingsbased
uponsecond-hand ethnographic information, thereis indirectevidenceto
suggestthatthe classificationsthemselvesare adequatelyreliable.Two or
moreratersusingthe sameethnographic materialas a basisof theirratings
canattaina respectable degreeof agreement. DriverandSchuessler (I957)
utilized280 ratingcharacteristics for theirNorthwestIndianTribes,and
found that the inter-rater reliabilityvariedfrom .67 to .87, a perfectly
acceptablerange.BarryandPaxton(Ig7I) demonstrated adequatereliability
within£ancyclassification codes.If sucha degreeof reliability existswithin
usedin thisstudy,the Ethnogruiphic
the classifications StZalswouldappear
to be a valuablecontribution research.
to cross-cultural
Anotherlimitation,on occasioninherentin the originalmonographs, is
thelackof information aboutthepresence or absenceof someof Murdock's
culturaltraits.An encouragingdevelopment, however,is the consistent
edortMurdockand his co-workers are makingto includemorecomplete
information as time permitsthroughan analysisof new monographs and
a re-analysisof olderones.In termsof the culturesincludedin the present
analysis,one effectof thesemissingdatawas the decisionto excludethe
Rennelland Rotumacultures,sincethey were characterized by a lack of
information, apparently fromoriginalmonograph sources.
I00 ETHNOLOGY