MATERNAL DE PRESSIVE SYMPTOMATOLOGYAND FAMI LYCOHES IO N:
A COMPARISONOF PERCEIVEDFAMI LYCOHESION INMOTHE R- CHI LD DYADS
by LoriMcDonald
A thes is sub mi t t e d to the School of Grad uat eStudies inpartial fulfilmentof the requirementsfor the degree of
Maste r of Science
Departmentof ps y ch olo gy Me mo r i al Universi tyof Newfoundland
December, 19 9 4
St.John's Newfoun d land
.+.
NationalLibrary01Canada ~~ naIJoMre
DirechondesacQlJisibcnsef desservicesbibIiographiQues
""--
~~l
The aut hor has granted an irrevocab le non-exclu sive licence allowingthe National Ubrary of
Canada to reproduce, loan,
distr ibute or sell copies of his/herthesis by eny meens end Inany form or format,making thisthesis avail able to Interested persons.Theauth orretai ns ownershipof thecopyright in his/herthesis.
Neith erthe thesisnorsubsta ntial extrac tsfromit maybeprintedor otherwise reproduced without his/ her permission.
ISBN 0-612-01888-1
Canada
L'auteu raeeeerdeune lice nce irrevoca ble et non exclus ive permeltent
a
10 Bibllothequo nationale du Canada de reprodulre,preter,distribuerou vendredescopiesdesa these de quelque manl~!re et sous quelqueform~que cesoit pour mettredesexemplairesde ceUe these it la disposition des personnes Interessees.L'auteurconservela propr ll!tl!du droit d'auteur qui protege sa these.Nila thesenides extraits substantlels de celle-cl ne dolvent etre lmprimes au autrement reproduits sans son autorlsatlon.
Abstra ct
Perception s.')f famil y co h es ion wer e compare din two grou psof mothe r-ch i l d dy ad s. Two mea s u r esof cohesion (th e FES an d FACES III) were obtaine d from 17 mothers who repo rted dsp re es ive symp t oma to l o gy an d their children (ages 11 to 17 years). Ea ch fami l y waD matche dwith acommuni t y comp a ri s o n familyonagean d sexofch i l d, familyAize ,and si n g le ver sus du a l parent sta tus. Threehyp o theses werete sted: (a) target m,)thers woul d perceiv e lower faIll1ly cohesion tha n wo u ld compar i s on mothers; (b) childre n of target moth ers would perc eive lowe r family cohesion than wO".ld the chil d ren of compa r i son mot h e rs; an d {cl ther e would be mor e congrue nce between ch i l d r en ' s ra t i ng s of cohesi on and the ir mothe rs' ratingsinth e targetgroupthan in the compa rison group . The results of a multivari at e ana l ys is ofvariancerevealedthat moth ersand chi ldre n in tbe target group did not differ in reportedcohesionfrom mot h ers andchildreninth e comparis on group. Howe v er, the group means we r e in the di re cti o n predicted in the first two hypotheses. In ad d i t ion, th e co r r e l a ti o n co efficients between mothers' and chi l d rs n' B cohe s i o n scores did not differ betws en groups for eit h e r
The results are discussed wit h respe ct to th e relationship betwe en cohe s i on , maternal depressive symptomatology, andch i l d adjustment.
i i
Acknowledgements
I would like totha n k my co-researc h er. JacquelineGoodwin.
for he r supportand an c o ura g eme n t (esp e c i a l l y for the lat e night pep cal l s) in co mpleting this research. Additional thanksare ext.nded toJ:JYIlluperviaor. Dr. Christine Arlett, an d theother memb ersof DIyth e s i scouall1t t e e,Dr.Mary Courage and Dr. Ray Pe nny. Also, special thanks are due to r:ry friend s,Igna c e Gysbersand De rekGoul d .SystemsAnaly sts, for thei r in valua bl e assistance in pre p ar ing this doc ument. wo uldlike to thankDr.Jo hnEvans an dDr .Al Kozma fortheir recoIlltnendat i on s f01· st a t is tica l analyses an d Dr. Andre e Liddell for al l owingme to use the Psychology Clini cfordat a coll e ction . For their l1s sis ta n c e in contacting fami li e s I would l ik e to extendIllyappreciationto Paul Hillyardand the section leade rs of Scouts Canada, Da n i t a Power of the St. Jo hn's Boy. and Girls Cl ub. To d d McDonald (BantaD\ A Developmental Coa ch)and Dr.Has sanKhalil iand other memb.rs of thePs y c h o l ogy and PsychiatryDepartmentsof theWaterford Ho s p i ta l . Ad d i t i o n althan ks areex tend e d to DeannaStokes- Sul livanofth e Lifestyles sectioDof theEveningTele gram and th e staf f of Cable 9 for their adverti" ing assi stanc e. Finally, gre a t ns t thanxs are ext ended to the fami li e s for givingfreely of their ti me tocon t ri b ute to thi s resear c h.
ii i
(
TABL EOF CONTENTS
Abstract • • . • i i
Acknowledgements ii i
List of Tables vi
Li s t ofFi gure s vii
Introduct ion • • • • • • • • • • • . • Int e rpers o n al context of depression • • Parentingbehaviours of depres sedmothers • • . • Maternal ve xeue paternal depre ss ive
symp t om8 tology • • . . • . . . • .• • . • . 9 Mater::la l depressionand adolescence • •• • . • • 10 Maternal depression,depressi v e symptomatologyand
distress • • • • • • • • • • • 11
Mat ernal psychological distress and ch ild adjust-lIIent • • • • • • • • • • . •• • . • • 13 FamilycchasLcue Defin itio ns • • • • • •• • • • 17 Relationship between family cohesion and child
funct ioning• • . • • • . • • • • • • • • • • 21 Depressedpersons' perceptionsof familyco h e sion 27 Co n grue n c e of perception ofcoh e s i o n among family
me mb e r.s . • • • • • • • • • • • • 31 The presentstudy • • • • • • • . • • 37 Method • • • • • • • • • • • . • • • . • • '11 Subjects • • • • • • • • • • • • • • • 41 Targetsampleselectioncriteria 41
Conurn.m i ty comparison sample 42
Sociodemographicfactors 43
Measures • • • • • • • • • . . • 43
Maternal functioning • . • 43
Familycohesion . • . • . • 44
Procedu re • • • • . . . • • . • 48
Targetsamp le • • • • • • • 48
Commun i t y comparisonsample 51
Results • • • • • • • • • • • • • 53
Prel iminaryanalyses • •• • S3
Family cohesion • • • • • • 53
Statistical analy s e e • 53
Statisticalassumptions 56
Testsof hypotheses • • • 57
Te s t sof statisticalpower 62
Descriptionof data 63
Di s c u s s i o n 70
References 78
iv
AppendixA
ss
Appendix B '8
Appendix C 100
Appendix0 102
AppendixE 10'
Appendix F 107
AppendixG 111
Appendix H 11.
Appendixr 118
Footnotes 121
Listof Tables
Table Page
Sample characteristics accordingto groups 54 Source table for theMANOVA • • • • • • • 58 Means and standarddeviations of FES and FACES II I coh esionscoresfor IlIOthe rsand children in each group. • • • . • • • • • • • • • . • . • • • 59 Comparisonof FEB andFACES IIIcorre lation coefficientsbetwe e n mothersand ch ildren in each group. . • • • • . . • . • • . . • • • . 61 Mean FES and FACES III cohesion scores formo t h e r s combinedacross gro up s and chil dre ncombined across group s • • • • • • • • • • • • • • • • • 65 MeanFES andFACES IIIcohesion scorsscombined aerosamo thers andchi ldrenin eeob group. • • • Fj7
vi
Li s t of Fi g u res Figu re
Age dist ributionof chi ldrenbygr oup Mekn FES cohe s i onscore s ofmothersand chi ld r e nbygroup • • • • • • • • • • •
Page 55
.8
Mean FACES 1:1:1 coh e s i o n sco re s ofmothers
andchil d r enbygroup • • • • • • • • . . • • • • • 69
vii
Introduction
Mater nalDepressive Symptomatology and Family Coheeion:
ACo mp a r i s o n of Perceived FamilyCohesion in Mother·child Dyads
In the last two decedee, numerous researchers have co n d uc t e d studies ofmaternaldepressionwiththe objective of identif yingits relationshipto ch i ldadjuetme ntand fami:!.y cohesion. Studies ofchildrenof depressed parents originated from the need for a psychiatric contro l group in otudiss of the offspring of achh:ophrenic parents (Samaroff, Seifer, &.
Barocas, 19 8 31 Sameroff, Seifer, &. Zax, 19 8 2; Seifer, Sameroff, &. Jones. 1981). These st udies led to th e serendipitous conclusion that children of depressed parents are at an equiva l ent. if not highe r, risk for distur banceas the childrenof schizophrenic parents (Downey&.Coyne, 1990;
Grunebawn &. Coh le r , 198 3 ; Samerof f et al., 19 8 2 ; Sameroffet al.• 19 831Seifer et al., 1981). In fact, Sameraff et al.
(1982) summarized their fin di ng s by conclUding tha t : It•• •i f one were to choose a diagnostic group wherechi ldrenweremost at risk, it woul d be neurotic depression rather than schizophreni a" (p. 58).
Research findingshave sinceindicatedth a t of fsp r ingof par e nts with depressive disorders have higher rates of physical,psychological, and behaviora l difficu l t i esth ando of fs p ring of paren t s witho ut depress ion . In pa r t icu l ar . cross~section al research ha s documented tha t ch i l d r en of
parents with unipolar de pressionappear to be at a greater r iek for a wide range of problems that include affective disorders,in t e r n a l i z i ng and externalizing pr o blems,deficitR in social and aoademio oompetenoe, and physical health problems (Beardslee, Be mp o r ad, Kel ler, &: Klennan, 1993;
Billingss,Moos, 198 3 , 19 8 5 0 1 Blatt&.Homann, 19 92 1 Downey&.
Coyne, 19901 Fendrioh,Warner , &:Weissman, 19 9 0;Jaenicke et a1., 1987; Lee &:Gotlib, 198 9; Riohters &:Pel legrini, 19891 Weissman et al.,1987). Two extensive lo n gi t u d i n al studies of children of depressed mothers, the Rochester Longitudinal Study (RLS,Sameroff et a1.,1983) and the NationalIn sti t u t e of Mental Health (NI MH) Colo rado Collaborative Studies (Davenport, Zahn-Waxler ,Ad1and&:Mayfield,1984;Gaensbauer, Harmon, Cytryn &. McXnew, 1984) suggest that maternal de p r e s s i o n is re l a t e d to increasedri s k for child adjustment dif fio ulties (seeals o Coyne,Kessler, st al ., 19 B7 ).
Childrenof unipolarparen tsare at an increasedriskfor affec tivedisorder , haVing th r e e ti me s th erat e of affective disordersoverall, andsix timesthe rate ofmajordepressive dis ord e r s when compared with oontrolchildren(Coyne,Burchill
&.Sti les , 1990 ; Do wn e y & Coyne, 1990 ). Other studies have
do c ume n t ed an eleva ted frequency of prob lems wi thi n this populationof childrenin infancyand early chi ldhood(Cohler, Gallan t. Grunebaum . Weiss, " Gamer, 1977; Do wn e y lit Coyne, 19 90 ; Sameroff et al. , 19 B3f Seifer et al., 1981 ), in the primaryschool years (Bea rds leeet a!., 19 B3 ; Fisher, Kokes,
Harder , &Jon e s, 19BO; Hammen, 1991;Hammen , Ad r i a n, et a1., 19B7; Hammen,Go r do n , et a1 ., 198 7a , 1987b; Jaenicke et a1. , 1987; WeIner, We I ner, Mc Cra ry, r. r.eonexd, 197 7) and in adolescenc e (Hamme n , 1991 ; Kau ffman , Grun ebaum, Cohl er , &
Game r , 1979 ; We i s s man , 19 8 3 ). Di ff i c u l t i e s ha v e becom e apparent through self-report, and th roug h th e re p orts of peers, teachers, an d parents.
Co ns i de red collectiv ely , th e s e studie spro v idepersuasive evidence of a link·between pare ntal depre ss ion and ch ild diffi cul ties. An ev enbleaker picture of the adj u s t me n t of childrenof depr e ssed mothers ie crea ted by research that showed tha t, despite improvemE:ont in mother's dep rf;lss! ve symptomatology, difficultiesin childadj u s tm entwe r e sti ll evi d e n t 10 mo nths later (Lee and Gotl i b, 19 89; 1991). Bi l l ing s and Mo o s (19 85 c) also ob se r v e d t.ha t.parenta wi.th remi tt ed depre s sion con t inued re p ort adj us t ment difficultiesin theirchildren in theirl-year longitudinal study . The s echi ldrenshowedlittle improvementand continued to ev i d e n ce more dysfunctionth a n childrenofco nt ro l paren ts.
These fi nd i n g s in di c at e that the influence of matern al depre ssionon child adjustmentmay be of a prolonge dnature, Intervention ot h e r than simply tr e a ti n g the moth er ' s depression may be required in order to alleviate chi ld difficulties .
The pres ent study explored mot h e r s ' and children' s perceptionsof th ei r family envirorunent:in famili e sin whi ch
mothers rep o r t ed depressive sympt omatology. Alarge body of l iterature exists tosupport the notion that one aspect of family environment , cohes i on, is assoc i a t e d with several aspElcts ofwell-being in family llIemb ers. Xl: i8noteworthy, th e re fo re , th at depee ased patients, as well as their no ndepr e ssed SPOUS ElS, have been found to pe rceivetheir family environmentsasbeing lower incohe sionthan matched co ntro l s (Billings , Cron k i te&.Moos, 19 831 Mitchell, Cr o n k i te&.Moos , 1983). Interestingly ,onlyone studyhas examinedperception s of family cohe s i on by the children of depres s e d mothers (Fe n d r ichetal., 1990). Th erefore, the primary interest of the current study was to measure children's pe r c e p t i o n s of co h es i o n in families in which there lived a mother who reported depressiv esymptomatology. Thisint e r es t waski nd l e d hy previous correlationalresearchthatdocwnen teda numberof in t e re s t i n g relationshipsbetwe e n per c eived fami l y cohe s ion and childou t c o me.
The followingrevi ewexamines the in te r p e r s o n a l context of depression,wi t h a descriptionof the parenting behaviours characteristic of depressed mothers. Th e la t te r de s c r i p t i on is provid e d to delineate the interpersonal mean s by which ma t e r n a l depression may affect offspring. A disc ussi onensues which focuses on the importance of research concerning maternaldepressivesymptomatology and adolescent populations.
The rEllationshipbetween per c e i v ed famil y cohesionand child out comeis subsequentlyexaminedto illustratethat cohesion
is a positive factor in the overall adjustment of children.
A studyconducted by Fendrich et a!. (1 9 9 0 )is then discussed, with empha si s onth e finding th a t theof f s p r i n g of depressed parentsrep orted lowe r familycohes i o n thandid th e off s p r i n g of nondepre s s ed paren ts. Th e presen t paperargues that this perception of lowcohesionmay berel atedto parentingstyles characteristic of depressed mothers. Additional research findingsare present ed sugg e s t i ng thatdepressed mothere also percei v etheir families to havelo w cohes i on. A dis ouss ion of the degre e of concor-danc e in perceptionsof cohesi o n among family membersfollows,with emphasisplaced on a compar i s o n of clin i c a l an d nonclinical popula t i o ns .
Interpersonal con tex t of depressio n
So me researchershav e insiste dthatth e targetof study sh o u l d be th e interpersonal sy s t e m of th edep re s sed person, not jus t the person' sintrapersonalstruc ture (BurgeoXHammen, 1991;Coyneat a1., 1990,Coyne,Kahn, oXGo tlib ,1987 , Coyne, Kessler at a1.. 198 7) . The interpers o na l co n tex t of depression refers to con d i tions suc h as fami l y str ess an d perceived familysuppor t that may precede .precipi t ate , co- occur with, or followfromdepression.
It hasbeen sugges tedthat bothclinicaldepressionand childbehaviourproblemsareoft e n expreesione of a distressed interpersonalcontext (Coyn e , Kahn , &Got lib, 1987) . On the other hand, numerous rese archershavefocusedon the negative imp a c t of depressionon close relationships. In particular,
one sugg e stionis that maternal depressi:m in t erf e r e s with the formation and maintenance of positive family re l a t i o ns h i p s (Coyne, Kahn, &Go tl i b, 1987). It is argued in the follo wi ng discussion that depressed mothers of t e n exhibit distinct parentingbeha v iours th a t may ha v e serious implications for the mothr.u;-chi ldrelationsh ip.
Par e n t i ng behaviours of depressed mothers
Studies of the impact of the parent ing behaviour of depresse.dmotherson children's adjustment have focused on the negative aspects ofli v i n g with a depressed parent. In sum, the findingssu gges t tha t parenting of depressedmothe rsis related to children's depressive symptomatology, school- rela teddif fic ulties , and social eompeeenee (assessedbype e r ratings andpr o blem solving abilities).
Weissman (1 9 8 3 ) found th a t al l of the social relationshipsof dep rel!llsedmothersWl!lre impaired wh e nco mp a r e d to thoseof ama tc h e d contro l group,and tha t of these, their rela t i o n s h i p s with their ehildren were the most impaired.
Optimally, the mother -childtie is charB'.ets rized by continuing maternalavailabi lity which is,ac cording to Cohle randMusick (15183), "appropriate to the stage of the child 'B own soc ial and cognitive development" (p o 149) and Cilso sufficiently fle x i ble tc adaptto the child'schanging developmen t a lneeds. Fisher eta1. (1980) suggested th a t "d e p r e s sed patients may displ ay hi gh de g r e e s ofnon-acknowledgementsuch that theydo not intera c t meaningfull ywiththechild~ (po354).
Depressed mothers a:oethoug htto experience -difficul t ies in the parenting role that ref lect the symptoms of th e i r disor de r" (Do\omey " Coyne, 1990, p. 61). Th e symptoms of maternal depression (i.e.,withdrawa land self-focus)may have serious implications fo r the mother-child relat.ionship.
According t.o Gizynski (1985) , the depressed mother is wi thd r a wn , apathetic, an d na rci s s i s t i c a l ly pr e o c c up i e d with her own melancholywhi ch makes i t difficul t forhe r to meet th e emotionaland physical needs of an infant. Anumber of researchershave found that individualsexperiencingunipolar dep ress ionare characterized by a heightenedstate of sel f- focused at tention (Anthony, 19831 Ferster, 1973; Ing r am , Lumry, Cruet, " Seiber, 1987). One effect of this increased sel f- focus in depressedmothe rs wo u l d be arel at ive la c k of awarenessand responsivenessto the emotionalneedsof thsi r children. Ac c ord i n g l y , i t wo ul d be expected that when depressed mothersare unable to re s po n d congruent ly toth e needsof theirchildren, th e childrenexperiencetheir mothers ae distant fromtbe m and-emot i ona lly un a va i l a bl e" (Cohle r &;
Musick, 19 8 3,p.148). Prolo n g e d self -focus, therefore, and the consequent unavailabili ty of thG parent may be one mechanismthrougb whicbdif f i cul ti e s inchil dren'sadj ustme nt ar e established.
Cross-sectionalstu diesexamining the ef fectofmaterna l de pre ss i o n on eu c i a l be h a v iour have revealed th a t during in teractionswit btheir young children , de pre o s e d mother sshow
poorer social behav iour and affective expression.
Spec i fi c all y, they have been found to express lit tl e positive af fe c t , respond more slowly, less contingently , and less consistent lyto the children, an d be less likelyto ut ilize infant -directedspeech (Bettes, 19 8 8,Cohn,Matias, Tronick, Conn ell , &.Lyons-Ruth, 1986; Cohn& Tronick, 198 9 ; Hops et al., 19 8 7 ; Lyons-Ruth,Connell, Grunebaum, & Bo t e i n , 1990).
Other studies hav ere p or t e d that depre ssedmothers aremore irritabletowardtheirin f a n t s than arecontrolmothers (Cohn et a1.,1986; Downey&:Coyne, 19 9 0 ).
Depress ed pa r e n t s themselves report that they are less affectionate, more emotional ly distant, irritable, and preoccupied,and that they experiencedifficultycommunicating with their child ren (Weissman & Pa y k e l. 1974) . using the ParentalAcceptance-Rej ection Qu e s t i onn a i r e , Colletta (1983) found thatadolescentmotherswho were depressed tendedto b fl hostile, indifferent , and rejecting of their children. Additionally, in a study of depressed mothers and their adolescents,Weissman (1983) found that depressedmothersha d considerablediffi cultieswith their offspring. These mothe:cs wereonly moderately involvedin th e i r children'slives, had di f f i c u l t y in cOIlln'lunicftting with the children , reported considerable friction, and expressed a loss of affection towardthe ir adolescents. In a similarvein, Sameroffat al.
(1 98 2 ) describeddepressedmothersas being less spontaneous,
less vocal, less positive, and more distant than co ntrol mothers when playingwith the ir4-month-old ch i l d r e n.
These findings support the vi e w tha t there is a generalized behavioral deficit associated with materna l depressionwhich affects the parenting of depressedmothers and their children's adjustment. It is suggested in the following section that th e parentingcebevaoue of depressed mothersha s impli c a t i o n s for childad justment.
Maternal vex'suspat ernaldepressive symptomatoloqv Th e interest of thepr e s e nt studyfocusesspecificallyon maternal depressivesymptomatology,as opposed tothe broader concept of pa r en t al depressive symptomatology which mu s t obviouslybe extendedto includepat ernS',1distress . Paternal depressivesymp t oma t o l ogy dOBSnot appear tobe as salient a fa c t o r in children's adjustlllent as is matsrna l dep ress ive symptoma tology.
In a study of 37 families in which at least on e biologicalparenthad a past or presentdepresBivedisorder, Keller et a1. (1986) found th atdepr e ssi.onin the mother was more st rongl y associated with impairment in adaptivE'!
functioningin the children than waadep ress ionin the father. Adaptive functioning was assessed by an AdaptivB Function Rati ngwhich included scores forde al i ng with relationships and school. Ad di ti o n all y, a seriesof papersentitled"Child Competence and Psyc h i a t r i c Ris k" present a mase of data concerningchildren whoseparentshad a psyc h i a tr l c dis ord er
10 and were includedin the Un i v e rsi t y of Ro ches t er Ch i l dand FatnilyStudy (URCAFS). Th ese papers report on somElof the re la t ionshi p s fo und between child compe tencean d fami ly and par ental va r i a b les (Fisheret a1.198 0 ;Harder,Kok e s , Fisher,
&Straus s,1980, Kok e s ,Ha r d e r , Fisher, .. St:J::a uss ,1980). One
of the find ings to emer g e was that the occ u r r en c e of depres s i on,in c o n g r uo u s af fec t,an d withdrawal in moth ers,but not in fathers, was related to lower co mpeten ce in their ch i ldr e n.
While the impo r t a n ce of pa t erna l in f l uenc e on child adjustmentisackn owledged , paternaldepr s sBlvesymptoma tol ogy ap pea r s tobe of les e si gn i fica nc e than is matern ;ll dep r e ss ive symptomatology. More over, since epidemiological studies indicat e that depre l!llsion is mo re prevalent inwomen thanin men (Wetzel. 1984 ),mothersweresel ecte d as thefo c u s of th e present study.
~@p ressi onand adolescenc e
Thexosea ec b haspo i ntedtoa widerange of deleterious effe ctsof parentaldepres s ion on childrenof all age levels. Although the effects of parental depress i on on child adj u s tm e n t may varydependingon the age of thech i l d, i t is clear that impl icat i onshavebeenfoundacrossthe age range from in f a n c y to adoles cence.
Numel;Ousre s earche r sha venot e d that thepr o b l e ms between depressed mothera an dtheir of f s pri n g become exac e r b ated as children:mo v e into the adolescentyears (Cummings &.Da vies,
11 19 94 ; Gizynski,19 8 5 ; Johnson&::Irvin, 1983). Ado lescence is a perio d of ra p i d cognitive, social. and physica l chango;
consequently, theadaptational ca p a c i t i e s of th e indi....idual may be stressed which increases th e ir vulnerabi lity to st r e s s ors such as maternal distress (Bur t , ccnen, &.Bjorc k , 19 8 8; Feldma n, Rubeilste in. &: Rubin, 19B81. Mo r e o v e r, adolescence is a period in which theindividual struggles to achieve separationand individuation from the family. The process of becomi n g more independent oft e n involves di f f i cul t behaviours such as testing of ;)arental li mi t s, re b e l l i o n and defiance (Gizynski.1985; Johnson&:Irvin. 19 83). It follows that the depressedmother's difficultieswith the parenting rolewoul dbe in tensifiedby the maladaptive be h a v i o urs of the adolescent, thereby inc r e a s i ng th e "opportunity for pathological interaction"between themother-adolescentdyad (Johnson&:Irvin, 198 3. p,118). Becauseof thepo t e n t i a l ly stressfuldemands ofthe adolescent period on bothmother sand adol escents, thepr e s e n t studyfocuse d on th ep~e-adolesc ent and adolescent ag e ra ng e.
Ma t e rnal depression depressive symptomatologyand distreaz.
Researche r s ha v e argu ed that diagn os a ble de press i o n is conceptu a l ly and emp i r i c ally di s ti n c t fromboth de p ressiv e symp toms and milder, more transient fo rm s of ps y c h olo g i c al distress(Coyne,1994;Fe c hn e r-B a t e s, Coyne,&Sc.:n.wenk , 19 9 4 ).
Some of th e ear l y studies of depres sion reli e d solely on me a sur es of self - repo rteddistre ss. or hospita l re c o r de when
12 describi ng the ir ta r g et samples asdep r essed. These studies sho uld be interpreted cautious l y to refle c t thei r nondi agnostic sampling methods. Specifically. the samples should be considared as representative of distressed rather than depre s s e d indiv idua ls. Unclea r di a g no sis hille made it diffic u l t toin t egra t e fi n di n g s of ea r l ie r studies of ch i l d adjustment and pare n ta l de pre s s i on wi t h more inv e s t i g a t i on s that have taken a more ri g orous approach (e.g., structu re d, di a g n os t i c interviews) in estab li shing the presence of dep ression.
Sole reli a n c e on sel f - repo r t measureswhe n descri bing nonclinical samples as depress e d is considere d problematic be c a us e personsmay achieve elevate d scores on sel f- rep o r t questionnaires without meeting the criteria ne c e s s ary for a di a gno s i s of depres sion . According to Co yn e (1994 ) diagno s a ble depres sionis le s s cOll1lllon thanthe presence of el evat ed scores in self - repor ts such as the Beck Depres s i on In v en t o ry (BDI; Beck, Ru s h, Sh a w, & Emery , 197 9 ). In non c linica l popul a tions , a con s i de rablepropo r tionofeleva ted scor eson se lf- r epo r tquestionnaireshas be e n showntore fle ct nonspec i f ic psycho logical dis t res s rather than cl inical depression (Fechner-Bateset al.,19 9'1.). There fore, rel ianc e on cut- o f f sc ores on self -reportinventoriesas cr iteri a for inclus i o n of subje c ts in ade p re ss ed samp le wi l lre s ul tin a considerab lenumber of falsediagnoses ofdepression for
13 individuals who can be better de s c r i b e d as exp e rie n c i n g psychological distress.
Maternal psychologicaldi s tressandchi l d adjustment The co n cep t of maternal psychological distress has also been.used to explore difficulties in children's adjustment acrossdi f f e r en t maternal diagnosticcategories (Co yn e , 1994,
Lee &.Gotlib, 1989; Le e &:Gotlib, 1991). Lee and Gotlib
(1989) examined the relationship between childadjustmentand maternalpsychopathology in a study offo u r groups of mother- child dyads. Maternal psychiatric status was determined throughtheuse of a diagnosticsemistructuredinterview. The gr o u p s included Cl i n i c all y depressed psychia tr ic patients, nondeprsssed psychiatric patients, nondepressed medical pa t i e n t s , and nondepressed nonpatients. One of the issues addressed in th i s study concerne d the specificityof child adjustmentdifficul,tiesto maternaldepression. Accordingly, two relevant hypotheses were tested: the depression- speci ficity hypo t h e s i s (Le., that only th e childre n of depresse d patients wou ld show adj ustmentdifficu l t ies)I and the psycho logica l dis tr e s s hypothesis ( i.e.I that thech i l d r e n inbo th psychiatricgroupswouldshow greaterdifficu ltythan wouldchildren in the nonpsychiatric groups). Th e re sul t s showed th a t the children of depressed mothers ha d significantlymorein ternal izingproblems thandi d child renin the nonps y chiatri c groups (i.e. , nondepressedmedical and nondepressed nonpatientgroups). Thorewereno differenceson
14 any of the internaliz i ng prob lems betwe e n the chi ld renof de p r e s s ed and nondepressed ps y c h i a t r i c patient mothers. Th i s finding suggested that child internal izingdifficu l tieswere not spocificto diagnosable dep ress ionbutra t h er wererelated to maternal psychological di s tr e s s.
Other re c e n t studies supp o r t th e argument th a t childhood adjustment dif f icultiesare rela tedtono n s p e c i fi c materna l distress and not just to diagnosable depression. In investigation of the ro l e of ne g a tiv e se l f-co ncepts in childhoodin ind uc ingvulnerabi lity tode pre ssi o n, Jaenicke at al. (19 87 )compa redchildrenat risk because of mat ernalmajor affective disorder (e.g., unipolar de pre s s i on and bi po lar disorder) withchi l dr en of medical lyi l l and contro l mothe rs on var i o u s indices of cognitions about the self. They found th a t child ren's negative self-concep t was significantly related to both maternal uni po lar depression and bipola r disorder. Thus childrenof psychologica l lydistressed mo t he r s di s pla y ed more negative views about th eIRs elv e s tha.n did chil d renofmedic ally i l l and controlmothe rs. Anc.therstudy by Hammen, Gordon et al . (1 987b) also compared chil d ren' s behaviour problems ,schoolfunction ing, andsocialcompetence .. cross these four groups. Th e re s u l t s showed that the child ren of unipolar depres s ed moth ers exhibited more imp airmentthanchi ld r en ofbipolarmothe rs,althoughchildren in th e af fec t ive disorde r groups had th e hig h e ilt rates of psyc hiatricdiagnoses . These findin gssu ggestthatchildren'llI
15 diagnostic sta tus wa s related to maternal psyc hological distress.
zn a (i-mo nth follow-updesign, Bur g e andHamnlen (1991 ) investiga te d the qual ity of ma tern al interacti on and task invo l v eme n t 88 predic t o r e of measure s of the chil dr e n' s affective di a gn o ses, echo o l beha v iour, and academi c competence. Theyexaminedasamp l eofchi l d r e n (ag e e B to16 years) con sider ed to be athighan d lowrisk for dep r e s s ion froma pr evious study (HillDIlIe n , Adrian at a1.. 198 7) . The chil dren we re cona i dered to be athigh riskfor dep r ess ion bec a us e ofmatern a lde p ressivesympt o ma t o logy(a ssessed bythe BOI) and/ or ch ron ic stres s. Low risk for depre s sionwas determined on th e basis of the absence of hath of these va r iables. Mothersand theirchildren were askedtodi scuss a common topic of disagreement and attempt to reach an agr e emen t. The discuss ion was vide o t a ped and ma ter n a l utt eranc es were transcribed and coded on the basis of two ca te go r ies: (a ) t:askin v o lve. men t orconununica ti o nclar ity and (b )affe ctivequality. Subsequel1tly.twoscores were compu t ed whichincludedthe degreeof positivematerna l cODUlluni cation (e.g•• positive feedback and confirmatory re marks ) and an ind ex of task involvemen t (e.g.• stating op i n ions. giving directions. andasking forfeedback or information ).
The resu:'.ts suggested that both nega tive or cr i tical interactionquality and lack of maternaltask involvementwere associated wi t h children's depressive symptomatology and
16 malad a p t i v e school behaviour at. the time of fol low-up assessment (Burge&.Hammen, 1991). The y found tha t mother s' depressive symptomatolClgywas associated with difficulty in maintaining ta s k fo cu s. whereas maternal ongoingstressfu l li f e conditions were as so c i ated wit h negative quality of cOmJllunications with th e child . The authors proposed that chron icstres smay impai r adi s t r e s s e d mother's patienceso thatsheis easily angeredduring discussionswi thher child th a t irvolve confl i c t andmore likelyto engage in cr itical interaction wi t h the ch ild. The authors added that the results point to the role of psychosoc ial fac tors, and impaired paren ting inparti cu l a r, in increasing children's risk for diaorder. Furthermore, th e y stated that: "Rather th an viewinga mothe rasexh i b iting defects that directlyLead to chi ldren' s risk, it is more help f ul to conceive of th e lIlother as caught in a vicious cycle of enviro:nmental and Interpersona1 circumstances that imp a i r her ability to functio nad apt i v ely with her children " (Burge&:HalIUtlen, 19 91 , p. 17 9 ).
Additionalstudieshave docume n teda cor re la tionbetween materna1 depressed mood and depre s sion in ch i ld r en, in thv ab sen ceof di agnos a b l ematernal depression (e.g. ,Lefkowitz&
Tes i ny , 1985; Seligmanet aI., 1984). It is alsonoteworthy thattheparentingdifficultiesthat ha v ebeen identified in stud:le s c.f de p r e s sio n are notspecific to mothers who have depressed mood. Similar pa r e n t i n g difficulties havebe e n
17
found in mothers coping with several different streasors, including divorce (Hetherington, Cox & Cox , 1982) . Th e s e findings suggestthatth epa r en t i n g difficulties observedin depressedmothersmay be cOll'lll'lonto mothers whoare di stressed because of other stregBOrS(e. g. ,marital conflict)that may becor rela t e s of depression .
Researchers who studyth e impact of maternal depressive symptomatology on offspring have come to recogni ze the interdependencebetween distressedpersons and their social context. Grotevant and Carlson (1989) ha v e proposed that:
"Problemsno longer reside within thein d i v i d u al; they may be symptoms of other problems in the famil y or the broader environment" (p.7). Accordingly, the link be t we e n materna l depressive symptomatolosyand the individua1'8 mostimme d i a t e 80c ialenvironment, the family , has emerged asa viablearea for resea rchinvestiga tion.
Fa mil y cohesion: Definitions
Thedime ns ionofco he sion has been incorporatedinto the wo r k of pr o f e s sio n a l s of many discip1ines, including psychiatrists, psychologists, family th era p i s t s, fami ly sociologists, group the r a p is t s , and anthropologists . It is no t e wo r th y that each disciplinere c o gn h e s its own distinct de fi n i t i on of cohesion (for a review of definitions of co h e s i on by the res p e cti v e di s c i plin e s seeOlson et al ., 19 83 ) . Wi t h i n the are a of family research, a number of theoretical modelsha v e been developedto delol,:,ribeand asses s
18 fami lyfu n c tio n i n g , and each placesconsider ab leemphas ison theme a s u r e me n t of cohesion. The multidiscipl inarynatureof family therapy and research may provide partial explanation bothfor thenwnberof models tha texist,and for thedistinc t theories andde f i n i ti on s of co h e sio n thatacc01IIpany each mo d el (Skinner, 19 87). Du e to th e association of the wor d
"co hesion" wi t h var ious de f i n i t i o ns , i t is neces sary to provide the specific de f i n i t i o n {s ) to beuse d in the pr e sen t study.
L'AtateandBagaro z z! (19 93)defin e d cohes ion as It• ••the emotional b01l.dingthat familymembershavetoward cne aoother and the degreeof individua l au tonomyapers o n experiencesin thefamil y system" Ip,168). Th i s definitionis in accor dance wit.h a re c e n t Three-dimensional Circumplex Mo d el of Family Functioning (Olson,19 93 ) , inwhich thereare fourle v el s of cohesion ....",nging fro m"disengag ed " (verylow) to "separated "
(low to moderate) to "con nected" (modera te to hi g h ) to
"enmeshed" (ver yhigh ). Th e Family Adaptabilit y and Cohesion Scales- Th ird Edition (FACES I I I; Olson, Por t ne r, &Lavee, 1985 ) hasbe e n developedto measure co h e s i on asit is defined by th eCircump lexModel of Fami ly Functioning. A hypothesis derive d from thismodel is that optima l familyfunctioning occursin the "ba lanced"or centra l le v el s of cohesion{i.e., se p a r a te dandconnected} whereas the "unb alan ced"or extreme le v el s (i.e., disengagedand enmeshed) aregenerally eeen as pr o b l e matic. From a theore t i calpers p ective, ext remeson this
"
continu um ar e tho u g ht to be patholo gic a l in the aenn e tha t extreme cohesion is tho ugb t topre v ent dif f ere n tiat i o n and in d ivi du at i onof the chi ldwhe r ea s lo w cohes ion isthoughtto be relat ed to exces sive au to nomy. es trangement from family memb ers and a li mi te d commitme nt of fami ly members anot he r (Minuohi n , 1974 ).
Re c e n t stud ies have cont ribute d to an ongo ing deb ate conc ern i n g the th eoryof curvilinea r i t yasi t relat e s to the Cohesion dimens i o n of the FACES III (for more deta i l ed discus sionsof thi s debate se e Cl u f f,Hic ka , &.Mads en , 1994:
Gre en, Ha rri s, For t e . &.Robinso n, 1991; Hamps on. Beave rs, &.
Hu lgus,1988). 'I'he debat eis focus e dontheargumenttha tthe FACES IIIisnota tru e measure of th e Circwnplex mode l . In a samp le of 2440 male me mbe r s of theVi rginia Nati onalGuard and their families , Gr een at a L, (1.9 9l ) exami n e d the relation shipbetwe en FACES IXXand two vali d an d reli a bl e measure sof family an d individ ual wel l - be i ng. Their results sh owe d that FACES III Co h es ion was relat ed to mea sures of well-be ing in a l in ea r manner. More o v er, th e Adaptability Subscale was unrelated to themeas u r e s of well-be ing , thus balancedfami l i e swerenomo relikel ythanothe r family types to report highsc ore son the wel l -being measures. Gree n et a1. (1 9 91) conc lud e d that FACES IIImay not be an accu ra te measure of the Circumplex Model. In fact, in the scoring and interpretati o n manual tha t accompani e d th e PACES IIX, Ol s o n and Tiesel (1991 ) reported that the FACES III Coh e s".on
'0
di mensi on was l in e ar an d that hi g h scores "are rein t erpret e d as 'very connected'" rather than erunellhed (p . 1). Olson (1991)con ceded th a t , on the basis ofprev i o us research (e.g.• Gre e n at a1. . 19 91), a l inear rel a t i o n s h i p exists between FACES III and individual and fam i l y well-being. Th u s , hi g h FACES III scores represent Balanced fami ly types and low BeOl"SS representExtremetypes.
Morerecently,Cluffat a1. (1994) proposed that th ere arene g at ive , confounding variablesattheextreme l y hi g h ends oftheCohesion dimensionwh i c h are endo rse d by"d y s f unc t i o n al subjects". Thus,dysfunction~.lfamilies are expected to have eit he r low or extremely high Cohesion eccz-es , creating a
"pseudoc urvi linea reffect " (p.467 ). However.no evidencehas been put forth teo support this pr.op o s i tio n by Clu f f et al.
(1994); thereforethe present study will followthe direct ion set fo rth by the most recentre c o mmenda t i ons of Olson (1991) andthe findi ngsof Green et al. (1991). In ehcz-t., FACES II I Cohesionwillbe treated as a lillear measure in thepresent study ,with high scoeearepresentingbetterfamily func t i o n i n g th a n low scores.
Proponen tsof a dif fe rentapproach to famil yfunctioning purpor t that behaviour is "•••a joint func t i o n of the person an d the en v iro nme n t " (Sk i nne r , 1987 , p , 433). According l}' , the social climate of the fami ly , asmeasured by theFamily Environment Sc al e (FES ;Moo s &Moos,1986). has been take nas a means of characteriz i ng unique human environments on
21 dimensions of relationship, personal development, and system maintenance (Skinner, 1987). Within the relationship dimension, the authors have defined cohesion as "... the degreeof commitment, help, and support fatnily memb ersprovide for one another" (Moos&Moos, 1986,p , 2). Compared with the definition of cchesion provided by L'Abate and Bagarozzi (1993), this definition soems to focus less onthe closeness of family members on an emotional level and more on tho helpfulness and r:!ommitment of family members toward ons another.
The present study investigated cohesion in light of these two definitions and their respective models and measurement devices. Although the models differ with rl!lspect to theory and measurement. they share one common.element: the recognition of cobesLcnas an important construct of family functioning.
Relationship between family cohesion and child functioning As the following review demonstrates, research findings have supported the role of family cohesion as a positive factor in child development, regardless of age level. A large segment of this research has been concerned with investigating the perceptions of family cohesion by parents or their offspring in clinical and nonclinical youth samples. Brown and Mann (1990) used Pearson product moment correlations to evaluate the relationship between familycohesion, as measured by the FACES (Olson, Sprenkle, &: Russell, 1979), and
22 adolescent decision-making. The results indicated that adolescents in more cohes ivefamilies snowedgre aterdecieion- making competencethantho se in le s s coh es i vefamilies .
Nume rousstudi e sha v e sh own thatpar e ntalpercepti onsof high co h e s i o n are aeao c ia t e d with more favourable fam i l y adjustment underad v ers e circumstances. Forinstance, hi gh FEScohes ion,asreported bypa rents . hasbe enassociatedwith parents' ab i l i t y to cope withchildrenwith autism (Br i s to l , 1984) an d mental retardation (Nihira. Meyers, &:Mink , 1980) . Parental perceptions of coh esi vefamily environmentshaveals o been l in ked to the promotion of cognitive development (Gs rtne z y, 1987) and general competence amongprimaryschool children(lun a t o, 1999;Garmezy,1997 ). FeldJnan et a!. (19 9 S) proposedthata su p p o r t iv e family mi l i e u is likelyto confirm the adolescent 'sself ima ge as a competent, wor t h wh i l e, and desirablein d i v i d u a l.
The re sults of the study by Billings and Moos 1199Sc) showed that chi l d r e n of parents with remit~ed unipolar depressionshowedsignifi c antlylowerlevels of psychological, physical . and behavioural pr o b l e ms than did children of parents with nonremit.t ed depression. Furthermore, these parents reported significantly higher FES cohesion than did parents with nonremitted depression.
Collect:lvely, theabovestudiespointto a relationship between hi g h family cohesion and po s i t i ve outcomes for children and adoll!lpcents. Several studieshave documented a
23 relationshipbetween lowfamilycohesionandthe presence of familydisturbance . For example, parental perceptionof low FES cohesionhasbeen fo und in familieswi thabusive parent s (Perry, Well s, "Doran , 198 3 ) , euce eenee ab u s e r s (Fi lst~ad, McElfrelJh, s, Anderson, 19 81 ) , and medical patients with chronic phYliical symptoms of obscure etiology (Waring &.
Russel l, 19 6 0 ). Th e re e:ul t s of the foll o wi n g studies have revealed a linkbetween undelidrable ou t c ome s for children and theirperceptionsof the family ss having lo w FES or FACES cohesion (Bu r tat a1.,1988:Feldmanet a1..1988; Fandrichat a1.. 1990; Moos&.MOOB, 1986:Walker &:Greene. 1987).
Walker and Greene (1987)investigatedthe roleof family cohesion (measuredby the FACES II)in protectingadolescents frompsychophysiological symptoms associated withne g a t ive li f e events. The adolescent sampl e ccnafsnedof12 3 males and females at an outp a tient medica l clinic. The resu lts indicatedthatadolescentswho perceived their familiesas lo w incohesion generallyreported more symptomsthan tho s e wi t h high cohesion, exceptwhenthe la t t er had a high incidenceof negativ e li f e ev e nts. Furthermore, theresults showed that la c k of family cohesion may be associated with hi g h symptomatologyevenintheabsenceof negativel if eeven t s.
A study by Burt et a1. (19 8 8 ) also uncovered a re l a t i o n ship between adolescents' psycholog i cal functioning (i.e ., depression, anxiety and self-esteem) and their perceptions of FES cohesion. Cross- sectional analy ses
2 .
revealed tha I: cohes ion was signif i cantly and positi v e ly correlate dwi th ado le s ce nts' sel f - e s t eemand si gn ificantlyand
nega t ive l y co rr elat ed with their de pre ss i o n and anxiety.
similarly ,in acOlllDl.uni t ysample of ear lyadole sce nts,Feldma n at al. (19 8 8) found through reg ression an:sl ys ell th at adolescentsrep orted JI\Ore depress i v eaffe ctifthey perc eiv ed the irfamiliestobe lowin FACES IIIcohesion. Consist e n t fiDdings were revealed in a study of family risk factors in
depressed pare r.t s an d theiroffspring(Fa ndrich et al.,19 9 0). In thi s study,Ch.t"qu a reanaly s e s showed thatthe childrenof depressed parentsweresignificantly morelikely to report low FACESfamily cohesion an d were more likely to be diagnosed wi thmajordepression an dconduc t disor de r th anwere children of nond e p res sed pa r en tI!.
KleiDJD.an. Han dal, Enos. Se a righ t. an d Ros s (1989 ) investigat edtherelat i o nshi pbetwe enFES subsc a les.including co hesion, andad o lescent distre s s in a sample of96 6 high sc hool student s. Adoles cent distress was de f i n ed aa the presence of a psycho log i caldisord er and wa D ass e s sed usin g the Langn e r Sympto m Su rvoy (LSS, Langner. 1962) and the GeneralHealthQue sti onnaire (GHQI Goldberg,1972in Kl eirunan et al.,19 8 9 ). Significant negative cor re l a t i o n coefficient s were obtained between the r'ES Cohesion Subscale andboth measureaofdistress . Specifica l ly, familycl i ma t e s that were perceivedashighin cohesionwere relatedtoleaa distress
25 and better adjustment for adolescent.s of co.ll ages and both
A number of researchershave concentrated on tl'oecoh e s i o n
of thl!l family unit in efforts to uncover theco r re l a te s an d
predictors of adolescent suicidal behaviour (Asarnow.Carlson,
&. Guthrie, 19871 Asarnow &; Carlson, 1988; Asarnow, 1992,
Garrison, Addy, Jackson, McKeown, &. Waller, 1991; King.
Raskin,Gdowski, Butkus,&.Opipari. 1990;Miller.King, Shain,
&. Naylor, 1992; Mitchell &. Rosenthal, 1992; Rubenstein,
Heeren, Housman, Rubin, &.Stechler, 1989). Investigations into the correlates of suicidal behaviour in populations of
psychiatric youth have found a strong relationship between suicidal behaviour and youths' perceptions of their family environments as unsupportive and lacking in cohesion (Asarnow, 1992;Asarnow. et al. , 1987; Asarnow& Carlson, 19881 Miller at aI., 1992). The findings of these studies point to a possible protective influence of cohesion with respect to adolescent suicidal behaviour. It has been suggeated that suicidal behaviour may occur coincident with an experience of isolation (i.e., lack of cohesion) withinthe family system (Miller et al •• 1992).
In a study of children of deprll:ased parents. Billings and Moos (1983) found that exposure to enviro1Ul\ental stressors coupled with a less cohesivefamily enviro1Ul\ent (as perceived by the parent) was related to a higher rate of child disturbance (measured by the Health and Daily Living Form
26 developed for this study). They concluded that streasore and an unsupportive family environment placed the childrenls health at risk, and speculated that low aeeeee and high support lIIay function as "protective" factors. This conclusion was supported by the finding of a much lower rate of disturbance among children exposed to low strese and high family support. One year later. these authors found that parental reports of low cohesion were found to be as strongly correlated with child functioning as was the severity of the parents' depressive symptomatology (Billings&Moos, 19850).
They concluded that a relative lack 0':: family stressors and high cohesion may be the cornmon elements that buffer the effect of parental depression on children's health and adjustment. These authors surmised that the children of depressed parents may be doubly disadvantaged since social resources may have indirect stress -bu.ffering effects in a'idition to direct positive effects on functioning. That is, since a.n association has been found between low cohesion (as perceived by the depressed parents, not the children) and nonromitted parental depression. the potential stress- buffering effects of family cohesion are precluded.
In short, based on reports from both children's and parents' perspectives, the presenceo~parental depression is related to perceptions of low family cohesion and negative outcome for offspring. As several of the above studies illustrate. a growing body of research has fooused on
27 assessing familycohesion from the perspective of adoll::tBcents.
Exploring adolescents' perspectives of fami1::cohesion is an important effort because other research overwhelmingly points to the importance of family cohesionin children' sdeve:Lopment at al lage levels. The presence of family cohesion h&1Ibeen shown to be 8ssociat€ld ~1ithbetter overall adjustment of family members, while its absence has been correlatedwith adverse familycircumstances and di3turhance (e.g ., childhood anxiety and depression. stress-related psycho-physiological symptoms, and adolescentsuicidal behaviour, for psychiatric and nonpsychiatric youth). Thus, if viewed along a stressor- protector continuum, the presence of family cohesion may be viewed as a potential "protector" within the context of negative life circumstances. Similarly,the absence of family cohesion may, in and of itself. function as a "stressor"
within the family system. When viewed in this light. family cohesion may be one of the factors which lessens the negative impact of parental deprEi.!lsion on child adjustment (CW!UlIings&.
Davies, 1994). The potential for an adaptive role of cohesion in families in which there is a depresAed mother points to the importance of this construct as a topic for research.
Therefore. the prosent study was designed to provide further information about perceptions of family cohesion in families in which mothers experienceddepressive sytllptomatology.
Depressed persons' perceptions of family cohesion Being connected to otherswi t h i n a supportive social
2.
ne t ...ork is essential to well- being(Wetzel. 1984). According to Coyne (1976), depressed individuals' poo r social in t e r a c t i ons and alienation of acquaintances and intimates erodetheir sources of social support. It is noteworthy that depressed patIents report that they have fewer close relationsh ips and le s s supportive fami ly relationships (Bill ings at al., 1983; Bil l ings &. Moos, 1985b; Wetzel Ii<
Redmond, 1960); while depressed patients who ha v e more numerous and more supportive Bo c i a l re s o u r c e s (family, support,and friends) have shown"be tte r- t h an- e x p e c t e d " post- trea tment functioning (Billings Ii<Moos. 1985a, p. 151). Accordingto Wetzel (1984), pexaon -environment interactlona maybe critical in understandingdepressionsincenei t he 't' the person nor the environment can be accurate ly assessed in isolation. Empirical support for the salience of pereon- environtnentinteractions wasdocumented in a stUdy in which FEBcohesion was f.ound to be the most discriminatingvariable separating a depres sed group of women froma nonde p ressed group(Wetze l «Redmond, 1980).
Since fam i li e s are seen as a major suppo r t system for individuals, examin ing the family cohesion of depressed parents using the perception of familymembers (as opposedto outside observer te c hn i qu es ) is of clinical, as wel l as th e ore ti c al, into res t. Billingsand Moos (1983) found that depresse d parents perce ived le s s cohesionin their families th an did control parents. In two studie s of 42 4 depres sed
adults at intake (Billingu et a1., 19B3) and one year later (Billings &. Moos, 1985b), depreDsed-parent families were characterized by less cchee Lcn enen were control families. At one-year follow-up, the family environments of parents with remitted depression were comparable to those of control families on the Cohesion Subscale of the FRS (as rated by the parents), while in contrast, the family environments of the parents with nonremitted depression were much lower in cohesion (Billings" Ml')os, 1985a). In a comparison study of community couples versus couples in which one of the partners was clinically depressed. Mitchell et a1. (1983) found that depressed patients experienced more strees and perceived a less positive f<Unily environment. Furthermore, the nondepressed spouses of the depressed patients werefOU~'1dto experience greater levels of strain and lower levels of family support than were control subjects. Family support was measured by the Family Relations Index (FRI) which is comprised of the Cohesion, Expressivenee'e and Conflict Subacales of the Family Environment Scale (Moos&:Moos,1996).
Mitchell and Moos (1984) conducted a longitudinal study of the relationship between stress and support (as measured by the FRI) in a sample of 233 clinically depressed patients. The results indicated that individuals who reported more severe depression also reported more negativeevents, fewer positive events and lower levels of family support.
FurthetrmClre, increases in level of strain (a composite
3.
vari ab le construc ted from stressorB that included medical i1lnes 8 .children' .health problems,work stressandnegative phy sical home environment) werecorrelatedwithdecr easesin tamilysup por t,"venafter controllingfordepressionand SES. Theabove fin dingsare consistentwith the workofRoehl andOkun(1984.) who found th at lo w FES cohesIon waa relatedto more life lItrellsorsand mor e depr t 8 S a d MOod in a sampleof fema l e stude n ts. As th e lev e l of perceived FES fami ly coh e s i on ro s e . the number of negat i v e lif e events wasless stronglyas s o ciated wi t hdepression ( i.e •• the estima te dslope of th e rogre s s i o n of depress i on symp toms on negat iv e l if e eveutsdecroased). Researche rs havedoc umen t e d thatpeople wi thhigher le v els of familycohesion arele s8li kely to show depress ive symp t o ms (Bil l ingsr..Hoos. 1985a.198 5b )and that social supportbuffers the effectsof life atrellls inin lllta n c."
where depressiondoes occu r (Bolahanlr.Moos, 1981). Clearly.compluodto nondepres sedindividual s, de p r esse d individualshave been found to experiencemorestre s sand to perce ive les s familycohesion. These stud ies poin t to the po ssibi l i ty of a st ress- b uffe ri ngeffectof fami l yco h e s i o n in thefac e of lif estresso rsanddepre ssion . Int eres tingly. t.h e spouses of depre s s edindi viduals al so repo r tmore£Itress and lossperceived coho s ion than do the spouses of nondeprs s s ed in d ivi d u a l s . It is noteworthy th a t re s earchers who ha v e fo c uee don thenatur e of famil y cohesion in fami l ieswit h a depres s e d pa r e n t have assessed perception s of cohe s ion
31 primari lyonth e basis of parental reports, and have ignored the child' sperspective. l"or this reason, thepresent study was designed to obtain perceptions of family cohesionfr o m children , as wel l as from their distressed mothers. The question th u s arises: Do the offspring of mothers wh o are experiencing depressive 8Y1l\ptomatology also perceive low fami ly cohesion?
Congruenceof perception of cohesion among family members Animportant characteristic of family environment ia the extent to which £EWl11 y members agree in their perceptions (Moos &.Moos, 1986). As the following review illustrates, numerous re s e a r c h ers have focused on gathering individual familyme:mbers' perceptionsof family cohesionover the last de c ad e . One common finding to emerge :':rom studies or.
nonclinical famil ies is that adolescents tend to report significantlylower cohesionscores thendo either of their parents. It is noteworthy that the lack of concordance between family members perceptionsis not confined to cohesion ratingsbut is found formost ratings of adjustment or family fun ct i on i n g.
Hamp s on, Beavers ,and Hulgus (l9B9)di s cus s e d the results of an unpublishedstudy conductedby the first author. This study utilized the observational (Bedve r s Interactional Competence and StyleScales) and Self-Report Inventory(SFI) of th e Beavers Systems Model taeevers&;Hampson , 19 93 ). The SFl is an instrument thatprovides a Competencesceeefor ea c h
32 family me mbe r and a Cohesion sco r e which is used as estimate of Family Style. According to Beaversand Hampson (1 99 3 ) coh esion "... addresses clo seness, togetherness, and te n d e nc i e s to en joy time togethe r , as such it is an approximation of some of the major familythemes related to style" (p,77 ). Strong positive correlationsbetweentheSFI Competence Scale and the Cohesion Sub s c a l e sof FACES II , III, an d the FES have be en reported (Be a v e r s s, Hampson, 1990, Hampso n at a1.. 198 8 ). Th e results indi cated th a t clinic fami l i e s showe d less within-familyvarianceon family style ratingsand SFI cohesion than did the nonclinic familjes.
Beavers and Hampson (1 9 90) discussed the results of another study of a large samp l e of no n c l i n i c a l families.
Unfortunately,they did not providea ref erencefor the study, nordid th eyreporttheirst a t i s t ica l finding s. Nevertheless , their discussionindicated that therewas lower variabili ty in SF I sc ores in families observed andra t e d as more competent than in less competent families. These findings are inconsistentwith the findings of the previous reportwhich re v e a l e d less within-family variance in SFI scores in the clinic and less healthy families. The exception to this in consistencywas shown in the higher degree of Within-family variance on SFI scores for mothers an d ad olescents in the least healthyfamilieein th e study byHamps o n et 01.1. (1989). Clearly, th e in c onsi stenciesin the previouslystated results point to the need for further inves tigation into the
33 pat t e r n i n goffami l y scores. both forclinicaland non cl i n i cal populations .
A:ind i n g ofpar ticula r interest inth est u dy by Beavers and Hampson (1990) was that on D'Ir.lst of the scales, adolescents typica lly rated their family's competence as significantly lower than did their parents. rn thecaBSof less competent.
cl in icalfamilies, there wa s morecongruencein familyra t ing s betweenparents and their adolescents. The authorsconcluded tha t: "••. an ado lescent ra t er's perspective can tel l us re l a tivel y li t tl e in and ofit s e l f , the rater may be a typical adolescent (Bcoring the family lowe r ) from an adequate or midrangefamily,or a "clear"perceiver of familydysfunction in a borderli nefamily" (p. 65).
Anumber of researchers have obtained similar findings concerning adolescents' lower reports of family cohesion relativeto their parents. Olson at a1. (1983) measuredthe viewsof adolescents and their parentein a familystudy using the FACESII. The re sul t s indicatedthatboth male and female adolescents re por t ed si gn i f i c a n t l y lowe r levels of fronily cohesion than di d eit h er of their parents. The authors interp reted this finding in li g b t of the notion that adolesc ents viewth e i r familyasle s s cohesive inanattempt todiff ere n t i a t e themselvesfromth e i rfamily . Asth e ir focus of j.denti tyduring thisperiodshiftsaway from theirfamily andto wa r d theirpeergroup, they must minimizethosepositive aspects of familyl ifs that encouragedependency (Ol s on et.
,.
a1.. J.~83). In a similar study of 281 community families, Nol le r and Callan (1986) used the FACES II to assess perceptions of fami l y cohesion in adolescents and their pa rent". Th e y found that pnrentsre p o r t e d the family as more cohesive than did their adolescer.tsand thiseffect became more pronounced as the age of the adolescent increased;in the youngest age group of adolE:l"lcents (L e . , 13~yaar-olds), adolsEJce ntand parent perceptions were similar.
Overall, the results of the above studies suggestt.hat the levelof agreemen t in perceptionsof cohesionamongfamily members may be related to the observer-ratedpsychological atatus of the fami ly (Le., clinical versus nonclinical healthy verSUD non-heal thy). In addition, the le v el of agreemen t may also be re l at e d to the presence of psyohopathologyin an individualfamily member. Finally, the presenoe ofan adolesoent in the familyhas alsobe e nfo und to contributesignificantlytodi s c r e p a nc i e s in perceptionsof oohesion among family members.
In a stu~y investigating perceptions of fami l y interactions in depressed and nondepressed university stude nts, Oliver, Handal, Finn, and IroJrdy (19B7) found that depressedstudentsperceived thei rfwniliesmore negativelyon the FES Subscale of Cohesion. Moreover,nondepressed siblings ofdepressedstudentsrated theirfamil y as significantlyless cohe s i v e t:han did nondepresslild siblings of nondapressed studenta. This raises the issuethat unfavourab le perc e p t i o n s
35 of familysupport may notbeE.pecificto the individual wh o is depressed. The state-specificity hypothesis proposed by Oliverat aL, (1987) maintains that negative perceptions of the fam i l y by depressed inciividua l s would be Been b1.Sa reaction to their own condition of being depressr..d, while negative perceptions of the familyby nondepressed members (i.e. , offspring)wo ul d be seen ae a reaction toth e prellance of a deprnssedmember in the family. In both instances the negative perceptions of famil y interactions are seen as reactions to the specificstate of depressio n, althouqh they maybe mediated throughdifferentmechanisms. In other words, fo r the mothe r, i t ispar t of her symptomatology to feel isolated and not supported , whereas the child ren may be re s p o n d ing to their mother' s la c k of interaction. For di f fere n t reasons both ma y sh ar e the perception of Lack of cohesion in the family.
A sing lestudy ....as foundin which perceptionsof family cohesi on in ebe offsp ring ofde pre s s e d parentswereexamined (Fend r ic h et &1., 19 9 0 ). Using the first version of th e FACES , th einvestigators fo und th a t re l at i ve to chi l d renof nond epre s s e d'pa r e n t s, th e childrenof depxe aaedparentswere significantl ymorelikely to report lo we r cchee Lcn scoresand were xnore li k ely to be di a g n o s e d with major depression.
Notwithst'1ndingth e val ue ofth i f\ findingwithrespect toth e perce p tion sof cohesionin the offspring of depressedparents, ce r t a i n methodol ogicalweaknessesexist in this study.
First, the study was based on offspring between the ages of 5 and 23 and consequently, i t does not take into account developmental factors that may affect perceptions of family cohesion. For instance. studies have shown that cohesion scores decreased as the age of the adolescent respondent increased (Papini.Roggman, &0Anderson, 19!H) and also became less similar to those reported by their parents (Noller&0 Callan, 1986).
Second, the otudy was bas'9d on disproportionate sample sizes. Specifically, i t compared 153 offspring from 63 familieswith one or more depressed parent with 57 offspring from 26 families with neither parent depressed. I t is possible that having a larger sampleof familieswith erie or more depressed parents may have increased the likelihood of obtaining significant differences between the groups.
The Fendrich et a1. (1990) study ex i s t s as ava l u a o l e research effort because i t is the first study that closely eXBJIlined the perceptions of cohesion in children "fi th depz eee ed parent(s). The present study was designed with a similar interestI however, certain refinements incorporated into the methodology. First, in order to minimize the influence of maturational factors in perceptionr.l of family cohesion, the present study focused on a narrower age range than that utilizedin the study by Fandrich et al . (1990) . aecond, the sample sizes were made aqubralent to eliminate unwanted effecta due to disproportionate sample
37 sizes. Although th e y are no t fla ws in methodology, cn e eeare two addi tional poi nt s wor thmentioningconcern ingthe Fand ric h ee al. (1,990) study. Parental perceptions of familycohesion werenot obtained inth e study; thisadditional information wou l dhave permit tedcomparisoneof perceptionsbetweenfamily
melllbers. In addition, th e ctudy included one or more
depressed par en t s within the depreassd group. Thus i t was imp o s s ibl e to dets:rmine the re l a t i o n shi p between mater nal depressionand children'sperceptions of cohesion. Therefore, the present study was designed to invest igate these two concerns by comparingmothers' and their children'scohesion scores, and by examining the presenceof maternal depressive symptomatologyin the absenceof paternal distre ss.
Thepresent study
The questi on s addressed by this study concern the perceptionsof familycohesion by the offspring of motherswho were experiencingdep ressivesymptomatology. Anexamination of the perceptions of familycohesion by a mother who was ex periencingdep r essivesymp toma to logyandher offs p ring wa s of specia l inte restbecauseboth were providing ra t i n g s of the same family,yet thechi l dr e nwere not themselvesidentI f i e d as dist ressed . Therefore, consistentwiththe aforemention ed findings (e.g., Fendr ich et a1., 19 9 0 ) and th e sta te- sp e c if i city hypothes is'Ol iveret aI., 1987),i t was expec te d that mot he rswho were experienci ngdepressive symptomatology andthei r offspri ngwou ld perce i v e lowsr familycohee ion th a n
38 acommunitycomparisongroup. Addit ionally,itwaspr edicted th a t targetmothers andth eir childrenwouldshowacong ruen t perc e pt i o n of their familyas havinglow co hesion .
The design of the present study involved obtaining assessments of perceivedfalllilycohesionbychi l drenand their mothe rs in families wi th a mother who was experiencing depres s i v e symptomatologyand incommunity-control families. Standa rdized instrumen ts were used to assess dep r e s s ive sympt omatologyandfamilycohesion . Motherswerein clud e din the targ et group if they experiencedsignifica n t depressive symptomatology for a period of at le Ast one llIonth. Thus in fo rmat ionconce rning li f e t i me hi,storyofaf f e c t i v e disorder wasnot obtained;instead,emphasiswa s placedonobtaining an indexof the currentseveri tyof dep r e s siv e symptoms. This approach wa s ta k e n inli g h t of a previousfind i n g that it is the mother ' scur rentdepre osivesympt.omatologymoreoftenth a n herlifetimehi s t o r yth a t predictedchi ld ren'sadverse schoo l and social function i ng (Hammen, Ad r i an et al . , 19 87).
Compar i sons of perceptions of cohesion were made bet.ween tar get an d compari son mothers and between the chi l d ren of targ e t moth ers and the childre n of comparison mothe rs. Children'e ra tings of cohesion were compa red wit h thei r mothe r s' ra tin g s , both overal l and wit hin the target and compa rison group s .
since socia l background fa ctors are related to adul t depre esion aswel l as toc~d.ldren's func t i o n i ng, they ne e d to
39 be co nsi de re din evaluating th el ink betweenpa renta ldistres s and chil dren' shealth. Beingof a low socioeconomicst an di n g is associated with a higher incidence of adult de pre s s i o n (Browns,Harris, 1978, We tz el , 19 8 4 ) Iln dD.greate rl ikel ihood ofha vi n g chil d renwhose functioning is impaired(Weissman&.
Mye rs , 191 81. Low socioeconomi c status hasalsobeen li n k e d to greater expoau re to environmental stres s ors (Ch a n d l er , Million, &.Shermis, 19 8 5 ) and to lesssupportive family and social re s our c e s (MOOB&.Hoos. 1986 ). Inthere s e a r c h l inked withthe developmentof the FEB, Moos andMo o s (19 86) found th a t the educationa l andoc c upa tion a l statusof each of the pa r tners in normal families were positively related to cohesion. In addition, scoreson the CohesionSubacaletended to decrease as familysizeincreased. Given theestabl ished rsl a tionshi p s between family backg round factors (i .e., educational an d occupational sta t u s of par tners and fami ly size) and the variab les of interest in th e present stu dy (maternal depress ive symptomatology and family cohesi on ), estimatesof socioeconomic st a t u s (i.e., familyincome) were obtain edfor ea ch family. Subsequently, thetarg e t samplewas comp ar e d wit ha socio Mdemog r a phical ly mat c hed grouf?fcontro l families on variab les in c luding ago and sex of children, familysi z e, and singleversus tWOMpa r e nt families.
The following hyp othe s e s were made: First , con s i ste n t with pre vi :,)us l.'esea:'·'Jh, itwa s predicted th a t target moth tore wou ld perc e i v e lower family cohes ion than would compa r ison
40 mothers. Second. on the basis of a previous fi n d ing (Fendrich at a1.. 1990), i t was predicted that children of target motherswouldpercei ve lowarfamily cohesion than would the children of compa r i s on mothers. Thi rcl , consistent with the state-specificity hypothes is, it was p:redicted that there would he significantly more co n g ru e n c e between children's ratings of cohesion an d their mothero' ratings inth e target groupthan in the comparison group. Theprediction thatthere would be les s congruencebe t we en childrenand mother 'sra t i ng s of family cohesion in thecommun i t y comparisonsample was made on the basis of previouslydiscussedresearch findings(Noller
" Callan, 19861Olson at al••19 B3 ).
41 Method
Subjects
Two groups of17 mother-childdyads served4S subj acts. The en t ire samplewas cOJI\prlsed of a targ etand a community comparison grou p. Th e target group was recruited through therapis t re f e r r al s, television advertisements, and a newspaper advertillsment. The comparison group was recruited throughteleviaion advertisements.local community groups, and word-af-mouth. Motherswere 18 years of age or olde r and li v e d at home wi that least cne childbetween the ages of 11 and 17 years. In target familiesin whichthere was morethan one cbild, the youngset within this age range was selectedas the target child. The mean agee for the target and co:rnmunity comparison children were 13.06 years an d 13.24 years, respectively. Bo t h groups consistedof 7 maleand 10 female children and 9 married and 8 singlemothers. Al lof tho data col lectedin this study were based on sel f- reportmeaeu res.
Ta r g e t sample selection criteria. The target sample consisted of mothersand chi ld renin 17 families withmothers who wereexperiencing depressive symptomatology. Of the 78 fami li e s that agreed to participate in the study, 17met tho targetsampleinclusion criteria. Th e existenceof depress ive symptomatologywas deterDIined by the fol lowingcriteria: (a) de pres s e d mood for most of the day, more days thannot for at le a s t one month, as 1.ndicatedeitherby subjectiveaccount or observationby othersJand (b) the presence,dur ingdep ressed
