olson's circumplex model a review and extension
TRANSCRIPT
Olson's Circumplex ModeL A Review and Extension
Wesley R. Burr 'rTimothy A. Lowe
This paper reuietus the l iterature abou,t Olson's Circumplex model,and attempts to resolue two problems that haue been discouered byattempts to use the modeL. The tu;o probLems are that: tI i themodel theorizes that the relationships in it are crrt: i l inear, butsome data and some other theorizing suggest that at Least one ofthe relationships is l inear; and 12) the role of'communication in themodel is problematic. This paper proposes that the solution to thecontrouersy obotLt the shape of the relationships is that therelationships probably haue a different shape than preuiousLl,thought. The solution to the problem of communication beinggenerally excluded from the moclel is to add a proiile method ofcommunicating inlbrmation about the three factors in the modeL.Mouing to a prolile system aLso makes it easy to expand the modelto include other factors, and one way of expanding it is presented.
Olson's Circumplex model is one of rhe most valuable middle-rangetheories in coniemporary family science. It is oniy a decade oid tOlson,
1976), but it has integrated a large number of partially overlapping conceptsinto a coherent theoretical model. It has pulled together a half century oftheory and research about adaptabil ity and cohesion, and the model is
conceptually clear, easy to understand, and easy to communicale. It aiso has
been heuristic in uhat it has led to over 300 related studies rOlson. 1986p.337). It has stimulated continual improvement in the measurement of thevariables, as the FACES inslrument has now gone through three revisions, andthere is evidence that its reliabil i ty and vaiidity are improving iQ[5sn. 1985).The theory also has helped meet an important, need in rhe field by buildingbridges between theorists, researchers, and practit ioners lOlson. 1976).
* Wesley R. Burr is professor of farnr ly sclence at Br igham Young Untverst ty, Provo,
Utah 84602. Timothy A. Lowe is a graduate student rn the department of fami ly
sciences, Br igham Young Universi ty. Reprrnt requests should be sent to Wesle! ' Burr .
- \ppreciat ion is expressed to Karhleen Hale for assistance in revtewing research about
the role of adaptabi l i ty , and to Thomas W. Draper and Thomas B. Holman for thetr
suggesuons.
Key words: c i rcumplex, adaptabi l i ty , cohesion, communi.cat lon. krndness, I is teninq,
consensus, control , anger, support .
IFamily Scrence Review, Vo1 1, November, L9B7]
Figure 1: Olson's Circumplex Nlodei' Table 1:
<- Lor - COH€SlOtl - |ltgh->
n|a3 ttrtlB EC-gtlo sEl6 l--l ,ls
[771n*
FSI*.
AI o.*II
fllghI
AoA arrluPTAIILI rf,rnrTY
I
T.oUII
Il i0I
-\Iarital Satisii,- 'tLocke \\ i.: ' i . :,
\Iarit,ai Satisf:c(ENRICH
Family Sat isfa. ' :
Quaiiry of Liie
Empirical Recs'
The lareeOlson and ) lcCThe main irncrelat ionships epoint aiong rhrfrom the darrrelat ionships uvar iabies rvoulincrease from ,.there is a s l ieadaptabil it l ' r:dependent vr.!-:t hcin no n sc , . i
The most .lcc:posi t ive and s; : .
Barnes -i:sample of Ols,rstudy had con:curvi l inearrn ' .The discnmir: :belrn'een adep,r.communicet lon.
As with most theories. there are unresolved problems rvrth Olson's
circumplex model, and there are ways it can be improved. This paper is an
attempl to review the l iterature about the model. clarif l ' severrl of the
controversies and limirations of the theory, and make several suggestions
which we hope wil l improve the modei and stimulate additional rheoretical,
conceptual, and empirical work. Hopefuliy, lhe suggestions also wii l make the
modei more useful to practit ioners such as therapists, famrly i i fe educators,
extension speciaiiscs, and famiiy scientists using it in business. induscry and
government.
THE CONTROVERSY ABOUT THE SHAPE OF RELATIONSHIPS
One of the controversies the model has stimulated is that chere are
different opinions about the shape of the relationships in the theory' Olson
( 1976) has proposed thar adaptabil ity and cohesion are related to family
effectiveness with symmetrical and curvil inear relationships. He views rhe
independent variables as having four categories as shown in Figure l, and
theorizes that famiiies in the balanced area have the grealest probabil ity of
being effective. Families in the extreme condil ions are thought to have
higher probabil it ies of ineffecciveness.
The controversy deals with rhe relationships that adaptabil ity and
cohesion have with famiiy effectiveness. The data from some studies -cuggestthese reiationshrps are curvri inear. but other srudies have found relativell '
I inear relationships. Aiso. some theorizing about the effects of adaptabil ity
suggests it has a curvil inear relationship, but other lheorizing suggests It IS a
Iinear relacionship.
FAMILY SCIENCE REVIEW November, 1987-\ovemDer, iy) .
Table 1: Means on the Satisfaction Variables of the Individuals in EachCategory of Family Cohesion and Family Adaptability in Olsonand McCubbin's 1983 Survey.
)Iarital Satisfaction(Locke Wal lace)
Maritai Satisfaction(ENRICH)
Family Satisfaction
Quality of Life
Adaptability Levels
Rigid Struc Flex Chaotic
+5.8 50.9 5.1. I 56. .1
31.3 36.0 39. r .11. .1
.r0.7 15.6 49.6 54.1
122.0 129.6 134.9 110.7
Cohesion Leveis
Disen Separ Conn Enmesh
45.2 50.5 54.1 59.0
3 r .3 35.7 39. 1 -1 1.3
39.9 -1,1.8 .19.8 55.1
L2t.1 128.7 135.2 l -10.8
Empirical Reosons For the Controuersy
The largest and most careful study of the relationships in the model wasOlson and McCubbin 's survey of 1.000 fami l ies (Olson. NlcCubbin, et a l . , 1983).The main findings from their study are reproduced in Table l. and therelabionships are all positive. None of the covariation is negative r.t enypoint along the range of variation in adaptabil ity or cohesion. It is clearfrom the data in Table 1 that lhe relationships are not i inear. If therelationships were l inear, it would mean that the scores on rhe dependentvariables would increase equally as the scores on adaptabil ity and cohesionincrease from one category to another. From data in Table l. i t appears that,there is a slightly curved rather than linear relarionship. The changes inadaptabil ity and cohesion xre associated with greater differences in thedependent variables when the independent variables are in the lower end oftheir range of varia[ion than when they are in rhe higher end of their range.The mosf accurate way to describe these relafionships is that, they arepositive and slighrly curvil inear.
Barnes and Olson (1985) did a follow-up study that analyzed a sub-sample of Olson and NlcCubbin's (1983) survey of 1000 families. and theirstudy had conflicting findings. The data for rhe parents fit Olson's model ofcurvil inearity. but the data for the adolescents had a positive relationship.The discriminant analysis in rhe study also argued for positive reiarionshipsbetween adaprabil ity, cohesion" family satisfaction, and the quality of theircommunication.
November, 1987 FA}IILY SCIENCE REVIEW
One other study has data that allows strong inferences about the shape
of the rdailonsdps )n the modd. Ib )s Russd)'s (1979) dara on 3) "lorma)"
families. This study supports the curvii inearity suggested b-v Olson as it
found that families who had baianced scores on adaptabil it l ' and r'ohesion
were having less diff icuity with their adoiescent child and those rvho had
more extreme scores were having more difficuity.
A number of studies have compared pathologtcal or clintcai s'oups rvrth
more normal groups and found the ineffective groups had hrgher proportions
with extreme scores (Craddock. 19831 Garbar ino e! a i . . l9r{ : Clark. -96{: end
Rodick et e l . , 1986, for example). A few studies have iound no, j t f ferences
between cl in ical and normai groups (Alexander et . r1. . l r ; - l . , l r .nconciuslve
patterns (Sprenkle & Olson, 1978).
Some of the studies that have compared pathological rnd nornei ::oupsprovide useful information about measurement, clinicel. and predtction issues.
but these studies are of i i tt le value in helping us understand :he rnsp€ 'ri the
relationships in the theory. They have two problems that prevent :hem irom
being useful in making inferences about the relacionships. The tlrst :s :hat
the studies combine all four of Olson's ex[reme types into one group. end thrs
makes it impossible to make inferences about the relattonships eciaot.rbil i ty
-rnd cohesion have with effectiveness.
The second problem is that comparing ciinical and non-ciinrcei samples
to see differences in rhe proportions who have balanced and extreme scores
does not provide information about how variation in the independent venables
is associated with changes in the probabil ity of effectiveness. The irnding
that clinical and non-clinical samples differ in the proponion '*'ho are
balanced and extreme can appear if the relationships are posttive. end it can
appear if rhe relationships are curvil inear. For example. r,r 'hen e large number
of the ineffective families are in any one of the four extreme categories,
which is f requent ly rhe case (Rodick et .a i . , 1986, for example' i t rs impossrbie
lo know whal the relationships are when either or boch of che rndependent
variables vary.
The above sludies all have quantitative data, and the evidence irom them
seems !o be contradictory. There is also a sizeable bodl' of clinrcal i i terature(see Olson el a l . , f979, pp. 7-9) that uses qual i tat ive data. end these' lata are
more consistent in arguing for the curvil inearity proposed in the ctrcumpiex
model. These studies suggest that families who have unusurliv lorv or hrgh
adaptabil ity or cohesion |end to have higher probabil itres ,rf ,Iff icuit, '-. E,. 'en
though these clinicaliy derived observations have not been letecteci in rnost of
the quantitative research. this cype of daba is as important as rhe quantltatlve
data in building and improving theories, and these clintcai inferences need ro
be given serious cOnsideration in any attempt [o resolve the controversv.
The conciusion we draw from lhese studies is that tne emDlrlcal er-tcience
currenbly available is controversiai and contradictory. We think that lt may
be possible to resolve [he controversy, bur before suggestlng a resoiutron. the
theoreticai reasons for the issue should be identif ied.
Tlworettcoi Recs,
Olson :9:theorv in dever,r1979: Olson etgenerai svstemsthe iamriv cs J, 19; I . rnd \\ 'er-'
l :1r3. o. ; { l - ;1.conesion ,- ' l 'e:i '-e is ' . 'er i ' aonvrncl i :
Beaver-s :ha-. .uggeste,i rL< lur te di i fer* :bets'een ri le Be.i3re :gnorec .i::,1prop')Sec t i l . l t :aHe . i rgues :hatf:.mii1' cl inrcranindrr-rduais ,,rr
edaptrveness :s , :
-\ rhiro :iunctronall l- ,rrlebetrveen edapt:r' . ln industr ia i rzehighl-'* adaDtrveercessive r t is , :
c iass l ines is :cteen-age memo€leen-age memksuch degree of er
Thus. prer 'rather than heisvstems anal l 'scurvr i ineantr ' . rJ.CCOUnt.
A Proposed Reso
The controtrvo chanqes erethe Beavers 3r,i i fference oi rpeciaptabriir l ' " rscontrnuum r':1it i:edaptabr i i t l ' :s . :
a qual i tat ive , r r ,
\ovember. l9SiFAMILY SCIENCE REVIEW \ovember, 198?
Theoretical Reasons for the Controuersy About the Shape of the Relationships
Olson (1976) used the perspective and insights provided b;z sysremstheory in developing and then refining and expanding his modei (Olson er al.,1979; Olson et a l . , 19831 Olson, 1986). He rvas especial iy " inf luenced bygeneral systems theory as described by Von Bertalanffy (1968) and applied tothe fami ly by Jackson (1965), Haley (1962, 1963, 1964), Speer r1970). Hi l l(1971), and Wertheim (1973, 1975)" (Olson e[ a i . , .1983 p.71). His explanar ions(1983, p. 7O-72, for example) of rhe reasons the extremes of adaptabi l i ry andcohesion creale probiems in femilies provides ir. sJ*stems oriented rarionale thatis very convincing.
Beavers (1977, f981. f982; Beavers & Voel ler . 1983), on rhe other hand,has suggested a more psychodynamic and developmentallv oriented model rhatis quite different from Olson's. There are several importanr differencesbecween the Beavers and Olson models, but if t irese more complicated aspectsare ignored and the differences are boiled dorvn to the relationships. Beeversproposes rhat the relationship between adaptabil it-"- and effectiveness is i inear.He argues lhal "adaptabil iry as a confinuum is common |o rhe work of mostfamily clinicians and research workers. Whether one is speaking ofindividuals or families, highly adaptive means capable, and rnoderareadaptiveness is deflrned as not so good" t 1983:87).
A third theoretical approach was taken by Vincent q 1966) in afunctionally oriented sociologica.l anirrysis. He theorized thar the relarionshipbelween adaptabii ity and famiiy eff ,ctiveness is curvil inear. He proposed rhat"an industrialized society character'zed by rapid social change necessirates ahighly adaptive family svstem" t 966:36), but when the adaptabil iry isexcessive it is dysfunctional. "An e<ampie or i l lustration which curs acrossclass l ines is to be found in the internal adaptiveness of the family to irsteen-age members. When familial adaptation to the needs and wancs of irsteen-age members reaches the point or degree rvhere parental controi is lost,such degree of adaptat ion becomes dysfunccional . . . " (1966:36).
Thus, previous theoretical formulations seem to add to the controversyrather rhan heip. There are theoretical models from clinical impressions,systems analysis. and sociological analyses that argue for l inearitv andcurvil inearity, and the plausibil i ty of each of them needs to be taken intoaccounc.
A Proposed Resolution of the Controuersy
The conlroversy about the shape of the relationships can be resolved iftwo changes are made. First. we need to reaiize that the differences betrveenthe Beavers and Olson models is a conceptuai difference rather rhan adifference of opinion about the shape of one relationship. Beavers definesadaptabil ity "as an emerging, ever eKpansible capabii iry to be placeci on acont inuum ranging from dysfuncr ional to opt imal" {1983:89). This means tharadaptabil ity is defined as capabil ity or adaptive capacity. and this ciearly hasa qualirative or good-bad qurrl ity to it. We agree with him rhat rvhen
November, 1987 FAMILY SCIENCE REVIEW
Figure 2: Relationships Olson ProposedAdaptability and Cohesion asVariables.
in the CircumplexContinuous Rather
IIodel Viervingrhan Categor icai
FamilyEffectiveness
High
Med
Low
FamilyEffectiveness
NIed
eddit ive mannerdirection in corl:otal distance ir:
Viewine l rcossible to takevar iables, : .n, i :t 'OOt[Ov€f S\- . Ti
: - r because: l ' .e:' . 'er iat ion in: . . tiY-rPmP l lnr ,pr , '
- r t r ' ' - .
Our t '+s,r , - 'l ls inbutrons :n (
- 'u:'rent :no,iei:e iet :onsi . r io retr ,i rgher. The .eddrt ionai lese3:l lgh as the :Oir.o i r t inn<hin r . '
"
percent i le rn: :e
-eseis ,t i e,iro::eiTectiveness :-.edeprabr i i tv . T1:shorvs lhe rn,t, ie,ietected in Trcedaptabil icl- er:o
Our inrerprend eciucarlon:.- i, i l lerent ior coir.1. ud i t a iso =imode is ios'er- :cflar mode. Ii iulhat : ravrng hraredeprabi i i t r ' . I :range of e 'ohes:trhel ' seek rn :nc. t
These sce: l:n:he eer:re: : tSroDoSe, i b l ' t ) ,s,'J ta! . r , faot :c i i i : rncs: of :he : -e: : tch.lnge in ,l ir:c::
t€searcr crcb.r: ' is€nsi i rve eaoui : ' l' is : r :bur:or:s. T' ]cCurS SeC3USe : i
have e stronge!- :
Scre: : :ber. - ! i -
High
Adaptability Cohesion
adaptabil ity is defined in this way it has to have a positive relatronshrp *rrh
effectiveness in families. We also believe that the Beavers deirruuons are so
lautological that terms have litt le value. The Beavers s)'st€m rs sayrng irtt le
more than adapting well is relaFd |o fiunctioning rvell. Olson's concept rs
very different. It is a continuum ranging from rigrd co flerrbh. and rt does
not have the dysfunctional to optimal component thar is burlt into Beasers'
variable. Thus, even though boch models use the term adaptabrliry. tbe-v are
very different concepts. This eliminaces one unportalrG p.rt oi che
controversy.
The second change suggested here has to do with finding thc pornt et
which adaptability becomes excesslve. The key lo this chang: b ia tbe *'a1'
we think adaptability and cohesion vary. We suggesl tb.at il ir pcstble lo
resolve the concroversy if adaptability and cohesion are vprcd ta Gmtnuous
variables rather than categoricai variables. They cen be opcrerindized as
categorical variables in research sUudies. but it is helpful o *rr them
conceptually and theorerically as con[inuous factors. From a sFlCOS theorv
point of view, people make slight, changes in their bondinS bqrrxiages.
coaiit ions. and closeness: and there are man)- subtle differeG h bt nuch
change occurs in ruies and system processes. The=e syst€o Fc ch.nge
in a conlinuous manner rather lhan .;ust havtng four ceqfb' d rt '*':lisharpen rarher bhan confuse the thinking of famrly scbBin a r- rhem 'rscontinuously varying Processes.
Figure 2 is an aftempr to diagram the relarrcnslir (h b proposed
using conrinuous variables. Viewed in rhis $'av, the frrLBare from
the mean on either variable, the greater the probctftf 1t s*tireness.
ThemodelalsosuggestSthatthetwoindependentnd*|c inen
Nloderate Low ] Ioderate Hign
l0 F"\MILY SCIENCE REVIES
additive manner. This means that the distance from the mean in eitherdirection in both distributions can be added together and the greater rhetotal distance from lhe mean Lhe greater the l ikelihood of ineffectiveness.
Viewing adaptabil ity and cohesion as continuous variables makes itpossible to take into account smaller intervais in the variation of rhese twovariables, and this provides the key that we think is needed to resolve thecontroversy. The reason we think this approach can resoive the controversyis because the reiationships are probably positive for mosl of the range ofvarialion in the independent variables, bur the direction changes in theexrreme upper end of the range of rheir variation.
Our resolut ion of the controversy is to suggesl thar rhe modes in lhedistribucions in Olson's model are not at the 5Oth percentile as implied b-"- thecurrenl modei and the research about the mociel. The mode in rherelationship between adaprabil ity end family effectiveness is probablv muchhigher. The eKitct locat ion of the mode can onl-y be determined wirhaddirional research that wil l allow these discriminations. bur ir may be ashigh as the 80th or 90th percentile of a random sample. We suspecr thar thisreiationship probably does not become strongly negacive unti l ebove rhe g5th
percentile in the con[emporary Western culture. Also, rve suspecr that rheIevels of adaptabil ity that are associated with the highesr probabil ity ofeffectiveness is not a large part of the total range of variarion inadaptabil ity. Therefore, our attempt to diagram this relationship in Fisure 3shows the mode to be fairly pointed. Also, we have included rhe slight curvedetected in Table 1 by dirrgramming a fairly strong relationship betweenadapcabil ity and effectiveness in the low end of its distribution.
Our interpretation of the data from previous research. and our clinicaland educationai impressions suggest thal the reiationship is probabiy somervhatdifferenl for cohesion and effectiveness. This relationship is drarvn in Figure3, and it also shows two directions in the relarionship. We suspec[ rhac rhemode is lower for cohesion than adaptabil ity, and we think rhar ir is a moreflat mode. If future data corroborate these theoretical speculations it meansthat having high cohesion is not as importanL for effectiveness as having highadaptabil ity. I l also means thar famiiies can choose among a much widerrange of cohesion and sti l l expect to effectively attain the outcomes or goalsrhey seek in their famiiy system.
These speculations seem to be an effective resolution of the differencesin the earlier theorizing and research. They allow for the curvii inearityproposed by Olson and Vincent, and they also recognize that the covaria[ionthat adaprabil ity and cohesion have with effectiveness is positive throughmost of lhe range of variation in the independent variables. The reason thechange in direction in the reiationship has not be detected rn the quantirativeresearch probabiy is because researchers have not used measures that aresensitive enough [o detect the negative covariation in the top part of thediscributions. The resolution also allows for the slight curvii ine:rrirv thatoccurs because changes in adaptabil ity and cohesion below their mid-pornthave a stronger relationship than above rt.
\ovember, 1987 FAMILY SCIENCE REVIEIV l l
Figure 3: Proposed Relationships in Olson's Circumplex Model.Fie.rre -l:
FamilyEffectiveness
FlamilyEffectiveness
High
Med
Low
High
Adaptabiiily Cohesion
The suggestions being made here about lhese relationships have several
implications for the circumplex modei. One implication rs rhat the :hree
conditions of balanced, mid-range, and extreme probabl-"* shouid ire cira*'n
differentiy on the circumplex graph than rhey have been. Figure { is en
abtempL to use the newiy suggested relationships ro drew :hese :hree
conditions so they wil l be more consistent with differences in the probabriit l '
of families being effective. This drawing uses [he same cut-off s-]-s@m Olson
and McCubbin used in their 1983 study tOlson and NIcCubbin. 1963'end in the
manual for FACES III (Olson et al., 1985) in shat there is an ettemot to hr.i 'e
reiatively equal quarti les in each of rhe four conditions of each vertabie *'hen
a random sample of American famiiies is used. The balanced condiuon rs
much higher on rhe distribution of adaptabil ity, and it includes a smdler total
amount of the variation in adaptabil ity than cohesion, The erueme condi:ron
includes a larger proportion of the low end of adaptabrlit l ' . The vaiue of
these changes in the circumplex graph will only be knorvn as addruonai ciata
are gathered to see if these speculations are consistent rvt[h data ed as this
revised model is used in therapeutic and educational applicatrons-
Before leaving [he issue of the nature of these relatronsiups. .r c3t'e3t :s
probabiy in order. The relationships thai are proposed here are undoubtecilr '
influenced by a number of unspecified contingencies. One of rhese rs
suggested in Vincent's (1966) anaiysis. He proposed :hat the emount ,ri
change that is occurring in a culture probably influences the relarr,onshrps.
For example, it is l ikely that the high level of socral change rn the trentreth
century Western cuiture tends to push the mode in the reieuonsiup betrveen
adaptabiiity and effectiveness relatively hrgh in the adaptabriity t'artaole-
Converseiv. : thar-e -r rrore itlre n'lo,ie forpornteO ,Jur:r:gfor eciaoabri:r1-ra Fizure 3.
. \ PF.OBLE\I
-\ .eCOndcommunlc: l t l r ln
l t r 2 rn : :c:ors:
f niorr'.:nrreii ' .fac:or :nooei.sugge:reci :rl:ra: :ec.e Ols,rr :dr t fe:-e:^. : : : :e: : : ,theorr 'pers!€c:Ttev ere xor . !cnSr f,.c ie !-:st!cssa:;-. i;,:::or:. :. ioptitll,r :'i i:n.lt
I t ' .vo ' - : . , : : .a-s 11:,_;u:co:T:e
- \o:e: t : i : - - . \
Moderate ][oderate
t2 F.\} I ILY SCIE}-CE REVIEW SoYcnoer. -9Si
Figure 4: Olson's Circumplex Model Using the Reiationshrps Proposed inThis paper.
(-r--oax.cr|(x _Hatr
ry nD
-D -x
l---lu lTnnw [IN:m
Conversely, it is l ikely that cuitures that are experiencing less change rvil lhave a more normal distribution in rhis relalionship. Also, it, is [kely thabthe mode for fhe relationship between cohesion and effectiveness is morepointed during periods of relative stabil ity, and during these periods the modefor adaptabil ity is probably more flat than the relationship we have proposed
in Figure 3.
A PROBLEII WITH THE ROLE OF CONIITUNICATION IN THE I,IODEL
A second problem with the circumplex model has to do with the role ofcommunication in the theory. Olson has stated that the theory has threemain factors: adaptabil ity, cohesion, and the quality of communication.Unfortunately, though, uhe practical resuit has been that it has become a twofactor model. Communication is ignored aimost ail of the time. Olson hassuggested that communication is different because it is a "facil i tating"var iable (Olson and NlcCubbin. 1983, p.+9), but rh is is not a helpfui way todifferentiate the roles of rhe three independent variables. From a s]'stemstheory perspective, the same thing can be said for adapcabil ity and cohesion.They are not outcomes or goals. Thev are transformalion processes or svs[emcharacteristics that help famiiies attain desired outcomes or goals such assatisfaction, affection. stress management, developmental tasks. consensus. andoptimai human growth.
It rvould be possible to make a case for cohesion being viewed sometimesas an outcome in familv systems because families sometimes talk about
\ovember, 198? F.\}I ILY SCIENCE REVIEW
,nl *II
"rA
AePTAIILlt l r I l rctY
ri.IIIlBI
\i/
r . t
wanting "closeness"; but adaptability is almost never se€n as an outcome' It
would be very unusual for a famiiy lo want to become an -adaptabb family"'
Communication aiso could be viewed as a family goal. but 116r of, the time it
is seen by families and family scient'ists as a facriitating prrcs' Thereiore'
the reasoning that communication should be rreeted 'Jrfferendy in rhe :heory
because it is a factlitating variable is not defensibie- [t should be rreateo the
same as the other explanatory variabies'
We suggest lhat lhe reason communicetion hes been gleated "lifferentlv
is because of the limitations rhat are inherenr in using :he crrcumpiex x..]ph.
The circumplex graph gels too confusing rvhen it rs elpandtd !o rnore iilen
rwo variables. We suggesc that rhe best sol tr t ion to:his probiem:-r l ' r - l '1t .1 f ,
profiie approach ro giaphing rhe characteristics of fernrires- \\ e ' 's -rnr to
emphasize [hat rve are nol suggesting rhar a protile s!.st€m r-epi.]ce :he
circumpiex graph. The circumplex graph is a useful cer.rce. end rt scould be
used whenever it is helpful. we are suggesting that ccdrrg 3 Profile g:ephrng
system would be helPful.
A profile method of graphing data can be us€d easily srth 6. :. or l0
factors, and rhis means that it would be easy to unciude one rlr more
communication variables simultaneously rvith adaptabiiity and cg,b€5rcn' Ir also
opens up [he possibility of including several other vanables- For exemple,
the M1VIPI uses a proFrle meihod of summarizing scores' and rt h's l{ fac'"ors'
AN EXANTPLB OF A PROFILE \IETHOD L]SED Ni
A FAMILY ENRICH}IE)'TT PROGBAIT
We have been experimenling rvich the ulc d r pronle that has
adaptability, cohesion, and a communicstion variabb I &tl of its valables'
The system we have been using is illusrrated tn Fl$aql 5 rd 6' The scores
ur" "f,o*r,
on [he vertical axis, and erght variab lr dcoufreci b1' the
horizonal columns. The scores on all the facrcc Ur rrdardized io vary
berween 0 and 100 to help make chem compenbb- 1b frtors that lre
included in rhis profile are the variables tcins nJiGd in ln ongorng
evaiuation t.ur"ur.h program evaluating tha cfuirgs of a iamiiy
enrichment program. -
The enrichment progran
-
& fuznt and' Teen
Worhshop (PAT) that was developed bf' thc lfbr eod severai other
col leagues (Burr et ai . , 1986).
The eight facrors in Figures 5 end 6 are all 6!'cd tlth a seif-report
questionnair-e called the Famitl' ProriJe 'FP\. F-{CE:i III rs inciuoe'1 :n :he
i,amity prolite to measure edaprabrlitl' end corresin, trrl dE other srs ilctors
are measured by adaptarions of several earlir instnff{rs1 Copres oi the
questionnaire are urr"ilubl" b1' rvritrng to tba SGGiOr a!t}61'' Da::' rbout
reliability and validity of the measunes of tba 3lr factors lie :'rot i'et
available. but data ur" b.ing gathered and tests of reliabrhty enci saiiciitl' rvtil
be made as soon as possibie. The con6ePcual &frnruon of the ot:er six
factors are:
Kirdne ss r:consrcierate:ror golng \A
Consensl .sseeking :nrc.riled -r '*nethoo I I I .
Conlrc I .s:o chi i , l len
- l r s;- i;
i .nd J.:rge:-
:u; 'n ln:e!-
co:nt , t t ' . ' :e
Srippo -
:::e
elcn,r :heri ; l r 'CPi iPi
, le!-rne,t :n:-ectors :eBarber enr
The , ines
lnrtrcrpereo lnroubieo iemriierrne ,rf :he ber6at rpplv ior er
Figure 5 s iSorths r:lr-e -r s6e rnother rnii
Feeotrons. lno6e iemriv -rsbtening es :he:Fres were si:Tl!o suDport lnonlcoglnrze :5aiD(ner :hougrt :bno rf :nf,rr=l:Gbel' courc i ' :ac:de i rberarei i ' t inn<ierst:.ndrng. . l
F i i ; re o =fgniir ' : i iJ.! :s lPa"renis. , ice :.f:parenrs. : re :dfar ier percerr '* j
Sore=:er. . ! r l1.t F.{}I ILY SCTE:{CE R€VIElT \o lenrer. 1987
Kindness is lhe amount famiiy members respond in a caring, loving,considerate and patient manner lo each other, especially when things arenot going weil in the family.
Consensu.s is the amount the famiiv tries to use a consensus-seeking method of making decisions. This method is somerimescalled a win-win method, and Gordon (1970) refers to it as hismethod III.
Control is how adequately the family is able to gradually transferto chi ldren the responsibi l i ty for valrous aspects of rheir own I ives.
Anger is a measure of how weil the famiiy is able to keep tempersand anger from being disruptive in the famiiy and horv rvell theyturn anger into construct ive uses. I t is adaprcd from ] Iace's i198 l )pornt of view of the role of anger in enrichment programs.
Support measures the amounc family members provide nurrurance toeach other through physical contact, helping each other achievesuccesses, and companionate interact ion. I t is used the way i t wasdefined in the Roll ins and Thomas review (1979), and includes thefactors derected in several recent factor analvses of suDDor[(Barber and Thomas. 1986).
The lines in Figures 5 and 6 show the data from [wo families thatparticipated in a PAT lVorkshop. These families were lwo of the moretroubled families that participated, and the data in these tables i l lustratesome of the benefits of using the profi le system ro help understand famijiesthat apply for enrichment programs.
Figure 5 shows the data for a family that we wil l cail the Smiths. TheSmibhs have a single mo[her and three children, and data were gathered fromthe mother and a 13 year old daughter. The solid l ine shows the morher'sperceptions, and the dotted l ine shows the daughter's. The mother perceivedthe family as being fairly good in most areas, but low in cohesion andlistening as these scores were belween 20 and 30. Some of the daughter'sscores were similar rc the mother's, but the daughter had much lower scoreson support and kindness. This information helped the workshop leadersrecognize that rhe daughter felt the family was not as effecrive as themother chought; and thab the daughter felt alienated and unappreciated. Thiskind of information helped the workshop leaders and participants find wa.v*schey could adapr the rvorkshop to help meef these specific needs. This famiil 'deliberately designed activit ies that would heip create interaction,understanding, and feelings of being included by the daughter.
Figure 6 shows the profi le of a famiiy that we wil l call the Jones. afamily that is quite differen[ from the Smiths. The Jones famiiy had twoparenls, one son, and four daughters. Figure 6 shows the scores for bothparents, the teenage son. and the teenage daughter. The scores show rhefather perceived the family to be fairly healthy, as he had no low scores.
\ovember, I987 FANIILY SCIENCE REVIEW
Figure 5: Profile of Smirh Family.
Scores
100
90
80
70
60
50
40
30
20
10
0Adaptability Kindness Consensus Anger
Cohesion Listening Control Support
He had a fairiy flat profile with most of the scores being around 70. The
mother and daughter, on the orher hand. had much lower scores. especrally
on cohesion, kindness, l istening, and support. \[ost of the son's scores were
between lhese extremes.
The profi les in Figure 6 reveal that the members oi thrs famrll- defined
their famiiy situation very differentl l ' . The rvorkshop leeders rrere concerned
about the low scores and the differences in the scores. Thel' .r lso detected
non-verbal messages and body language from the daughter that communicated
she was usualiy angry and detached. Init iall;- none of the famrll ' provided
additional information, but evenrually- the mother reveaied :hat the father was
abusive to the daughter. Arrangements were then made to have the famiiy
begin a therapy program.
IO F.{}TILY SCIENCE RE\ IE\ \ \ovember, 1987
Figure 6: Pro
5cores
idd
I \TPRO\
There :slp eridrng e prcturnF.er :Tr;rr*raor.:t:es o{(Lson e:.r--.aa.lgce.:- :a:lie :e::;e of rfhe(hc:lre res(trt ,)n '-he :arFqi:c:i'.-e DoF
Tle .i:nr:trrat :ar-e Leer!!OUO e-ii :,iur
lgrce:c: - -:5 l
JO
I r l
;{ l
l0
t0
: ( l
Figure 6: Profile of Jones Family.
Scores
r00
90
80
70
60 .r ',/i\/ r \-?r
Moth
{daptabil ity Kindness Consensus Anger
Cohesion Listening Control Support
TMPROVING A THIRD ASPECT oF THE CIRCUN{PLEX :vToDEL
There is another aspect of the circumplex model thar can be improvedby adding a profile method of summarizing data. The marn hypotheses in rhecircumplex model suggests that baranced families tend i" have higherprobabilities of effectiveness and extreme families have lower probabiiii ies(Olson et.al., 1983). Grouping several different rypes of families inro rhe"balanced" calegory is not a problem beceuse variarion in rhe middle part ofche range of variation with most variables has little value in famii-v- science.Predictive research since the lg30's has shown that the furrher people areout on the taiis of a distribution the grea[er the diagnosric. evaiuative, andpredictive power (Burgess and Cottrell, lgJg).
The limitation with fhe circumpiex model is rhat the rheorericai ideasthat have been generated and most, of [he research irbout the modei tend togroup all four forms of extreme conditions mgerher. when this is done ic
iSon's{0
30
10
November, 1987 FAMILY SCIENCE REVIEW t7
conceals variance that is potentially vaiuable. In this case, families with
certain combinations of extieme conditions may have different probabil it ies of
various types of ineffectiveness. For example, it may be that certain
combinations of the independent variables are associated rvith high
probabil it ies of undesirable outcomes such as violence. substence abuse' or
alienation, and other combinarions of extreme conditions are 3'ssociated rvith
other outcomes, such as deviance or inabil ity lo estabiish long-term
relat ionships. Rodick 's t1986) data. for example. suggest thet del inquenc\- mey
be more fiequent in chaoricaiiy enmeshed families rhan in en1' of the other
extreme types. Profi l ing these and oLher variables ma1- heip us uncierstand
horv other combinations of factors are associated rvith imput't 'r 'nI outcornes in
family systems.
The circumplex model distinguishes conceptuallv betrveen four :1-pes of
extreme conditions, but for some reason lhe theorizrng rnd rese:.rrch so flar
Fnds to ignore these distinctions. Theorists i lnd researchers :enci :o focus
on the diflerences between balanced and extreme l '3ther than iocuslng on
different types of extreme condirions. Ib is l ikel-"* thlt ecicitng 'r profi le
approach wil l make it easier to focus on ways the veriou: er'-reme conditions
aie associated differently with a number of different t1'pes of nelTectiveness.
ADVANTAGES, CAUTIONS AND FUTURE NEEDS
The two profiles in Figures 5 and 6 illustrate that prohles can ire very
helpful in a number of wayi in therapy and in enrichment Progr3tns. They
can help leaders know how to adapt or modify- programs u) me3 the needs
and goals of the participants. They also can help idenufy famrlies that
shouli be screened- out of enrichment programs because tbcy oced therapl"
and they can be used to monitor progress and evalual'e ihrl cfictiveness of
programs.
Profiles about family variables, such as the one begrg dacloped for the
PAT Workshop, also .utt b" valuable in a variety of oab hCily science
settings. For example, lhey could be used in family lift 1!fbo c'ourses'
Students beginning marriage and famill' classes could co6tb qElrnnaires
that would lead to profiles, and this could help rba !!, rmderstand
themselves and their families. It aiso could help tbn ld lb insructors
in setting goals for their course. and rn planntng fuarp I| cfr Profiles
of famif *data
also can be used in diagnosis. c*i Flr€ss. and
evaluation research in marriage end famill ' therapy'
This is helpful because it is the tanis that usurlt
information. Al.o, the relarronsi"rips rhat are of i
changes in the probabilities of ourcomes' In I
thinking is more helpful to pracitioners than F:atdemanding assump[ions such as needing ncul Ihomoscedasticity. Thinkrng about fps chr{r
There are severai other advantages of using
about famiiies. They allow scholars to focus on tb
FA}TTLY
detaions.
Gt useful
ltofiles rre*'av of
6at make
1'. andJ.re
r.1987l8
rscrare.J rlt}rHptui $'av '€ :
Proiil€s i*pendenr rarenabie nod€mFn-<afor-1'dElre' lo Ll
ril as prooierr
lrllen p:-rt:Ib s:or-i3s:ic
.ts-i€f:3:gJ
of ree-.on
;Gcs€ :::n{f,I
aJ 6pg-..pn et'tu
of snerri lt'r
hoirlcs rltiey han
ol :alll
frb eno uire \l Oe relsons
6.kL
Tb. &reilala n
crodreclAre
reourb used etro
caPefTDcrther rril
Tbe n€rt. ' ]5€ol
Y Itl
r::l hr.'oo.gia
cffccoreiyof rcii
Ecirmlo6rq6sibat'o
associated with changes in lhe probability of outcomes is a comfortable and
helpful way to rhink theoretically, empirically, clinically and educationally.
Profiles also focus attention simultaneousiy on several different
independent variables, and this helps schoiars avoid the overly simplistic two-
variable models that are so common. It also heips them identify
compensatory factors and the interaction of variables, and it is more
conducive to thinking in terms of strengths (Stinnett and Defrain. 1985) as
weil as problems.
When profi les are used, scholars also tend ro find themselves thinking
with stochastic or probabil istic models. They reason that changes in profi les
are associated with changes in the probabil ity of different ou[comes. This
type of reasoning is helpful in the contemporarv scene because it allows
precise thinking but avoids rhe host of problems that come with differences
of opinion about causation, de[erminism, free agency, and positivism thal are
concerns of some scholars (Osmond, 1981; Harre ' , Davide, and Nicola, 1985;
Thomas and Wilcox, 1987).
Profi les were first used in the famiiy field by Burgess and Wallin (1953),
but they have not been used extensively. There are a number of
psychological instruments Lhat provide profi le data about individual
characteristics, but we are only e',vare of [wo instruments that provide
profi les of family variables. The tv'o are Moos' (1983) family environment
scale and the Marital Inventory (MI tHolman et al., 1981). We sugges[ that,
for bhe reasons cited above, greabe. use of profi les would be very helpful to
the f-reld.
The development of effective, standardized instruments rhat profi le
family data wil l be a long-term process. It wil l involve a careful revierv of
theoretical models which can direct the process of selecting the variables t 'hal
should receive atlention, and it, will include the use of factor analy'sis to
determine redundance and orthogonalitv of factors. The instruments rvil l need
to be used enough to get norms and determine their reliabil i ty and validity,
and experience suggests they wrll need to be developed and then revised
before they will be very effective.
The next step is to engage in more of lhe systematic analysis of
concepts, theory, and measurement processes that' Ied Olson to identify
adaptabil ity and cohesion as two central variables in famiiy processes. This
analysis wil l help us identify a manageable group of other variables that wil l
provide complementary data. Next, the need wii l be to develop lnstrumenls
that effectively measure [hese factors, develop norms for them. :1nd acquire
evidence of reliabii i ty and validity. Fortunauely, the theory in the field and
rhe technology for instrument deveiopment are now sophisticated enough that
it is possible lo move ahead with this t.vpe of work at a rapid pace.
\ Iovember, 198? FAMILY SCIE}iCE REVIEW l9
REFERENCES
Barber, B. K., Thomas, D. L. (1986). Dimensions of Fathers'and \ Iothers'Supportive Behavior: The Case for Physical Affection. Jottrnal oi trIarriageand the Family,48, 783-794.
Barnes, H. L. and Olson, D. H. (1985). Parent-Adoiescent Communicat ion andthe Circumplex NIodel. Chi lc l Deuelopment, 56, ' t38-+'17.
Beavers, W. (1977). Ps-rchotherapl o.n: l Growth: A Famii : Srste,ns - \nprtoch.New York: Brunnerillazel.
Beavers, W. R. (1981). A systems modei of famil l - lor f rrmrl l ' therapists.Journal of Marriage and Family Therapy, 7, 299'307.
Beaverq, W. R. (1982). Healthy, Nl idrange, and Ser-erei1- Dvsfunct ionalFamilies. In F. Walsh (Ed.), Normal Famill ' Processes. -\erv York:Guilford Press.
Beavers, W. R. & Voel ler, M. I \ . (1983). Family models: Comparing andcontrasfing the Olson circumplex model with the Beavers s]'stems model.Family Process, 22, 85-98.
Burgess, E. W. & Cottrei l , L. S., Jr. (1939). Predict ing success or 1 'oi iure inmarriage, Englewood Cliffs, NJ: Prentice Hall.
Burgess, E.W., & Wal l in, P., (1953) Engagement and J[arr icge. Phi ladelphra: J.B. Lippincott Company.
Burr, W. R., Smith, S.W., Marshal l , C., and Henry. G.. ' l9E6 Parent ond TeenWorhshop, Provo: Family Living Center at BYU.
Clark, J. ( f984). The family types of schizophrenics. neurot ics. and "normals."Unpublished Doctoral Dissertation, Universit-'- of \Iinnesota. St. Paul.
Craddock, A. E. (1983). Family cohesion and adaptabrl i tv rs iactors in theaetioiogy of sociai anxiely." AustraLian Journoi o1'Ser. Ilarnoge & Familv,. r , 181- 190.
Garbarino, J. , Sebes, J. & Schenbach. C. 198{, . Famrhes at r isk fordestructive parenr-child relations in adolescence. Ct:lci De,-eiopment. 55,17.1- 183.
Gordon, T. (1975). Parent Effectiueness Traintng. \ew York: Plume.
Haiey, J. (1962). Famiiy experimencs: - \ nerv t1 'pe of iamri l ' expenmentatron.Family Process, 1, 165-293.
F.{,} l ILY SCIE\CE REVIE\T20
Irby. J. i9{Srratton.
Ir ley. J. ' i96{fclarlr P.ccr
Irrre'. D.. Da" nr.6odtk:;oa
R:3: lSa:cr S':;'lc
lne-'r. T. B-Femri;.. Theti { : .
D.D
D.R 1If.ounuurn I
RH..r?ibuger E{bzp1. Ber
*rat in scfils. [A: G
D. H.3i!T
D. H. 11&al-', P.ccc
D. H_. IL .rgs. : I
D. H.. PrD H- Olson.5l i l0s.
D. H.. RDrrul .r,I}d l
D i{., sDn:.ri r,r.oqrF .r.no rp
Haley, J. (1963). Strategies of psychotherapy. New York: Grune andStratton.
Haley, J. (1964). Research on family patterns: An Instrument .\'Ieasurement.Family Process. 3, 1l-65.
Harre' , D., Davide, C., & Nicola, D. (1985). l lot iues and rnechanisms: An' introduction to the psychology of action. London: Methuen and Co. Ltd.
Hiil, R. (1971). Modern syslems theory and the family: A confrontation.Social Science Infbrmation, 1 0, 7 -26.
Hoiman, T. B. & Burr, W. R., ( . 1981) Beyond the Beyond: The Grorvth ofFamily Theories in the 1970's Journai of Marriage and the Family 42:729-7 4t .
Jackson, D. D. (1965). Family rules. Archiues general psychiatry, l2:, 58g-594.
Nface, D. R. (1982). Close Companions: The fuIarriage Enrichment Handbooh.Continuum Publishing Co.
Moos, R. H., & Moos, B. S. (1983). Family environmentai scale. In E.Filsinger (Ed.), Marriage and fanily assessment: A source book for lamilytherapy. Beverly Hills, CA: Sage.
Olson, D. H. (1976). Bridging research theory and application: The triplethreat in science. In D. H. Olson (8d..), Treating ReLationships. LakeMills, IA: Graphic.
Olson, D. H. (1985). Commenlary: Struggling with congruence acrossbheoretical models and methods. Family Process, 24, 203-207.
Olson, D. H. (1986). Circumplex model VII: Validation studies and FACES III.Family Process, 25, 337-35L.
Olson, D. H., McCubbin, H. I . , Barnes, H., Larsen, A.. N1uxen, M., & Wilson,NI. (1983). Families: What makes them worh. Los Angeles: Sage.
Olson, D. H., Portner, J., & Lavee, Y. (1985). f'ACtS III Manual and Norms.D. H. Olson, Family Social Science, University of ll innesota, S[. Paul, ]IN55 108.
Olson, D. H., Russel l , C. S. and Sprenkle, D. H. (1983). Circumplex model ofmarital and family systems. Family Process, 22, 69-83.
Oison, D. H., Sprenkie, D. H., & Russel l . C. S. r1979). Circumplex model ofmarital and family sysrems: I. Cohesion and adaptabiliry dimensron. family'fypes and appl icat ions. Family Process,18, 3-28.
November, 1987 FAMILY SCIENCE REVIEW 21,
Osmond, M. W. (1981). Rethinhing fami$ sociologv ;',o^ operspectiue: Applications and implications. Paper presented et
rod ical-critical\CFR Theory
cConstruction and Research Methodology Workshop. -\[rhx'eukee.\ \- I . October.
Rodick, J. D., Henggeler, S. W., & Hansen, C. L. 1986'. - \n esaiuarton of thefamily adaptability and cohesion evaluation scaies and the circumplexmodel. Journal of Abnormal Child Psychologl, 11. ;;'6;.
Roi l ins,8.C., & Thomas, D. L. (1979). Parental support . pn$'er. end controltechniques in the socialization of children. Contempror' tkeories about
the family (Vol 1). New York: \Iacmillan Pubiishrng-
Russel l , C. S. (1979). Circumplex model of manni end femri l 's1-stems: I I I .Empirical evalualions with families' Famih Process. I8. 29'15.
Speer, D. (1970). Family systems: ) [orphostasis end morpnogenesis. or ishomeostasis enough? Family Process. 9, 259'258.
Sprenkie, D. H., & olson, D. H. (1978). c ircumplex model of mantd s_vstems:An empirical study of ciinic and non-clinic coupies- Journai cf l[arriage
and Family Counseling, 4, 59-7 4.
Sfinnett, N., & DeFrain, J. (1985). Secrets oi StronS Fentltes. Boston:
Little Brown.
Thomas, D. L., & Wilcox, J. E.from philosophy of scienceSteinmetz (Eds.), HandboohPlenum.
Vincent, C. (1966). FamiliaMarriage and the Family, 28,
(1987). The rise in family theor-r': Cririquesand hermeneutics. ln !l- Sussman & S.on Marriage and ttu Foumrly. \erv York:
spongla: The adapuve frrnctroo. Journal of29-36.
\ \ - i tn :hepopuier 'ro:1ceIbti ivrdua,l cc r:tlttu3Lrons a3.n5Gcr:be ro'A irbr iam:l:es xtlt-erence ln Ibteracrrons offrnrll' reiarrond.ivrdual-ievelb lend ro 'Jr1965 . in an eu
.{lthoughb ps1'choiog:ca
{el ' r ' : r :s: C
Von Bertalanffy, L. (1968). Ceneral systems theota. J{er Yorb Brazrller.
Werthheim, E. (1973). Family unit therapy and tbe stience of t1'pology offamily systems. Family Process, 12, 36 f -376.
Werthheim, E. (1975). The science and t]'pology of famrl-'- s)'stems II.Further theoretical and practical considera[ions. Fonu\ P'pcess. 11, 285-308.
22 F.{} I ILY SCIE\CE RE\ ' IEW \ovember, 1987