Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Family characteristics of anorexic, bulimic, psychiatric control, and nonpsychiatric control female adolescents Taylor, Lori Anne 1992

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1992_fall_taylor_lori_anne.pdf [ 2.44MB ]
JSON: 831-1.0086645.json
JSON-LD: 831-1.0086645-ld.json
RDF/XML (Pretty): 831-1.0086645-rdf.xml
RDF/JSON: 831-1.0086645-rdf.json
Turtle: 831-1.0086645-turtle.txt
N-Triples: 831-1.0086645-rdf-ntriples.txt
Original Record: 831-1.0086645-source.json
Full Text

Full Text

FAMILY CHARACTERISTICS OF ANOREXIC, BULIMIC,PSYCHIATRIC CONTROL, AND NONPSYCHIATRIC CONTROLFEMALE ADOLESCENTSbyLORI ANNE TAYLORB.A., The University of British Columbia, 1984M.A., The University of British Columbia, 1987A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE STUDIES(Psychology)We accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIAMay, 1992Lori Anne Taylor, 1992Signature(s) removed to protect privacyIn presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature)Department of 4 oL-OThe University of British ColumbiaVancouver, CanadaDate EJL , 199—DE.6 (2/88)Signature(s) removed to protect privacyiiAbstractThe aim of the present study was to investigate thecharacteristics and interaction patterns in the families ofadolescent eating—disordered patients. Four groups of femaleadolescents and their mothers (restrictive anorexic, buliinictype, psychiatric control, and nonpsychiatric control) wereassessed on a number of self-report instruments: The FamilyEnvironment Scale, Dyadic Adjustment Scale, Work and FamilyOrientation Questionnaire, Sex Role Ideology Scale, FoodFitness and Looks Questionnaire, and Body Esteem Scale.Support was found for the hypothesis that the families ofbulimic type and psychiatric control subjects arecharacterized as more dysfunctional than the families ofrestrictive anorexic and nonpsychiatric control subjects. Inparticular, restrictive anorexic and nonpsychiatric controlmothers and daughters characterized their families as morecohesive than did bulimic type and psychiatric control mothersand daughters. No differences were found amongst the fourgroups on expressiveness, conflict, independence,organization, control, or marital adjustment. These familyinteraction data were found to vary with the adolescent’slevel of depression, general psychiatric distress, andimpulsivity, but only for daughters, not for mothers. Littlesupport was found for the hypothesis that restrictive anorexicand bulimic type mothers and daughters are characterized ashigher in achievement orientation, traditional sex roleideology, and weight and appearance orientation thaniiipsychiatric control mothers and daughters. There were nogroup differences with respect to individual or familyachievement orientation; however, restrictive anorexic andnonpsychiatric control daughters did have higher school gradesthan psychiatric control daughters. No differences in sexrole ideology were found amongst the groups. Restrictiveanorexic and bulimic type daughters, but not mothers, ascribedgreater importance to weight and had more negative attitudestoward their own weight than psychiatric and nonpsychiatriccontrol daughters. No group differences were found formothers or daughters with respect to attitude toward one’s ownattractiveness or importance ascribed to appearance orfitness. Potential explanations for lack of congruence withthe theoretical literature are advanced, and the possiblespecificity of family pseudocohesiveness and problem denial toeating disorders is discussed.ivTable of ContentsAbstract iiList of Tables. vii.A.cJcnowledgement viiiINTRODUCTION 1Anorexia Nervosa 2BulimiaNervosa 4Diagnostic Issues 7Psychiatric Comorbidity 12Family Factors 15Demographics 17Psychopathology in Family Members 19Family Characteristics and InteractionPatterns: Major Theories 27Family Characteristics and InteractionPatterns: Clinical Data 32Family Characteristics and InteractionPatterns: Controlled Studies 39Summary and Hypotheses 58METHOD 62Subjects 62Measures 70Family Environment Scale 70DyadicAdjustmentScale 71Work and Family Orientation Questionnaire 71Sex—Role Ideology Scale 72Food Fitness and Looks Questionnaire 73Body Esteem Scale 73VBackground Information Forms 74Weight Status 74Interview 75EatingAttitudesTest 75BriefSymptomlnventory 76Procedure 77REStJITS 79Clinical Characteristics 82Eating— and Weight—Related Measures 84Psychiatric Distress—Related Measures 87Tests of Study Hypotheses 89Family System/Interaction Hypothesis 90Family Sociocultural Milieu Hypotheses 92Additional Analyses 100Analyses of Covariance 100Subsidiary Analyses: Family System 104Summary 105DISCUSSION 109Family System/Interaction Hypothesis 109MainAnalyses 109SubsidiaryAnalyses 120Family Sociocultural Milieu Hypotheses 127Achievement Orientation 127SexRole Ideology 130Weight and Appearance Attitudes 130Additional Sociocultural Milieu Measures 133viGeneral Conclusions 134Limitations of the Present Study 137Questions and Recommendations for Future Research 139REFERENCES 141APPENDIX 157viiList of TablesTable 1: FES Subscale Definitions 49Table 2: Demographic Data Means 66Table 3: Treatment—Related Information 67Table 4: Daughters’ Eating- and Weight-RelatedMeans (Standard Deviations) 85Table 5: Daughters’ Psychiatric Distress-RelatedMeans (Standard Deviations) 88Table 6: Family System/Interaction Means 91Table 7: Dyadic Adjustment Scale Means 92Table 8: Achievement Orientation Means 94Table 9: Daughters’ School Grades Means 95Table 10: Sex Role Ideology Scale Means 95Table 11: weight and Appearance Attitudes Means 97Table 12: Mothers’ Weight-Related Means 99Table 13: Additional Sociocultural Milieu MeasuresMeans (Standard Deviations) 100Table 14: Correlations Between Dependent Variablesand Potential Confounds-- Daughters 102Table 15: Correlations Between Dependent Variablesand Potential Confounds—— Mothers 103Table 16: Composition of Study’s Groups 158Table 17: Low versus High Depression Group Means 160Table 18: Low versus High Distress Group Means 161Table 19: Low versus High Impulsivity Group Means 163Table 20: Low versus High Eating DisorderSymptomatology Group Means 164viiiAcknowledgementI would like to express my appreciation to my researchsupervisor, Dr. Dimitri Papageorgis, especially for sharinghis vast fund of knowledge with me and for being so thoroughat every stage of this project. Special thanks also go toDr. Ralph Hakstian for his invaluable statisticalconsultation, and to Dr. Anita DeLongis and Dr. Ron Manleyfor their support, encouragement, and clinical expertise.I am grateful to the professionals who, despite theirbusy schedules, assisted me in subject recruitment. Theseinclude Dr. Roger Tonkin, Ms. Carolyn Hammond, Ms.Marguerite Amos, Ms. Winnie Ting, Dr. Geoff Carr, Dr. PatManley, and many others. Sincere thanks are also given toall of the family members who volunteered their time toparticipate in this study.Finally, I wish to express my deep gratitude to myfamily and dearest friends who have given me their love andsupport throughout this project’s duration. My heartfeltthanks go out to Gene Flessati, Anna Fritz, AngelaLamensdorf, Bill and Margaret Lowe, Cindy Neilly, Denis andJean Taylor, and Rob Taylor.1IntroductionAnorexia nervosa and bulimia nervosa are considered to bemultidetermined disorders of eating. There has been muchinterest over the past decade in possible contributing factorswithin the families of those with eating disorders.Theoretical formulations, clinical data, and controlledstudies have suggested that the family environment ofrestrictive anorexic individuals is cohesive and harmoniouswhile the family environment of bulimic individuals isdisengaged and conflictual. The vast majority of controlledstudies have employed adult women as subjects and have notincluded a psychiatric control group. It has also beenpostulated that the families of eating-disordered subjects maybe characterized by high achievement orientation, traditionalsex-role ideology, and high weight and appearance orientation;however, little controlled research has addressed theseissues. The present study was conducted in an effort toextend and contribute to the eating disorder family literatureby investigating the characteristics and interaction patternsin the families of female adolescent anorexic and bulimicsubjects, as perceived by both the adolescent and her mother.In addition, a psychiatric control group was employed so as toyield information on the specificity of family factors toeating disorders.Before proceeding to a review of the literature on familyfactors in anorexia nervosa and bulimia nervosa, the two2eating disorders will be defined and diagnostic issues will bediscussed.Anorexia NervosaAnorexia nervosa is an eating disorder in which anindividual is fearful of gaining weight or becoming fat, feelsfat in one or more areas of the body, and, as a result,engages in behaviors to decrease weight or pursue thinness.The primary weight-reducing behavior engaged in is veryrestrictive dieting, usually involving severe limits oncaloric intake and complete avoidance of certain foods.Extensive exercising, self—induced vomiting, or the use oflaxatives or diuretics may also be used for the purpose ofweight reduction. Through such behaviors, anorexicindividuals lose a significant proportion of their body weightand maintain their weight at a level below that expected fortheir age and height.Anorexics may show other features in addition to thevarious weight—reducing behaviors described above. Forinstance, they may collect recipes, hoard food, followmonotonous or unusual diets, perform eating rituals, orprepare elaborate meals for others but not partake of themthemselves (e.g., American Psychiatric Association, 1987;Bemis, 1978). Binge-eating episodes may also occur. Anotherfeature is that the individual usually denies or minimizes theseverity of the problem, and is uninterested in or resistantto therapy (e.g., American Psychiatric Association, 1987). Itis of note that the term anorexia is a misnomer as an actual3loss of appetite is rarely present except, at times, duringthe later stages of the disorder (American PsychiatricAssociation, 1980, 1987).In the present study, the diagnostic criteria of therevised edition of the Diagnostic and Statistical Manual ofMental Disorders (DSM-III-R; American Psychiatric Association,1987) were used to operationalize the syndrome of anorexianervosa. These diagnostic criteria are as follows:A. Refusal to maintain body weight over a minimal normalweight for age and height, e.g., weight loss leading tomaintenance of body weight 15% below that expected; or failureto make expected weight gain during period of growth, leadingto body weight 15% below that expected.B. Intense fear of gaining weight or becoming fat, eventhough underweight.C. Disturbance in the way in which one’s body weight, size,or shape is experienced, e.g., the person claims to “feel fat”even when emaciated, believes that one area of the body is“too fat” even when obviously underweight.D. In females, absence of at least three consecutivemenstrual cycles when otherwise expected to occur (primary orsecondary amenorrhea).Anorexia nervosa is more prevalent in women than in men:The American Psychiatric Association (1987) estimates that 95%of those with anorexia are female. (Similar statistics arereported for bulimia nervosa and, therefore, the femininepronoun will be used in this thesis when referring to4individuals with eating disorders.) The onset of anorexia isusually in early to late adolescence, and researchers suggestthat the incidence of the disorder has been rapidly increasingin industrialized societies over the last 20 to 30 years(e.g., Bends, 1978; Mitchell & Eckert, 1987). The prevalenceof anorexia nervosa is estimated to be less than 1% (e.g.,American Psychiatric Association, 1987; Bends, 1978; Pope,Hudson, & Yurgelun-Todd, 1984). Finally, the mortality ratein anorexia nervosa has been estimated to be from as low as 3—5% to as high as 18—25% (e.g., American PsychiatricAssociation, 1987; Bends, 1978).Bulimia NervosaMany labels have been attached to the syndrome referredto in this thesis as bulimia nervosa, including bulimia(American Psychiatric Association, 1980), buliniarexia(Boskind-Lodahi & White, 1978; Boskind-White & White, 1987),the abnormal normal weight control syndrome (Crisp, 1981), anddietary chaos syndrome (Palmer, 1979). “Bulimia nervosa” wascoined by Russell (1979) and has been adopted for use in theDSM-III-R (American Psychiatric Association, 1987). This termis preferred by the author as it highlights the relationshipof bulimia nervosa to anorexia nervosa, and as “bulimia” issometimes used to refer solely to the symptom of binge—eating.Bulimia nervosa is an eating disorder in which anindividual has episodes of binge eating and also engages inbehaviors to reduce weight or to prevent weight gain. Thebulimic individual may be underweight, overweight, or, most5often, of normal weight (e.g., American PsychiatricAssociation, 1987; Garner, 1986; Schlesier-Stropp, 1984).Following a binge-eating episode, the bulimic individual mayself—induce vomiting, take laxatives or diuretics, exercisevigorously, or begin dieting or fasting to counteract thefattening effects of the binge. However, these behaviors mayalso be engaged in independently of the binge-eating episodesfor purposes of weight control.Binge eating has been reported to vary greatly onquantitative indices. The duration of a binge may range from15 minutes to 8 hours ( = 1.2 hours); the frequency of binge-eating episodes may vary between 1 and 46 per week ( = 11.7);and during an average binge—eating episode, 4800 calories maybe consumed (Johnson, Lewis, & Hagman, 1984). It is unclear,however, whether such quantitative indices are critical indefining what constitutes a binge. Instead, it has beensuggested that the subjective experience of feeling out ofcontrol during a binge may be more important in its definition(Johnson et al., 1984; Johnson, Lewis, Love, Stuckey, & Lewis,1983).While the presenting symptoms of the bulimic individualare often the binge eating and purgative behaviors (i.e.,self—induced vomiting and laxative abuse), it must be kept inmind that the desire to be thin is also a central feature ofthe disorder. Like the anorexic, the bulimic fears fatnessand is preoccupied with her weight and shape (Crisp, 1981),complains of being too fat when others do not perceive her to6be so (Boskind—Lodahl & White, 1978), and is determined toweigh less than some predetermined threshold (Russell, 1979).Garner (1986) has stated that individuals with bulimia nervosamay appear to be less resistant to treatment than individualswith anorexia nervosa because they wish to eliminate theirbinge—eating behavior whereas, in fact, they are similarlyresistant to giving up their dieting behaviors. Garner (1986)and other authors (e.g., Crisp, 1981) have suggested that thebulimic individual often wishes she were anorexic.As with anorexia nervosa, the diagnostic criteria of theDSM-III-R were used in the present study to operationalize thesyndrome of bulimia nervosa. These diagnostic criteria are asfollows:A. Recurrent episodes of binge eating (rapid consumption of alarge amount of food in a discrete period of time).B. A feeling of lack of control over eating behavior duringthe eating binges.C. The person regularly engages in either self—inducedvomiting, use of laxatives or diuretics, strict dieting orfasting, or vigorous exercise in order to prevent weight gain.D. A minimum average of two binge eating episodes a week forat least three months.E. Persistent overconcern with body shape and weight.Bulimia nervosa, like anorexia nervosa, is more prevalentin women than in men: Estimates suggest that from 90%(Striegel—Moore, Silberstein, & Rodin, 1986) to 99% (Johnson,Stuckey, Lewis, & Schwartz, 1983) of bulimic individuals are7female. The onset of bulimia nervosa is usually inadolescence or early adulthood (American PsychiatricAssociation, 1987). A mean age of onset of 18 years has beenreported in three major studies of bulimia nervosa (Johnson etal., 1983; Mitchell, Hatsukami, Eckert, & Pyle, 1985; Pyle,Mitchell, & Eckert, 1981). It is further reported that self-induced vomiting usually begins approximately one year afterthe onset of binge eating.Finally, the prevalence of bulimia nervosa has been foundto vary with the diagnostic criteria employed. Mitchell andEckert (1987) reported that 26% to 79% of women and 41% to 60%of men report binge eating, depending on how it is defined.This illustrates the need to require more than the symptom ofbinge eating for a diagnosis of bulimia nervosa. In general,not utilizing any frequency criteria for binge eating orpurging yields prevalence rates of bulimia of from 8 to 10%,whereas requiring weekly or greater frequencies of bingeeating and purging yields prevalence rates of from 1 to 5%(e.g., Johnson, Lewis, Love, Stuckey, & Lewis, 1983; Mitchell,& Eckert, 1987; Pope et al., 1984; Pyle, Mitchell, Eckert,Halvorson, Neuman, & Goff, 1983).Diagnostic IssuesIn the preceding two sections, the current diagnosticcriteria for anorexia nervosa and bulimia nervosa weredescribed. However, these sets of criteria are in the processof evolution .and there is some debate in the literature as towhich distinctions among eating disorders are the most valid.8Currently, potential revisions to the diagnostic criteria foreating disorders are being proposed for the forthcoming DSM-IV(Wilson & Walsh, 1991).One controversial issue is whether anorexia nervosa andbulimia nervosa should be considered separate disorders (e.g.,Abraham & Beumont, 1982; Johnson et al., 1983; SchlesierStropp, 1984). In particular, there is some debate overwhether bulimia nervosa should be a distinct diagnosis. Thediagnostic criteria for bulimia nervosa were only developed in1979 and 1980. The attention to bulimia nervosa came partlyas a result of researchers finding bulimic symptoms in theiranorexic subjects (Schiesier-Stropp, 1984); however, bulimianervosa is also found in obese and, especially, normal—weightsubjects (e.g., Johnson et al., 1983). The most recentdiagnostic manual (DSM-III-R; American PsychiatricAssociation, 1987) changed the disorder’s label from bulimiato bulimia nervosa to reflect its strong relationship toanorexia nervosa. Researchers point out that the attitudestowards eating and weight are very similar in anorexia nervosaand bulimia nervosa (e.g., Fairburn & Garner, 1986; Pyle etal., 1981; Russell, 1979), and that the same patient mayalternate between the disorders of anorexia nervosa andbuliinia nervosa at different times (e.g., Abraham & Beumont,1982; Garner, 1986).In contrast to the issue of whether anorexia nervosa andbulimia nervosa are separate disorders, another issue concernshow best to further subdivide these eating disorders. For9instance, anorexia nervosa patients may or may not exhibit thesymptoms of bulimia nervosa, and bulimia nervosa patients mayor may not have a history of anorexia nervosa. This leads todistinctions between “restrictive .anorêxics” and “bulimicanorexics,” and bulimics with and without a history ofanorexia nervosa. Researchers and clinicians suggest thatbetween 25% and 50% of anorexia nervosa patients developbulimic symptoms (e.g., Bruch, 1984; Garner, Rockert, Olmsted,Johnson, & Coscina, 1984; Mitchell & Pyle, 1982). For bulimianervosa samples, estimates range from 6% to 29% for a historyof anorexia nervosa (e.g., Fairburn, 1984; Johnson & Larson,1982; Pyle et al., 1981). Some authors suggest that bulimicsymptoms may develop as a function of chronicity in anorexianervosa (e.g., Crisp, 1981; Johnson & Larson, 1982), andothers report that the bulimic behaviors often begin early inthe disorder (e.g., Abraham & Beumont, 1982). Probably manyscenarios occur. For example, there may be anorexics whoexhibit bulimic symptoms after a long period of restriction,anorexics who begin binge eating with purging at the onset ofthe disorder, and patients in whom bulimic symptoms occur inthe absence of anorexia nervosa.How valid are the eating disorder distinctions? Withrespect to the distinction between bulimia nervosa with andwithout a history of anorexia nervosa, there is littleinformation. What information there is suggests that bulimicsubjects with a history of anorexia nervosa are very similarto bulimic subjects without such a history in terms of eating10habits, concerns about weight, and associated psychopathology(Fairburn, 1984; Fairburn & Garner, 1986; Mitchell, Pyle,Eckert, Hatsukami, & Soil, 1990). Mitchell et al. (1990) didfind, however, that bulimic subjects with a history ofanorexia nervosa had a lower ideal weight, were more likely toabuse laxatives, were less likely to self-induce vomiting, andreported having been treated for depression more often thanbulimic subjects without a history of anorexia nervosa. Thereis more evidence to suggest that there are differences betweenrestrictive anorexics and bulimic anorexics. For instance,Strober (1981) found that anorexics with bulimia had higherlevels of anorexic and other psychological symptoms thananorexics without bulimia. Parental personalitycharacteristics and family psychiatric morbidity have alsobeen found to differ between restrictive and bulimic anorexicsamples (Strober, Salkin, Burroughs, & Morrell, 1982), as haveother characteristics of the patients and their families.These findings will be discussed in detail in later sections.It has further been suggested that bulimic anorexics andnormal—weight bulimics are more similar to each other thanthey are to restrictive anorexics in terms of impulse—relatedbehaviors, family variables, premorbid maximum weight, andweight- and eating-related variables (Garner, Garfinkel, &O’Shaughnessy, 1985). Similarly, Strober and Humphrey (1987)suggested that the psychopathology, weight tendencies, andinteraction patterns of the family are distinctly differentbetween bulimic and nonbulimic eating disorder subtypes.11Recently, a study was conducted which compared underweight,normal-weight, and overweight bulimic women with each otherand with nonbulimic controls from all three weight categories(i.e., restrictive anorexics, normal controls, and obesecontrols; Shisslak, Pazda, & Crago, 1990). It was found thatbulimic subjects at all three weight levels exhibited greaterpsychopathology and lower self—esteem than nonbulimic subjectsat corresponding weight levels. Among the three bulimicgroups, underweight bulimic subjects showed the greatestpsychopathology. Thus, it appears that there are manydifferences between bulimic anorexics and restrictiveanorexics, and that there may also be some differences betweenbulimic subjects of different weight levels.Nonetheless, it should be pointed out that researchersalso find many similarities among eating disorder groups. Forinstance, in both the Strober (1981) and Garner et al. (1985)studies cited above, there were no differences among eatingdisorder subgroups on a number of eating, personality,psychopathology, and family variables. It appears that incurrent research, one must attempt to delineate both thecommonalities and the distinctions amongst eating disordersubgroups (e.g., Striegel-Moore et al., 1986).In the present study, hypotheses are advanced for thegroup of eating—disordered subjects as a whole, as well as forsubjects divided into eating disorder subtypes. There appearto be similarities between subjects with anorexia nervosa andbulimia nervosa, but there also appear to be differences,12especially between bulimic and nonbulimic subgroups. Thepresent study investigated subjects with restrictive anorexianervosa and subjects with bulimia nervosa. The bulimic—anorexic subgroup is of interest as well, but due to limitedresources the priority of this study was to examine the twomore distinct subgroups.Eating Disorder ComorbidityAnother issue of importance in research on anorexianervosa and bulimia nervosa is that of psychiatriccomorbidity. For instance, symptoms of depression arereportedly common in anorexic (e.g., Kalucy et al., 1984) andbulimic subjects (e.g., American Psychiatric Association,1987; Johnson et al., 1983; Russell, 1979). Based onindividual and/or parental interviews, approximately 45% of 26anorexic patients were diagnosed as having an affectivedisorder at follow-up an average of 5 years post—hospitalization (Cantwell, Sturzenberger, Burroughs, Salkin, &Green, 1977). No data were presented on presence or absenceof binge-eating. Lee, Rush, and Mitchell (1985) reported that52% of their sample of adult female bulimic subjects had apersonal history of unipolar affective disorder, and Blouin,Zuro, and Blouin (1990) found that 62% of their bulimic samplemet DSM-III criteria ,for lifetime history of major depression.Comparing eating disorder subtypes, researchers have not foundsignificant differences among restrictive anorexic, bulimicanorexic, and bulimic subjects in terms of depressive symptoms(e.g., Garner et al., 1985; Strauss & Ryan, 1988; Wonderlich &13Swift, 1990b); however, Strober (1981) found a significantdifference in terms of diagnosis of depression with 41% ofbulimic anorexic subjects and 9% of restrictive anorexicsubjects receiving DSM—III diagnoses of major depression.Another diagnosis frequently reported to co—occur withbulimia nervosa is substance abuse disorder (e.g., AmericanPsychiatric Association, 1987; Crisp, 1981; Pyle et al.,1981). Dykens and Gerrard (1986) found that bulimic subjectsused alcohol and drugs more frequently and at an earlier agethan control subjects, and Strober (1981) found alcohol use in23% of bulimic anorexic adolescents and 0% of restrictiveanorexic adolescents. Similarly, another study found thatbulimic anorexics showed greater drug and alcohol use thanrestrictive anorexics (Piran, Lerner, Garfinkel, Kennedy, &Brouillette, 1988).Personality disorder diagnoses have also been examined ineating—disordered groups. In a mixed eating disorder sample,and using DSM-III-R criteria, Wonderlich and Swift (1990a)found that 24% of subjects met criteria for borderlinepersonality disorder, 48% of subjects met criteria for otherpersonality disorders, and 28% of subjects received nopersonality disorder diagnosis. In a sample of 94 bulimicpatients, 46% were found to have borderline personalityfeatures as assessed by a self—report inventory which measuresDSM-III criteria for borderline personality disorder (Johnson,Tobin, & Enright, 1989). Another study was designed tocompare personality disorder diagnoses between restrictive and14bulimic anorexics (Piran et al., 1988). These researchersfound that 60% of the restrictive anorexic sample receiveddiagnoses of avoidant personality disorder and 55% of thebulimic anorexic sample received diagnoses of borderlinepersonality disorder. Thus, while there was heterogeneity ofdiagnosis within groups, a majority of restrictive anorexicsshowed social discomfort, fear of negative evaluation, andtimidity, while a majority of bulimic anorexics showedinstability of mood, interpersonal relationships, and self-image.This comorbidity in eating-disordered populations raisesthe issue of the nature or direction of this association. Itcould be that the psychiatric symptomatology predisposes theindividual to the development of an eating disorder or,conversely, that the eating disorder symptomatology maycontribute to the development of other psychiatric disorders.On the other hand, other family, biological, or socioculturalfactors could foster the development of both the eating- andnon—eating—related psychopathology. Some authors have arguedthat eating disorders may be variant expressions or alternateforms of affective disorders (e.g., Cantwell et al., 1977;Hudson, Pope, Jonas, & Yurgelun-Todd, 1983; Lee et al., 1985).While recognizing and examining such comorbidity, the presentstudy, rather than searching for similarities amongstdisorders, will focus on determining family factors specificto anorexia nervosa and bulimia nervosa as distinct from otherpsychiatric disorders.15Family FactorsMost writers agree that anorexia nervosa and bulimianervosa are multidetermined disorders. That is, they are seenas resulting from a complex interaction of biological,personality, psychopathological, family, and socioculturalfactors (e.g., Bruch, 1973; Garner & Garfinkel, 1980; Johnson,Connors, & Tobin, 1987; Mitchell & Eckert, 1987; Root, Fallon,& Friedrich, 1986; Strober & Yager, 1984). For instance,sociocultural factors such as pressures to be thin,discrimination against the obese, and female sex—rolesocialization may predispose women to the development ofeating disorders (e.g., Boskind-White & White, 1987; Garner etal., 1984; Striegel-Moore et al., 1986). In combination withmore specific family and individual factors (some perhapsgenetic and/or biological) to be discussed in detail later, aneating disorder may be precipitated by social—emotionalstressors such as the onset of adolescence with its pressuresof increased separation, identity formation, sexuality, andbodily changes. Finally, biological factors such as theeffects of starvation (e.g., Garner, Garfinkel, & Bemis, 1982;Kaplan & Woodside, 1987; Keys, Brozek, Henschel, Mickelsen, &Taylor, 1950) and of persistent binge-eating and vomiting(e.g., Garner et al., 1984), along with individual, family,and sociocultural factors, may serve to perpetuate the eatingdisorder. Thus, it is highly unlikely that any single factorwould account for all the aspects of anorexia nervosa orbulimia nervosa (e.g., Bemis, 1978; Chiodo, 1987). This is16the context in which the present study on family contributingfactors must be viewed. While one may investigate one set offactors separately, it must be remembered that this is anartificial approach, and that no factor acts independently inthese “etiologically complex syndromes” (Strober & Humphrey,1987).The literature on possible family contributing factors inanorexia nervosa and bulimia nervosa will now be reviewed. Inparticular, demographic factors, psychopathology of individualfamily members, and family interaction patterns andcharacteristics will be examined. While research findings maypoint to factors that contribute to eating disorders, they mayalso reflect effects on the family of having a daughter withan eating disorder. Certain family characteristics observedmay be the results of a current family crisis rather than ofpreexisting family dysfunction. For instance, Kay, Schapira,and Brandon (1967) remind researchers to imagine how a“normal” parent would behave when faced with a daughter intenton self-starvation. And Bruch (1973, 1978) suggests thathaving an anorexic daughter may increase family anxiety,concern, annoyance, and resentment, and lead to thedevelopment of power struggles. Nonetheless, even if a familycharacteristic has developed in response to having an eating—disordered daughter and, thus, did not predispose the daughterto develop anorexia nervosa or bulimia nervosa, this does notnecessarily diminish its importance: The characteristic in17question may be contributing to the maintenance of thedaughter’s eating disorder.Demographics. Many authors report that adisproportionately large percentage of individuals withanorexia nervosa have middle or upper socioeconomic classbackgrounds (e.g., Bends, 1978; Bruch, 1973; Edwards, 1987).They are said to have been brought up in educated andprosperous homes (Bruch, 1984) in which family members arehigh achievers• (Cooper, 1987), and status has been achievedrather than inherited (Hall, 1978). In samples of 41, 50, and56 ariorexic patients, 50% to 70% of patients’ families havebeen found to be from the upper two socioeconomic classes, ascompared with population norms for these two classes of 14% to18% (Hall, 1978; Kalucy, Crisp, & Harding, 1977; Morgan &Russell, 1975). However, Heron and Leheup (1984) examinedcase records of anorexic and control patient adolescents andfound no differences in socioeconomic class.Some authors believe that there is an overrepresentationof the higher socioeconomic classes in individuals withbulimia nervosa as well as in those with anorexia nervosa(e.g., Boskind-White & White, 1987; Shisslak et al., 1987),whereas others believe that individuals with bulimia nervosacome from more diverse socioeconomic backgrounds (e.g.,Cooper, 1987). Consistent with Heron and Leheup’s (1984)results above, it has been suggested that both anorexianervosa and bulimia nervosa are beginning to occur inindividuals of more diverse socioeconomic backgrounds,18including those of the lower socioeconomic classes (e.g.,Bruch, 1988; Selvini—Palazzoli, 1978; Shisslak et al., 1987).A survey of women in communities of different socioeconomicstatuses found, contrary to expectations, that anorexianervosa and bulimia nervosa were more common in lower incomerespondents than in upper income respondents (Pope, Champoux,& Hudson, 1987). More recently, researchers found nodifferences in social class or education between samples ofbulimic and nonbulimic women (Dolan, Lieberman, Evans, &Lacey, 1990). Bruch (1988) suggests that, regardless of theirsocioeconomic class, the families of anorexic individuals haveaspiration levels which are high relative to the socioeconomicclass they occupy.Apart from socioeconomic status, the demographiccharacteristics of the families of individuals with eatingdisorders are rarely noteworthy. No consistent differenceshave been reported with respect to family size, religiousaffiliation, or birth order (Dolan et al., 1990; Hall, 1978;Lacey, Gowers, & Bhat, 1991). Dolan et al. (1990) foundbulimic women’s parents to be older at the time of theirdaughter’s birth than nonbulimic women’s parents, whereasother researchers found no such difference (Kog &Vandereycken, 1985). Bruch (1973) has suggested that, ifthere were any demographic differences, it would be thatanorexic individuals often come from female—dominated familieswith few sons. Hall (1978) reported that the families of 50female anorexic patients included 112 daughters and 48 sons,19and it was recently reported that all-female sibships wereoverrepresented in a large sample of bulimic women (Lacey etal., 1991).If demographic characteristics are found whichdiscriminate between the families of eating-disordered andnon-eating-disordered individuals, the mechanism by which theypredispose individuals to develop anorexia nervosa or bulimianervosa must be explicated. For instance, certain attitudesmay be associated with higher socioeconomic status, such asachievement orientation or perfectionism, or, as some authorshave suggested, pressures on women to be thin may be greaterin the upper socioeconomic classes (e.g., Striegel—Moore etal., 1986; Wooley, Wooley, & Dyrenforth, 1979).Psychopathology in family members. Various disordershave been reported to occur with increased prevalence in thebiological relatives of individuals with anorexia nervosa orbulimia nervosa. Eating disorders themselves are thought tobe more prevalent in relatives of eating—disorderedindividuals than in control populations (e.g., Mitchell &Eckert, 1987; Strober & Humphrey, 1987). For instance,researchers examined 30 pairs of female twins in which theproband had anorexia nervosa and found concordance rates foranorexia nervosa of 55% in monozygotic twins and 7% indizygotic twins (Crisp, Hall, & Holland, 1985). This is incomparison with a maximum prevalence rate of anorexia nervosaof 1% in young women. Another study, however, described 11female adolescent anorexic patients who were members of a20same-sex twin pair (5 monozygotic and 6 dizygotic) and foundno concordance for anorexia nervosa (Waters, Beumont, Touyz, &Kennedy, 1990). It was unclear, however, on what basisdiagnoses were determined in this study. Examining bulimianervosa in female twins, researchers have found concordancerates for bulimia nervosa of 27% in dizygotic twins and 83% inmonozygotic twins (Fichter & Noegel, 1990). Hsu, Chesler, andSanthouse (1990), with a much smaller sample and with nodirect contact with the non-patient twin in almost half of thecases, found concordance rates for bulimia nervosa of 0% fordizygotic twins and 33% for monozygotic twins. Thus, on thebasis of the twin studies, it appears that genetic factors mayplay a role in the pathogenesis of anorexia nervosa andbulimia nervosa. This will be discussed further at the end ofthis section.Another study compared the rates of anorexia nervosa,bulimia nervosa, and subclinical anorexia nervosa in thefirst— and second-degree relatives of 60 female anorexicadolescents and 95 female nonanorexic adolescent psychiatricinpatients (Strober, Morrell, Burroughs, Salkin, & Jacobs,1985). There were higher incidences of anorexia nervosa,bulimia nervosa, and subclinical anorexia nervosa in therelatives of anorexic adolescents than in the relatives ofcontrol adolescents. Some evidence of specificity was foundin that all 4 diagnoses of severe restrictive anorexia weremade in relatives of restrictive anorexic subjects, and 7 ofthe 9 diagnoses of bulimia nervosa or bulimic anorexia21occurred in relatives of bulimic anorexic subjects. Furtherdata were then collected (Strober, Lampert, Morrell,Burroughs, & Jacobs, 1990), and the results corroborated theearlier finding in that there was a higher lifetime prevalenceof eating disorders (anorexia nervosa in particular) in therelatives of adolescent probands with anorexia nervosa than inthe relatives of probands with affective disorder or othertypes of psychiatric disturbance. Another study comparedadult female probands with and without bulimia nervosa andfound higher rates of eating disorders (bulimia nervosa inparticular) in the first-degree relatives of the bulimicprobands (Kassett, Gershon, Maxwell, Guroff, Kazuba, Smith,Brandt, & Jimerson, 1989).It has also been suggested that obesity is common in theparents of bulimic and bulimic anorexic individuals,especially in the mothers (American Psychiatric Association,1987; Kog & Vandereycken, 1985). Pyle et al. (1981) foundthat 23 of their 34 bulimic subjects had at least 1 obesefirst-degree family member; this included 13 mothers and 6fathers. No control group was employed, however. Otherresearchers measured the heights and weights of 30 anorexicindividuals’ parents and did not find them to differ fromthose of 30 control subjects’ parents (Halmi, Struss, &Goldberg, 1978).Other psychiatric disorders have been investigated in therelatives of individuals with eating disorders. For instance,the American Psychiatric Association’s (1987) diagnostic22manual notes that higher than expected frequencies of majordepression are found in the first—degree relatives of bulimicindividuals. Similarly, major depression and bipolar disorderare reported to be more frequent in the first—degree relativesof individuals with anorexia nervosa (American PsychiatricAssociation, 1987). There have been uncontrolled studieswhich have reported family histories of affective disorder inpatients with anorexia nervosa (e.g., Cantwell et al., 1977;Kalucy et al., 1977) and bulimia nervosa (e.g., Lee et al.,1985; Pyle et al., 1981). Rivinus et al. (1984) found moredepression in first— and second-degree relatives of anorexianervosa patients than in the corresponding relatives of normalcontrol subjects. Similarly, family history of affectivedisorder has been found to be greater in the families ofpatients with anorexia nervosa and/or bulimia nervosa than inthe families of patients with schizophrenia or borderlinepersonality disorder (comorbidly depressed subjects excluded)(Hudson et al., 1983). Recently, Strober et al. (1990) foundhigher rates of affective disorder among relatives of anorexicadolescents than among relatives of adolescents in a mixedpsychiatric control group, but only if the anorexic adolescenthad a coexisting affective disorder, thus suggesting aspecific transmission of liability. In contrast, otherresearchers, comparing bulimic and nonbulimic probands, havefound higher rates of affective disorder in the relatives ofbulimic probands than in the relatives of nonbulimic probandsregardless of history of affective disorder in the bulimic23proband (Kassett et al., 1989; Reck, Pope, Hudson, McElroy,Yurgelun-Todd, & Hundert, 1990), thus suggesting a possiblecommon diathesis in bulimia nervosa and affective disorder.In general, rates of family history of affective disorder havebeen found to be similar for eating—disordered patients andpatients with affective disorders (e.g., Hudson et al., 1983;Reck et al., 1990; Strober et al., 1990).Research has also been conducted comparing the familiesof restrictive anorexic adolescents with those of bulimicanorexic adolescents (Strober, 1981; Strober et al., 1982).Using a structured diagnostic interview, mood disorder wasfound to be more prevalent in the first— and second—degreerelatives of bulimic anorexic subjects than in those ofrestrictive anorexic subjects (Strober et al., 1982). Nine of11 diagnoses of bipolar disorder were given to relatives ofbulimic anorexic subjects. Prevalence of mood disorder was23% in the mothers and 14% in the fathers of bulimicanorexics, as compared with 6% of the mothers and 3% of thefathers of restrictive anorexics. In a regression analysis,maternal and paternal depression, along with paternalimpulsivity, were found to predict the greatest severity ofbulimia.Substance abuse, particularly alcoholism, has also beeninvestigated in the family members of individuals with eatingdisorders. Uncontrolled studies have reported familyhistories of alcohol abuse in anorexic (e.g., Cantwell et al.,1977; Kalucy et al., 1977) and bulimic probands (e.g., Lee et24al., 1985; Pyle et al., 1981). Rivinus et al. (1984) foundhigher rates of substance use disorders in first— and second—degree relatives of anorexia nervosa patients than in thefamilies of normal control subjects. Similarly, Kassett etal. (1989) found higher rates of alcohol abuse in the first-degree relatives of bulimic probands than in the first-degreerelatives of normal control probands. In a review of theliterature, Kog and Vandereycken (1985) reported that morealcoholism is found in the families of bulimic subjects thanin the families of anorexic subjects (Kog & Vandereycken,1985). For instance, the research cited above on depressionin the families of restrictive anorexic and bulimic anorexicadolescents also examined the prevalence of substance abuse(Strober, 1981; Strober et al., 1982). Alcohol and drug abusewere found to be more prevalent in the families of bulimic—anorexic subjects.Various characteristics besides depression and substanceabuse have been investigated in the families of patients witheating disorders. In an uncontrolled study of 56 families ofanorexic subjects, phobic avoidance was noted in 30% of themothers. and 11% of the fathers, marked obsessional traits werenoted in 14% of the mothers and 29% of the fathers, andmigraine headaches were noted in 30% of the mothers (Kalucy etal., 1977). In a controlled study, first-degree relatives ofbulimic probands scored higher on 3 of 11 DSM-III-Rpersonality disorder categories (histrionic, schizotypal, andobsessive—compulsive) than relatives of nonbulimic probands25did (Carney, Yates, & Cizadlo, 1990). In another controlledstudy, the fathers of bulimic anorexic subjects had higherscores on MNPI Infrequency, Psychopathic Deviate, andHypomania scales than restrictive anorexics’ fathers (Stroberet al., 1982); the MNPI content scales Family Problems andHostility were also higher in bulimic anorexics’ fathers. Therestrictive anorexics’ fathers had higher scores onMasculinity-Femininity and Social Introversion scales. Th,emothers of bulimic anorexic subjects scored higher on MMPIHypochondriasis, Depression, Family Problems, and Hostilityscales than the restrictive anorexic subjects’ mothers; andthe mothers of the restrictive anorexic subjects had higherelevations of Masculinity—Femininity, Social Introversion, andthe content scale Phobias. Strober (1981) also found that 50%of bulimic anorexic subjects’ fathers and 50% of their mothershad diagnosable disorders according to a structured diagnosticinterview, as compared with 14% of restrictive anorexicsubjects’ fathers and 18% of their mothers.In summary, there is some well—controlled researchsuggesting that eating disorders may be more prevalent in therelatives of anorexic and bulimic patients than in therelatives of control patients. It has been suggested thatobesity is common in the parents of bulimic individuals, butthe one controlled study reviewed investigated the weights ofanorexic subjects’ parents and did not find them to differfrom the weights of control subjects’ parents. Depression andalcohol abuse may be more prevalent in the families of eating—26disordered patients than in the general population, and thereis some evidence that these problems are more prevalent inbulimic and bulimic anorexics’ relatives than in restrictiveanorexics’ relatives. Finally, on the NMPI, bulimicanorexics’ fathers have been characterized as impulsive,bulimic anorexics’ mothers have been characterized asdepressed, and both have been characterized as hostile andinvolved in family conflict. Restrictive ariorexics’ mothersand fathers were depicted as introverted, and the mothers asphobic.An issue to consider in this research is whetherpsychopathology, if any, in the parents might be contributingto the etiology of the eating disorder in the daughter, orwhether the daughter’s eating disorder might be contributingto the development of various disorders in the parents. Alsoto be noted in this research is that it is never the case thata disorder is seen in 100% of the anorexic or bulimicsubjects’ mothers or fathers, nor even that a positive familyhistory for a disorder is found in 100% of the subjects. Assimilarly noted in the literature on personalitycharacteristics of anorexic and bulimic individuals, Yager(1982) notes that there is great personality diversity in thefamily members of individuals with eating disorders.Research on the psychopathology of family members raisesthe issue of genetic influences. Disorders such as depressionand alcoholism are thought to have some genetic basis (e.g.,Mitchell & Eckert, 1987), and the concordance rate in anorexic27and bulimic inonozygotic twins raises the issue of hereditaryfactors in eating disorders. The interesting question thusarises of what factor might be inherited that could predisposeone to develop an eating disorder. Regarding bulimia nervosa,it has been suggested that affective instability or asusceptibility to loss of control might be inherited (Hsu etal., 1990). With respect to anorexia nervosa, it has beenpostulated that phobic or avoidant personality traits-- suchas high harm—avoidance, low novelty seeking, and high reward-dependence -- might be the inherited liabilities (Strober etal., 1990). This research, however does not rule out theinfluence of environmental or nongenetic factors. Mostauthors agree that researchers must consider how geneticfactors interact with environmental factors, such as familyinteractions or attitudes, and the sociocultural context(e.g., Crisp et al., 1985; Strober & Humphrey, 1987). It isunlikely that either straightforward inheritance or directmodeling alone could account for the development of anorexianervosa or bulimia nervosa (Strober & Humphrey, 1987).Family characteristics and interaction patterns: Ma-brtheories. Several authors have formulated theories regardingthe family characteristics and interaction patterns ofindividuals with eating disorders. The late Hilde Bruch(1973, 1978, 1984, 1988), writing from a psychodynamicperspective, suggested that the family interactions ofanorexics only appear to be happy and harmonious. Familymembers are said to deny the existence of problems, portraying28family life and relationships as perfect. Bruch claimed thattensions lie beneath this facade of normality, but thatexpression of feelings, especially negative ones, is notallowed in these families. She asserted that, throughout thefuture—anorexic’s childhood, the parents do not respond to thechild’s expressions of needs, wants, or feelings, but insteadrespond to their own erroneous perceptions of the child’sneeds. Independence is discouraged in the child, whileexcessive closeness, over—conformity, and obedience areperceived as ideal.Bruch also claimed that these families have a success,achievement, and appearance orientation. The parents are saidto have high achievement expectations and to view the future—anorexic child as academically and socially superior to hersiblings. Many of the mothers may have been frustrated intheir career aspirations by marriage, and become conscientiouswives and mothers instead. Finally, Bruch suggested that theparents of anorexics may be unusually weight—conscious andpreoccupied with dieting and appearance.Minuchin and his colleagues suggest that anorexia nervosais a psychosomatic disorder which maintains and is maintainedby dysfunctional family systems or structures (e.g., Minuchin,Rosman, & Baker, 1978; Rosman, Minuchin, Baker, & Liebman,1977; Sargent, Liebman, & Silver, 1984). Families whichencourage somatization are said to be characterized by thefollowing five attributes: enmeshment, overprotectiveness,rigidity, lack of conflict resolution, and involvement of the29child in parental conflict so as to avoid or suppress it.Enmeshment is characterized by overinvolvement, excessivecloseness and loyalty, poor individuation, inference ofothers’ feelings and thoughts, and lack of privacy.Overprotectiveness is exhibited in a high degree of concernfor each other’s welfare, hypersensitivity to signs ofdistress, and highly nurturant interactions. Enmeshment andoverprotection result in reduced autonomy. Rigidity isreflected by a commitment to maintain the status quo and aninability to cope with change. Lack of conflict resolutionmay be the result of conflict avoidance or diffusion, or ofopen conflict which family members are unable to resolve. Theparents of anorexic children are said to avoid conflict moreoften than the parents of children with other psychosomaticdisorders. Minuchin and his co-workers also report that theanorexic’s entire family exhibits special concerns aroundeating, diets, table manners, and food fads. Schwartz,Barrett, and Saba (1984) have reported that the families ofbulimic individuals also exhibit the five attributes ofMinuchin’s psychosomatic families, plus consciousness ofappearances, isolation, and attaching special meanings to foodand eating.Selvini-Palazzoli (1978) has proposed formulations aboutthe families of anorexics from a systems perspective whichoverlap with those of Bruch and Minuchin. The parents aresaid to be concerned with appearances and norms, and to givethe superficial impression of having a mature emotional30relationship with each other. They do not acknowledge theirdisillusionment with each other. Criticism is not permissiblein the family. The family group is more important than theindividual, with self-sacrifice being viewed positively andself-indulgence negatively. The parents satisfy theirperceptions of the child’s needs, not the child’s actualneeds, with the result that the child feels she has no controlover her life, and any increase in her autonomy becomesanxiety-provoking for both her and her parents. The family isisolated and overprotective, and rigidly resists change.Selvini-Palazzoli also reported that all 12 of the familiesshe had treated had tried to maintain traditional sex-rolevalues especially with regard to division of labor within thefamily.More recently, a systems formulation of buliiuia nervosahas been advanced (Fallon & Root, 1986; Root et al., 1986).These authors suggest that bulimics’ families are enmeshed andhave difficulties resolving autonomy/intimacy conflicts.There are family rules regarding which feelings arepermissible, and when and how to express them. Intensefeelings are difficult for these families to cope with andconflict is avoided or left unresolved, Weight and appearanceare important, and there are inultigenerational patternsrevolving around food, dieting, and weight. These authorsalso suggest that the family is critical as a messenger ofsocietal expectations of the feminine role. The family issaid to mirror the inequities of power between men and women31in society, with the fathers being powerful and the motherspowerless. The daughters are said to be determined to livelives different from those of their mothers.Root et al. (1986) have also described three types ofbulimic families which are said to share the abovecharacteristics but to differ along other dimensions. Theyare the Perfect Family, the Overprotective Family, and theChaotic Family. Characteristics of the Perfect Familyinclude: an emphasis on appearances, family reputation,family loyalty, and achievement; discouragement of expressionof negative feelings; perfectionism; need for approval andacceptance by others; and rigid family rules. Attributes ofthe Overprotective Family include age-inappropriateoverprotection of children, and lack of trust in peopleoutside the family. The Overprotective and Perfect Familyformulations are somewhat similar. Finally, the ChaoticFamily is characterized by: inconsistent rules; physical oremotional unavailability of one or both parents; sexual,physical, or emotional abuse; substance abuse and impulsivity;frequent and inappropriate expressions of anger; and distrust,pseudoautonomy, and depression.Other authors have suggested that similar family typesexist in the families of anorexic individuals. Martin (1983)reported that 20 families of her sample of 25 families ofanorexic patients were enmeshed, overprotective, and lackedconflict resolution. She was able to categorize these 20families into two subgroups on the basis of the mechanisms32they used to handle conflict. One subgroup was labeledDenial, with families claiming to have no problems; the otherwas labeled Escalation, with families claiming to have toointense and too numerous problems to resolve. Denial familieswere characterized as perfect families with conscientious,overachieving members. The two subgroups seem parallel withthe Perfect and Chaotic Family types, respectively. Thesocioeconomic status (SES) of the Denial families was higherthan the population average, whereas the SES of the Escalationfamilies did not differ from that of the general population.The patients were also younger in the Escalation families thanin the Denial families; however, there were no differences induration of illness.Strober and Yager (1984) have similarly observed twosubgroups of anorexic families. One subgroup is characterizedby excessive cohesion, limited outside contacts, low emotionalexpressivity, and a lack of permissiveness. The othersubgroup’s attributes are high conflict, broken homes, angerand marital discord, and threats of abandonment. Again,parallels with Perfect/Overprotective and Chaotió Family typesare noteworthy.Family characteristics and interaction patterns:Clinical data. The above theories suggest the followingfamily issues may be associated with eating disorders:enmeshmerit, overprotection, and independence/dependenceconflicts; lack of conflict resolution; “perfect” versuschaotic family presentations; marital relationship problems;-33achievement orientation; traditional sex roles; and importanceof weight and appearance. Similar characteristics andinteraction patterns are apparent in the clinical literature.For instance, regarding enmeshment, overprotection, andindependence/dependence conflicts, many authors have describedthe families of anorexic and buliinic individuals as loyal andcohesive (e.g., Dare, 1985; Roberto, 1986; Wooley & KearneyCooke, 1986; Wooley & Wooley, 1984). This is said to causeproblems during adolescence when progress towards separationfrom the family, autonomy, individuation, and identityformation is expected (e.g., Kalucy et al., 1984; Wooley &Kearney-Cooke, 1986; Wooley & Wooley, 1984). The family mayonly permit certain feelings to be felt. and certain cognitivemodels to be adopted, preventing the child from recognizingher own feelings or developing a sense of personal identity(e.g., Guidano, 1988; Guidano & Liotti, 1983). Psychoanalyticand developmental object relations writers hypothesize thatparenting problems begin in infancy and include expecting thechild to conform to the parents’ needs, and not toleratingaggression or attempts to separate (e.g., Sours, 1974; Stern1986). It has also been suggested that the parents have neverindividuated from their own parents (Humphrey & Stern, 1988;Stern, Whitaker, Hagemann, Anderson, & Bargman, 1981).Enmeshment issues have also been described in variousclinical samples. For instance, Morgan and Russell (1975)reported that excessive emotional dependence, especiallybetween mother and daughter, was common in their sample of 4134anorexic patients. Similarly, Kaffman (1987; Kaffman & Sadeh,1989) found overly dependent or enmeshed mother-daughterrelationships in over three-quarters of a sample of 66anorexic (restrictive and bulimic combined) patients, andcompared this with a rate of 34% in an unspecified psychiatriccontrol group. Data were obtained through interviews andunspecified questionnaires. From their experience with aparents of anorexics support group, Lewis and MacGuire (1985)reported that closeness between mothers and daughters was atheme in the groups, but that they had observed morepronounced closeness between autistic children and theirmothers. Finally, Norris and Jones (1979) evaluated 10anorexic patients and their families via interviews, aquestionnaire, family therapy, and parent groups, and reportedon characteristics found in at least 9 families. Patientswould not say anything negative about their families, dyadicrelationships were enmeshed, members claimed, to share valuesand goals, and independence strivings of the adolescent werenot expressed.Various authors have also suggested that communicationproblems are common in the families of anorexic and bulimicindividuals (e.g., Guidano & Liotti, 1983). Parents mayconceal problems and avoid expressing personal emotions andopinions (Guidano, 1988), or family members may be poor atcommunicating about emotions (Rakoff, 1983). The family maynot tolerate the expression of negative feelings and insteadencourage their suppression (e.g., Edwards, 1987; Orbach,351986; Wooley & Wooley, 1984). Dare (1985) suggested that theproblem is lack of conflict resolution which may occur in thecontext of conflict avoidance or unending arguments. In theirsample of 10 anorexic patients and their families, Norris andJones (1979) found a lack of conflict resolution. Noordenbos(1987) gave questionnaires to 108 anorexic individuals and 79dieting women and interviewed 37 of the anorexics. Parents ofanorexics were reported to be less able to express positiveand negative emotions, and to talk less and more negativelyabout bodily development and sexuality. The anorexics wereless able to express emotions and opinions, and were moreconforming and approval—seeking.The distinction between perfect and chaotic families hasbeen linked to eating disorder subtypes. Most authors suggestthat the families of bulimic and buliiuic anorexic individualsare more conflicted, disorganized, and abusive., whereas thefamilies of restrictive anorexic individuals are more cohesiveand pseudoharmonious (e.g., Garner et al., 1984; Kalucy etal., 1984; Kog & Vandereycken, 1985; Wooley & Wooley, 1984).However, it has also been suggested that buliiuics’ familiespresent in an idealized and problem—free manner (Humphrey &Stern, 1988), and Kaffman and Sadeh (1989) reported that, ofsix pairs of sisters concordant for anorexia nervosa, fivepairs were comprised of one restrictive anorexic and onebulimic anorexic. In Norris and Jones’s (1979) sample of 10anorexic patients and families, an idealized family myth ofcloseness, absence of conflict, and harmony was exhibited. In36a sample of 56 families with an anorexic member, the earlydevelopmental history was presented as trouble-free (Kalucy etal., 1977). And in a parents of anorexics support group,after 2 years parents were still unwilling to consider thatcertain family dynamics might be contributing to the eatingdisorder (Lewis & MacGuire, 1985). In a sample of 172 bulimicwomen, 47% were found to have a history of child sexual abuse(29%) or child physical abuse (29%) (Root & Fallon, 1988), andin another sample of 35 bulimic women, 34% reported a historyof family child sexual abuse (Bulik, Sullivan, & Rorty, 1989).Similarly, another research group has found a history ofchildhood sexual abuse in 31% of a sample of 158 eatingdisordered clients (Palmer, Oppenheimer, Dignon, Chaloner, &Howells, 1990). No associations were found between the rateor type of abuse and the eating disorder subtypes. Mothers ofbulimic women have also been reported to have been victims ofchild sexual abuse at a higher than expected rate (Root etal., 1986).Finally, in terms of family interaction patterns, theparents’ marital relationship has been described by variousauthors. It has been suggested that the parents ofindividuals with eating disorders rarely divorce despitehaving poor relationships (Guidano & Liotti, 1983). Parentsof bulimics are believed to have conflictual relationships andhigher rates of divorce, whereas parents of anorexics arebelieved to show pseudoharmony and covert conflict in theirrelationships (e.g., Gordon, Beresin, & Herzog, 1989; Roberto,371986; Schwartz et al., 1984; Wooley & Wooley, 1984). Inclinical samples of parents of anorexics, authors reportmarital relationship problems (e.g., Lewis & MacGuire, 1985;Norris & Jones, 1979; Taipale, Tuomi, & Aukee, 1971). Forinstance, Morgan and Russell (1975) reported that there wasserious disharmony between the parents in 24% of their sampleof 41 anorexics. And Kalucy et al. (1977) reported that in41% of their sample of 56 families of anorexics, parents neveror rarely had sexual relations. Kaffman (1987) reporteddivorce rates of 6% of restrictive anorexics’ parents and 33%of bulimic anorexics’ parents, as compared with a generalpopulation divorce rate of 19%. However, failing marriages ormarital conflict were reported in 50% of the restrictiveanorexic sample, versus 25% of the bulimic anorexic sample.There are also clinical data and descriptions regardingcharacteristics of the families of individuals with eatingdisorders besides their interaction patterns. For instance,the families of anorexics are said to have high achievementexpectations (e.g., Edwards, 1987), and the families ofbulimics are said to value success and achievement, and havehigh standards of achievement in many domains (e.g., Humphrey& Stern, 1988; Roberto, 1986). In their sample of 10 anorexicpatients and their families, Norris and Jones (1979) reportedthat the families valued status and achievement and had highexpectations regarding work, sports, and moral issues.Similarly, mothers of 13 anorexics were reported to have highexpectations for their daughters (Taipale et al., 1971).38The transmission of cultural sex role attitudes by thefamily may also be an important factor in eating disorders(Wooley & Wooley, 1984). It has been suggested that theparents of bulimics may strongly emphasize women conforming totraditional female sex role characteristics (Mizes, 1985), orthat parents may show rigid polarization of sex—rolestereotypic traits (Wooley & Kearney—Cooke, 1986). Fathersmay be powerful, successful, and emotionally distant, whilemothers may be powerless and nurturant, and have abandonedcareers (e.g., Boskind-White & White, 1987; Wooley & Kearney—Cooke, 1986). Similarly, anorexic women may have been raisedin families with strong patriarchal values or stereotypicparental gender role models (Edwards, 1987; Gordon et al.,1989). Noordenbos (1987) found that anorexic women reportedthat there was a strong traditional division of roles betweentheir parents.The family may also augment societal values on thinnessand appearance (Boskind-White & White, 1987; Garner & Bemis,1984). For instance, the anorexic individual’s family membersmay be concerned with issues of weight, fitness, eating, anddieting (e.g., Bends, 1978; Goodsitt, 1974; Taipale et al.,1971), and these issues may have special family meanings ofself-control, self—esteem, and emotional expression (Kalucy etal., 1984). Similarly for bulimic individuals, family membersmay reinforce the adolescent’s obsession with her shape,weight, and appearance (e.g., Boskind-White & White, 1987;Humphrey & Stern, 1988; Roberto, 1986; Striegel-Moore et al.,391986). In a sample of 56 families of anorexics, disturbedeating habits were found in 23% of the families (Kalucy etal., 1977). There was a history of low adolescent weight,weight phobia, or anorexia nervosa in 16% of mothers and 23%of fathers; and 27% of mothers and 16% of fathers were dietingas adults. Kaffman and Sadeh (1989) reported that 83% ofrestrictive anorexics’ and 58% of bulimic anorexics’ parentsshowed preoccupation with food, weight, and dieting forreasons of health or aesthetics. And in a parents ofanorexics support group, mothers and some fathers reportedlyshared their daughters’ fears that they would lose control andbegin overeating (Lewis & MacGuire, 1985).Family characteristics and interaction patterns:Controlled studies. While the above theories and data aresuggestive, little can be concluded from them because ofmethodological problems such as lack of control groups ornormative data, unspecified dependent measures, and theretrospective nature of patient reports. However, there hasbeen some better—controlled research conducted on thecharacteristics and interaction patterns of the families ofeating—disordered individuals.Heron and Leheup (1984) compared the case records of 16adolescent anorexics and 40 adolescent control patients. Theyfound no differences between groups in the number of intactfamilies (12 of 16 anorexics, 27 of 40 control patients).There were significant differences between groups in: degreeof closeness in the family (15/16 anorexics versus 10/4040control patients had families in which members spent most oftheir spare time together); exclusivity of family (13/16anorexics versus 5/40 control patients); external stresses onthe family (4/16 versus 32/40); and satisfaction with familyrelationships (14/16 versus 9/40 happy with their family).Goldstein (1981) compared 11 anorexics’ families withdata on families in which an offspring developedschizophrenia. The anorexics’ parents did not exhibit thecombination of high communication deviance and high negativeaffective style common to the schizophrenics’ parents.Criticism was not observed in any of the anorexics’ familydiscussions. Goldstein also compared the anorexics’ familieswith 5 nonanorexic inpatients’ families and found theanorexics’ parents to be more dependent and insecure. NeitherGoldstein (1981) nor Heron and Leheup (1984) described indetail how their constructs were operationalized.Some researchers have attempted to operationalizeMinuchin’s variables of enmeshment, rigidity,overprotectiveness, and lack of conflict, resolution. Forinstance, Kog and Vandereycken (1989) compared 30 eating-disordered patients’ families with 30 normal controls’families and found that the eating-disorder families showedmore conflict avoidance on behavioral tasks than the controlfamilies. In addition, anorexic daughters perceived theirfamilies as more cohesive than bulimic and control daughtersdid, whereas bulimic daughters perceived more disorganizationin their families than anorexic or control daughters did.41Harding and Lachenmeyer (1986) administered the StructuralFamily Interaction Scale (Perosa, Hansen, & Perosa, 1981) to30 adult anorexics and 30 college control subjects. Thisscale was intended to measure Minuchin’s constructs ofenmeshment, overprotectiveness, and rigidity. No differenceswere found.Sights and Richards (1984) administered structuredinterviews to 6 bulimic and 6 nonbulimic college women withand without their parents present. Blind raters coded thetranscripts using the Parental Characteristics Rating Scale,an instrument developed for the purposes of the study.Bulimics’ mothers were judged to be more domineering andcontrolling, and to have higher expectations of theirdaughters. Both parents of the bulimic women were thought tobe more demanding and likely to compare siblings openly. Witha larger sample of 38 bulimic women and 40 normal controlwomen and also employing a questionnaire specifically devisedfor use in the study, Dolan et al. (1990) found that bulimicsreported less parental attention and affection towards themand greater parental marital conflict than the comparisongroup. No differences were found regarding emphasis onacademic achievement or importance of traditional femaleroles.Recently, a number of research groups (McNamara &Loveman, 1990; Steiger, Liquornik, Chapman, & Hussain, 1991;Waller, Slade, & Calam, l990b) have investigated the familyfunctioning of eating-disordered subjects using the Family42Assessment Device (Epstein, Baldwin, & Bishop, 1983). Thismeasure has seven subscales——general family functioning,problem—solving, roles, couuuunication, affectiveresponsiveness, affective involvement, and behavior control——and has respondents describe their family as it was while theywere growing up. McNamara and Loveman (1990) compared thereports of 30 bulimic, 61 repeat dieter, and 59 nondieterundergraduate women and found that the bulimic subjectsdescribed their families as more enmeshed and intrusive, moreemotionally disengaged or unresponsive to members’ needs, lessskilled in communication and problem—solving, and lessstructured or rule-governed than the control subjects.Another research group found similar results with the FamilyAssessment Device with 30 control, 34 bulimic, and 14 anorexicadult female subjects and their parents (Wailer et al.,1990b). Anorexic and bulimic subjects rated their families asmore dysfunctional than the control subjects did on all of thesubscales of the test; however, the eating—disorderedsubjects’ mothers rated their families as more dysfunctionalon only two subscales, and there were no group differencesamongst fathers. Steiger et al. (1991) also found nodifferences between adult restricter and binger eating—disorder subtype subjects on the Family Assessment Device, butdid find that eating disorder subjects perceived theirfamilies as being less well—functioning than control subjectsdid.43Sixteen families with a bulimic anorexic daughter (meanage 18 years) and 24 families of women with no psychologicalproblems were observed in mother—father—daughter triads usingthe Marital Interaction Coding System (MICS; Robin & Weiss,1980) and the Structural Analysis of Social Behavior (SASB;Benjamin, 1974) (Humphrey, Apple, & Kirschenbaum, 1986). TheMICS is a cognitive-behavioral coding system and the SASB isbased on interpersonal theory. Both the cognitive—behavioraland interpersonal coding systems discriminated between groups.In particular, buliiuic anorexics’ families were seen as morenegative, less positive, and more contradictory inconununication. There were no differences in problem—solving.Humphrey (1989) also employed the SASB observational systemwith 16 restrictive anorexic, 16 bulimic, 18 bulimic anorexic,and 24 normal control subjects (mean age 18 years) and theirparents. Restrictive anorexic subjects’ parents were reportedto communicate a mixed message of nurturant affection combinedwith neglect of their daughter’s needs; bulimic subjects’relationships with their parents were reportedly hostile; andnormal control subjects’ parents were observed to be helpfuland positive toward their daughters. No distinct pattern wasobserved in the bulimic anorexic group.Wonderlich and Swift (1990b) employed the SASB selfrating scales with 11 restrictive anorexic, 26 normal—weightbulimic, 11 bulimic anorexic, and 29 control adult femalesubjects. They did not find the eating disorder subtypedifferences they had predicted regarding control/submission or44attack/hostile withdrawal. They did find that bulimia subtypesubjects perceived their parental relationships as morehostile than control subjects did. Also using the SASB self-rating scales, Humphrey (1986b) found eating disorder subtypedifferences amongst 20 restrictive anorexic, 20 bulimic, 20bulimic anorexic, and 20 control subjects (mean age 19 years).Bulimic subjects perceived greater deficits in parentalnurturance than did subjects in the other three groups.Bulimic and buliiuic anorexic subjects also perceived lessparental empathy and nurturance than did control subjects.Both anorexic and bulimic subjects perceived greater parentalblaming, rejecting, and neglecting than control subjects did.Lucido and Abrainson (1988) gave the Sexual EventsQuestionnaire, which measures adverse sexual experiencesbefore age 12 years, and the Bulimia Test, which measures DSM—111-defined bulimia, to 125 bulimic or nonbulimic women.There were 63 questionnaires returned from 16 bulimic and 47nonbulimic women (mean age 34 versus 23 years). Childhoodsexual experiences were reported by 69% of bulimic and 70% ofnonbulimic women. However, bulimic women reported more sexualexperiences with fathers or brothers, a greater number ofsexual experiences, and more fear or shock reactions to theevents. For instance, 2 or more negative sexual experienceswere reported by 46% of bulimic versus 6% of nonbulimicsubjects. Also, 100% of the bulimic women kept theexperiences secret as compared to 64% of the nonbulimic women.45Researchers compared. 30 adult restrictive anorexic womenwith 38 bulimic anorexic women on the Childhood EventsQuestionnaire (CEQ) (Piran et al., 1988). The CEQ (Barnes,Ennis, & Trachtenberg, 1985) measures events occurring insubjects’ families during childhood. Bulimic anorexics’families in childhood were characterized by greater financialproblems, unemployment, interpersonal violence between parentsand towards children, disagreement with parents, and substanceabuse in a parent or sibling. Adult-child sexual abuseoccurred in 8% of the bulimic anorexics’ families and criminalconvictions occurred in 5% of the bulimic anorexics’ families;these events were never reported in the families ofrestrictive anorexics. No socioeconomic class differenceswere found between the groups.Five recent studies have been conducted using theParental Bonding Instrument (PBI; Parker, Tupling, & Brown,,1979). This instrument measures how individuals remembertheir parents from childhood, and consists of Care (warmth,affection, empathic responsiveness) and Protection (control,overprotectiveness, intolerance of autonomy) subscales. Onestudy included 56 bulimic women and 30 control women (Pole,Wailer, Stewart, & Parkin-Feigenbaum, 1988). Buiimic womenperceived their mothers as less caring than nonbuiimic womendid. Similarly, Fichter and Noegel (1990) found that 27buiimic twins perceived less maternal and paternal care, andmore maternal and paternal overprotection than control•subjects. Other researchers administered the PBI to 3546anorexic and 37 bulimic women, and compared their scores topublished data on 40 normative subjects (Palmer, Oppenheimer,& Marshall, 1988). Anorexic and bulimic women had lowerperceived maternal Care scores and bulimic women had lowerpaternal Care scores. Another research group gave the PBI to31 anorexic, 34 bulimic with a history of anorexia, 33 bulimicwithout a history of anorexia, and 242 control adult femalesubjects (Calam, Wailer, Slade, & Newton, 1990). The eating-disordered women recalled less maternal and paternal care andmare paternal overprotection than did control women.Contrasting the individual clinical groups, bulimic subjectswith a history of anorexia were found to perceive theirfathers as less caring, while bulimics without such a historyperceived both parents as less caring. Finally, the PBI wasgiven to 15 restrictive anorexics, 9 bulimic anorexics, 21normal-weight bulimics, 13 bulimics with a history of.anorexia, and 24 non—eating—disordered women (Steiger, Van derFeen, Goldstein, & Leichner, 1989). The eating-disorderedwomen perceived less paternal caring.Other researchers have administered the Family AssessmentMeasure (FAN; Skinner, Steinhauser, & Santa—Barbara, 1983) tothe families of eating-disordered individuals. This measurehas the following subscales: Task Accomplishment, RolePerformance, Communication, Affective Expression, AffectiveInvolvement, Control, Values and Norms, and SocialDesirability. In one study, the FAN was administered to 41adolescent anorexics (restrictive and bulimic mixed) and their47parents, and 24 nonanorexic adolescents and their parents(Garfinkel et al., 1983). In the anorexic group, the mothersand daughters scored in a significantly more pathologicaldirection on Task Accomplishment, Role Performance,Communication, and Affective Expression, and had significantlylower scores on Social Desirability, than the nonanorexicgroup mothers and daughters. There were no differences forfathers. Garner et al. (1985) administered the FAN to 59bulimic anorexic, 59 normal—weight bulimic, and 59 restrictiveanorexic adult women. The two bulimic groups had scores aboveT—values of 60 on six of the seven FAN subscales (excludingRole Performance); the restrictive group had no elevationsabove 60. Differences between groups were significant for allseven subscales. The Social Desirability subscale wasanalyzed separately, showing the restrictive anorexic group tohave higher scores than the bulimic groups.In the Garfinkel et al. (1983) study above, the EatingAttitudes Test (EAT), Restraint Scale, and body sizeestimation and satisfaction measures were also administered toparents and daughters. Parents showed no differences withrespect to abnormal attitudes to weight control and dieting,or body size estimation or satisfaction on these measures. Amore recent study, however, did find differences in weight—and eating-related behaviors and attitudes between a group of39 mothers of adolescent daughters who reported a level ofdisordered eating comparable with clinical bulimic samples anda group of 38 mothers whose daughters reported a low level of48eating disturbance (Pike & Rodin, 1991). Compared withmothers of non—eating—disordered daughters, mothers whosedaughters were eating disordered had higher scores on acomposite of the Drive for Thinness, Bulimia, and BodyDissatisfaction subscales of the Eating Disorder Inventory,began dieting at a younger age, thought their daughters shouldlose more weight, and rated their daughters as less attractivethan the daughters judged themselves. The mothers did notdiffer with respect to current Body Mass Index, maximum weightloss, desired weight loss, or ratings of their ownattractiveness and weight.Finally, the Family Environment Scale (FES) has been usedin a number of studies. The FES has 10 subscales: Cohesion,Expressiveness, Conflict, Independence, AchievementOrientation, Intellectual-Cultural Orientation, Active-Recreational Orientation, Moral—Religious Emphasis,Organization, and Control (see Table 1 for definitions).Johnson and Flach (1985) administered the FES to 105 adultbulimic patients and 86 nonbulimic college students. Thebulimic subjects perceived their families as lower onCohesion, Expressiveness, Independence, Intellectual—CulturalOrientation, Active—Recreational Orientation, and Moral—Religious Emphasis, and as higher on Conflict. There were nodifferences in Achievement Orientation, Organization, andControl. In a regression analysis, Organization andAchievement Orientation were important predictors of severityof bulimia, along with three eating—related measures.49Table 1FES Subscale Definitions1. Cohesion the degree of commitment, help, andsupport family members provide for oneanother2. Expressiveness the extent to which family members areencouraged to act openly and to expresstheir feelings directly3. Conflict the amount of openly expressed anger,aggression, and conflict among familymembers4. Independence the extent to which family members areassertive, are self—sufficient, and maketheir own decisions5. Achievement the extent to which activities are castOrientation into an achievement—oriented orcompetitive framework6. Intellectual- the degree of interest in political,Cultural social, intellectual, and culturalOrientation activities7. Active— the extent of participation in socialRecreational and recreational activitiesOrientation8. Moral-Religious the degree of emphasis on ethical andEmphasis religious issues and values9. Organization the degree of importance of clearorganization and structure in planningfamily activities and responsibilities10. Control the extent to which set rules andprocedures are used to run family lifeNote. Based on Moos & Moos, 1986 (p. 2).50Another research group employed the FES with a sample of24 normal-weight bulimic (without a history of anorexianervosa), 13 bulimic anorexic, and 41 normal control adultfemale subjects (Shisslak, McKeon, & Crago, 1990). They foundthat both bulimic groups characterized their families as lesscohesive, less expressive, less oriented toward social andrecreational activities, and more conflictual than the controlgroup did. The bulimic anorexic subjects perceived theirfamilies as less encouraging of independence than the normal—weight bulimic and control subjects did.Strauss and Ryan (1987) administered six subscales of theFES (Cohesion, Expressiveness, Conflict, Independence,Organization, and Control) to 19 restrictive anorexics, 14bulimic anorexics, and 17 women without eating disorders. Theanorexic groups perceived less expressiveness and cohesionthan the control group, and the iestrictive anorexicsperceived more conflict than the control subjects. Strober(1981) examined FES scores in the families of 22 bulimicanorexic and 22 restrictive anorexic adolescents. The parentsjointly completed the FES trying to disregard changes due tothe anorexia nervosa. The bülimic anorexics’ familyenvironments were characterized as less cohesive and organizedand as more conflicted than the restrictive anorexics’families. The Short Marital Adjustment Test was alsoadministered with instructions to depict the marriage as itwas before the onset of anorexia. Parents of the bulimicanorexics reported higher marital discord. And in a51semistructured family interview, bulimic anorexic girls wererated as more distant from their parents than restrictiveanorexic girls.Stern et al. (1987) administered the FES to 114 women anda parent of each (all but 2 were mothers). There were 20restrictive anorexics, 13 bulimic anorexics, 24 normal—weightbulimics without a history of anorexia, and 57 age—matchedcontrol subjects. Subjects were instructed to respond to theFES based on how the family was when the daughter was livingat home. In all the eating disorder groups, daughters ratedtheir families as less expressive than control daughters did.Restrictive anorexic daughters had lower Active—RecreationalOrientation scores than control daughters, and bulimic andbulimic anorexic daughters had lower Cohesion scores thancontrol daughters. The parents of bulimic anorexics perceivedmore family conflict than control parents. Finally, bulilnicdaughters perceived greater Achievement Orientation thancontrol daughters, and the parents of bulimics perceivedgreater Achievement Orientation than the parents of bulimicanorexics. There was also a significant interaction such thateating-disordered daughters rated their families as higher onAchievement Orientation than their parents did, whereascontrol daughters rated their families as lower on this scalethan their parents did.Two studies have involved administering the FES alongwith the Family Adaptability and Cohesion Evaluation Scale(FACES; Olson, Bell, & Portner, 1978). The FACES is intended52to measure the degree of perceived family cohesiveness andadaptability, and to describe a range of family funätioningfrom chaotically disengaged to rigidly enmeshed. Ordiuan andKirschenbaum’s (1986) study included 25 bulimic women and 36control women. On the FES, bulimics had lower Cohesion,Expressiveness, and Active—Recreational Orientation scores,and higher Conflict scores than control subjects. On theFACES, bulimics were lower on Cohesion and SocialDesirability. In a discriminant function analysis, FESIndependence discriminated best between groups. Humphrey(1986) administered these scales to 16 bulimic anorexics andtheir parents and 24 nondistressed family triads. The meanage of the daughters was 18 years. The FES and FACES itemswere factor analyzed and only factor scores were compared inthe study; thus, results are not directly comparable to thoseof other studies. There were 8 FES and 8 FACES factors.Bulimic anorexics’ families were characterized by mothers,fathers, and daughters as less involved and supportive (FESand FACES), more isolated and nondisclosing (FES and FACES),and as being more detached and having poorer boundaries(FACES). Bulimic daughters characterized their families asmore conflictual (FES and FACES) and unstructured (FES).Another research group administered the FACES to 41 eatingdisordered and 27 control adult female subjects (Wailer,Slade, & Calam, 1990a). They found that the eating-disordergroup perceived their families as lower in adaptability andcohesion than the control group did.53Some more recent research with the FES has examinedwhether the perceived family environment of bulimic womenvaries with level of depression, borderline personalitydisorder features, or history of childhood sexual abuse and,therefore, may not be directly associated with the eatingdisorder per se (Blouin, Zuro, & Blouin, 1990; Bulik et al.,1989; Johnson, Tobin, & Enright, 1989). Blouin et al. (1990)administered the FES to 81 depressed bulimic, 18 non-depressedbulimic, and 37 normal control subjects. They found that thedepressed bulimic women characterized their families as lesscohesive, less active in recreation, and more controlling thanthe nondepressed bulimic or control women did. Depressedbulimic subjects also perceived their families as lessexpressive, less encouraging of independence, and moreachievement—oriented than control subjects did.Johnson et al. (1989) administered the FES to 43 bulimicpatients with borderline personality features and 27nonborderline bulimic patients. The borderline personalitybulimic women described their families as less cohesive, lessexpressive, more conflictual and controlling, lessindependence—encouraging, and less oriented towardintellectual or recreational activities than the nonborderlinebulimic women did. And, finally, Bulik et al. (1989) comparedthe responses of bulimic women who had (n=12) or did not have(n=23) a history of family childhood sexual abuse and foundthat the sexual abuse group described their families as lesscohesive, less emphasizing of moral or religious issues, and54more conflictual. It is of note that the level of bulimicsymptomatology was very similar between groups in all threestudies, regardless of status with respect to depression,borderline features, or sexual abuse history, again suggestinga lack of association between eating disorder status andreported family environment.Woriderlich and Swift (1990a, 1990b) have conductedresearch similar to the above but employing the StructuralAnalysis of Social Behavior rather than the FES, and examininga mixed anorexic and bulimic eating disorder group rather thanbulimics only. For instance, these researchers comparedeating-disordered female adult subjects with borderlinepersonality disorder (n=ll), with other personality disorders(n=22), or with no personality disorder (n=13) to 29 normalcontrol subjects (Wonderlich & Swift, 1990a). They found thatthe borderline personality disorder eating—disordered subjectsperceived greater hostility in their parental (especiallymaternal) relationships than subjects in the three othergroups. Other personality disorder eating—disordered subjectsperceived greater hostility in their parental relationshipsthan normal control subjects on some measures. Similarly,dividing eating disorder subjects into high (n=34) and low(n=14) dysthymia groups, these investigators found that onlythe dysthymic subjects perceived greater hostility in theirparental relationships than did controls (Wonderlich & Swift,1990b) .55To summarize the controlled studies, many researchershave employed the FES with adult female samples and resultshave consistently portrayed the families of women with bulimiaand bulimic anorexia as less cohesive and expressive thancontrol families. Women with bulimia nervosa have also fairlyconsistently been shown to perceive their family environmentsas more conflictual and less active and recreationallyoriented than controls. The majority of the FES research hasfocused on adult or young adult women with bulimia nervosa orbulimic anorexia. Two studies included restrictive anorexicwomen and consistently found that these women described theirfamilies as less expressive than controls (Stern et al., 1987;Strauss & Ryan, 1987). Other findings with the FES have beenless consistent, and no researchers have administered theinstrument to subjects in early adolescence. The one study onearly adolescents involved administration of the FES to theirparents (Strober, 1981). It is of note that this has been theonly one of three possible FES studies to findrestricter/binger differences. In total, three studies haveexamined the responses of parents on the FES: A fairlyconsistent finding has been that the parents of bulimic andbulimic anorexic subjects portray their families as lesscohesive and more conflictual than the parents of restrictiveanorexic or control subjects do (Humphrey, 1986; Stern et al.,1987; Strober, 1981). Finally, recent evidence has suggestedthat the family environments described by bulimic women maynot be associated with their eating disorder status per se,56but rather with their level of depression, borderlinepersonality disorder features, or history of family sexualabuse.Researchers using other measures have provided evidencethat the families of restrictive anorexics appear harmoniousand satisfied whereas the families of bulimics and bulimicanorexics appear conflictual and chaotic. There have beeninconsistent results in studies directly comparing restrictiveand bulimic eating disorder subtypes regarding the existenceof restricter/binger differences in family interaction. Therehas been one attempt to measure the sex role attitudes inbulimic families: No difference from control families wasfound (Dolan et al., 1990). The two attempts at measuringeating- and weight-related attitudes in the parents ofadolescents with eating disorders have yielded inconsistentresults (Garfinkel et al., 1983; Pike & Rodin, 1991).Achievement attitudes have been measured via the FES andresults have been inconsistent. In one study (Sights &Richards, 1984), bulimics’ mothers were rated as having higherexpectations of their daughters than nonbuliiuics’ mothers;however, there were only 6 families in each group so resultswere tentative. Another study found no differences inperceptions of family achievement orientation between bulimicand control subjects (Dolan et al., 1990).A few methodological issues regarding this research areimportant to note. For instance, one must distinguish betweenstudies which investigate the eating—disordered woman’s57perception of her family and those which actually examine thebehavior or attitudes of other family members as well. Lessthan half (30%) of the studies reviewed in this sectioninvolved both the eating-disordered daughter and one or bothof her parents. Another consideration is whether the daughteris an adolescent who still lives with her parents, or is anadult and/Or no longer living at home. The vast majority(85%) of studies reviewed herein involved adult subjects.Related to this is the issue of whether the subject isresponding on the basis of her present situation, or is beingasked to retrospectively describe her childhood, or respond asif she were still living at home or did not have an eatingdisorder. To overcome the problem of retrospectivity ofsubject reports, one could have adolescents who live withtheir parents respond to measures so as to reflect theircurrent experience. This would also yield information on thefamily characteristics present in the early stages of aneating disorder.A final problem is the lack of psychiatric control groupsin these studies, making it unclear whether findings arespecific to eating disorders or reflect patterns which wouldcontribute to a wide range of disorders. Regarding the FESresults, it has been suggested that the results found forbulimics’ families may be similar to what would be found forother distressed families (Johnson & Flach, 1985; Ordman &Kirschenbaum, 1986). The family interaction patternformulations themselves may not be specific to eating58disorders. For example, Bruch’s theories were meant toencompass certain cases of obesity and schizophrenia, andMinuchin’s constructs pertain to psychosomatic diabetic andasthmatic as well as anorexic cases. The formulations do notexplain why a disorder of eating in particular should appear(e.g., Strauss & Ryan, 1987; Strober & Humphrey, 1987; Yager,1982). Perhaps the postulated family characteristics ofweight and appearance consciousness, traditional sex roles, orachievement orientation would be present in the families ofeating-disordered individuals more often than in the familiesof adolescents with other disorders. Currently, however,there is no controlled evidence regarding this hypothesis.Summary and HypothesesRegarding family factors associated with eatingdisorders, it had been suggested that anorexics and possiblybulimics came from the upper socioeconomic classes and/or fromfamilies with high aspiration levels; however, more recentresearch suggests that individuals with eating disorders maycome from diverse socioeconomic backgrounds. Eating disordersmay be more prevalent in the relatives of eating—disorderedindividuals than in the general population. Also, the parentsof buliinics and buliinic anorexics may have higher rates ofdepression and alcoholism and be more impulsive than theparents of restrictive anorexics. Similarly, bulimicindividuals themselves may be more depressed and impulsive,and use alcohol more than restrictive anorexic individuals.59Theoretical formulations and clinical data have suggesteda number of family characteristics and interaction patterns aspossibly contributing to anorexia nervosa and bulimia nervosa.Family interaction patterns include enmeshment,overprotection, and independence/dependence conflicts; lack ofconflict resolution; perfect versus chaotic familypresentations; and marital relationship problems. There issome support from controlled studies that the families ofrestrictive anorexics appear cohesive and harmonious whereasthe families of bulimics and bulimic anorexics appearconflictual and disengaged. Recent evidence suggests that thefamily environment associated with bulimia may be a functionof subjects’ level of depression, borderline personalityfeatures, or sexual abuse history rather than a function ofthe eating disorder. The family characteristics suggested bythe theoretical and clinical literature include achievementorientation, traditional sex roles, and importance of weightand appearance. Little controlled research has addressedthese issues and that which has has yielded inconsistentresults. Methodological issues in the research on familyfactors include lack of psychiatric control groups,retrospectivity of reports of adult subjects, and lack ofdirect study of parents.In an attempt to contribute to the literature on familyfactors in anorexia nervosa and bulimia nervosa, the presentstudy included four groups of female adolescents and theirmothers: restrictive anorexic, bulimic type, psychiatric60control, and nonpsychiatric control. Including controlfamilies in which the daughter is presenting for treatment forreasons other than an eating disorder yields informationregarding the specificity of family characteristics toanorexia nervosa and bulimia nervosa. The selection ofadolescents rather than adult women with eating disordersovercomes to a large extent the problem of retrospectivity ofself-report and enables one to investigate family variablesduring the early stages of an eating disorder. Includingmothers as well as daughters provides two perceptions of thefamily environment and allows comparison of these differentperceptions. Another contribution of this study was toexamine the family characteristics of achievement orientation,traditional sex roles, and importance of weight andappearance, as well as the family interaction patterns morecommonly investigated.Thus, the aim of the present study was to investigate thecharacteristics and interaction patterns present in thefamilies of adolescent eating—disordered patients, asperceived by both the adolescent and her mother, and todetermine which, if any, of these family factors are specificto anorexia nervosa or bulimia nervosa. Consistent with thetheoretical formulations and some of the research resultsreported above, the specific hypotheses are as follows:1. The families of bulimic type and psychiatric controlsubjects will be characterized as more dysfunctional than thefamilies of nonpsychiatric control subjects, whereas thefamilies of restrictive anorexic subjects will becharacterized as more similar to the families ofnonpsychiatric control subjects.2. The restrictive anorexic and bulimic type mothers anddaughters will be characterized as higher in achievementorientation, traditional sex role ideology, and weight andappearance orientation than the psychiatric control mothersand daughters.6162MethodSublectsFour groups of female adolescents——restrictive anorexic,bulimic type, psychiatric control, and nonpsychiatric control—-participated with their mothers in this study. There were 97mother-daughter pairs in total, the majority (94%) of whomwere Caucasian. The recruitment strategies, criteria forinclusion, and resulting number of pairs per group will now bedescribed.Nonpsychiatric control subjects were recruited throughnotices posted in high school newsletters, local newspapers,and community centers. To be included in the study, daughterscould not meet DSM—III—R criteria for anorexia nervosa orbulimia nervosa, have a history of such a diagnosis, or scoreabove the cut-off point (i.e., 20) on the Eating AttitudesTest (EAT—26; Garner, Olmsted, Bohr, & Garfinkel, 1982), awidely used measure of the symptoms of anorexia nervosa andbulimia nervosa. In addition, daughters could not have ahistory of treatment for psychological problems. Similarly,no sibling in the family could have a history of treatment forpsychological problems or be known to have had an eatingdisorder. On these bases, 8 of the 32 mother-daughter pairsrecruited were excluded from the analyses of the study——5because the daughter had a history of treatment forpsychological problems and 3 because a sibling had been or wasreceiving treatment for psychological problems.63Psychiatric control subjects were recruited throughmental health centers and hospital-based psychiatric treatmentservices where they were being seen at that time for non—eating-related psychological problems (excluding schizophreniaor bipolar disorder). Presenting problems includeddepression, anxiety-related disorders, and defianceoppositionality. To be included in the study, daughters hadto meet the same criteria regarding lack of eating disorderdiagnosis and symptoms that the nonpsychiatric controldaughters did. Similarly, their siblings could not be knownto have had an eating disorder. They were not excluded,however, on the basis of a sibling having a history oftreatment for psychological problems. The application ofthese criteria resulted in 5 of the 25 mother-daughter pairsrecruited being excluded from the analyses of the study. Onedaughter had a probable history of anorexia nervosa, one had ahistory of bulimia nervosa, and three had scores above thecut-off on the Eating Attitudes Test.Restrictive anorexic and bulimic type subjects wererecruited from a hospital—based adolescent eating disordertreatment center where they were being seen at that time fortreatment. To be included in the restrictive anorexic group,daughters had to meet the DSM-III-R diagnostic criteria foranorexia nervosa, but not for bulimia nervosa (current orpast). At the time of participation, however, many subjects’weights were no longer 85% below that expected for their ageand height as participation was usually requested after64treatment had been ongoing for some time. For this reason,the criteria were adjusted such that a documented weight priorto participation of less than 85% of that expected wasrequired. To be included in the bulimic type group, daughtershad to meet or have recently met DSM-III--R criteria forbulimia nervosa. The reason for including subjects who nolonger met the full diagnostic criteria was that, as mentionedabove, many participants had been in treatment for some timeand had begun to make progress in overcoming some of theireating disorder symptomatology. Utilizing the diagnosticcriteria for eating—disordered subjects as delineated above, 6of the 40 mother—daughter pairs recruited were excluded fromthe analyses of the study. Two daughters met criteria foranorexia nervosa but also reported bulimic symptomatology, andfour daughters did not meet criteria for anorexia nervosa orbulimia nervosa.The inclusion criteria described in the precedingparagraphs were developed with the aim of achieving fourdistinct groups based on the status of the female adolescentstherein. The resulting number of mother-daughter pairs pergroup are as follows: restrictive anorexic n=20; bulimic typen=14; psychiatric control n=20; and nonpsychiatric controln=24. The clinical characteristics (eating/weight-related andpsychiatric syinptomatology-related) of the female adolescentswithin these groups will be presented in the Results section.Demographic data and treatment-related information regardingthe groups will now be presented.65Demographic data for the four groups are presented inTable 2; treatment-related information for the three clinicalgroups is presented in Table 3. One—way analyses of variance(ANOVAs) were conducted on the eight continuous measures(daughter’s age, mother’s age, average parental education,total parental income, number of children, treatment duration,previous treatment duration, and duration of disorder), withalpha set at .05. The Type I error rate (alpha) was not mademore stringent (i.e., adjusted for the number of measures) asit was important to detect any potentially confoundingvariables. As the number of subjects per group was unequal,the homogeneity of variance assumption was assessed byBartlett’s test and, where found untenable, corrected for bythe Welch procedure (Glass, Peckham, & Sanders, 1972; Howell,1982). Significant ANOVAs were followed by Tukey multiplecomparisons with alpha set at .05. The Tukey—Krameradjustment for unequal Ns was employed (Kirk, 1982). TheTukey test was also adjusted via the Games-Howell procedure incases of violation of the assumption of homogeneity ofvariance (Kirk, 1982). For the dichotomous measures (singleversus married maternal status, history of hospitalization,and history of psychotropic medication), analyses of varianceof proportions were conducted with alpha set at .05.Significant results were followed by the multiple comparisontechnique recommended by Narascuilo (1966), which holds alphaat the .05 level experiment—wise.66Table 2Demographic Data Means (Standard Deviations)Restrict. Bulimic Psych. Nonpsych.Anorexic Type Control ControlAge— 15.05 16.36 14.90 15.04Daughter (1.23) (1.28) (1.52) (1.76)Age— 41.90 45.64 41.25 43.54Mother (5.65) (6.03) (5.16) (3.62)AverageParental 13.78 14.32 13.75 16.08Education (2.95) (2.44) (2.78) (2.31)TotalParental 82.50 78.93 43.75 79.83Income (64.01) (60.56) (25.83) (46.92)(thousands)Number of 2.75 2.64 2.40 2.42Children (1.12) (0.84) (1.05) (0.88)Single 20.0% 14.3% 45.0% 16.7%Mothers67Table 3Treatment—Related Information Means (Standard DeviationsRestrictive Bulimic PsychiatricAnorexic Type ControlTreatment 10.40 7.14 10.55Duration (13.82) (5.71) (9.16)PreviousTreatment 4.25 1.64 10.65Duration (6.39) (3.99) (17.01)Duration 22.15 23.21 35.20Disorder (14.04) (11.47) (19.74)History ofHospital. 60.0% 14.3% 10.0%History ofMedication 30.0% 21.4% 15.0%Note. Durations are in months.68With regard to the demographic data, the above analysesshowed significant group differences on three of the sixmeasures. The ANOVA for daughter’s age was significant ((3,74) = 3.21, <.05), with multiple comparisons showing thatbulimic type daughters were significantly older thanpsychiatric control daughters. The ANOVA for mother’s age wasnot significant. The average parental education ANOVA wassignificant ((3, 74) = 3.95, p<.05), with nonpsychiatriccontrol parents completing a significantly higher averagenumber of years of education than restrictive anorexic orpsychiatric control parents. The adjusted-for-heterogeneityANOVA for total parental income was significant (‘(3, 35) =5.10, <.Ol), with parental income for nonpsychiatric controlsubjects being significantly greater than that for psychiatriccontrol subjects. As outlying extreme scores may influencemean income values, medians were also calculated for totalparental income: These were 66.0 for restrictive anorexic,64.0 for bulimic type, 42.5 for psychiatric control, and 70.5for nonpsychiatric control. The ANOVA for number of childrenwas not significant, nor was that for marital status.Regarding the treatment—related data, analyses revealedsignificant group differences on two of the five measures.The ANOVA for duration of disorder was significant ((2, 51) =3.99, .O5), with multiple comparisons indicating thatpsychiatric control subjects’ duration of disorder wassignificantly longer than restrictive anorexic subjects’. Thecorrected—for-heterogeneity analyses for treatment duration69and previous treatment duration did not reach significance.Similarly, the analysis for history of psychotropic medicationdid not reveal significant group differences. The analysisfor history of hospitalization, however, was significant(x22) = 15.87, p<.005), with multiple comparisons indicatingthat significantly more restrictive anorexic daughters had ahistory of hospitalization than did bulimic type orpsychiatric control daughters.Finally, it should be noted that it had originally beenplanned to collect data for each family from fathers as wellas from mothers and daughters. The sole inclusion criterionfor parental participation was that the parent had been livingwith the daughter since her birth. Due to the high rate ofdivorce in the sample, this inclusion criterion resulted in amuch lower availability of fathers than mothers. Thus, byexcluding fathers no longer living in the home andstepfathers, the potential number of fathers available was asfollows: restrictive anorexic n=l1; bulimic type n=11;psychiatric control n=8; and nonpsychiatric control n=19. Inaddition, there was also a lower rate of participation forfathers than for mothers such that the actual number offathers participating was: restrictive anorexic n=9; bulimictype n=6; psychiatric control n=3; and nonpsychiatric controln=1O. This small number of fathers would yield unstable meansand provide insufficient power for analysis. Thus, it was notpossible to provide comparative family data on fathers in thisstudy.70MeasuresIn order to test the hypotheses of the study,operationalizations of the following constructs were needed:family system/interaction and family sociocultural milieu(i.e., achievement orientation, sex role ideology, andattitude toward weight and appearance). Familysystem/interaction was defined by scores on the Cohesion,Expressiveness, Conflict, Independence, Organization, andControl subscales of the Family Environment Scale (FES; Moos &Moos, 1986), and by the Dyadic Adjustment Scale total score(DAS; Spanier, 1976). Achievement orientation was defined byscores on the Work and Family Orientation Questionnaire (WOFO;Helmreich & Spence, 1978) and by the Achievement Orientationsubscale of the FES. Sex role ideology was defined by scoreson the Sex-Role Ideology Scale (SRIS; Kahn & Tilby, 1978).Attitudes toward weight and appearance were defined by scoreson the Food Fitness and Looks Questionnaire (FFL; Hall,Leibrich, & Walkey, 1983) and the Body Esteem Scale (BES;Franzoi & Shields, 1984). A background information form, abrief interview, the Eating Attitudes Test (EAT-26; Garner etal., 1982), and the Brief Symptom Inventory (BSI; Derogatis &Mehisaratos, 1983) were used to characterize subjects andensure appropriate group composition.Family Environment Scale. The Family Environment Scale(FES; Moos & Moos, 1986) consists of 90 items to whichsubjects must respond true or false The following ten 9—itemsubscales comprise the test: Cohesion, Expressiveness,71Conflict, Independence, Achievement Orientation, Intellectual-Cultural Orientation, Active-Recreational Orientation, Moral—Religious Emphasis, Organization, and Control (see Table 1, p.49, for definitions). The FES can be administered toindividuals of age 11 years and older (Moos, 1987).The internal consistencies (Cronbach aiphas) of thesubscales have been reported to range from .61 to .78 ( =.71), and 2-month test—retest reliabilities of .68 to .86 ( =.78) have been reported (Moos & Moos, 1986). The FES is alsosensitive to environmental change in families. (Moos, 1987),and discriminates between normal and distressed families (Moos& Moos, 1986). In the FES manual, Moos and Moos (1986)present extensive evidence of. the subscales’ validity.Dyadic Adiustment Scale. The Dyadic Adjustment Scale(DAS; Spanier, 1976) is a 32-item measure of the quality ofmarital relationships. It yields a total score as well asfour factor-analytically derived subscales: DyadicSatisfaction, Dyadic Cohesion, Dyadic Consensus, andAffectional Expression. Cronbach aiphas were reported of .96for the total scale and .73 to .94 for the subscales (Spanier,1976). The DAS has been shown to discriminate between marriedand divorced individuals (Spanier, 1976).Work and Family Orientation Questionnaire. The Work andFamily Orientation Questionnaire (WOFO; Helmreich & Spence,1978) is a measlre of achievement motivation consisting of 23items which form four factor-analytically derived subscales.One subscale, Personal Unconcern (4 items), has not shown72adequate reliability or validity and was not be used in thepresent study. The three remaining subscales are: Work (thedesire to work hard); Mastery (preference for challengingtasks); and Competitiveness (enjoyment of interpersonalcompetition). The items are responded to on a 5-point Likertscale with points labeled “Strongly agree,” “Slightly agree,”“Neither agree nor disagree,” “Slightly disagree,” and“Strongly disagree.”Cronbach alphas were reported of .61 to .76 for the threesubscales of the WOFO (Helmreich & Spence, 1978). Constructand predictive validation evidence was presented and suggeststhat achievement motivation is not a unitary construct and,thus, use. of a total score is not recommended (Helmreich &Spence, 1978). More recently, J. T. Spence (personalcommunication, June 6, 1989) reported that the three factorsubscales of the WOFO have been verified in confirmatoryanalyses and that acceptably high alphas for the threesubscales have also been repeatedly verified.Sex—Role Ideoloqy Scale. The Sex-Role Ideology Scale(SRIS; Kahn & Tilby, 1978) is a 30—item measure ofprescriptive beliefs about behavior appropriate for men andwomen. Sex role ideology is conceived as a dimension with afeminist and a traditional pole. Statements are responded toon 7—point scales labeled from “Disagree strongly” to “Agreestrongly.” A median reliability of .79 (amongst restrictedand wide range samples) was reported based on split-halfreliability and item-total correlations (Kahn & Tilby, 1978).73Test-retest reliability was .87. A more recent study reportedinternal consistencies of .82 to .85 for single—sex samples(Cota & Xinaris, 1989). The scale has discriminated betweengroups of feminist and traditional women (Kahn & Tilby,1978), and showed evidence of construct and predictivevalidity (Leichner & Kahn, 1981).Food Fitness and Looks Questionnaire. The Food Fitnessand Looks Questionnaire (FFL; Hall et al., 1983) was developedto measure attitudes to and importance of weight, appearance,eating, and fitness postulated to exist in the families ofpatients with eating disorders. Fifty-two statements areresponded to on a 5-point Likert scale with points labeled“Definitely agree” to “Definitely disagree.” Five subscaleswere derived by factor analysis. Three of these subscales (35items) are relevant to the hypotheses of the present study:Weight, Appearance, and Fitness. The Cronbach aiphas forthese subscales are .88, .92, and .84 respectively (Hall etal., 1983). Some validity data were also presented; however,the scale was only administered to mothers of nonpatientdaughters.Body Esteem Scale. The Body Esteem Scale (BES; Franzoi &Shields, 1984) is a multidimensional measure of people’sattitudes toward their bodies. The scale consists of a listof 35 body parts or functions which subjects must rate on a 5-point scale from “Have strong negative feelings” to “Havestrong positive feelings,” according to how they feel abouttheir own bodies.74Factor analyses led to the development of three subscaleswhich differ somewhat for male and female subjects. They arelabeled Physical Attractiveness, Upper Body Strength, andPhysical Condition for men and Sexual Attractiveness, WeightConcern, and Physical Condition for women. The correspondingCronbach aiphas for women are .78, .87, and .82 (Franzoi &Shields, 1984). Corresponding three-month test-retestreliabilities are .78, .79, and .81 for women (S. L. Franzoi,personal communication, November, 1991). Validity evidencewas presented which supported the multidimensional structure(Franzoi & Herzog, 1986; Franzoi & Shields, 1984).Background information forms. Background informationforms given to daughters included items regarding demographiccharacteristics, eating disorder symptoms (operationalizedDSM—III-R diagnostic criteria), and impulse-related behaviors(as adapted from Garfinkel, Garner, & Moldofsky, 1980). Theimpulse—related clinical features coded on a yes/no basis wereas follows: cigarette use, alcohol use, street drug use,stealing, self—harm, suicide attempt, mood swings, and sexualintercourse. The total number of these behaviors reported wasemployed as an indication of impulsivity or acting out.Background information forms given to mothers included itemsregarding demographic characteristics only.Weight status. Indices of relative weight status werederived from the information obtained regarding mothers’ anddaughters’ weights. (As will be described in the Proceduresubsection, while mothers’ weights were self—reported,75daughters’ weights were measured either by the investigator orby the primary clinician.) For both mothers and daughters,the Body Mass Index (BMI) was calculated as weight inkilograms divided by height in meters squared as delineated bythe Metropolitan Life Insurance Company (1984). The BMI is ameasure of degree of overweight which is highly correlatedwith direct measures of body fat (Metropolitan Life InsuranceCompany, 1984). For mothers, the percentage of average bodyweight for height was calculated based on the MetropolitanLife Insurance Company’s (1983) table of standard weights.This method is inappropriate for the daughters in the study asit is intended for those aged 25 years and older. Therefore,the percentage of average body weight for height and age fordaughters was calculated using adolescent norms as adaptedfrom Forbes (1972).Interview. A brief interview was used to solicitinformation regarding daughters’ duration of disorder andtreatment, and previous treatment history; to ensureappropriate group status in the present study; and todetermine the composition of the family.Eating Attitudes Test. The Eating Attitudes Test (EAT26; Garner & Garfinkel, 1979; Garner et al., 1982) is a 26—item index of the symptoms of anorexia nervosa and bulimianervosa. Respondents must indicate whether each item refersto them “always,” “usually,” “often,” “sometimes,” “rarely,”or “never.” The three responses in the most “anorexic”direction to an item are scored 3, 2, and 1, while the76remaining three responses are scored 0. The scale is commonlyused with adolescent subjects (e.g., Rosen et al., 1988;Williams, 1987).The EAT-26 has shown acceptable reliability, withCronbach alphas reported of .90 for an eating—disorderedsample and .83 for a female comparison sample (Garner et al.,1982). Validation studies have shown the EAT—26 todiscriminate between eating—disordered and non—eating—disordered samples, to detect undiagnosed eating disordercases (Garner et al., 1982), and to be correlated withmeasures of restrained eating and dieting behavior (Rosen,Silberg, & Gross, 1988).Brief Symptom Inventory. The Brief Symptom Inventory(BSI; Derogatis & Melisaratos, 1983; Derogatis & Spencer,1982) is a 53-item psychological symptom inventory.Respondents indicate how much they are distressed by variousproblems on a 5-point scale (0 [not at all) to 4 [extremely]).The inventory may be administered to respondents 13 years ofage and older.The BSI yields 9 primary symptom scale scores and 3global indices of distress. The General Severity Index (GSI)is described as the single best indicator of overall currentdistress level derived from the BSI. For the purposes of thepresent study, the General Severity Index (GSI) and Depressionscale score were employed.The BSI is a shortened version of the Symptom Check List90—Revised (Derogatis, 1977). Its symptom scales correlate77from .92 to .99 ( = .96) with the corresponding scales of theparent instrument (Depression scale .95). Cronbach aiphas forthe 9 symptom scales range from .71 to .85 ( = .78;Depression scale .85), and 2-week test-retest reliabilitiesrange from .68 to .91 ( = .81). Test—retest reliabilitiesare .90 for the GSI and .84 for the Depression scale.Evidence of construct and criterion-related validity was alsopresented (Derogatis & Spencer, 1982).ProcedureClinical subjects (i.e., restrictive anorexic, bulimictype, and psychiatric control) were initially approached bytheir primary clinician and given an initial contact letterwhich described the study. If they gave permission to becontacted, the investigator was given their names and phonenumber and called them to further describe the study andrequest their participation. Nonpsychiatric control subjectswho saw the notice regarding the study, and wished toparticipate, contacted the investigator directly.Subjects participated in the study in one of thefollowing locations: at the center in which they had theirappointments with their primary clinician; in their familyhome; or at the investigator’s university—based psychologydepartment. When subjects arrived at the study, theinvestigator described the study to them, informed them of theconfidentiality of results (including inter-family memberconfidentiality), and familiarized them with thequestionnaires. Written consent forms were then given to78subjects to sign. After providing consent, subjects filled inthe questionnaires in the following order: FES, WOFO, FFL,SRIS, BES, EAT-26 (daughters only), BSI (daughters only), DAS(parents only), Background information form (separate daughterand parent forms). The investigator was present and availableto answer questions the subjects might have had. The briefinterview was then conducted with daughters, and non-eating-disordered daughters were weighed. (Restrictive anorexic andbulimic type daughters’ weights were obtained from theirprimary clinician.) Subjects then had any questions they mayhave had regarding the study answered. If they requested it,a general summary of the results of the study was sent to themupon the study’s completion. Daughters were also given asmall gift as a token of appreciation for participating in thestudy (e.g., tickets for the local cinema, planetarium, oraquarium)79ResuitsA factoriai design with repeated measures on one factorwas employed in the present study. Specifically, a 4(Group) X2(Relation) between-within groups design was utilized. It isa correlational study in that the levels of the factors areorganismic rather than experimentally manipulated variables.The four levels of the Group factor (with their associatednumber of mother—daughter pairs) are as follows: restrictiveanorexic (n=20); bulimic type (n=14); psychiatric control(n=20); and nonpsychiatric control (n=24). The two levels ofthe Relation factor are daughter (n=78) and mother (n=78).This design enables one to analyze differences amongst thefour groups, differences between mothers and daughters, andinteractions between the two factors.The hypotheses of the present study are divided into twogroups--family system/interaction hypothesis and familysociocultural milieu hypotheses. These hypotheses areprimarily tested with the 4 X 2 design described above, withthe result that the following variables are entered into theanalysis: FES-Cohesion, FES-Expressiveness, FES—Conflict,FES-Independence, FES-Organization, and FES-Control; WOFOWork, WOFO-Mastery, WOFO-Competitiveness, and FES-AchievementOrientation; SRIS; FFL-Weight, FFL-Appearance, FFL-Fitness,BES—Attractiveness, BES-Weight Concern, and BES-PhysicalCondition. To obtain comparability with the existing.literature, the three remaining subscales of the FES(Intellectual-Cultural Orientation, Active-Recreational80Orientation, and Moral-Religious Emphasis) were also includedin the analysis, with the result that there were a total of 20dependent measures in the main analyses of this study.Because of the large number of dependent variables usedin this study, a multivariate approach to the analysis wastaken. A multivariate analysis of variance (MANOVA),utilizing the Wilks’s lambda test statistic with the standardF-approximation and with alpha set at .05, was conductedbefore proceeding to univariate F-tests. MANOVA is commonlyemployed as protection against an excessive experiment—wiseType I error rate in cases of multiple dependent measures;however, there is some disagreement in the statisticalliterature as to whether MANOVA is appropriately used in thismanner (e.g., Harris, 1975; Huberty & Morris, 1989). It isargued that the initial MANOVA may not hold the experiment-wise error rate at alpha. Thus, in the present study, aninitial MANOVA was conducted for purposes of comparabilitywith extant literature but was not assumed to provide Type Ierror rate protection. Also, due to unequal L{s, the analysiscould not be assumed to be robust to violations of thehomogeneity of variance—covariance matrices assumption(Hakstian, Roed, & Lind, 1979) and, therefore, Box’s N testwas conducted to assess the viability of this assumption.Significance on the multivariate analysis of variance wasfollowed by univariate analyses of variance (ANOVAs) todetermine which individual dependent measures yieldedsignificant differences. If each of these ANOVAs were81conducted with alpha set at .05, the family-wise Type I errorrate would be unacceptable; therefore, the Bonferroniinequality was employed to hold the family-wise error rate atan upper bound of 10% (i.e., .10 divided by the number ofdependent measures). Due to the unequal js, Box’s test, ageneralization of Bartlett’s test for homogeneity of variance(e.g., Kirk, 1982), was conducted at the univariate level totest the assumption of multisample sphericity of the variancecovariance matrices. Significant ANOVAs were followed bymultiple comparisons conducted via the Tukey method with alphaset at .05 experiment—wise. The Tukey—Kramer adjustment formoderately unequal Ns was employed (Kirk, 1982).The above approach to the analyses of this study is theresult of a consideration of potential Type I and Type IIerrors. In a study with multiple dependent measures, there isan inflated risk of Type I errors; specifically, the actualalpha has an upper bound of alpha multiplied by the number ofdependent measures. Particularly with correlational research,in which random assignment is impossible, interpretingmultiple tests with .05 significance levels may result in nonreplicable and spurious findings. On the other hand,correlational clinical research studies are also prone to lowpower and, therefore, an inflated risk of Type II errorsbecause of the difficulties in subject recruitment withresultant low Ns, and because of the lack of control overextraneous variation. In the present study, the approach tothe analyses was planned with the goal of compensating for the82inflated risk of Type I errors without forfeiting all chancesof achieving a significant result should the null hypothesisbe false. Hence, the family-wise error rate of the multipleANOVAs was set at .10 rather than .05 (to decrease theprobability of Type II errors), and the Bonferroni inequalitywas employed to modify alpha (to decrease the probability ofType I errors). Similarly, a conservative multiple comparisonprocedure was used, but with a standard rather than modifiedalpha.The statistical analyses of the hypotheses of the studywill be described in detail shortly. Before proceeding tothis description, however, analyses of the clinicalcharacteristics of the groups will be presented. Followingpresentation of the results of the tests of the study’shypotheses, some additional analyses of potential interestwill be briefly described. All analyses followed the generalplan of approach described above. The only notable exceptionis that in some cases a one-way design rather than two—waybetween—within design was appropriate and, therefore,Bartlett’s test of the homogeneity of variance assumption wasemployed at the univariate level rather than Box’s test formultisample sphericity.Clinical CharacteristicsIn this section, analyses of the eating- and weightrelated and psychiatric distress-related clinicalcharacteristics of the female adolescents in the four groupswill be presented. The purpose of these analyses is to83determine the extent to which the recruitment strategies andselection criteria resulted in the desired group compositions.The ten eating— and weight—related measures were: height,current weight, percentage of average body weight for age andheight (%ABW), minimum percentage of average body weight forage and height (minimum %ABW), body mass index (BMI), maximumweight, minimum weight at current height, ideal or desiredweight, dissatisfaction with current weight, and score on theEating Attitudes Test (EAT-26). The three psychiatricdistress—related measures were: the Brief Symptom InventoryGeneral Severity Index score (BSI-GSI), the Brief SymptomInventory Depression scale score (BSI—Depress), and the totalnumber of impulse-related behaviors reported (IMPULSES).A one—way MANOVA was conducted on these 13 clinicalcharacteristics’ measures. This MANOVA was significant (F(39,184.34) = 4.43, p<.OO1); however, the Box’s M test was alsosignificant suggesting a violation of the assumption ofhomogeneity of variance-covariance matrices. Thisheterogeneity was examined as suggested in Hakstian et al.(1979). Examination of the four generalized variances withrespect to j revealed the negative condition whereby thesmaller samples are associated with the greater dispersion,and a liberally-biased test will result. Comparing theresults obtained with the results of the empirical samplingdistributions presented in Hakstian et al. (1979), however,suggested that the present MANOVA, given its high level ofsignificance and mildly to moderately unequal js, was unlikely84to reflect a situation of a true null hypothesis. Also, usingPillai’s criterion, which is reported to be more robust thanWilks’s lambda (Tabachnick & Fidell, 1983), the MANOVA wasstill significant at the .00l level. Univariate tests will,therefore, be reported. These tests were conducted with amodified alpha of .008 (i.e., .10 divided by 13).Eating— and weight—related measures. Means for the teneating— and weight—related measures are presented in Table 4.The ANOVA5 for height and for maximum weight, and thecorrected-for-heterogeneity ANOVA for dissatisfaction withcurrent weight, were nonsignificant. The remaining sevenANOVAs were significant.The results with current weight ((3, 74) = 8.33, p<.001)showed that restrictive anorexic subjects had significantlylower weight than female adolescents in the other threegroups, as did the corrected-for—heterogeneity results withpercentage of average body weight (E’(3, 36) = 15.56, p<.0O1).Similarly, the corrected-for-heterogeneity result for BodyMass Index (‘(3, 35) = 19.46, p<.OOl), a measure ofoverweight, showed that the restrictive anorexic subjects hadsignificantly less body weight or body fat than the subjectsin the three other groups. In addition, the results withminimum weight ((3, 74) = 18.81, p<.00l) and for minimumpercentage of average body weight ((3, 74) = 25.45, p<.001)showed that the minimum weight of the restrictive anorexicsubjects was significantly lower than that of the subjects inthe other three groups. The result for ideal weight ((3, 74)85= 8.17, p<.OO1) showed that the restrictive anorexic femaleadolescents’ desired weight was significantly less than thepsychiatric and nonpsychiatric control subjects’. Finally,the results of the corrected—for—heterogeneity analysis of theTable 4Daughters’ Eating- and Weight-Related Means (StandardDeviations)Restrict. Bulimic Psych. Nonpsych.Anorexic Type Control ControlHeight— 64.75 64.71 64.40 64.50inches (2.63) (2.97) (2.70) (2.78)Weight— 104.55 129.43 130.80 129.29pounds (15.02) (16.90) (24.30) (19.83)%ABW 89.30 110.50 111.75 110.33(9.18) (18.41) (19.96) (15.71)Minimum 73.10 94.36 101.55 100.46%ABW (10.41) (10.40) (14.13) (11.70)BMI 17.40 21.79 22.10 21.83(1.60) (3.64) (3.73) (3.20)Maximum 116.45 139.71 133.30 132.25Weight (18.36) (17.69) (24.00) (21.36)Minimum 85.30 110.86 118.85 117.50Weight (16.91) (10.55) (18.26) (16.63)Ideal 96.60 108.57 115.05 117.17Weight (17.74) (11.55) (14.38) (14.04)Dissatis— 7.55 17.50 14.55 9.79faction (17.71) (14.87) (20.96) (9.00)EAT—26 34.30 35.36 7.65 5.00(18.19) (16.40) (5.78) (4.29)86EAT-26 scores (‘(3, 32) = 29.42, p<.OO1) showed that therestrictive anorexic and bulimic type subjects’ scores weresignificantly greater than the psychiatric and nonpsychiatriccontrol subjects’ scores.To put some of the means in Table 4 in perspective, somecomparative data will be presented. Garner et al. (1982)reported data on the EAT—26 for undergraduate—aged women. Themeans (and standard deviations) they presented are as follows:restrictive anorexic 33.7 (18.7); bulimic anorexic 38.4(15.0); and control 9.9 (9.2). The percentages of averagebody weight for age and height presented in Table 4 are basedon adolescent norms as derived from Forbes (1972). Much ofthe eating disorder research employs adult women as subjectsand, therefore, percentages of average body weight for heightare often reported which are based on the Metropolitan LifeInsurance Company (1983) standard tables. Correspondingcurrent percentages of average body weight for height for theadolescent female subjects in the present study would be asfollows: restrictive anorexic 76.3%; bulimic type 94.5%;psychiatric control 95.5%; and nonpsychiatric control 94.4%.In terms of the BMI, the average range of values (associatedwith Metropolitan Life Insurance Company standards) for womenof comparable height to the adolescent subjects in the presentstudy is 21 .to 24 (Metropolitan Life Insurance Company, 1984).Finally, some data of relevance only to the eatingdisorder groups will be described. The current mean number ofbinge-eating episodes per month for the bulimic type group was8710.71; however, 3 of the 14 subjects were not currently binge—eating. For the 11 (79%) subjects who were currently binge-eating, the mean was 13.64 episodes per month. Of the bulimictype subjects, 86% reported having self-induced vomiting (79%)or having used laxatives (43%). Of the restrictive añorexicsubjects, 45% reported a history of self-induced vomiting(35%) or laxative use (20%). At the time of the study, 36% ofthe bulimic type subjects reported self-induced vomiting (29%)or laxative use (29%). Of the restrictive anorexic subjects,35% reported currently engaging in self-induced vomiting (20%)or laxative use (15%).Psychiatric distress—related measures. Means for thethree psychiatric distress—related measures are presented inTable 5. The effect for number of impulse-related behaviors(IMPULSES) was nonsignificant. The corrected-for-heterogeneity ANOVA for the BSI-GSI was significant (‘(3, 34)= 7.81, p<.OO1). Multiple comparisons showed that the bulimictype subjects reported significantly greater levels of generalpsychiatric distress than the nonpsychiatric control subjectsdid. The corrected-for-heterogeneity effect for BSI-Depresswas also significant (‘(3, 31) = 6.86, p<.005), with bulimictype subjects reporting significantly more depression thannonpsychiatric control subjects.Comparing the results of the present study on the BriefSymptom Inventory (BSI) with the published norms for femaleadolescents on the BSI (Derogatis & Spencer, 1982) renders the88present study’s means more interpretable. In terms of T—scores (mean of 50, standard deviation of 10), the means forTable 5Daughters’ Psychiatric Distress-Related Means (StandardDeviations)Restrict. Bulimic Psych. Nonpsych.Anorexic Type Control ControlBSI— 1.16 1.53 1.15 0.62GSI (0.88) (0.71) (0.84) (0.43)BSI— 1.33 2.06 1.25 0.63DEPRESS (1.32) (1.17) (0.93) (0.67)IMPULSES 1.85 2.86 2.15 1.71(1.69) (2.25) (1.73) (1.55)the BSI-GSI translate as follows: restrictive anorexic T =58; bulimic type T = 62; psychiatric control T = 58; andnonpsychiatric control T = 49. The means for the BSIDepression scale translate as: restrictive anorexic T = 58;bulimic type T = 64; psychiatric control T = 57; andnonpsychiatric control T = 50. Thus, while nonpsychiatriccontrol subjects’ depression and distress levels arenormatively average, clinical subjects’ depression anddistress levels are approximately one standard deviation abovethe norm.89Tests of Study HypothesesTo initiate the tests of the study’s hypotheses, a 4 X 2MANOVA was conducted on the 20 dependent measures ofrelevance. Due to the problem of singularity of variance—covariance matrices, Box’s M test had to be conducted on twosets of 13 dependent measures. These tests yieldednonsignificance for mothers and daughters, suggesting thathomogeneity of variance—covariance matrices could be assumed.The general questions addressed by this MANOVA were: 1)Are there differences in the family characteristics reportedamongst the four groups? and 2) Do the differences reportedamongst the four groups vary as a function of whether thesource of the report is mothers or daughters? The firstquestion was addressed by testing the main effect of Group,while the second question was addressed by testing theinteraction effect of Group by Relation. A third questionaddressed by the MANOVA, but of less interest in the presentstudy, was: Do the reports of mothers and daughters differoverall? (i.e., the main effect of Relation). It would be ofinterest, however, if the relationship between mothers’ anddaughters’ reports differed as a function of group membership(i.e., the interaction effect).The results of the 4 X 2 MANOVA were significant forGroup ((60, 164.92) = 1.81, R<.OO3) Group by Relation ((60,164.92) = 1.74, p<.OO4), and Relation ((20, 55) = 6.54,p<.OOl). Thus, univariate tests were conducted with amodified alpha of .005 (i.e., .10 divided by 20).90Family system/interaction hypothesis. Six of the 20•dependent measures in the overall MANOVA above were relevantto the family system/interaction hypothesis: FES-Cohesion,FES—Express iveness, FES-Conf 1 ict, FES-Independence, FESOrganization, and FES-Control. The specific hypothesis wasthat the families of bulimic type and psychiatric controlsubjects would be characterized as more dysfunctional than thefamilies of nonpsychiatric control subjects, whereas thefamilies of restrictive anorexic subjects would becharacterized as more similar to the families ofnonpsychiatric control subjects. “Dysfunctional” isoperationalized as less cohesive, expressive, independent, andorganized, and as more conflictual and controlling.The means for the six dependent variables for daughtersand mothers are presented in Table 6. The ANOVA for Cohesionwas signif,jcant for Group ((3, 74) = 6.18, p<.002) and forRelation ((l, 74) = 16. 39, p<. 001), but not for the Group byRelation interaction. Multiple comparisons indicated thatrestrictive anorexic and nonpsychiatric control mothers anddaughters characterized their families as more cohesive thanbulimic type and psychiatric control mothers and daughtersdid. Overall, mothers characterized their family environmentsas more cohesive than daughters did.The ANOVAs for Expressiveness and Independence were notsignificant for Group or for the Group by Relationinteraction. The main effect of Relation was significant forExpressiveness ((l, 74) = 56.24, p<.00l) and for Independence91((1, 74) = 8.62, p<.OO5). Mothers characterized theirfamilies as more expressive and encouraging of independencethan did daughters.None of the effects for Conflict, Organization, orControl reached significance.Table 6Fimilv Svstm/Tntraction Man (Standard Deviations’IVARIABLE Rest. Bul. Psych. Non.RELATION Anor. Type Cont. Cont.Cohesion Daughter 6.35 4.43 4.15 6.29.(2.58) (2.88) (2.58) (2.69)Mother 7.20 5.64 6.10 7.67(1.82) (1.99) (2.40) (1.55)Express. Daughter 4.10 3.29 3.35 4.63(2.40) (1.54) (1.98) (2.00)Mother 5.75 5.29 5.25 6.79(2.05) (1.59) (1.71) (1.84)Conflict Daughter 4.05 4.64 5.20 3.83(2.21) (1.99) (2.14) (2.48)Mother 4.00 4.36 4.35 3.04(2.08) (1.78) (2.25) (2.12)Independ. Daughter 5.65 5.07 5.75 6.88(2.16) (2.20) (1.45) (1.23)Mother 6.25 7.07 6.15 6.83(1.45) (1.33) (1.42) (1.09)Organiz. Daughter 5.55 5.21 4.65 5.71(2.52) (2.55) (2.25) (2.14)Mother 5.90 5.36 5.20 5.67(2.05) (2.68) (2.46) (2.48)Control Daughter 4.85 4.86 5.80 4.71(2.48) (2.25) (2.84) (2.14)Mother 5.05 4.57 5.00 4.50(2.06) (2.34) (2.03) (2.09)92The results of mothers’ responses to the DyadicAdjustment Scale (DAS) were also relevant to the familysystem/interaction hypothesis. A one-way ANOVA, with modifiedalpha of .005 so as to be congruent with the foregoinganalyses, was therefore conducted. It should be noted thatthe number of subjects available for this analysis wasdecreased by the number of single mothers in each group. Thefollowing s resulted: restrictive anorexic n=16; bulimic typen=12; psychiatric control n=ll; and nonpsychiatric controln=20. The ANOVA was nonsignificant. The means are presentedin Table 7.Table 7Dyadic Adlustment Scale Means (Standard Deviations)Restrictive Bulimic Psychiatric NonpsychiatricAnorexic Type Control Control(n=16) (n=12) (n=ll) (n=20)107.81 109.83 105.73 112.05(14.72) (18.34) (30.53) (19.28)Family sociocultural milieu hypotheses. Fourteen of the20 dependent measures included in the overall MANOVA were ofrelevance to the family sociocultural milieu hypotheses. Fourwere relevant to the achievement orientation hypothesis:WOFO-Work, WOFO-Mastery, WOFO-Competitiveness, and FESAchievement Orientation. One was relevant to the sex roleideology hypothesis: the Sex Role Ideology Scale. And six93were relevant to the weight and appearance attitudeshypothesis: FFL-Weight, FFL—Appearance, FFL-Fitness, BESAttractiveness, BES-Weight, and BES-Condition. Three othermeasures were included for comparability with extantliterature: FES-Active—Recreational Orientation, FESIntellectual-Cultural Orientation, and FES-Moral—ReligiousEmphasis. The specific hypotheses were that the restrictiveanorexic and bulimic type mothers and daughters would becharacterized as higher in achievement orientation (individualand family), traditional sex role ideology, and weight andappearance orientation than the psychiatric control mothersand daughters. No specific hypothesis was made regarding thenonpsychiatric control subjects; however, their inclusion inthe design was important for providing normative data.The means regarding the achievement orientationhypothesis are presented in Table 8. The main effects ofGroup for WOFO-Work, for WOFO-Mastery, for WOFOCompetitiveness, and for FES—Achievement Orientation were allnonsignificant. Similarly, none of the interactions weresignificant. For the main effect of Relation, the ANOVAS forWork and Mastery were nonsignificant, whereas the ANOVAs forCompetitiveness ((1, 74) = 15.45, p<.OOl) and AchievementOrientation ((l, 74) = 9.51, p<.004) were significant.Mothers rated themselves as less competitive than daughtersrated themselves, and mothers characterized their families asless achievement oriented than did daughters.94Table 8Arhivcmnt Orintation Means (Standard Dviations’VARIABLE Rest. Bul. Psych. Non.RELATION Anor. Type Cont. Cont.WOFO— Daughter 20.70 18.50 19.10 19.54Work (3.39) (4.18) (4.35) (3.13)Mother 20.80 20.50 20.95 21.13(2.88) (2.79) (3.90) (2.91)WOFO— Daughter 18.35 15.43 18.65 19.88Mastery (3.12) (4.99) (4.98) (3.90)Mother 16.85 18.29 19.15 19.96(5.49) (4.60) (3.47) (5.27)WOFO— Daughter 12.70 12.07 9.85 11.63Compet. (4.77) (4.20) (5.59) (3.99)Mother 9.80 10.00 8.60 8.54(4.23) (4.74) (4.58) (5.14)FES— Daughter 6.20 6.00 5.85 5.63Achieve. (1.70) (1.80) (1.79) (1.53)Orient. Mother 5.65 4.71 5.10 5.50(1.73) (1.49) (1.33) (1.96)To provide some context to the tests of the achievementorientation hypothesis, a one—way ANOVA on daughters’ averagegrades in school was conducted (with modified alpha of .005for congruence with the preceding analyses). Means arepresented in Table 9. This ANOVA was significant ((3, 74) =5.97, p<.002), with multiple comparisons showing that theschool grades of restrictive anorexic and nonpsychiatriccontrol daughters were significantly higher than those ofpsychiatric control daughters.95Table 9Daughters’ School Grades Means (Standard Deviations)Restrictive Bulimic Psychiatric NonpsychiatricAnorexic Type Control Control8.40 7.57 7.05 8.42(1.43) (1.22) (1.10) (1.18)Note. Grades scored on the following scale: 1 = 0 — 10%; 2 =11 — 20%; 3 = 21 — 30%; 4 = 31 — 40%; 5 = 41 — 50%; 6 = 51 —60%; 7 = 61 — 70%; 8 = 71 — 80%; 9 = 81 — 90%; and 10 = 91 —100%.The means for the Sex Role Ideology Scale are presentedin Table 10. The results of the ANOVA showed that the effectsof Group, Relation, and the Group by Relation interaction wereall nonsignificant.Table 10e Pn1 Tdo1oav Sca1 Maris (Standard Deviations’)Restrict. Bulimic Psych. Nonpsych.RELATION Anorexic Type Control ControlDaughter 147.75 151.14 151.40 152.25(21.02) (26.22) (21.28) (24.44)Mother 139.35 152.79 146.40 157.42(26.00) (21.88) (25.94) (26.32)Means regarding the weight and appearance attitudeshypothesis are presented in Table 11. The Group by Relation96interaction effects were significant for FFL-Weight ((3, 74)= 12.14, p<.OOl) and BES—Weight ((3, 74) = 8.84, p<.OO1), butnot for FFL—Appearance, FFL-Fitness, BES-Attractiveness, orBES-Condition. Following the significant interactions, themain effects were not examined; rather, tests on simple maineffects were conducted with appropriate adjustments forrepeated measures to the within-cell mean square and degreesof freedom as delineated by Winer (1971; pp. 529-532).The simple main effects tests of Group for FFL-Weightrevealed that the group effect for Mother was nonsignificant,whereas that for Daughter was significant ((3, 146) = 13.11,p<.OO1). Subsequent multiple comparisons showed thatrestrictive anorexic and bulimic type daughters ascribedsignificantly greater importance to weight than didpsychiatric and nonpsychiatric control daughters. The simplemain effects tests of Group for BES-Weight similarly indicatedthat the Group effect for Mother was nonsignificant, whereasthat for Daughter was significant ((3, 143) = 11.17, <.OO1).The multiple comparisons showed that restrictive anorexic andbulimic type daughters had significantly more negativeattitudes toward their own weight than did psychiatric andnonpsychiatric control daughters.The main effect of Group was nonsignificant for FFLAppearance, FFL—Fitness, and BES-Attractiveness, but wassignificant for BES-Condition ((3, 74) = 7.75, p<.OO1).97Table 11Weiqht and Appearance Attitudes Means (Standard DeviationsVARIABLE Rest. Bul. Psych. Non.RELATION Anor. Type Cont. Cont.FFL— Daughter 35.65 35.29 25.00 26.38Weight (7.57) (5.37) (7.48) (6.18)Mother 24.45 28.00 27.60 29.00(6.70) (7.37) (7.03) (6.72)FFL— Daughter 47.60 45.71 43.50 48.13Appear. (8.78) (11.34) (8.86) (10.50)Mother 44.45 41.64 42.70 43.71(6.89) (9.48) (8.57) (8.25)FFL— Daughter 26.85 25.00 24.50 25.83Fitness (6.24) (6.74) (6.25) (4.91)Mother 23.45 23.36 21.75 26.17(4.10) (5.65) (5.36) (5.44)BES— Daughter. 39.10 38.64 43.05 46.08Attract. (5.93) (5.97) (7.64) (5.56)Mother 45.70 45.21 44.85 46.29(6.28) (5.75) (5.26) (6.48)BES— Daughter 19.00 16.43 26.95 29.96Weight (9.21) (6.95) (10.68) (7.54).Mother 29.10 30.50 26.75 29.29(5.88) (7.14) (8.98) (8.66)BES— Daughter 27.25 27.00 29.30 35.63Condition (6.45) (5.51) (7.41) (4.72)Mother 29.60 32.86 30.80 33.42(5.93) (4.11) (6.54) (6.74)98Multiple comparisons showed that nonpsychiatric controlsubjects reported significantly more positive attitudes towardtheir own physical condition than did subjects in the otherthree groups. The main effect of Relation was nonsignificantfor FFL-Appearance, FFL-Fitness, and BES-Condition, but wassignificant for BES—Attractiveness ((l, 74) = 18.11, p<.OOi).Mothers reported significantly more positive attitudes towardtheir own physical attractiveness than did daughters.To put the results of the weight and appearance attitudeshypothesis for mothers in context, a one—way MANOVA on thefollowing five variables was conducted: height, currentweight, percentage of average body weight for height (%ABW),body mass index (BMI), and ideal or desired weight. Means forthese variables are presented in Table 12. This MANOVA wasnot significant and, therefore, univariate tests were notconducted.Finally, the means for the three additional socioculturalmilieu measures are presented in Table 13. The ANOVA showednonsignificant interaction effects for FES—Active—RecreationalOrientation, FES—Intellectual-Cultural Orientation, and FESMoral-Religious Emphasis. Although the effects of Group forFES—Active-Recreational Orientation and FES-Moral-ReligiousEmphasis were nonsignificant, the effect of Group for FESIntellectual-Cultural Orientation ((3, 74) = 6.97, p<.OO1)99Table 12Mothers’ Weight-Related Means (Standard Deviations)Restrict. Bulimic Psych. Nonpsych.Anorexic Type Control ControlHeight— 64.45 64.43 64.60 64.13inches (2.04) (2.31) (2.37) (2.15)Weight— 139.80 141.36 152.85 139.46pounds (23.66) (24.88) (25.44) (21.85)%ABW 105.70 106.43 114.75 105.96(17.19) (15.68) (18.17) (14.75)BMI 23.70 23.79 25.70 23.83(4.03) (3.45) (4.39) (3.36)Ideal 126.15 133.07 132.80 126.33Weight (11.14) (23.10) (11.05) (12.69)was significant. Multiple comparisons showed thatnonpsychiatric control mothers and daughters characterizedtheir families as significantly more interested inintellectual and cultural activities than did mothers anddaughters in the other three groups. Regarding the maineffect of Relation, the ANOVA for FES-Intellectual-CulturalOrientation was significant ((1, 74) = 19.52, p<.O01),whereas those for FES-Active-Recreational Orientation and FESMoral—Religious Emphasis were not. Mothers characterizedtheir family environments as significantly more interested inintellectual and cultural activities than did daughters.100Additional AnalysesAnalyses of covariance. As will be recalled from theMethod section, there were a number of significant differencesamongst the groups in terms of demographic and treatment-related data. In particular, there were significant groupdifferences regarding daughter’s age, daughter’s duration ofdisorder, daughter’s history of hospitalization, averageparental education, and total parental income. In the testsof the study’s hypotheses, significant Group effects werefound for five dependent variables: FES—Cohesion, FESIntellectua1-Cultural Orientation, BES-Condition, BES-WeightTable 13Additional Sociocultural Milieu Measures Means (StandardDeviations)VARIABLE Rest. Bul. Psych. Non.RELATION Anor. Type Cont. Cont.FES— Daughter 6.20 5.21 5.30 6.71Active— (2.55) (2.26) (2.23) (1.83)Recreat. Mother 6.00 5.07 4.70 6.50Orient. (2.43) (2.24) (2.52) (1.79)FES— Daughter 5.00 4.50 4.50 6.46Intell.— (2.34) (1.61) (2.40) (2.06)Cultural Mother 5.80 5.86 5.10 7.71Orient. (2.14) (1.56) (2.27) (1.63)FES— •Daughter 4.05 3.86 3.95 3.92Moral— (2.63) (2.80) (2.72) (3.01)Relig. Mother 4.55 4.64 4.55 4.08Emphasis (1.96) (2.87) (2.65) (2.86)101(daughters only), and FFL-Weight (daughters only). To addressthe possibility that some of the demographic or treatment—related effects might be confounded with the Group effects andmight, therefore, be biasing the results, an analysis ofcovariance (ANCOVA) was conducted for any significantdependent variable which was also significantly (alpha = .05)correlated with a potentially biasing demographic ortreatment—related variable. An exception was made in thatcorrelations with history of hospitalization were not examinedas differences on this variable seemed inherent to the groupsand, therefore, not appropriate for analysis of covariance.The correlation coefficients between the other variables, fordaughters and mothers, are presented in Tables 14 and 15respectively.Examining the correlations suggests that age should becovaried from BES-Weight, education should be covaried fromFES—Intellectual-Cultural Orientation and BES-Condition, andincome should be covaried from FES-Intellectual-CulturalOrientation and FFL-Weight. Thus, five ANCOVAs wereconducted. The homogeneity of regression coefficientsassumption was not tested as ANCOVA is reported to be robustto violations of this assumption (Glass et al, 1972; Winer,1971), and Type I errors are unlikely to result fromheterogeneous slopes alone (Glass et al., 1972).For BES-Weight and FFL-Weight, the Group by Relationinteraction effects were significant and tests of simple main102effects revealed significant Group differences for daughtersonly. Therefore, one-way ANCOVA5 were conducted in this caseTable 14Correlations Between Dependent Variables and PotentialConfounds — DaughtersPOTENTIAL CONFOUNDDEPENDENT Age Duration Parental ParentalVARIABLE Daughter Disorder Education IncomeFES— —.009 —. 124 .023 —.080Cohesion p=.469 p=187 p=.420 p=.244FES— —.111 .174 .342 .088Intellect. p=.166 p=.104 p=.OO1 p=.222BES— —.179 —.084 .309 .057Condition p=.O59 p=.274 p=.003 p=.310BES— —.170 .138 .162 —.095Weight p=.068 p=.161 p=.078 p=.205FFL— .116 —.113 —.124 .305Weight p=.i55 p=.2O9 p=.140 p=.003Note. Figures based on n=78 for Age, Education, and Income.Figures based on n=54 for Duration.103Table 15Correlations Between Dependent Variables and PotentialConfounds — MothersPOTENTIAL CONFOUNDDEPENDENT Age Duration Parental ParentalVARIABLE Mother Disorder Education IncomeFES— —.090 —.081 .176 .036Cohesion p=.216 p=.280 p=.062 p=.377FES— .137 —.111 .579 .247Intellect. p=.116 p=.213 p=.000 p=.014BES— .149 —.205 .155 .103Condition p=.096 p=.068 p=.088 p=.184BES— .262 —.162 .073 .128Weight p=.O1O p=.12l p=.264 p=.131FFL— .066 .030 .076 .158Weight p=.282 p=.414 p=.253 p=.084Note. Figures based on n=78 for Age, Education, and Income.Figures based on n=54 for Duration.and compared with results of one—way ANOVAs. The one—wayANCOVA for BES-Weight with age as a covariate was significant((3, 73) = 9.06, p<.OOl) at the same level as the one-wayANOVA (F(3, 74) = 10.05, p<.001). Similarly, employing incomeas a covariate in a one-way ANCOVA for FFL-Weight yieldedsignificance (F(3, 73) = 11.90, p<.001) of a comparable levelto that in the one-way ANOVA ((3, 74) = 13.37, <.001).Thus, the age or income differences did not appear to beaccounting for the results.104Similarly, covarying education from BES-Condition in theanalysis of variance yielded a still-significant F-ratio.This ANCOVA ((3, 73) = 5.39, p<.003) was significant at aslightly lower level than the ANOVA ((3, 74) = 7.75, p<.00l),but was still significant vis-a-vis the modified alpha.Comparing the results of ANCOVA and ANOVA in the case ofFES-Intellectual-Cultural Orientation did reveal a situationof confounding bias. The original ANOVA Group effect wassignificant ((3, 74) = 6.97, p<.001), whereas the ANCOVAusing education as a covariate was not significant given themodified alpha ((3, 73) = 3.64, = .017). Covarying incomeonly slightly decreased the significance of the Group effect((3, 73) = 6.27, p<.002)Subsidiary analyses: Family system/interactionhypothesis. It was of interest to explore other possiblesources of variation in the data in order to aid in theinterpretation of the results of the tests of the familysystem/interaction hypothesis. In particular, the possibleeffects of high versus low depression, psychiatric distress,impulsivity, and eating disorder symptomatology were ofinterest. Ideally, it would have been possible to enter thesevariables as third factors in the design; however, because ofthe small number of subjects and the dissimilar ranges ofscores amongst the groups, this would have resulted in greatdisproportionality of cell size, low power, and questionablemeaningfulness of results. Therefore, preliminary analyseswere conducted in the form of four 2(High/Low) (Depression,105Distress, Impulsivity, or Eating Disorder Syinptomatology) X2(Relation) between-within groups MANOVAs with appropriatefollow-up tests congruent with the other analyses of thestudy. The results of these analyses are presented in detailin the Appendix.In summary form, the results of the analyses were asfollows. The MANOVAs were significant for Depression,Distress, and Impulsivity, but not for Eating DisorderSymptomatology. Subsequent ANOVA5 and tests of simple maineffects revealed several significant effects for daughters,but not for mothers. Specifically, daughters in the HighDepression and High Impulsivity groups characterized theirfamilies as significantly less cohesive and more conflictualthan did daughters in the Low Depression and Low Impulsivitygroups, respectively. Further, daughters in the HighDepression and High Distress groups described their familiesas significantly less encouraging of independence than diddaughters in the Low Depression and Low Distress groups,respectively.SummaryThe results can be summarized as follows. In terms ofgroup composition, the restrictive anorexic daughters’ currentand minimum weights, percentages of average body weight, andbody mass index scores were, as expected, lower than those ofthe daughters in the other three groups. The desired weightsof the restrictive anorexic daughters were lower than those ofthe psychiatric and nonpsychiatric control daughters. On the106Eating Attitudes Test (EAT-26), a measure of eating disordersymptomatology, the restrictive anorexic and bulimic typedaughters’ scores were higher than the psychiatric andnonpsychiatric control daughters’ scores. There were no groupdifferences on height, maximum weight, or dissatisfaction withcurrent weight. The psychiatric distress—related measuresshowed that bulimic type daughters reported greater levels ofdepression and general psychiatric distress than didnoñpsychiatric control daughters. There were no differencesbetween the groups in terms of number of impulse—relatedbehaviors reported.Regarding the family system/interaction hypothesis of thestudy, restrictive anorexic and nonpsychiatric control mothersand daughters characterized their families as more cohesivethan did bulimic type and psychiatric control mothers anddaughters. There were no differences amongst the four groupson expressiveness, conflict, independence, organization, orcontrol; nor were there any differences amongst mothers onreports of marital adjustment. Overall, mothers characterizedtheir families as more cohesive, expressive, and encouragingof independence than did their daughters.In the subsidiary analyses, other sources of variationwere found in the family system/interaction data fordaughters, but not for mothers. Daughters in the HighDepression group characterized their families as lesscohesive, more conflictual, and less encouraging ofindependence than daughters in the Low Depression group did.107Daughters in the High Impulsivity group characterized theirfamilies as less cohesive and more conflictual than daughtersin the Low Impulsivity group did. And, finally, the daughtersof the High Distress group described their families as lessencouraging of independence than daughters in the Low Distressgroup did. There were no group differences for mothers ordaughters in terms of expressiveness, organization, orcontrol. Also, there were no differences between High and LowEating Disorder Symptomatology groups for mothers ordaughters.With respect to the achievement orientation hypothesis,there were no group differences on work, mastery,competitiveness, or family achievement orientation, butrestrictive anorexic and nonpsychiatric control daughters didhave higher average school grades than psychiatric controldaughters had. Overall, mothers characterized their familiesas less achievement oriented and described themselves as lesscompetitive than daughters did.There were no differences in sex role ideology amongstthe groups.Regarding the weight and appearance attitudes hypothesis,restrictive anorexic and bulimic type daughters, but notmothers, ascribed greater importance to weight and had a morenegative attitude toward their own weight than psychiatric andnonpsychiatric control daughters did. Nonpsychiatric controlmothers and daughters had more positive attitudes toward theirown physical condition than mothers and daughters in the other108three groups did. There were no group differences in terms ofattitude toward one’s own attractiveness, or importanceascribed to appearance or fitness. Overall, mothers had amore positive attitude toward their own attractiveness thandaughters did. There were no differences amongst mothers interms of weight—related variables.Finally, with regard to the additional socioculturalmilieu measures, nonpsychiatric control mothers and daughterscharacterized their families as more interested inintellectual and cultural activites than mothers and daughtersin the other three groups did. However, differences amongstthe groups in average parental education could account forthis finding. There were no group differences in terms offamily participation in social and recreational activites orin degree of family emphasis on religious issues and values.Overall, mothers characterized their families as moreinterested in intellectual and cultural activities thandaughters did.109DiscussionIn general, there was some support for the familysystem/interaction hypothesis. There was, however, littlesupport for the family sociocultural milieu hypotheses. Theseresults will be discussed and interpretations regarding thefindings will be advanced. Limitations of the present studywill be delineated, and methods to overcome these difficultieswill be suggested. Finally, questions for future researcharising from the results, and possible ways of testing these,will be discussed.Family System! Interaction HypothesisMain analyses. The family system/interaction hypothesis-—that the families of bulimic type and psychiatric controlsubjects will be characterized as more dysfunctional than thefamilies of nonpsychiatric control subjects, whereas thefamilies of restrictive anorexic subjects will becharacterized as more similar to the families ofnonpsychiatric control subjects-—received some support. Inparticular, the results for perceived family cohesion provedto be robust, with both mothers and daughters of the bulimictype and psychiatric control groups characterizing theirfamily environments as less cohesive than mothers anddaughters in the restrictive anorexic and nonpsychiatriccontrol groups. That is to say, the bulimic type andpsychiatric control subjects reported that family membersprovided less commitment, help, and support for one another in110their families than did restrictive anorexic andnonpsychiatric control subjects.Support for the family system/interaction hypothesis wasnot obtained on the measures of expressiveness, conflict,independence, organization, control, or marital adjustment.It should be noted, however, that the nonsignificance ofresults on the Expressiveness and Independence subscales wasdue to the modified alpha level employed in the present study.Thus, judgement should be suspended regarding the existence ofeffects on these variables until further, more powerful,studies are conducted. Similarly, the effect for the Conflictsubscale approached significance and, therefore, futureresearch should not be curtailed on the basis of the presentstudy’s negative result. It is also of note that, overall,mothers characterized their families as more cohesive,expressive, and encouraging of independence than daughtersdid. it is a common and well-known finding that parents ratetheir families more favorably than children do (e.g., Moos &Moos, 1986); therefore, this result will not be exploredfurther here.The results on the Organization and Control subscales andon the Dyadic Adjustment Scale did not support the hypothesis.According to the literature, one would expect the families ofbulimic type subjects to be characterized as disorganized(e.g., Garner et al., 1984; Kog & Vandereycken, 1989; Root etal., 1986)., and the families of eating-disordered subjects tobe characterized as controlling (e.g., Bruch, 1973; Minuchin111et al., 1978; Selvini—Palazzoli, 1978). Similarly, one wouldexpect the marital relationship to be characterized asconflictual in the parents of bulimic type subjects (e.g.,Schwartz et al., 1984; Strober & Yager, 1984). Other studiesemploying the FES, however, have also not found groupdifferences on the Organization and Control subscales (Johnson& Flach, 1985; Ordman & Kirschenbaum, 1986; Shisslak et al.,1990; Stern et al., 1987; Strauss & Ryan, 1987). It may bethat these subscales do not tap the constructs of interest.For instance, the items on the Organization subcale revolvearound the importance of organization and structure inplanning family activities and responsibilities. Perhapsthese items do not capture the important emotionaldisengagement component of the “chaotic” construct.Similarly, the items on the Control subscale reflect theextent to which set rules and procedures are used to runfamily life. This may not adequately measure the intendedtheoretical constructs which revolve around the child’s needsnot being responded to and, therefore, her independent controlover her life being discouraged. In short, the FES subscalesmay assess organization and structure in a family in a‘concrete sense, whereas the importance of these constructs maylie in their more subtle, dynamic aspects. The FamilyAssessment Device and the Parental Bonding Instrument, as wellas other subscales of the FES, appear to have been moresuccessful in tapping these theoretical constructs (e.g.,Fichter & Noegel, 1990; McNamara & Loveman, 1990).112The lack of results on the Dyadic Adjustment Scale isincongruent with the finding of Strober (1981) that theparents of bulimic anorexics rated their marital relationshipmore negatively than the parents of restrictive anorexics.Similarly, bulimic subjects have described their parents’marriages as more conflictual than control subjects (Dolan etal., 1990). It may be that the present study did not providean adequate test on this measure due to the inclusion of bothmarried and remarried subjects, exclusion of separated anddivorced subjects, and resulting lowered number of subjectsfor this analysis. It is also possible, however, that thereported satisfaction of the marital relationship is not acrucial aspect of the family system/interaction hypothesis.With respect to the failure of the effects for FESsubscales Expressiveness, Conflict, and Independence to reachsignificance, a number of potential explanations can beadvanced. Significant differences on the Independencesubscale of the FES have not been consistently found inresearch with adult women. It may be that, as with theOrganization and Control subscales, this subscale does not tapthe specific theoretical construct of interest. That is tosay, whereas the subscale is intended to reflect the extent towhich family members are assertive, self-sufficient, and maketheir own decisions, the lack of encouragement of independencedescribed in the literature has more to do with the child’sneeds not being acknowledged or heeded. The effect on thissubscale did, however, approach significance and, therefore,113the possibility that it does tap the construct of interest inan adolescent population should not be ruled out. Also,bulimic or bulimic anorexic women have scored significantlylower than control women on this subscale in some otherstudies (Johnson & Flach, 1985; Shisslak et al., 1990).The lack of significant findings on the Expressivenessand Conflict subscales is definitely incongruent with theexisting research literature. Bulimic and restrictiveanorexic women have consistently been found to describe theirfamily environments as less expressive than control women(Johnson & Flach, 1985; Ordman & Kirschenbauiu, 1986; Shisslaket al., 1990; Stern et al., 1987; Strauss & Ryan, 1987), andbulimic women have fairly consistently portrayed theirfamilies as more conflictual than control women (Johnson &Flach, 1985; Ordman & Kirschenbaum, 1986; Shisslak et al.,1990). One reason for the discrepant findings of the presentstudy on the Expressiveness subscale may be the modified alphalevel which was employed in an attempt to provide some controlover the Type I error rate. However, this would not explainthe failure to replicate the Conflict effect and, uponexamination, it appears that the absolute value of thedifferences attained with the adolescent subjects on theExpressiveness and Conflict subscales are smaller than thoseobserved in the adult subject literature.It may be that the nonsignificance of results in thepresent study on these two subscales is due to the young ageof the subjects employed. For instance, Calam et al. (1990)114found, within a group of adult eating-disordered and non-eating—disordered subjects, that older subjects perceivedtheir parents in a more negative light than younger subjects.They suggested that early adulthood may be a time whensubjects re—assess their families and come to see theirparents more critically. It could be that, in the presentsample, clinical subjects are as yet unable to perceive orassess family dysfunction which may in fact exist. This mightapply to the restrictive anorexic group in particular asdenial and minimization are thought to be associated featuresof this disorder (e.g., American Psychiatric Association,1987). Alternatively, it may be that it is only withincreasing time and chronicity that the families of clinicalsubjects become dysfunctional. Smaller effect sizes couldalso have been due to the adolescent nonpsychiatric controlsubjects perceiving their family environments more negativelythan an adult control group might, due to the struggle forseparation and autonomy occurring in adolescence. Thus, morenegative perceptions of the adolescent nonpsychiatric controlgroup, or less negative perceptions of one or more of theadolescent clinical groups, may have resulted in less robustfindings on the Expressiveness and Conflict subscales than arefound in the adult literature.Finally, returning to the Cohesion subscale, the resultsof the present study were robust and consistent with thetheoretical and empirical literature regarding anorexianervosa and bulimia nervosa. Congruent with Hilde Bruch115(e.g., 1973), the families of restrictive anorexics werecharacterized as equally cohesive and supportive as thefamilies of nonpsychiatric controls. Similarly, Minuchin’stheory (e.g., Minuchin et al., 1978) predicts closeness,loyalty, and concern in the families of anorexics. Thecohesion measure may also reflect the perfect/chaotic familydistinction formulated by Root et al. (1986), with therestrictive anorexics of the present study characterizingtheir families as more united and nurturant, while the bulimicsubjects characterize their families as more distant andemotionally disengaged. Perhaps the Cohesion subscale of theFES is the most robust of the subscales because it taps a coreemotional aspect of the theoretical formulations in terms ofemotional support versus disengagement.In terms of the empirical research employing the FES, theresults of the present study are congruent with the consistentfindings of other researchers regarding the lower familycohesiveness of buliluic subjects as compared to noripsychiatriccontrols (e.g., Johnson & Flach, 1985; Ordman & Kirschenbaum,1986; Shisslak et al., 1990; Stern et al., 1987). The presentstudy also empirically supports the theoretical prediction ofhigher perceived cohesiveness in the families of restrictiveanorexics than in the families of bulimics, which previouslyhad been supported by the reports of parents of restrictiveand bulimic anorexics (Strober, 1981), but not by the reportsof restrictive anorexic and bulimic or bulimic anorexic womenthemselves (Stern et al., 1987; Strauss & Ryan, 1987).116Strauss and Ryan (1987) found that restrictive anorexic andbulimic anorexic women both perceived lower family cohesionthan controls, and Stern et al. (1987) found that buliinicwomen, but not restrictive anorexic women, reported lowercohesiveness than control women. Similarly, researchemploying other measures than the FES with adult subjects hasproduced inconsistent results regarding restricter/bingerdifferences, with some researchers finding such differences(Garner et al, 1985; Humphrey, 1986b, 1989; Kog &Vandereycken, 1989; Piran et al., 1988), and others not (Calamet al., 1990; Palmer et al., 1988; Steiger et al. 1989;Steiger et al., 1991; Waller et al., 1990b; Wonderlich &Swift, 1990b). Wonderlich and Swift (1990b) suggest thatolder restrictive anorexic subjects may be less likely thanyounger anorexic subjects to perceive their families ascohesive and nurturing. Thus, the significant differencefound between restrictive anorexic and bulimic type subjectsin the present study may, in part, be a function of the youngage of the subjects. Perhaps as restrictive anorexics getolder, they become more similar to bulimics in terms ofperceived family cohesion.Thus, the present study has shown that, during the earlystages of an eating disorder, restrictive anorexic daughtersand their mothers characterize their families as similar incohesion to normal controls, whereas bulimic type daughtersand their mothers characterize their families as similar incohesion to psychiatric controls. The interpretation of this117finding depends, in part, on whether the similar level ofperceived cohesion observed in restrictive anorexic andnonpsychiatric control subjects reflects a situation ofsimilar actual family cohesion or a situation of denial orminimization on the part of the restrictive anorexic group.If there are no actual differences in family cohesionbetween restrictive anorexic and nonpsychiatric controlgroups, it might be argued that the family environment doesnot play a contributing role in anorexia nervosa. However,there are numerous factors which suggest that the high levelof cohesion reported by the restrictive anorexic group may bea product of denial or reflect a desire to appear cohesiverather than a truly cohesive family environment. Forinstance, denial and minimization are reportedly common inanorexia nervosa patients (e.g., American PsychiatricAssociation, 1987). Similarly, Bruch (1973) characterized thefamily members of anorexics as denying the existence ofproblems and having tensions hidden beneath a facade ofnormality, and Gordon et al. (1989) described how the empathyand relatedness apparently observed in an anorexic’s familyare actually distortions or simulations of normal concern.Also, one could argue from what is known about the severe andoften intractable nature of anorexia nervosa, from themultitude of associated characteristics such asineffectiveness and self-hatred (e.g., Bruch, 1973; Garner &Bemis, 1984), and from the intent on self-starvation andincreased control that a problem—free family environment would118be improbable. Finally, it was the impression of the currentauthor that there was more incongruence between the self-reports of the restrictive anorexic daughters and theirmothers and information obtained from other sources than therewas in the nonpsychiatric control daughters and mothers.The results of the present study are correlational and,therefore, no inference regarding causation can be made. Itis of interest, however, to speculate on the possibledirection and mechanism of the association between eatingdisorder subtype and perceived family cohesion. Oneinterpretation is that the different symptomatology of theeating disorder subtypes may differentially affect familycohesion. For instance, the starvation, low weight,hospitalization, or threat of impending death of therestrictive anorexic may function so as to increase familyconcern and cohesion to nonpsychiatric levels. Thisinterpretation would fit with Minuchin’s (Minuchin et al.,1978) systemic formulation that the anorexic symptoms maintainand are maintained by the dysfunctional family structure by,for instance, serving to avoid family conflict or separation.On the other hand, having an anorexic daughter may have servedto increase family tension, power struggles, and frustration(e.g., Bruch, 1978; Kay et al., 1967) and, thus, to havelowered perceived family cohesion to nonpsychiatric levels.Similarly, regarding the bulimic type and psychiatric controlgroups, having a daughter in psychiatric treatment may havedecreased the cohesiveness of the family.119An alternative interpretation of the results of thepresent study is that the reported cohesion of family memberscontributes to the development of the eating disordersubtypes. For instance, the syndrome of anorexia nervosa maybe more likely to develop in a “perfect” or cohesive familyenvironment where self-control, self—discipline, and self—denial are modeled. Similarly, a family which denies theexistence of problems may encourage the development ofanorexia nervosa, a disorder in which the existence or gravityof the problem is denied. The onset during adolescence wouldmake sense in that separation would be stressful in a highlycohesive family environment. This would especially be trueif, as has been suggested by numerous authors, the familyenvironment is also one in which the child’s needs andfeelings have not been acknowledged, encouraged, or respondedto (e.g., Bruch, 1973; Selvini—Palazzoli, 1978) and,therefore, the child does not know what her needs or feelingsare and does not feel in control or ready for autonomy. Ifthe family appears cohesive and supportive, but actually doesnot respond to the true needs of the child and, therefore,neglects her, the child may feel confused and worry thatsomething is somehow wrong with her or that love andacceptance may be withdrawn. The syndrome of anorexia nervosathen might function to help the adolescent to feel in control,approved of, and worthy, and to give her an identity via asymptom which Gordon et al. (1989) describe as sufficiently120similar to normal self-control that the patient and her familydo not, at least initially, perceive it to be a problem.The reported level of family cohesion in the presentstudy was different in restrictive anorexics than inpsychiatric controls, therefore suggesting that there may be aspecific link between anorexia nervosa and high perceivedcohesion. The possibility that high cohesion would facilitatethe development of certain other disorders cannot be ruledout, however, as the present study employed a heterogeneouspsychiatric control group. With respect to the bulimic typegroup, there were no significant differences in cohesion fromthe psychiatric control group; therefore, a specific familycharacteristic was not found. This is not to say, however,that the low family cohesion could not contribute to thedevelopment of bulimia nervosa in a nonspecific manner as partof a multivariate causal model (e.g., Garber & Hollon, 1991).For instance, the lower cohesion of the bulimic type subject’sfamily may model a more uncontrolled coping style, and mayresult in feelings and needs being less easily denied. Thesymptom of binge-eating may, therefore, be more likely todevelop.Subsidiary analyses. To aid in the interpretation of theresults and, in particular, to render them more comparablewith the results of studies which had come out since theinception of the present study, some subsidiary analyses wereconducted. As suggested by the more recent research, othersources of variation were found in the family121system/interaction data for daughters. In particular,cohesion, conflict, and independence were found to vary withthe daughter’s level of depression, impulsivity, andpsychiatric distress such that the highly depressed,impulsive, or distressed daughters reported their familyenvironments to be more dysfunctional than the less depressed,impulsive, or distressed daughters. Level of depressionyielded the most robust results. It is of note that therewere no differences in these analyses for mothers. Also,dividing daughters into high and low eating disordersymptomatology groups failed to produce significantdifferences in family interaction.The recent research has shown that the familyenvironments described by bulimic and mixed eating disordersubjects are associated with the subjects’ level of depression(Blouin et al., 1990; Wonderlich & Swift, 1990b), personalitydisorder features (Johnson et al., 1989; Wonderlich & Swift,1990a), and family childhood sexual abuse history (Bulik etal., 1989). The highpercentage of eating disorder subjectswho fall into these categories has also been pointed to asevidence of the substantial coinorbidity and heterogeneitywithin eating disorder groups (e.g., Johnson et al., 1989).There are also strong associations amongst depression,borderline personality disorder, and sexual abuse history(e.g., Bulik et al., 1989; Johnson et al., 1989). In thepresent study, measures of depression, general psychiatricdistress, and impulsivity were available, and were similarly122found to be related to the adolescent subjects’ reportedfamily environment patterns. In addition, heterogeneity andcomorbidity were also apparent within the adolescent groups ofthe present study (see Table 16). Regarding comorbidity, itis also of note that 20% of the present study’s psychiatriccontrol group had to be excluded on the basis of high eatingdisorder symptomato logy.In the present study, there were no differences in familyfunctioning between high and low eating disordersymptomatology groups. Taken together with the abovefindings, this could lead one to postulate that familyfunctioning is not related to eating disorders per se at all,but rather is a function of eating—disordered subjects’comorbidity. However, in the main analyses of the familysystem/interaction hypothesis, differences in family functionwere found between the restrictive anorexic and bulimic typegroups, suggesting that a qualitative rather than quantitativedivision is of importance. Still, it could be argued thatthis restricter/binger difference is only a function of thehigher levels of depression and distress apparent in thebulimic type group (see Tables 5 and 16). The finding thatthere are significant differences in family functioningreported by mothers in the main analyses but not in thesubsidiary analyses, however, suggests that there may also bean association between eating disorders and familysystem/interaction which is not dependent upon subjects’comorbidity.123That differences in reported family interaction arepresent in the subsidiary analyses for daughters but not formothers suggests that the finding may be a result of state-dependent differences in perception. That is to say,depressed or distressed daughters may be more likely toperceive their family environments negatively, either throughnegative distortion/bias or through increased awareness anddecreased censoring (i.e., lack of positive bias). Similarly,more impulsive or acting out daughters may be less likely tocensor their responses in the service of social desirability,or may be more aware of family dysfunction. Such state—dependent effects could not, however, account for thedifferences in mothers’ reports found in the main analyses,raising the possibility of a direct association between eatingdisorders and reported family functioning. Another findingwhich suggests there may be a direct association betweeneating disorders and family environment is that, despitesimilar percentages and levels of depression, impulsivity, anddistress observed in the restrictive anorexic and psychiatriccontrol groups (see Tables 5 and 16), mothers and daughtersreported significantly different levels of family cohesiveness(see Table 6). Thus, there may be some specificity ofassociation of family environment with eating disorders, atleast for the restrictive anorexic subtype.Thus far, the bulimic type group has not beendiscriminated from the psychiatric control group. However, itwill be recalled that the bulimic type group reported124significantly greater depression and psychiatric distress thanthe nonpsychiatric control group, whereas the restrictiveanorexic and psychiatric control groups did not significantlydiffer in depression or distress from either thenonpsychiatric control group or the bulimic type group (seeTable 5). Also, examining Table 16, it is apparent that thebulimic type group is composed of approximately twice as manyhighly depressed and highly distressed subjects as therestrictive anorexic and psychiatric control groups are.Despite this greater level of distress and depression, therewere no significant differences in family environment betweenthe bulimic type and psychiatric control groups. On thisbasis, one might have expected lower reported family cohesionin the bulimic type group than in the psychiatric controlgroup. Thus, with the aim of stimulating further research,the highly speculative possibility will be raised here thatthe family environments of bulimics may be similar to thefamily environments of restrictive anorexics in terms of thedesire to appear cohesive and to deny the existence ofproblems, but that the bulimic family members are not able tomaintain this myth as well as the restrictive anorexic familymembers due to their greater affective instability.The literature on family factors in anorexia nervosa andbulimia nervosa suggests that there may be similarities infamily environment. For instance, both anorexics’ (e.g.,Norris & Jones, 1979) and bulimics’ (e.g., Humphrey & Stern,1988) families have been characterized as presenting in an125idealized and problem—free manner. Similarly, enmeshment andoverprotection are attributed to the families of anorexics(e.g., Minüchin et al., 1978) and bulimics (e.g., Root et al.,1986; Schwartz et al., 1984). Further suggestive of familysimilarities are: the observations of anorexic and bulimicmembers within the same family (e.g., Kaffinan & Sadeh, 1989);the statistic that approximately 25% to 50% of restrictiveanorexics eventually develop bulimic symptoms (e.g., Garner etal., 1984); and reports that the same individual may alternatebetween the disorders of anorexia nervosa and bulimia nervosa(e.g., Garner, 1986). In addition, the eating disorders ofanorexia nervosa and bulimia nervosa are very similar, withthe major difference being the presence or absence of binge—eating. Bulimic individuals are similarly resistant toceasing dieting as restrictive anorexic individuals are(Garner, 1986). Finally, the associated psychopathology ofanorexics and bulimics is similar. For instance, both havedesires to be in control and deny interpersonal needs (e.g.,Armstrong & Roth, 1989; Cooper, 1987; Sallas, 1985), show lowself—esteem (e.g., Bruch, 1984; Mizes, 1985), have high needsfor approval (e.g., Garner & Bemis, 1984; Weiss et al., 1985),and may struggle with issues of autonomy and identity, oftenbeing unable to articulate their inner worlds (e.g., Armstrong& Roth, 1989; Garner & Olmsted, 1984).Thus, there have been many similarities noted betweenanorexic and bulimic individuals and their families. However,there are also differences consistently noted between these126individuals and families. In particular, bulimic individualsand their family members are reported to be more affectivelyunstable, depressed, and impulsive (e.g., Garner et al., 1984;Hsu et al., 1990; Strober et al., 1982), whereas anorexicindividuals and their family members are described as moreavoidant, overcontrolled, and introverted (e.g., Bruch, 1973;Piran et al., 1988; Shisslak et al., 1987; Strober et al.,1982, 1990). It has been suggested that such differences maybe genetically transmitted (e.g., Hsu et al., 1990; Strober etal., 1990).It is speculated, therefore, that a similar familyenvironment of problem—denial and pseudo—cohesion may,depending on the context of family members’ predominantgenotypes, be associated with either anorexia nervosa orbulimia nervosa. In the context of familial affectiveinstability, decreased ability to maintain the desired familycohesion may result, and the adolescent’s eating disorder maycome to serve-—in addition to the functions of anorexianervosa of increasing control, gaining approval, increasingself—worth, and securing an identity——the function ofexpressing, releasing, or numbing feelings through the act ofbinge-eating (cf. Garner, Garfinkel, & Bemis, 1982; Mizes,1985). The bulimic, because of her greater affectiveinstability, may be unable to maintain the rigid control ofthe anorexic. Similarly, because of the bulimic’s, or one ormore of her family members’, emotional lability, the familymay not be able to maintain the desired appearance of family127cohesiveness. It is postulated that, just as the bulimic mayaspire to be anorexic (e.g., Crisp, 1981), the bulimicindividual and her family members may aspire to have theapparent family environment of an individual with anorexia.The psychiatric control subjects, on the other hand, may nothave such a strong desire to deny problems or to appearcohesive.Family Sociocultural Milieu HypothesesThe family sociocultural milieu hypotheses--that therestrictive anorexic and bulimic type mothers and daughterswould be characterized as higher in achievement orientation,traditional sex role ideology, and weight and appearanceorientation than the psychiatric control mothers and daughters——were not supported. These results will now be discussed.Achievement orientation. Regarding the achievementorientation hypothesis, no group differences were found on theWork, Mastery, or Competitiveness subscales of the WOFO, or onthe Achievement Orientation subscale of the FES. Therestrictive anorexic and nonpsychiatric control daughters did,however, have higher average grades in school than thepsychiatric control daughters.The lack of differences amongst daughters on the WOFO isin contrast with the literature on the cognitive andpersonality characteristics of those with eating disorders.This literature suggests that anorexic and bulimic individualsare perfectionistic and achievement—oriented (e.g., Garner etal., 1982; Garner & Bemis, 1984; Heron & Leheup, 1984;128Thompson et al., 1987). It may be that the WOFO does notcapture the essential aspects of the construct of interest.The WOFO was intended to reflect the desire to work hard, thepreference for challenging tasks, and the enjoyment ofinterpersonal competition. Perhaps the anorexic or bulimicindividual does not rate herself as high on these measuresbecause, due to her negative self—evaluation andperfectionistically high standards (e.g., Garner & Beiuis,1984), she does not perceive herself to be achieving orsuccessful, but rather to simply be doing what in her eyes isnecessary to attain acceptance by self or others (e.g., Bruch,1984; Garner et al., 1982). Even more extremely, due todichotomous reasoning, she may believe that anything short ofspecial or perfect performance renders her worthless (e.g.,Bemis, 1985; Bruch, 1978). Thus, the anorexic or buliinicindividual may not self-report desiring, preferring, orenjoying working hard, despite appearing to others to bedriven to achieve. Garner and Bemis (1984) note that anorexicindividuals often feel inadequate, and that they have not metpeople’s expectations, despite often outstanding actualaccomplishments. This is congruent with the results of thepresent study which showed that, in spite of having a seriouseating disorder, restrictive anorexic daughters were achievingschool grades comparable to nonpsychiatric control daughters’and higher than those of psychiatric control daughters.The lack of significant differences on the FESAchievement Orientation subscale and on the mothers’ WOFO129subscales is incongruent with the clinical literature whichsuggests that the families of eating-disordered individualsare highly achievement-oriented and have high expectations oftheir daughters (e.g., Bruch, 1973; Edwards, 1987; Humphrey &Stern, 1988). What little empirical research has been done inthis area, however, has not generally supported theachievement orientation hypothesis (e.g., Dolan et al., 1990).The majority of studies employing the FES AchievementOrientation subscale have not found significant groupdifferences (e.g., Johnson & Flach, 1985; Ordman &Kirschenbaum, 1986; Shisslak et al., 1990; Strober, 1981).One potential explanation of this discrepancy between theclinical and empirical literature is that, as was argued fordaughters above, the family members are not aware that theirachievement strivings are excessive. Another potentialexplanation is that, because eating—disordered individualshave tended to come from the higher socioeconomic classes(e.g., Boskind—White & White, 1987; Hall, 1978), highachievement orientation may have been more apparent or moreeasily assumed than it would be now as eating disorders arebeginning to occur in a broader range of socioeconomic classes(e.g., Dolan et al., 1990; Pope et al., 1987). In the presentstudy, there were no significant differences in averageparental education or total parental income amongst the threeclinical groups. In general, eating—disordered subjects inthe present study came from the middle to upper-middle class130but there was high variation with all socioeconomic classesbeing represented.Sex role ideology. No significant differences in sexrole ideology were found amongst the groups in the presentstudy. This is incongruent with the clinical literature whichdescribes the families of those with eating disorders asadhering to traditional sex role values (e.g., Gordon et al.,1989; Root et al., 1986; Selvini—Palazzoli, 1978; Wooley &Kearney-Cooke, 1986). However, the one empirical family studyof sex role attitudes also found no significant differences(Dolan et al., 1990). Similarly, the empirical literature onthe sex role attitudes of eating—disordered subjectsthemselves has yielded contradictory results (cf., Ordman &Kirschenbaum, 1986; Rost, Neuhaus, & Florin, 1982;Srikameswaran, Leichner, & Harper, 1984). Thus, it appearsthat the sex role ideology of the families of anorexic andbulimic individuals may not differ from that of controlgroups. This does not, however, preclude the possibility thatthe sex role socialization and stereotypes of the culture atlarge play a necessary role in predisposing women to bevulnerable to the development of eating disorders (e.g.,Boskind—White & White, 1987; Striegel—Moore et al., 1986).Weight and appearance attitudes. Restrictive anorexicand bulimic type daughters, . but not mothers, were found toascribe greater importance to weight and to have more negativeattitudes toward their own weight than psychiatric andnonpsychiatric control daughters. This finding for daughters131was to be expected as such overvaluation and concerns aboutweight are part of •the symptomatology of eating disorders.Thus, this result provides confirmatory evidence of the groupselection criteria, but does not provide evidence in supportof the family weight and appearance orientation hypothesis.That there were no significant differences amongst mothers interms of importance attributed to weight, esteem regardingone’s own weight, or current or ideal weight, fails to supportthe family weight and appearance orientation hypothesis.Similarly, there were no differences amongst mothers ordaughters in terms of importance ascribed to fitness orappearance, or evaluation of one’s own physicalattractiveness. Nonpsychiatric control mothers and daughtershad more positive attitudes toward their own physicalcondition than mothers and daughters in the other threegroups; however, examining the means (see Table 11) suggeststhat this result was more a function of the lower esteem ofthe daughters in the clinical groups than of that of themothers. (The interaction effect for this measure wassignificant at the .05 level but not at the modified alphalevel employed in the present study.)These findings fail to support the suggestions oftheoretical and clinical authors that the parents of thosewith eating disorders are characterized by preoccupation withweight, dieting, and appearance (e.g., Bruch, 1973; Garner &Bends, 1984; Minuchin et al., 1978; Root et al., 1986). Twoempirical studies have addressed these issues and have found132conflicting results. Garfinkel et al. (1983) found nodifferences between the parents of eating—disorderedadolescents and the parents of non—eating—disorderedadolescents in terms of attitudes toward weight and dieting,or body size estimation or satisfaction. Pike and Rodin(1991), however, found that mothers of nonclinical eating-disordered adolescents had more eating—disordered behavior andevaluated their daughters’ weight and appearance morenegatively than mothers of non-eating—disordered adolescents.Similar to the results of the present study, no differenceswere found in mothers’ current or ideal weights, nor inmothers’ evaluation of their own weight and appearance.Further research is needed in this area; however, onepotential explanation of the discrepant findings above is thatPike and Rodin’s (1991) eating-disordered sample was notcomprised of families in treatment for an adolescent’s eatingdisorder. They were, in fact, unaware of the reason for theirinclusion in the study. Perhaps knowing one’s daughter has aneating disorder and is receiving treatment for it (which thefamily is usually involved in, in some way) serves to decreasethe weight and appearance preoccupation which may have beenpresent in the family, or at least to make parents less likelyto report such preoccupation. Pike and Rodin’s (1991) resultsalso suggest that more specific measures, such as weight andappearance orientation being focused directly on the daughter,may be informative.133Another potential explanation for the lack of consistentresults in this area is that weight and appearancepreoccupation may be so prevalent in Western culture (e.g.,Boskind-Lodahl & White, 1978; Garner et al., 1984; Polivy &Herman, 1987) that clinicians observe it in their patients’families, but group differences are not found in empiricalstudies. It is of note that in the present study nosignificant differences were found amongst mothers ordaughters with respect to dissatisfaction with current weight(i.e., current weight minus ideal weight). Thus, mothers anddaughters across groups desired to lose approximately 8 to 20pounds, suggesting an almost universal dissatisfaction withcurrent weight. As Garner’s research group has pointed out(Garner, 1986; Garner et al., 1984), anorexia nervosa is aculturally syntonic disorder.Additional sociocultural milieu measures. Nonpsychiatriccontrol mothers and daughters characterized their families asmore interested in intellectual and cultural activities thanmothers and daughters in the other three groups did; however,this difference appeared to be accounted for by the higheraverage parental education level of the nonpsychiatric controlgroup. There were no group differences in terms of familyparticipation in social and recreational activities or indegree of family emphasis on religious issues and values. Thelack of significant group differences regarding active andrecreational orientation is incongruent with the majority ofthe FES literature which finds lower scores on the Active-134Recreational Orientation subscale in the families of eating—disordered subjects than in normal controls (e.g., Johnson &Flach, 1985; Ordman & Kirschenbaum, 1986; Shisslak et al.,1990; Stern et al., 1987). It is of note, however, that theeffect for the Active-Recreational Orientation subscale in thepresent study was significant according to conventional .05levels, but not by the modified alpha level currentlyemployed.General ConclusionsIn the present study, restrictive anorexic andnonpsychiatric control female adolescents and their motherscharacterized their family environments as more cohesive thanbulimic type and psychiatric control mothers and daughters.Thus, reports by family members of high levels of familysupport and togetherness may distinguish the families ofrestrictive anorexic adolescents from the families ofpsychiatric control adolescents. As the present studyemployed a heterogeneous general psychiatric control group, itis not known whether reports of high cohesion are specific toanorexia nervosa or may also be associated with other morenarrowly defined psychiatric disorders. No specificity ofcharacterization was found for the bulimic type group;however, it was speculated that bulimic type subjects’families may also value high cohesiveness but be unable tomaintain the appearance of cohesion due to greater familialaffective instability. Similarly, while the familysociocultural milieu variables of achievement orientation,135traditional sex role ideology, and weight and appearanceorientation were hypothesized as potentially specific toeating disorders, no significant differences were foundamongst the groups of the present study on these measures.Again, speculations were offered regarding this lack ofcongruence with the theoretical literature, with the aim ofproviding ideas to stimulate further research.As mentioned previously, anorexia nervosa and bulimianervosa are considered to be multidetermined disorders ofeating. Thus, while family cohesion values may be acontributing factor in the genesis of eating disorders, theinfluence of family environment will be exerted within amultidimensional context. An eating disorder will onlydevelop as a result of complex, and as yet undetermined,interactions of biological, personality, psychopathological,family, and sociocultural factors (e.g., Bruch, 1973; Garner &Garfinkel, 1980; Johnson et al., 1987; Strober & Yager, 1984).Some of these other factors may help to explain why notall children in a family develop a disorder of eating, or whyone child develops anorexia nervosa while another developsbulimia nervosa. For instance, the general socioculturalmilieu may help explain why girls and women are much morelikely than boys and men to develop an eating disorder. Also,the recent literature has put forth interesting suggestions asto how genetic differences in personality may contribute tothe origin of anorexia nervosa versus bulimia nervosa; inparticular, it has been suggested that more affectively136unstable genotypes may manifest in buliinia nervosa (Hsu etal., 1990), whereas more avoidant genotypes may manifest inanorexia nervosa (Strober et al., 1990). Similarly,biological differences in the regulation of weight set pointmay contribute differentially to anorexia nervosa and bulimianervosa (e.g., Garner et al., 1984; Keesey, 1986; StriegelMoore et al., 1986). It is of note, however, that evenmonozygotic twins, who share the same genetic material andfamily environment, do not show 100% concordance for eatingdisorders. Evidence suggests that unshared environmentalfactors are a more important source of variance in personalityand psychopathology development than shared environmentalinfluences (Strober et al., 1990). Thus, it is likely thatsuch factors as the different roles and experiences ofsiblings within a family, and the extrafamilial experiences atschool and with peers, play important, but difficult toassess, roles in the development of eating disorders. Thereis also, as Paul Meehi wrote, the “random walk” of life——theaccumulation of perhaps—minor events an individual is exposedto in his or her life, often only as a result of luck orchance (Meehi, 1978). The development of disorders as complexas anorexia nervosa and bulimia nervosa, as well as ofpersonality and psychopathology in general, is subject to themyriad of inexplicable and unknown influences which impactupon every human life, and of which we are so infinitesimallyaware.137Limitations of the Present StudyOne of the main limitations of the present study is thereliance on self—report measures. Especially given the denialor lack of self—awareness common in eating—disordered subjectsand possibly their parents, such reliance on self—report couldbe misleading. However, taking possible denial orminimization into account, it is of interest to determine whatsubjects perceive and/or are willing to report. The additionof more objective or observational measures could provideuseful contextual information to the reports of familymembers. For instance, nonsystematic and unstandardizedobservation in the present study sometimes yielded informationcontradictory to self-report, particularly in the restrictiveanorexic mothers and daughters. In general, it might also beuseful to start employing and/or developing more specificmeasures of family characteristics and functioning than arecurrently in use. In the present study, some null results mayhave been a function of multifaceted constructs not beingadequately assessed. Given the comorbidity of eating disordersamples, very specific measures may be necessary to tap subtledifferences between groups.The small sample size is another weakness of the presentstudy. Cross—validation of the results with a larger samplewould be useful. This is particularly a problem with respectto the bulimic type group. Another limitation is that theclinical groups had been receiving treatment for an average of9 months before participating in the study. This resulted in138eating disorder subjects participating after progress had beenmade and symptoms were often no longer at an acute or crisisstage. Thus, many restrictive anorexic subjects’ weights wereno longer less than 85% of that expected for age and height,although it is of note that their weights were 20% below thoseof girls in the other three groups (see Table 4). Similarly,the bulimic type group was composed of 8 subjects whocurrently met diagnostic criteria for bulimia nervosa, and 6who had met such criteria in the recent past but did not atthe time of participation. Strength of results, therefore,could have been diminished due to duration of treatment andpartial symptomatic recovery. On the other hand, the durationof treatment also serves to decrease the probability thatreported family interaction is solely a function of currentcrisis because of acute symptomatology.Finally, the results of the present study are of limitedgeneralizability. The nonpsychiatric control group iscomposed of self—selected volunteer subjects. There was ahigh rate of participation within the clinical groups;however, all clinical subjects were currently receivingpsychological treatment. Thus, results are only generalizableto families whose daughter is in treatment at an early stageof her disorder. While such a sample is invaluable in terms ofproviding information on the factors present in the initialstages of an eating disorder, it cannot be assumed to yieldresults generalizable to more chronic eating—disordered women139who do not receive treatment until after having suffered manyyears from anorexia nervosa or bulimia nervosa.Questions and Recommendations for Future ResearchThe recent research examining variations in familyenvironment amongst bulimic subjects with respect to level ofdepression (Blouin et al., 1990; Wonderlich & Swift, 1990b),borderline personality disorder features (Johnson et al.,1989; Wonderlich & Swift, 1990a), and history of childhoodsexual abuse (Bulik et al., 1989) is stimulating and thought—provoking. It would be of interest to extend this research infuture to samples of restrictive anorexic and psychiatriccontrol subjects and their parents. The results of thesubsidiary analyses of the present study, while preliminary,suggest that such research might be fruitful and may yieldinformation on the specificity of family variables to anorexiaand/or bulimia nervosa. Also, in terms of specificity,increasingly homogeneous psychiatric control groups could beemployed in attempts to determine family factors of importancein various disorders and to investigate combinations offactors which may be specific to eating disorders, eitheralone or as part of a subgroup of psychiatric disorders.Similarly, it would be of interest to compare groups ofborderline, depressed, or sexually abused subjects with andwithout eating disorders to gain an increasingly fine—tunedunderstanding of the family environments of differentpsychiatric populations.140Another question of interest concerns the direction ofinfluence with respect to the association between familyenvironment and eating disorders. Research designs whichcould illuminate this issue are extremely prohibitive in termsof the time and resources which must be invested in them;however, it would appear that there is currently sufficientbasic information available to make such research endeavorsworthwhile. Pike and Rodin’s (1991) recent study provides agood example of how a nonclinical eating—disordered (or at—risk) sample may be selected and yield interesting results.It would be of great interest to select such a sample andfollow it prospectively with a view to observing andpredicting which subjects develop diagnosable eatingdisorders. Similarly, it would be of interest to follow anadolescent sample of eating-disordered subjects and theirparents, such as that of the present study, to investigate ifthere are changes in family functioning as an eating disorderbecomes more chronic and/or as restrictive anorexic subjectsdevelop bulimic symptomatology.141ReferencesAbraham, S. F., & Beumont, P. J. V. (1982). How patientsdescribe bulimia or binge eating. Psychological Medicine,12, 625—635.American Psychiatric Association. (1980). Diagnostic andstatistical manual of mental disorders (3rd ed.).Washington, DC: Author.American Psychiatric Association (1987). Diagnostic andstatistical manual of mental disorders (3rd ed. rev.).Washington, DC: Author.Armstrong, 3. G., & Roth, D. M. (1989). Attachment andseparation difficulties in eating disorders: Apreliminary investigation. International Journal ofEating Disorders, , 141-155.Barnes, R. A., Ennis, J., & Trachtenberg, D. D. (1985).Childhood disturbances reported by self-harming adults.Paper presented at the Annual Meeting of the AmericanAssociation of Suicidology, Toronto, Canada.Bemis, K. M. (1978). Current approaches to the etiology andtreatment of anorexia nervosa. Psychological Bulletin,, 593—617.Bends, K. M. (1985). “Abstinence” and “nonabstinence” modelsfor the treatment of bulimia. International Journal ofEating Disorders, 4, 407—437.Benjamin, L. S. (1974). Structural analysis of socialbehavior. Psychological Review, 81, 392—425.Blouin, A. G., Zuro, C., & Blouin, J. H. (1990). Familyenvironment in bulimia nervosa: The role of depression.International Journal of Eating Disorders, , 649—658.Boskind-Lodahl, M., & White, W. C., Jr. (1978). The definitionand treatment of bulimarexia in college women: A pilotstudy. Journal of the American College HealthAssociation, fl, 84-86.Boskind-White, M., & White, W. C., Jr. (1987). Bulimarexia:The binge/purge cycle (2nd ed.). New York: W. W. Norton.•Bruch, H. (1973). Eating disorders: Obesity, anorexia nervosa,and the person within. New York: Basic Books.Bruch, H. (1978). The golden cage: The enigma of anorexianervosa. New York: Vintage Books.142Bruch, H. (1984). Four decades of eating disorders. In D. M.Garner & P. E. Garfinkel (Eds.), Handbook ofpsychotherapy for anorexia nervosa and bulixnia (pp. 7-18). New York: The Guilford Press.Bruch, H. (1988). Conversations with anorexics. New York:Basic Books.Bulik, C. M., Sullivan, P. F., & Rorty, M. (1989). Childhoodsexual abuse in women with bulimia. Journal of ClinicalPsychiatry, Q, 460-464.Calam, R., Waller, G., Slade, P., & Newton, T. (1990). Eatingdisorders and perceived relationships with parents.International Journal of Eating Disorders, , 479-485.Cantwell, D. P., Sturzenberger, S., Burroughs, J., Salkin, B.,& Green, J. K. (1977). Anorexia nervosa: An affectivedisorder? Archives of General Psychiatry, , 1087-1093.Carney, C. P., Yates, W. R., & Cizadlo, B. (1990). Acontrolled family study of personality in normal-weightbulimia nervosa. International Journal of EatingDisorders, , 659—665.Chiodo, J. (1987). Invited case transcript - Bulimia: Anindividual behavioral analysis. Journal of BehaviorTherapy and Experimental Psychiatry, 18, 41-49.Cooper, T. (1987). Anorexia and bulimia: The political and thepersonal. In M. Lawrence (Ed.), Fed up and hungry: Women,oppression and food (pp. 175-192). New York: PeterBedrick.Cota, A. A., & Xinaris, S. (1989, June). Psychometricproperties of the Sex-Role Ideology Scale. Paperpresented at the Canadian Psychological AssociationAnnual Convention, Halifax, Nova Scotia.Crisp, A. H. (1981). Anorexia nervosa at normal body weight!The abnormal normal weight control syndrome.International Journal of Psychiatry in Medicine, J,, 203-233.Crisp, A. H., Hall, A., & Holland, A. J. (1985). Nature andnurture in anorexia nervosa: A study of 34 pairs oftwins, one pair of triplets, and an adoptive family.International Journal of Eating Disorders, 4, 5—27.Dare, C. (1985). The family therapy of anorexia nervosa.Journal of Psychiatric Research, j, 435-443.143Derogatis, L. R. (1977). SCL—90: Administration, scoring, andprocedures manual for the revised version. Baltimore:Clinical Psychometric Research.Derogatis, L. R., & Melisaratos, N. (1983). The Brief SymptomInventory: An introductory report. PsychologicalMedicine, fl, 595—605.Derogatis, L. R., & Spencer, P. M. (1982). The Brief SymptomInventory (BSI): Administration, scoring, and proceduresmanual. Towson, MD: Clinical Psychometric Research.Dolan, B. M., Lieberman, S., Evans, C., & Lacey, J. H. (1990).Family features associated with normal body weightbulimia. International Journal of Eating Disorders, 9,639—647.Dykens, E. M., & Gerrard, M. (1986). Psychological profiles ofpurging bulimics, repeat dieters, and controls. Journalof Consulting and Clinical Psychology, , 283—288.Edwards, G. (1987). Anorexia and the family. In M. Lawrence(Ed.), Fed up and hungry: Women, oppression and food (pp.61-73). New York: Peter Bedrick.Epstein, N., Baldwin, L., & Bishop, D. (1983). The McMasterFamily Assessment Device. Journal of Marriage and FamilyCounseling, 9, 171—180.Fairburn, C. G. (1984). Cognitive—behavioral treatment forbulimia. In D. M. Garner & P. E. Garfinkel (Eds.),Handbook of psychotherapy for anorexia nervosa andbulimia (pp. 160-192). New York: The Guilford Press.Fairburn, C. G., & Garner, D. H. (1986). The diagnosis ofbulimia nervosa. International Journal of EatingDisorders, 5, 403—419.Fallon, P., & Root, M. P. P. (1986). Family typology as aguide to the treatment of bulimia. American Mental HealthCounselors Association Journal, , 221-228.Fichter, M. N., & Noegel, R. (1990). Concordance for bulimianervosa in twins. International Journal of EatingDisorders, , 255—263.Forbes, G. B. (1972). Relation of lean body mass to height inchildren and adolescents. Pediatric Research, 6, 32-37.Franzoi, S. L., & Herzog, M. E. (1986). The Body Esteem Scale:A convergent and discriininant validity study. Journal ofPersonality Assessment, , 24—31.144Franzoi, S. L., & Shields, S. A. (1984). The Body EsteemScale: Multidimensional structure and sex differences ina college population. Journal of Personality Assessment,4k., 173—178.Garber, J., & Hollon, S. D. (1991). What can specificitydesigns say about causality in psychopathology research?Psychological Bulletin, UQ, 129-136.Garfinkel, P. E., Garner, D. N., & Moldofsky, H. (1980). Theheterogeneity of anorexia nervosa: Bulimia as a distinctsubgroup. Archives of General Psychiatry, fl, 1036-1040.Garfinkel, P. E., Garner, D. N., Rose, J., Darby, P. L.,Brandes, J. S., O’Hanlon, J., & Walsh, N. (1983). Acomparison of characteristics in the families of patientswith anorexia nervosa and normal controls. PsychologicalMedicine, j., 821—828.Garner, D. M. (1986). cognitive therapy for anorexia nervosa.In K. D. Brownell & J. P. Foreyt (Eds.), Handbook ofeating disorders: Physioloqy, psychology, and treatmentof obesity, anorexia, and bulimia (pp. 301—327). NewYork: Basic Books.Garner, D. M., & Bends, K. M. (1984). cognitive therapy foranorexia nervosa. In D. N. Garner & P. E. Garfinkel(Eds.), Handbook of psychotherapy for anorexia nervosaand bulimia (pp. 107-146). New York: The Guilford Press.Garner, D. M., & Garfinkel, P. E. (1979). The Eating AttitudesTest: An index of the symptoms of anorexia nervosa.Psychological Medicine, 9, 273-279.Garner, D. M., & Garfinkel, P. E. (1980). Socio-culturalfactors in the development of anorexia nervosa.Psychological Medicine, IQ, 647-656.Garner, D. M., Garfinkel, P. E., & Bends, K. M. (1982). Amultidimensional psychotherapy for anorexia nervosa.International Journal of Eating Disorders,,3—46.Garner, D. M., Garfinkel, P. E., & O’Shaughnessy, M. (1985).The validity of the distinction between bulimia with andwithout anorexia nervosa. American Journal of Psychiatry,142, 581—587.Garner, D. M., & Olmsted, N. P. (1984). Eating DisordersInventory: Manual. Odessa, FL: Psychological AssessmentResources.145Garner, D. N., Olmsted, N. P., Bohr, Y., & Garfinkel, P. E.(1982). The Eating Attitudes Test: Psychometric featuresand clinical correlates. Psychological Medicine, , 871—878.Garner, D. N., Rockert, W., Olmsted, N. P., Johnson, C., &Coscina, D. V. (1984). Psychoeducational principles inthe treatment of bulimia and anorexia nervosa. In D. N.Garner & P. E. Garfinkel (Eds.), Handbook ofpsychotherapy for anorexia nervosa and bulimia (pp. 513—572). New York: The Guilford Press.Glass, G. V., Peckham, P. D., & Sanders, J. R. (1972).Consequences of failure to meet assumptions underlyingthe fixed effects analyses of variance and covariance.Review of Educational Research, 4, 237-288.Goldstein, N. J. (1981). Family factors associated withschizophrenia and anorexia nervosa. Journal of Youth andAdolescence, , 385-405.Goodsitt, A. (1974). Anorexia nervosa [Letter to the editor].Journal of the American Medical Association, 230, 372.Gordon, C., Beresin, E., & Herzog, D. B. (1989). The parents’relationship and the child’s illness in anorexia nervosa.Journal of the American Academy of Psychoanalysis, ,29—42.Guidano, V. F. (1988). A systems process-oriented approach tocognitive therapy. In K. S. Dobson (Ed.), Handbook ofcognitive—behavioral therapies (pp. 307-354). New York:The Guilford Press.Guidano, V. F., & Liotti, G. (1983). Cognitive processes andemotional disorders: A structural approach topsychotherapy. New York: The Guilford Press.Hakstin, A. R., Roed, J. C., & Lind, 3. C. (1979). Two—sampleT procedure and the assumption of homogeneous covariancematrices. Psychological Bulletin, , 1255-1263.Hall, A. (1978). Family structure and relationships of 50female anorexia nervosa patients. Australian and NewZealand Journal of Psychiatry, j., 263-268.Hall, A., Leibrich, J., & Walkey, F. H. (1983). Thedevelopment of a Food Fitness and Looks Questionnaire andits use in a study of “weight pathology” in 204nonpatient families. In P. L. Darby, P. E. Garfinkel, D.M. Garner, & D. V. Coscina (Eds.), Anorexia nervosa:Recent developments in research (pp. 41—55). New York:Alan R. Liss.146Halmi, K. A., Struss, A., & Goldberg, S. C. (1978). Aninvestigation of weights in the parents of anorexianervosa patients. Journal of Nervous and Mental Disease,166, 358—361.Harding, T. P., & Lachenmeyer, J. R. (1986). Familyinteraction patterns and locus of control as predictorsof the presence and severity of anorexia nervosa. Journalof Clinical Psychology, 42, 440-448.Harris, R. J. (1975). A primer of inultivariate statistics. NewYork: Academic Press.Helmreich, R. L., & Spence, J. T. (1978). The Work and FamilyOrientation Questionnaire: An objective instrument toassess components of achievement motivation and attitudestoward family and career. JSAS Catalog of SelectedDocuments in Psychology, (2), 35.Heron, J. M., & Leheup, R. F. (1984). Happy families? BritishJournal of Psychiatry, i4, 136-138.Howell, D. C. (1982). Statistical methods for psychology (2nded.). Boston: PWS.Hsu, L. K. G., Chesler, B. E., & Santhouse, R. (1990). Bulimianervosa in eleven sets of twins: A clinical report.International Journal of Eating Disorders, 9, 275-282.Huberty, C. J., & Morris, J. D. (1989). Multivariate analysisversus multiple univariate analyses. PsychologicalBulletin, 1Q, 302—308.Hudson, J. I., Pope, H. G., Jonas, J. N., & Yurgelun—Todd, D.(1983). Family history study of anorexia nervosa andbulixnia. British Journal of Psychiatry, 142, 133—138.Humphrey, L. L. (1986a). Family relations in bulimic—anorexicand nondistressed families. International Journal ofEating Disorders, 5, 223-232.Humphrey, L. L. (l986b). Structural analysis of parent—childrelationships in eating disorders. Journal of AbnormalPsychology, , 395-402.Humphrey, L. L. (1989). Observed family. interactions amongsubtypes of eating disorders using Structural Analysis ofSocial Behavior. Journal of Consulting and ClinicalPsychology, Z, 206-214.147Humphrey, L. L., Apple, R. F., & Kirschenbaum, D. S. (1986).Differentiating bulimic—anorexic from normal familiesusing interpersonal and behavioral observational systems.Journal of Consulting and Clinical Psychology, 54, 190195.Humphrey, L. L., & Stern, S. (1988). Object relations and thefamily system in bulimia: A theoretical integration.Journal of Marital and Family Therapy, 14, 337-350.Johnson, C., Connors, M. E., & Tobin, D. L. (1987). Symptommanagement of bulimia. Journal of Consulting and ClinicalPsychology, , 668-676.Johnson, C., & Flach, A. (1985). Family characteristics of 105patients with bulimia. American Journal of Psychiatry,142, 1321—1324.Johnson, C., & Larson, R. (1982). Bulimia: An analysis ofmoods and behavior. Psychosomatic Medicine, 4.4, 341-351.Johnson, C., Lewis, C., & Hagman, J. (1984). The syndrome ofbulimia: Review and synthesis. Psychiatric Clinics ofNorth America, 2, 247—274.Johnson, C. L., Lewis, C., Love, S., Stuckey, M., & Lewis, L.(1983). A descriptive survey of dieting and bulimia in afemale high school population. In Understanding anorexianervosa and bulimia: Report of the Fourth Ross Conferenceon medical research (pp. 14—20). Columbus, OH: RossLaboratories.Johnson, C. L., Stuckey, M. K., Lewis, L. D., & Schwartz, D.N. (1983). A survey of 509 cases of self-reportedbulimia. In P. L. Darby, P. E. Garfinkel, D. M. Garner, &D. V. Coscina (Eds.), Anorexia nervosa: Recentdevelopments in research (pp. 159-172). New York: Alan R.Liss.Johnson, C., Tobin, D., & Enright, A. (1989). Prevalence andclinical characteristics of borderline patients in aneating—disordered population. Journal of ClinicalPsychiatry, 50, 9-15.Kaffman, N. (1987). Anorexia nervosa in the kibbutz: Factorsinfluencing the monoideistic development. In D. Hardoff &E. Chigier (Eds.), Eating disorders in adolescents andyoung adults: An international perspective (pp. 163-179).London: Freund.Kaffman, N., & Sadeh, T. (1989). Anorexia nervosa in thekibbutz: Factors influencing the development of amonoideistic fixation. International Journal of EatingDisorders, , 33-53.148Kahn, R., & Tilby, P. J. (1978). Development and validationof a sex—role ideology scale. Psychological Reports, 42,731—738.Kalucy, R. S., Crisp, A. H., & Harding, B. (1977). A study of56 families with anorexia nervosa. British Journal ofMedical Psychology, Q, 381-395.Kalucy, R. S., Gilchrist, P. N., McFarlane, C. M., &McFarlane, A. C. (1984). The evolution of a iuultitherapyorientation. In D. N. Garner & P. E. Garfinkel (Eds.),Handbook of psychotherapy for anorexia nervosa andbulimia (pp. 458-487). New York: The Guilford Press.Kaplan, A. S., & Woodside, D. B. (1987). Biological aspects ofanorexia nervosa and buhimia nervosa. Journal ofConsulting and Clinical Psychology, , 645—653.Kassett, J. A., Gershon, E. S., Maxwell, N. E., Guroff, J. J.,Kazuba, D. M., Smith, A. L., Brandt, H. A., & Jimerson,D. C. (1989). Psychiatric disorders in the first-degreerelatives of probands with bulimia nervosa. AmericanJournal of Psychiatry, j4, 1468-1471.Kay, D. W. K., Schapira, K., & Brandon, S. (1967). Earlyfactors in anorexia nervosa compared with non—anorexicgroups. Journal of Psychosomatic Research, U, 133—139.Keck, P. E. Jr., Pope, H. G. Jr., Hudson, J. T., McElroy, S.L., Yurgelun-Todd, D., & Hundert, E. M. (1990). Acontrolled study of phenomenology and family history inoutpatients with buhimia nervosa. ComprehensivePsychiatry, 275—283.Keesey, R. E. (1986). A set-point theory of obesity. In K. D.Brownell & J. P. Foreyt (Eds.), Handbook of eatingdisorders: Physiology, psychology, and treatment ofobesity, anorexia, and bulimia (pp. 63—87). New York:Basic Books.Keys, A., Brozek, J., Henschel, A., Mickelsen, 0., & Taylor,H. L. (1950). The biology of human starvation.Minneapolis: University of Minnesota Press.Kirk, R. E. (1982). Experimental design: procedures for thebehavioral sciences (2nd ed.). Monterey: Brooks/Cole.Kog, E., & Vandereycken, W. (1985). Family characteristics ofanorexia nervosa and buhimia: A review of the researchliterature. Clinical Psychology Review, 5, 159—180.149Kog, E., & Vandereycken, W. (1989). Family interaction ineating disorder patients and normal controls.International Journal of Eating Disorders, 8, 11—23.Lacey, J. H., Gowers, S. G., & Bhat, A. V. (1991). Bulimianervosa: Family size, sibling sex, and birth order: Acatchment-area study. British Journal of Psychiatry, ,491—494.Lee, N. F., Rush, A. J., & Mitchell, J. E. (1985). Bulimia anddepression. Jour’nal of Affective Disorders, , 231—238.Leichner, P., & Kahn, R. (1981). Sex-role ideology amongpracticing psychiatrists and psychiatric residents.American Journal of Psychiatry, 1342-1345.Lewis, H. L.., & MacGuire, M. P. (1985). Review of a group forparents of anorexics. Journal of Psychiatric Research,19, 453—458.Lucido, G., & Abramson, E. E. (1988, April). Adverse childhoodsexual experiences and buhimia. Paper presented at theThird International Conference on Eating Disorders, NewYork, NY.Marascuilo, L. A. (1966). Large-sample multiple comparisons.Psychological Bulletin, , 280-290.Martin, F. (1983). Subgroups in anorexia nervosa: A familysystems study. In P. L. Darby, P. E. Garfinkel, D. M.Garner, & D. V. Coscina (Eds.), Anorexia nervosa: Recentdevelopments in research (pp. 57-63). New York: Alan R.Liss.McNamara, K., & Loveman, C. (1990). Differences in familyfunctioning among bulimics, repeat dieters, andnondieters. Journal of Clinical Psychology, 4, 518—523.Meehl, P. E. (1978). Theoretical risks and tabular asterisks:Sir Karl, Sir Ronald, and the slow progress of softpsychology. Journal of Consulting and ClinicalPsychology, 46, 806-834.Metropolitan Life Insurance Company. (1983). 1983 Metropolitanheight and weight tables. Metropolitan Life InsuranceCompany Statistical Bulletin, , 2-9.Metropolitan Life Insurance Company. (1984). Measurement ofoverweight. Metropolitan Life Insurance CompanyStatistical Bulletin, , 20—23.Minuchin, S., Rosman, B. L., & Baker, L. (1978). Psychosomaticfamilies: Anorexia nervosa in context. Cambridge, MA:Harvard University Press.150Mitchell, J. E., & Eckert, E. D. (1987). Scope andsignificance of eating disorders. Journal of Consultingand Clinical Psychology, 55, 628-634.Mitchell, J. E., Hatsukami, D., Eckert, E. D., & Pyle, R. L.(1985). Characteristics of 275 patients with bulimia.American Journal of Psychiatry, 14, 482-485.Mitchell, J. E., & Pyle, R. L. (1982). The bulimic syndrome innormal weight individuals: A review. InternationalJournal of Eating Disorders, j, 61-73.Mitchell, J. E., Pyle, R. L., Eckert, E. D., Hatsukami, D., &Soil, E. (1990). Bulimia nervosa with and without ahistory of anorexia nervosa. Comprehensive Psychiatry,i, 171—175.Mizes, J. S. (1985). Bulimia: A review of its symptomatologyand treatment. Advances in Behavior Research and Therapy,2, 91—142.Moos, R. H. (1987). The Social Climate Scales: A user’s guide.Palo Alto: Consulting Psychologists Press.Moos, R. H., & Moos, B. S. (1986). Family Environment Scale:Manual (2nd ed.). Palo Alto: Consulting PsychologistsPress.Morgan, H. G., & Russell, G. F. M. (1975). Value of familybackground and clinical features as predictors of long—term outcome in anorexia nervosa: Four—year follow—upstudy of 41 patients. Psychological Medicine, , 355-371.Noordenbos, G. (1987). Possible preventive measures foranorexia nervosa. In D. Hardoff & E. Chigier (Eds.),Eating disorders in adolescents and young adults: Aninternational perspective (pp. 437-446). London: Freund.Norris, D. L., & Jones, E. (1979). Anorexia nervosa: Aclinical study of ten patients and their family systems.Journal of Adolescence, , 101-111.Olson, D. H., Bell, R., & Portner, J. Family Adaptability andCohesion Scale. St. Paul, Minnesota: University ofMinnesota Press.Oppenheimer, R., Howells, K., Palmer, R. L., & Chaloner, D. A.(1985). Adverse sexual experience in childhood andclinical eating disorders: A preliminary description.Journal of Psychiatric Research, j, 357-361.Orbach, S. (1986). Hunger strike: The anorectic’s struggle asa metaphor for our age. New York: Avon Books.151Ordman, A. M., & KirschenbaunL, D. S. (1986). Bulimia:Assessment of eating, psychological adjustment, andfamilial characteristics. International Journal of EatingDisorders, 5, 865—878.Palmer, R. L. (1979). Dietary chaos syndrome:; A useful newterm? British Journal of Medical Psychology, , 187-190.Palmer, R. L., Oppenheimer, R., Dignon, A., Chaloner, D. A., &Howelis, K. (1990). Childhood sexual experiences withadults reported by women with eating disorders: Anextended series. British Journal of Psychiatry, 156, 699-703.Palmer, R. L., Oppenheimer, R., & Marshall, P. D. (1988).Eating—disordered patients remember their parents: Astudy using the Parental Bonding Instrument.International Journal of Eating Disorders, 2, 101-106.Parker, G., Tupling, H., & Brown, L. B. (1979). A parentalbonding instrument. British Journal of ClinicalPsychology, 52, 1-10.Perosa, L., Hansen, J., & Perosa, S. (1981). Development ofthe Structural Family Interaction Scale. Family Therapy,8, 77—90.Pike, K. M., & Rodin, J. (1991). Mothers, daughters, anddisordered eating. Journal of Abnormal Psychology, QQ,198—204.Piran, N., Lerner, P., Garfinkel, P. E., Kennedy, S. H., &Brouiliette, C. (1988). Personality disorders in anorexicpatients. International Journal of Eating Disorders, 2,589—599.Pole, R., Wailer, D. A., Stewart, S. M., & Parkin-Feigenbaum,L. (1988). Parental caring versus overprotection inbulimia. International Journal of Eating Disorders, i601—606.Polivy, J., & Herman, C. P. (1987). Diagnosis and treatment ofnormal eating. Journal of Consulting and ClinicalPsychology, , 635-644.Pope, H. G., Champoux, R. F., & Hudson, J. I. (1987). Eatingdisorder and socioeconomic class: Anorexia nervosa andbuliiuia in nine communities. Journal of Nervous andMental Disease, 620—623.Pope, H. G., Hudson, J. I., & Yurgelun-Todd, D. (1984).Anorexia nervosa and bulimia among 300 suburban womenshoppers. American Journal of Psychiatry, lil, 292—294.152Pyle, R. L., Mitchell, J. E., & Eckert, E. D. (1981). Bulimia:A report of 34 cases. Journal of Clinical Psychiatry, 42,60—64.Pyle, R. L., Mitchell, J. E., Eckert, E. D., Halvorson, P. A.,Neuman, P. A., & Goff, G. M. (1983). The incidence ofbulimia in freshman college students. InternationalJournal of Eating Disorders, a, 75-85.Rakoff, V. (1983). Multiple determinants of family dynamics inanorexia nervosa. In P. L. Darby, P. E. Garfinkel, D. M.Garner, & D. V. Coscina (Eds.), Anorexia nervosa: Recentdevelopments in research (pp. 29—40). New York: Alan R.Liss.Rivinus, T. M., Biederman, J., Herzog, D. B., Kemper, K.,Harper, G. P., Harmatz, J. S., & Houseworth, S. (1984).Anorexia nervosa and affective disorders: A controlledfamily history study. American Journal of Psychiatry,14.1, 1414—1418.Roberto, L. G. (1986). Bulimia: The transgenerational view.Journal of Marital and Family Therapy, j., 231-240.Robin, A. L. & Weiss, J. G. (1980). Criterion-related validityof behavioral and self—report measures of problem—solvingcommunication skills in distressed and nondistressedparent—adolescent dyads. Behavioral Assessment, a, 339—359.Root, N. P. P., & Fallon, P. (1988). The incidence ofvictimization experiences in a bulimic sample. Journal ofInterpersonal Violence, 2, 161-173.Root, M. P. P., Fallon, P., & Friedrich, W. N. (1986).Bulimia: A systems approach to treatment. New York: W. W.Norton.Rosen, J. C., Silberg, T., & Gross, J. (1988). EatingAttitudes Test and Eating Disorders Inventory: Norms foradolescent girls and boys. Journal of Consulting andClinical Psychology, , 305-308.Rosman, B. L., Minuchin, S., Baker, L., & Liebman, R. (1977).A family approach to anorexia nervosa: Study, treatment,and outcome. In R. A. Vigersky (Ed.), Anorexia nervosa(pp. 341-348). New York: Raven Press.Rost, W., Neuhaus, N., & Florin, I. (1982). Bulimia nervosa:Sex role attitude, sex role behavior, and sex rolerelated locus of control in buliinarexic women. Journal ofPsychosomatic Research, , 403-408.153Russell, G. (1979). Bulimia nervosa: An ominous variant ofanorexia nervosa. Psychological Medicine, 9, 429—448.Sallas, A. A. (1985). Treatment of eating disorders: Winningthe war without having to do battle. Journal ofPsychiatric Research, 19, 445-448.Sargent, J., Liebman, R., & Silver, M. (1984). Family therapyfor anorexia nervosa. In D. M. Garner & P. E. Garfinkel(Eds.), Handbook of psychotherapy for anorexia nervosaand bulimia (pp. 257-279). New York: The Guilford Press.Schlesier-Stropp, B. (1984). Bulimia: A review of theliterature. Psychological Bulletin, , 247-257.Schwartz, R. C., Barrett, M. J., & Saba, G. (1984). Familytherapy for bulimia. In D. M. Garner & P. E. Garfinkel(Eds.), Handbook of psychotherapy for anorexia nervosaand bulimia (pp. 280-307). New York: The Guilford Press.Selvini—Palazzoli, M. (1978). Self—starvation: From individualto family therapy in the treatment of anorexia nervosa(A. Pomerans, Trans.). New York: Jason Aronson.Shisslak, C. N., Crago, N., Neal, M. E., & Swain, B. (1987).Primary prevention of eating disorders. Journal ofConsulting and Clinical Psychology, , 660-667.Shisslak, C. M., McKeon, R. T., & Crago, N. (1990). Familydysfunction in normal weight bulimic and bulimic anorexicfamilies. Journal of Clin.ical Psychology, 4, 185—189.Shisslak, C. N., Pazda. S. L., & Crago, N. (1990). Body weightand bulimia as discriminators of psychologicalcharacteristics among anorexic, bulimic, and obese women.Journal of Abnormal Psychology, , 380-384.Sights, J. R., & Richards, H. C. (1984). Parents of bulimicwomen. International Journal of Eating Disorders, 2., 3—13.Skinner, H. A., Steinhauser, P. D., & Santa—Barbara, J.(1983). The Family Assessment Measure. Canadian Journalof Community Mental Health, ., 91-105.Sours, J. A. (1974). The anorexia nervosa syndrome.International Journal of Psychoanalysis, j, 567-576.Spanier, G. B. (1976). Measuring dyadic adjustment: New scalesfor assessing the quality of marriage and similar dyads.Journal of Marriage and the Family, 3, 15-28.154Srikameswaran, S., Leichner, P., & Harper, D. (1984). Sex roleideology among women with anorexia nervosa and bulimia.International Journal of Eating Disorders, 3, 39-43.Steiger, H., Liquornik, K., Chapman, J., & Hussain, N. (1991).Personality and family disturbances in eating-disorderpatients: Comparison of “restricters” and “bingers” tonormal controls. International Journal of EatingDisorders, 10, 501-512.Steiger, H., Van der Feen, J., Goldstein, C., & Leichner, P.(1989). Defense styles and parental bonding in eating-disordered women. International Journal of EatingDisorders, 8, 131-140.Stern, S. (1986). The dynamics of clinical management in thetreatment of anorexia nervosa and bulimia: An organizingtheory. International Journal of Eating Disorders, 5,233—254.Stern, S. L., Dixon, K. N., Jones, D., Lake, M., Nemzer, E., &Sansone, R. (1987). Family environment in anorexianervosa and bulimia. In D. Hardoff & E. Chigier (Eds.),Eating disorders in adolescents and young adults: Aninternational perspective (pp. 125-134). London: Freund.Stern, S., Whitaker, C. A., Hagemann, N. J., Anderson, R. B.,& Bargman, G. J. (1981). Anorexia nervosa: The hospital’srole in family treatment. Family Process, ZQ, 395-408.Strauss, J., & Ryan, R. M. (1987). Autonomy disturbances insubtypes of anorexia nervosa. Journal of AbnormalPsychology, , 254-258.Strauss, J., & Ryan, R. N. (1988). Cognitive dysfunction ineating disorders. International Journal of EatingDisorders, 2, 19-27.Striegel-Moore, R. H., Silberstein, L. R., & Rodin, J. (1986).Toward an understanding of risk factors for bulimia.American Psychologist, 4, 246-263.Strober, M. (1981). The significance of bulimia in juvenileanorexia nervosa: An exploration of possible etiologicfactors. International Journal of Eating Disorders, j,28—43.Strober, M., & Humphrey, L. L. (1987). Familial contributionsto the etiology and course of anorexia nervosa andbulimia. Journal of Consulting and Clinical Psychology,55, 654—659.155Strober, N., Lampert, C., Morrell, W., Burroughs, J., &Jacobs, C. (1990). A controlled family study of anorexianervosa: Evidence of familial aggregation and lack ofshared transmission with affective disorders.International Journal of Eating Disorders., 9, 239—253.Strober, N., Morreli, W., Burroughs, J., Saikin, B., & Jacobs,C. (1985). A controlled family study of anorexia nervosa.Journal of Psychiatric Research, , 239-246.Strober, N., Salkin, B., Burroughs, J., & Morrell, W. (1982).Validity of the bulimic-restricter distinction inanorexia nervosa: Parental personality characteristicsand family psychiatric morbidity. Journal of Nervous andMental Disease, 120, 345-351.Strober, M., & Yager, J. (1984). A developmental perspectiveon the treatment of anorexia nervosa in adolescents. InD. N. Garner & P. E. Garfinkel (Eds.), Handbook ofpsychotherapy for anorexia nervosa and bulimia (pp. 363—390). New York: The Guilford Press.Tabachnick, B. G., & Fidell, L. S. (1983). Using multivariatestatistics. New York: Harper & Row.Taipale, V., Tuomi, 0., & Aukee, N. (1971). Anorexia nervosa:An illness of two generations? Acta Paedopsychiatry, ,2 1—25.Thompson, D. A., Berg, K. N., & Shatford, L. A. (1987). Theheterogeneity of bulimic symptomatology: Cognitive andbehavioral dimensions. International Journal of EatingDisorders, §, 215—234.Wailer, G., Slade, P., & Calam, R. (1990a). Familyadaptability and cohesion: Relation to eating attitudesand disorders. International Journal of Eating Disorders,., 225—228.Wailer, G., Siade, P., & Calam, R., (1990b). Who knows best?Family interaction and eating disorders. British Journalof Psychiatry, j, 546-550.Waters, B. G. H., Beumont, P. J. V., Touyz, S., & Kennedy, M.(1990). Behavioral differences between twin and non-twinfemale sibling pairs discordant for anorexia nervosa.International Journal of Eating Disorders, , 265-273.Weiss, L., Katzman, M., & Woichik, S. (1985). Treatingbulimia: A psychoeducational approach. New York: PergamonPress.156Williams, R. L. (1987). Use of the Eating Attitudes Test andEating Disorder Inventory in adolescents. Journal ofAdolescent Health Care, 8, 266—272.Wilson, G. T., & Walsh, B. T. (1991). Eating disorders in theDSM—IV. Journal of Abnormal Psychology, 100, 362-365.Winer, B. J. (1971). Statistical principles in experimentaldesign (2nd ed.). New York: McGraw-Hill.Wonderlich, S. A., & Swift, W. J. (1990a). Borderline versusother personality disorders in the eating disorders:Clinical description. International Journal of EatingDisorders, , 629—638.Wonderlich, S. A., & Swift, W. 3. (l990b). Perceptions ofparental relationships in the eating disorders: Therelevance of depressed mood. Journal of AbnormalPsychology, , 353-360.Wooley, S. C., & Kearney-Cooke, A. (1986). Intensive treatmentof bulimia and body-image disturbance. In K. D. Brownell& J. P. Foreyt (Eds.), Handbook of eating disorders:Physiology, psychology, and treatment of obesity,anorexia, and bulimia (pp. 476-502). New York: BasicBooks.Wooley, S. C., & Wooley, 0. W. (1984). Intensive outpatientand residential treatment for bulimia. In D. N. Garner &P. E. Garfinkel (Eds.), Handbook of psychotherapy foranorexia nervosa and bulimia (pp. 391-430). New York: TheGuilford Press.Wooley, 0. W., Wooley, S. C., & Dyrenforth, S. R. (1979).Obesity and women II: A neglected feminist topic. Women’sStudies International Quarterly,,81-92.Yager, J. (1982). Family issues in the pathogenesis ofanorexia nervosa. Psychosomatic Medicine, 44., 43—60.157AppendixSubsidiary Analyses: Family System/Interaction HypothesisFour 2(High/Low) (Depression, Distress, Impulsivity, orEating Disorder Symptomatology) X 2(Relation) between-withingroups MANOVAs were conducted. (The main effect of Relationwas not of interest and, therefore, was not examined.) Thedependent variables were FES—Cohesion, FES—Expressiveness,FES—Conflict, FES-Independence, FES-Organization, and FESControl. Depression was operationalized by scores on the BSIDepression scale. A standard T—score of 60, or 1 standarddeviation above the mean on the female adolescent norms, wasused to define high/low; thus, daughters with scores greaterthan 1.51 comprised the High Depression group and daughterswith scores less than or equal to 1.51 comprised the LowDepression group. Similarly, Distress was operationalized bythe BSI—General Severity Index. Daughters with scores above1.39 comprised the High Distress group and those with scoresless than or equal to 1.39 comprised the Low Distress group.Impulsivity was defined by the number of impulse-relatedbehaviors reported. The grand mean on this measure was 2.06;hence, daughters with scores greater than 2 comprised the HighImpulsivity group and daughters with scores less than or equalto 2 comprised• the Low Impulsivity group. And finally, EatingDisorder Symptomatology was operationalized by scores on theEAT—26. The recommended cut-off of 20 was employed such thatdaughters with scores greater than 20 comprised the High group158and daughters with scores less than or equal to 20 comprisedthe Low group.As an indication of the composition of the study’s fourgroups with regard to the four factors above, percentages arepresented in Table 16. There are, of course, no daughtersfrom the psychiatric or nonpsychiatric control groups in theHigh Eating Disorder Symptomatology group as the selectioncriteria excluded daughters with EAT-26 scores greater than 20from these groups.Table 16Composition of Study’s GroupsRestrict. Bulimic Psych. Nonpsych.Anorexic Type Control ControlDEPRESSION:High 35% 79% 35% 12.5%Low 65% 21% 65% 87.5%DISTRESS:High 40% 71% 30% 4%Low 60% 29% 70% 96%IMPULSIVITY:High 30% 50% 40% 33%Low 70% 50% 60% 67%E.D. SYMPTOMS:High 70% 79% 0% 0%Low 30% 21% 100% 100%The 2(Depression) X 2(Relation) MANOVA was significantfor the Depression by Relation interaction ((6, 71) = 4.51,159<.002) and for Depression (E(6, 71) = 2.70, <.03). Themeans for the six dependent measures for daughters and mothersare presented in Table 17. Subsequent ANOVAs, with modifiedalpha of .017 (.10 divided by 6), showed significantinteraction effects for Cohesion ((l, 76) = 9.24, p<.004),Conflict ((1, 76) = 10.04, p<.003), Independence ((1, 76) =10.46, p<.003), and Organization ((l, 76) = 12.06, p<.002).Tests of simple main effects revealed significant groupdifferences for daughters but not for mothers. Thus, theDepression effects for mothers on Cohesion, Conflict,Independence, and Organization were, all nonsignificant,whereas the effects of Depression for daughters on Cohesion(F(1, 135) = 12.55, p<.001), Conflict ((1, 134) = 11.98,p<.001), and Independence (corrected for heterogeneity; F’(l,41) = 12.97, p<.001) were all significant. The result forOrganization for daughters was nonsignificant. Neither theinteraction effect nor the Depression effect forExpressiveness or for Control were significant.The MANOVA for Distress was significant for Distress((6, 71) = 2.40, p<.04) and for the Distress by Relationinteraction ((6, 71) = 2.55, p<.03). The means are presentedin Table 18. The follow-up ANOVAs were significant for theDistress by Relation interaction for Conflict ((1, 76) =6.53, p<.O17) and for Independence ((1, 76) = 9.33, p<.004).The tests of the simple main effects of Distress on Conflictwere not significant for mothers or for daughters. The simple• main effect of Distress on Independence was significant for160daughters ((1, 152) = 21.07, p<.001), but not for mothers.The 2 X 2 ANOVA interaction effects were not significant forTable 17Low versus High Depression Group Means (Standard Deviations)Low HighVARIABLE Depression Depression, RELATION (n==50) (n=28)6.14(2.55)6.80(2.12)4.32(2.11)6. 00(1.95)3.76(2.19)3 .88(2.17)6.52(1.40)6.54(1.28)5.82(2.02)5.48(2.39)4.96(2.37)5.00(1.98)4 . 14(2.85)6.75(2.01)3.21(1.83)5.61(1.83)5.50(1.99)3 . 82(2.07)4.93(2.09)6 • 57(1.48)4.39(2.62)5.68(2.39)5.21(2.57)4.39(2.25)CohesionExpressivenessConflictIndependenceOrganizationControlDaughterMotherDaughterMotherDaughterMotherDaughterMotherDaughterMotherDaughterMother161Table 18Low versus High Distress Group Means (Standard Deviations)Low HighVARIABLE Distress DistressRELATION (n=53) (n=25)5.87(2.80)6.79(2.13)4.32(2.08)6.02(1.93)3.96(2.28)3.93(2.24)6.49(1.50)6.59(1.34)5.42(2.41)5.32(2.34)4.87(2.45)4.74(2.11)4.48(2. 65)6.76(1.99)3 .08(1.82)5.52(1.85)5.28(2.01)3.72(1.88)4.80(1.98)6.48(1.39)5.08(2.20)6.04(2.44)5.44(2.40)4.88(2.07)CohesionExpressivenessConflictIndependenceOrganizationControlDaughterMotherDaughterMotherDaughterMotherDaughterMotherDaughterMotherDaughterMother162Cohesion, Expressiveness, Organization, or Control.Similarly, the main effects of Distress were nonsignificantfor Cohesion, Expressiveness, Organization, and Control.The MANOVA for Impulsivity yielded a significantImpulsivity by Relation interaction effect (F(6, 71) = 3.14,p<.O1) and a nonsignificant Impulsivity main effect. Themeans for the Low versus High Impulsivity groups are presentedin Table 19. The subsequent ANOVAs revealed significantinteraction effects for Cohesion ((1, 76) = 7.78, <.OO8),Conflict ((1, 76) = 11.69, p<002), and Independence ((1,76) = 6.41, p<.017) Tests of simplemain effects showed thatthe main effect of Impulsivity on Cohesion was significant fordaughters ((1, 134) = 6.93, p<.Ol), but not for mothers; thatthe main effect on Conflict was significant for daughters((1, 132) = 7.29, p<.O1), but not for mothers; and thatneither the main effect on Independence for mothers nor thecorrected—for—heterogeneity main effect on Independence fordaughters were significant.The Impulsivity by Relation interaction effects forExpressiveness, Organization, and Control were allnonsignificant. Similarly, the main effect of Impulsivity wasnonsignificant for Expressiveness, Organization, and Control.Finally, the 2 X 2 MANOVA for Eating DisorderSymptomatology was nonsignificant for the main effect and theinteraction effect. The means for the Low versus High EatingDisorder Symptomatology groups are presented in Table 20.163Table 19Low versus High Impulsivity Group Means (Standard Deviations)VARIABLERELATIONLowlinpuls ivity(n=49)5.98(2.76)6.67(2.25)4.16(2.11)5.88(2.01)3.88(2.20)4.06(2.23)6.18(1.40)6.31(1.34)5.67(2.28)5.96(2.35)5.16(2.39)5.18(2.09)HighImpulsivity(n=29)4.48(2.69)6.97(1.74)3 . 52(1.99)5.83(1.75)5.24(2.15)3 .52(1.92)5.55(2.38)6.97(1.27)4 . 69(2.35)4.86(2.30)4.86(2.53)4.10(1.93)Cohesion DaughterMotherExpressiveness DaughterMotherConflict DaughterMotherIndependence DaughterMotherOrganization DaughterMotherControl DaughterMother164Table 20Low versus High Eating Disorder Symptoinatology Group Means(Standard Deviations).Low E.D. High E.D.VARIABLE Symptoxnatology Symptomato logyRELATION (n=53) (n=25)Cohesion Daughter 5.38 5.52(2.89) (2.69)Mother 6.81 6.72(2.09) (2.05)Expressiveness Daughter 3.98 3.80(1.97) (2.33)Mother . 5.98 5.60(2.01) (1.68)Conflict . Daughter 4.32 4.52(2.46) (1.85)Mother 3.85 3.88(2.18) (2.05)Independence Daughter 6.23 5.36(1.48) (2.36)Mother 6.42 6.84(1.34) (1.34)Organization Daughter 5.36 5.20(2.35) (2.36)Mother 5.60 5.44(2.48) (2.20)Control Daughter 5.30 4.52(2.42) (2.42)Mother 5.11 4.08(2.12) (1.87)


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items