Aspects of Self Concept

7/28/2019 Aspects of Self Concept

1/27

Jna Sca and Cnca Pscg, V. 29, N. 7, 2010, pp. 821-846

821

The research reported in this article was supported by the ollowing grants to AnnaM. Bardone-Cone: NIH 1 R03MH074861-01A1, University o Missouri PRIME Grant,and University o Missouri Research Council Grant.

Correspondence concerning this article should be addressed to Anna M. Bardone-Cone, CB #3270 Davie Hall, Department o Psychology, University o North Carolina atChapel Hill, Chapel Hill, NC 27599. E-mail: [email protected].

SELF-CONCEPTANDEATINGDISORDER RECOVERY

BARDONE-CONE ETAL.

aspeCts of self-ConCept and

eating disorder reCovery:

What does the sense of self look like

When an individual reCovers from

an eating disorder?

ANNA M. bArDoNe-CoNeUniversity of North Carolina at Chapel Hill

lAureN M. SChAeer

University of South Florida

ChriStiNe r. MAlDoNADo

University of Missouri

elleN e. itZSiMMoNS AND MeGAN b. hArNeyUniversity of North Carolina at Chapel Hill

MeliSSA A. lAWSoN, D. PAul robiNSoN, ANeeSh toSh,AND roMA SMithUniversity of Missouri School of Medicine

ts sac xamnd ans wn aspcs s-cncp and vassags ang dsd cv. indvdas m sn an ang dsd

a a Mdwsn cnc w cagzd as avng an acv ang dsd (n =53) as paa cvd (n = 15) as cvd (n = 20) sng a cm-pnsv cv dnn w cv ncdd psca, av-a, and pscgca cv and paa cv ncdd n psca andava cv. t s-cncps s gps w cmpad ac and 67 cns. t cvd gp ad g s-sm, g


2/27

822 Bardone-Cone et al.

s-dcdnss, and w vs mps pnmnn an ndvdasw w paa cvd s w m ca an ang dsd, as

w as g s-cac an acv ang dsd gp. cvdndvdas as kd n ms spcc dmans s-cncp (.g.,nmac, sca, c.) wn cmpad acv ang dsd gp.rss pvd vdnc a cvd ndvdas w cmpaa cns n a mass s-cncp. Paa cvd ndvdas, w-v, w m sma acv ang dsd cass, sggsng a mpvds-cncp ma an nga pa ang dsd cv. Addna,s ss pvd spp a m cmpnsv dnn -cv wc acknwdgs pscgca aspcs an ang dsd.

Sel-concept disturbances have been theoretically posited as corevulnerabilities or the development and maintenance o eating dis-orders, as well as or relapse (Bruch, 1981; Daley, Jimerson, Heath-erton, Metzger, & Wole, 2008; Stein & Corte, 2003). For example,researchers have ound that the presence o relatively ew positiveand more negative sel-schemas may be predictive o eating pathol-ogy (Stein & Corte, 2007); urther, low sel-esteem has been asso-ciated with poor outcome and relapse in a one-year ollow-up o

individuals with bulimia nervosa (Fairburn, Peveler, Jones, Hope,& Doll, 1993). Qualitative work also supports an important relationbetween sel-concept and disordered eating, with women in recov-ery rom an eating disorder describing reaching sel-acceptance, aswell as cultivating and maintaining a sense o sel-worth, as criti-cal to attaining and maintaining recovery (Federici & Kaplan, 2008;Patching & Lawler, 2009; Vanderlinden, Buis, Pieters, & Probst,2007). Those who relapsed identied sel-criticism and a pervasivesense o worthlessness as actors hindering their recovery (Federici& Kaplan, 2008). Thus, it appears that a more sustained recovery is

more likely among individuals with improved sel-concept. Con-versely, i sel-concept disturbances, such as low sel-esteem, persistollowing recovery o an eating disorder, risk or relapse may beheightened (Daley et al., 2008).

How might ongoing sel-concept disturbances contribute to re-lapse? Those no longer meeting criteria or an eating disorder butwith low sel-esteem may be at risk or returning to a ocus on ap-pearance (and concomitant behaviors, such as dietary restriction orweight loss) as a way to boost sel-esteem (Anderson & Maloney,2001). And recovery with a lingering sense o ineectiveness (lowsel-ecacy) may result in small slips via eating disordered behav-


3/27

self-ConCept and eating disorder reCovery 823

iors or thoughts snowballing into a more serious return o eating

pathology because the individual lacks condence in her abilitiesto maintain recovery. Also, given that ones sel-image guides be-haviors in social interactions and can elicit behaviors that conrmthe sel-image (Birgegard, Bjorck, Norring, Sohlberg, & Clinton,2009; Jones, 1986), a poor sel-image may generate behaviors thatreinorce the negative view o the sel, resulting in negative aectwhich is a robust risk actor or eating pathology (Stice, 2002). Inthe current study, we investigate what the sel-concept looks like atdierent stages o an eating disorder, with particular interest in howthose ully recovered rom an eating disorder experience the sel.

Components of self-ConCept

From a cognitive perspective, sel-concept can be dened as a seto knowledge structures about the sel (Stein & Corte, 2007, p. 59)and encompasses a wide range o constructs (Baumeister, 1999).While these sets o structures may come together to refect one un-derlying sel-concept, it is inormative to look at various aspects osel-concept separately given that they have conceptual dierences

and that there may be dierent ways to target these aspects in in-tervention and prevention approaches. According to Markus andWur (1987), sel-concept represents a dynamic multiaceted con-struct, rather than a unitary undierentiated structure. Sel-repre-sentations that comprise the sel-concept are not all alike; some maybe more positive, more negative, more salient, more predictive outure behaviors, or more accurate than others (Markus & Wur,1987). Indeed, researchers and clinicians alike have ound great util-ity in parsing out various aspects o sel-concept to urther elucidate

their signicance in both the etiology and treatment o eating disor-ders (Fairburn, 2008; Halvorsen & Heyerdahl, 2006; Jacobi, Paul, deZwaan, Nutzinger, & Dahme, 2004; Wilson, Fairburn, Agras, Walsh,& Kraemer, 2002; Wonderlich et al., 2008).

The particular aspects o sel-concept chosen or examination inthis study are those with support or therapeutic relevance in theeating disorders (sel-esteem, sel-ecacy, and sel-directedness) aswell as the imposter phenomenon, which is a conceptually compel-ling, but understudied, way to look at the sel. In brie, sel-esteeminvolves an evaluative element and sense o worth, and can be mea-

sured globally (e.g., overall, I think I am a pretty good person), as


4/27


well as in specic domains (e.g., I think Im a good worker). Sel-

ecacy is the aspect o sel-concept that links the sel to agency andcontrol (Bandura, 1977; Baumeister, 1999), with individuals high insel-ecacy eeling condent in their abilities to do what is neces-sary to attain their goals. Relatedly, the concept o sel-directednessreers to sel-determination and the ability to control, regulate, andalter behavior as needed in pursuing goals (Cloninger, Svrakic, &Przybeck, 1993). Sel-directedness also refects sel-acceptance, per-sonal responsibility, resourceulness, and the perception o the selas integrated and autonomous (Cloninger et al., 1993). Yet anotherway o studying sel-concept is to examine the degree to which indi-

viduals experience the sel as alse, in particular eeling that otherssee them as competent (exterior sel), while they themselves eelinadequate (interior sel). This construct has been reerred to as theimposter phenomenon and perceived raudulence (Kolligan &Sternberg, 1991).

These sel-concept constructs have been implicated in therapeuticchange in the eating disorders. For example, Fairburns (2008) en-hanced version o cognitive behavioral therapy or eating disor-ders (CBT-E) includes a module targeting low sel-esteem, concep-tualized as an obstacle to recovery. Additionally, there is evidencethat sel-ecacy mid-treatment is a mediator o change in eatingdisorder symptomatology (Wilson et al., 2002) and that increasesin sel-directedness rom pre- to post-CBT are associated with im-proved eating psychopathology (Grave, Calugi, Brambilla, Abbate-Daga, Fassino, & Marchesini, 2007).

researCh on the self and eating disorders

There is substantial support or eating disorders being associatedwith low sel-esteem (Gual et al., 2002; Jacobi et al., 2004; Peck &Lightsey, 2008). Regarding the recovery process and sel-esteem,there is some support or individuals in remission rom bulimia ner-vosa (BN), dened as the absence o eating disorder symptoms (e.g.,binge eating, purging) or six months, having sel-esteem scores thatare signicantly higher than those with current BN, but signicant-ly lower than healthy controls (Daley et al., 2008). In contrast, Blaaseand Elklit (2001) ound that individuals recovered rom an eatingdisorder were comparable to healthy controls on measures o sel-

esteem. While little work has examined domain-specic sel-esteem


5/27


in relation to eating disorder recovery, there is some evidence that

compared to symptomatic individuals, those recovered rom BN atan 18-month post-treatment assessment reported greater social sel-esteem (Troop, Schmidt, Turnbull, & Treasure, 2000).

Sel-ecacy also has a long-standing relationship with eating dis-orders, with historical reports identiying low levels o sel-ecacyas a striking eature o eating disorder patients (e.g., paralyzingsense o ineectiveness; Bruch, 1962, p. 191) and research ndingthat eating disorder individuals report greater levels o personal in-eectiveness and lower general ecacy compared to healthy con-trols (Etringer, Altmaier, & Bowers, 1989; Jacobi et al., 2004; Peck &

Lightsey, 2008; Wagner, Halmi, & Maguire, 1987). Studies assessingineectiveness and eating disorder recovery have yielded mixedresults with some studies nding no dierences between individu-als recovered rom an eating disorder and controls (Brambilla et al.,2003; Lileneld et al., 2000) and other studies nding that recoveredindividuals reported a greater sense o ineectiveness than controls(Kaye et al., 1998; Stein et al., 2002). There is also some evidence orthe prognostic value o ineectiveness; high initial ineectivenesshas been associated with poor prognosis or patients with anorexianervosa (AN; Bizeul, Sadowsky, & Rigaud, 2001).

Sel-directedness appears to be low across all eating disordertypes (Cassin & von Ranson, 2005). Klump et al. (2004) ound thatwomen with current eating disorders and women recovered roman eating disorder scored signicantly lower on sel-directednessthan controls. However, other work has ound that individuals ullyrecovered rom an eating disorder have signicantly higher levelso sel-directedness when compared to those partially recoveredor with an active eating disorder (Bloks, Hoek, Callewaert, & vanFurth, 2004; Bulik, Sullivan, Fear, & Pickering, 2000). From a pre-

diction perspective, pre-treatment and end-o-treatment sel-direct-edness appear to predict symptomatology in AN patients (Bloks,Hoek et al., 2004).

Although minimal research exists on the imposter phenomenonand eating disorders, this construct is conceptually compelling sinceindividuals with eating disorders oten strive to conceal their dis-ordered eating behaviors (e.g., binge eating, vomiting, extreme re-striction), which may contribute to a disconnect between their pub-lic and private selves. Striegel-Moore, Silberstein, and Rodin (1993)ound that women with BN reported greater perceived raudulence

than a nonclinical group with elevated eating disorder symptoms,


6/27


7/27


or more subscale o the EDE-Q greater than 1 SD o community

norms). The current study uses this comprehensive denition orecovery or the rst time in examining how ones view o oneselmay dier across stages o an eating disorder.

the Current study

In the current study, we examined aspects o the sel in relation toeating disorder recovery. We ocused on measures o sel-esteem,sel-ecacy, sel-directedness, and the imposter phenomenon and

dened recovery using physical, behavioral, and psychologicalindices in order to produce both ully recovered and partially re-covered groups. We hypothesized that individuals ully recoveredrom an eating disorder would have higher sel-esteem, higher sel-ecacy, higher sel-directedness, and lower levels o the imposterphenomenon than either individuals partially recovered rom aneating disorder or those meeting criteria or an eating disorder.

methods

PARTICIPANTS AND RECRUITMENT

Attempts were made to contact all current and ormer emale eatingdisorder patients (ages 16 and older) seen at the University o Mis-souri Pediatric and Adolescent Specialty Clinic (N= 273) between1996 and 2007, the year o data collection. This clinic is a prima-ry care and reerral clinic specializing in the care o children andadolescents (ages 10-25 years) that has physicians with expertisein eating disorders. O the 273 eating disorder patients, 96 (35.2%)

were successully contacted and recruited. Fity-ve (20.1%) o the273 were contacted but did not participate due to other time com-mitments or lack o interest. O the remaining patients, our (1.5%)were deceased and 118 patients (43.2%) could not be contacted dueto absent or incorrect mailing addresses or inability to make phonecontact. These rates are airly comparable to those o other studiesdoing a rst ollow-up o eating disorder patients over a range oabout 10 years (Reas, Williamson, Martin, & Zucker, 2000). In sum,o the 151 eating disorder patients we were able to contact, 63.6%


8/27


participated. Healthy controls were recruited rom two sources:

the clinic rom which the eating disorder patients were recruited(n = 17) and the university campus (n = 50). Eligible controls wereemales ages 16 and older with no current or past eating disordersymptoms.

For all participants recruited rom the clinic (ormer and currenteating disorder patients and healthy controls), current contact in-ormation was sought via patient records, public records such aswhitepages.com, court records, and marriage records, and paidtracking searches. Eligible participants were mailed a cover letterthat described the study and included the lead researchers phone

number or requesting more inormation or expressing interest. Upto two mailings were sent out and i there was no response at thatpoint, then attempts were made to contact the eligible participantvia phone to describe the study and solicit participation. Recruit-ment or the healthy controls outside o the clinic occurred throughfiers and introductory psychology courses. Those who respondedto the fier, which noted inclusion criteria o no current or past eat-ing disorder symptoms, called the lead researcher at which pointthe inclusion criteria were reiterated and the study was described.Those who were recruited rom introductory psychology classeswere contacted via phone i they met inclusion criteria based onscreening measures administered at the start o the semester.

STUDY PROCEDURES

Ater providing written consent, all participants rst completed aset o questionnaires and then, on a separate date, an interview. (Forparticipants under the age o 18, we obtained written assent rom

the minor and written consent rom a parent.) For the majority othe participants, the time between questionnaires and interview waswithin one week. Most participants completed the questionnaires(71.2%) and interview (82.9%) in person. Those who lived too araway to travel to the study site completed the questionnaires viamail and did a phone interview. Interviews were privately conduct-ed by one o three extensively trained individuals who participatedin over 50 hours o training videos, role plays, and discussions aboutinterviewing. Participants were provided nancial remuneration a-


9/27


ter completing the interview, except or the controls rom psychol-

ogy classes who received course credit. All aspects o this study wereapproved by the universitys institutional review board.

MEASURES USED FOR DEFINING EATING DISORDERSTATUS GROUPS

Structured Clinical Interview or DSM-IV, Patient Edition (SCID;First, Spitzer, Gibbon, & Williams, 1995).The SCID is a widely usedsemi-structured interview that has well-established reliability andvalidity (Segal, Hersen, & Van Hasselt, 1994). Axis I modules oreating disorders, AN, BN, and eating disorder not otherwise speci-ed (EDNOS), were administered to determine lietime and currentdiagnoses. EDNOS cases were those that met the SCIDs guidelinesor EDNOS (e.g., subthreshold AN, subthreshold BN, regular use oinappropriate compensatory behaviors, and binge eating disorder).A random subset (approximately 5%) o the audiotaped interviewswas examined or inter-rater reliability, yielding absolute agreementbetween the lead author and the other two interviewers or currentAN, BN, and EDNOS.

Eating Disorders Longitudinal Interval Follow-Up Evaluation Inter-view (LIFE EAT II; Herzog et al., 1993). We used portions o the LIFEEAT II asking about the presence o binge eating, vomiting, laxativeuse, and asting over the past three months.

Eating Disorder Examination-Questionnaire (EDE-Q; Fairburn &Beglin, 1994). The EDE-Q assesses disordered eating thoughts andbehaviors over the past our weeks, yielding our subscales: Re-straint (attempts to restrict ood intake), Eating Concern (eelingguilty and concerned about eating), Weight Concern (dissatisac-

tion with and overvaluation o weight), and Shape Concern (dis-satisaction with and overvaluation o shape). The EDE-Q is one othe most commonly used measures o disordered eating attitudesand behaviors in clinical and community populations (Anderson &Williamson, 2002) and its subscales have good internal consistency(alphas o .78-.93; Luce & Crowther, 1999) and convergent validity(Fairburn & Beglin, 1994; Grilo, Masheb, & Wilson, 2001). This ques-tionnaire is derived rom the Eating Disorder Examination (EDE) in-terview (Fairburn & Cooper, 1993), which was used in prior work in


10/27


dening recovery rom AN (Couturier & Lock, 2006). In the current

study, coecient alpha was .85 or Restraint, .85 or Eating Concern,.88 or Weight Concern, and .94 or Shape Concern.

Body Mass Index. Weight and height were measured ater the in-terview or all in-person interviewees and used to compute bodymass index (BMI). For those who completed the interview over thephone, we used sel-reported height and weight in the BMI compu-tations.

MEASURES OF SELF-CONCEPT

Sel-Esteem. Global sel-esteem was assessed via the RosenbergSel-Esteem Scale (RSES; Rosenberg, 1965). The RSES is a reliableand well-validated 10-item scale and is the most widely used mea-sure o overall sel-esteem (Heatherton & Wyland, 2003). In the cur-rent sample, a 5-point response scale was used and was .94.

Domain-specic sel-esteem was obtained via the Adult Sel-Per-ception Prole (ASPP; Messer & Harter, 1986), which is a 50-itemmeasure tapping into individuals sel-perception in a variety oconceptually labeled domains: sociability, job competence, nurtur-

ance, athletic abilities, physical appearance, adequate provider, mo-rality, household management, intimate relationships, intelligence,and sense o humor. Participants rated each statement on a 4-pointscale with higher scores refecting more positive evaluations. TheASPP has adequate psychometrics, including internal consistenciesranging rom .65 to .90 in a sample o women (Messer & Harter,1986), and in the current sample, s or the subscales ranged rom.74 to .90. Since the ASPP ocuses on adult competencies and hasbeen validated specically on adults, analyses involving this mea-sure were limited to those ages 18 and older (n = 145).

Sel-Efcacy. Sel-ecacy was assessed via the General Sel-Eca-cy subscale (GSES) o the Sel-Ecacy Scale developed by Sherer etal. (1982). The GSES uses a 5-point response scale and is composedo 17 items not tied to specic situations or behavior (e.g., WhenI make plans, I am certain I can make them work). The GSES hasadequate reliability ( = .86) and validity (Bosscher & Smit, 1998;Sherer et al., 1982). In the current sample, was .90.

Sel-Directedness. Sel-directedness was assessed with the Sel-

Directedness subscale o the short orm o the Temperament and


11/27


Character Inventory (TCI-R-140; Cloninger, 1999), which contains

20 items rated using a 5-point response scale. Sel-directedness isa dimension o character capturing the traits o purposeulness to-ward goals, personal responsibility, and sel-acceptance (Cloningeret al., 1993). The Sel-Directedness subscale o the TCI-R-140 hasgood internal consistency in prior work ( = .88), and in the currentsample, was .91.

Imposter Phenomenon. The imposter phenomenon was assessedwith the 20-item Clance Imposter Phenomenon Scale (CIPS; Clance,1985) which uses a 5-point response scale. The CIPS has evidenceor good reliability ( = .96) and convergent validity, and it distin-guishes between individuals independently identied as impostersand nonimposters (Chrisman, Pieper, Clance, Holland, & Glickau-Hughes, 1995; Holmes, Kertay, Adamson, Holland, & Clance, 1993).In the current sample, was .94.

ANALYTIC PROCEDURE

Given that the sel-concept variables o sel-esteem, sel-ecacy,

sel-directedness, and the imposter phenomenon were highly cor-related, with correlations ranging rom -.57 or sel-ecacy and theimposter phenomenon to .82 or sel-esteem and sel-directedness(all correlations signicant at p < .001), a multivariate analysis ovariance (MANOVA) was rst perormed or these our global sel-concept variables. A signicant multivariate eect was ollowed upwith univariate analyses and then pair-wise comparisons (Tukeystest) in order to see what was driving the eect. A separate MANOVAwas perormed or the domain-specic sel-esteem subscales o theASPP, with ollow-up ANOVAs and pairwise comparisons (Tukeys

test). In the univariate analyses, the assumption o homogeneity ovariance across groups was not met or sel-esteem and sel-ecacyrom the global constructs (i.e., signicant Levenes tests o equalityo variances) and also was not met or sociability, intimate relation-ships, morality, and adequate provider rom the domain-specicconstructs. While ANOVA is robust to violations o the homoge-neity o variance assumption, this is not so when sample sizes areunequal, as is the case in this study. Thus, or the analyses involvingthe variables where the homogeneity assumption was violated, we


12/27


report Welchs F-test and use the Games-Howell test or pairwise

comparisons, as recommended by Field (2009).1

results

ATTRITION ANALYSES

In order to examine whether the individuals who participated di-ered rom those who did not, we used clinic chart data to makecomparisons. The participants were not signicantly dierent romthe nonparticipants (those with whom contact was never made orwho declined to participate but agreed to let us use limited chartinormation) in terms o current age, age at rst clinic visit, BMIat rst clinic visit, eating disorder diagnoses, and number o yearssince last clinic visit. These ndings provide condence that, at leaston these measures, study participants were representative o thelarger eating disorder patient population at this clinic.

DESCRIPTIVE STATISTICS

The participants were categorized into one o our groups: healthycontrols, active eating disorder, partially recovered eating disorder,and ully recovered eating disorder. Healthy controls (n = 67) hadno history o an eating disorder, and active eating disorder cases (n= 53) had a current eating disorder diagnosis (AN, BN, or EDNOS).Individuals who met criteria or an eating disorder in the past butnot currently were divided into two groups by combining a weightindex, an assessment o eating disorder behaviors, and scores on theEDE-Q subscales to derive greater nuances in recovery.

The ully recovered group (n = 20) comprised women without acurrent eating disorder who had a BMI o at least 18.5, reported

1. For the MANOVAs, Boxs M Test o equality o variance-covariance matriceswas signicant, suggesting that the assumption that these matrices were equal acrossgroups had been violated, with this violation being more meaningul given the unequalsample sizes. Rather than randomly removing data rom the larger groups to equalizethe sample sizes (which would have reduced power), we proceeded to ollow-up thendings o signicant multivariate eects with the univariate analyses given that thepattern and magnitude o the means suggested robust group dierences and given thator the univariate tests we were able to report an alternative F-statistic (i.e., Welchs F)

in cases where the univariate homogeneity o variance assumption was violated.


13/27


no binge eating, purging, or asting in the prior three months, andscored within 1 SD o community norms on each o the EDE-Q sub-scales. The partially recovered group (n = 15) met all o the criteriaor ull recovery except or the psychological recovery criteria (i.e.,at least one EDE-Q subscale was greater than 1 SD o age-matchednorms).2 The BMI cut-o o 18.5 was chosen since the range o18.5-24.9 is considered normal weight by the World Health Organi-zation (Bjorntorp, 2002). The norms used or determining ully andpartially recovered status were the age-banded community norms

reported by Mond, Hay, Rodgers, and Owen (2006), so that or agiven eating disorder patient, her EDE-Q scores were compared tothose o women o a similar age. We chose to use 1 SD rom normsrather than 2 SD rom norms because 2 SD rom norms on the EDE-Q subscales oten included scores o 4 or higher, which are consid-ered clinically signicant (Mond et al., 2006).

Table 1 includes descriptive statistics o the our groups in terms oage, ethnicity, and socio-economic status. The groups did not dierin terms o ethnicity and socio-economic status, but did dier in age,F(3, 151)=15.44,p < .001, with healthy controls signicantly youngerthan the eating disorder groups. Controlling or age did not changethe pattern o signicance, so results without age as a covariate arepresented or parsimony. The ully recovered, partially recovered,

2. Eight o the 96 current and ormer eating disorder patients did not meet criteriaor a current eating disorder or either denition o recovery (i.e., partial recovery ophysical and behavioral recovery, but not psychological recovery, or ull recovery ophysical, behavioral, and psychological recovery). These were primarily individualswho had reported some (though minimal) binge eating or purging, typically onceor twice in the past three months. When analyses were run including these eightindividuals in the partial recovery group, the same pattern o results emerged as

presented in this work using the stricter denition o partial recovery.

taBle 1. dc sc

ac ed(n = 53)

prc ed

(n = 15)

frc ed

(n = 20)

hC(n = 67)

Ag 23.18 as(4.39)

23.53 as(5.80)

24.55 as(4.89)

19.46 as(1.88)

enc 92.5%Cacasan

93.3%Cacasan

95.0%Cacasan

89.6%Cacasan

Sc-ecnmcSas

16.68 as(2.79)

16.63 as(2.52)

16.60 as(3.14)

16.52 as(2.65)

Note. eD = ang dsd. Mans and sandad dvans a pd ag and sc-cnmcsas, w sc-cnmc sas fcng pans gs v dcan aand.


14/27


and active eating disorder groups did not dier in their pattern o

lietime eating disorder diagnoses (e.g., no signicant dierencesin the percentage with a lietime diagnosis o AN), the number oyears since the emergence o the eating disorder symptoms, or ageor BMI at start o treatment. O the active eating disorder group,17% currently had AN, 6% had BN, and 77% had EDNOS.

GROUP DIFFERENCES IN SELF-CONCEPT

Both MANOVAs were signicant (see Table 2) and were ollowed

up with univariate analyses to determine which sel-concept con-structs were driving the eects and to determine the pattern o di-erences between groups. The results presented below refect thendings rom the ollow-up ANOVAs and pairwise comparisons,with the statistics presented in Table 2.

The healthy controls and ully recovered group did not dier sig-nicantly in global sel-esteem, sel-ecacy, sel-directedness, orthe imposter phenomenon. Furthermore, inspection o the meansor these two groups revealed remarkably similar scores on thesemeasures o sel-concept. The ully recovered group was signi-cantly dierent rom the partially recovered group and the activeeating disorder group in terms o having higher sel-esteem, highersel-directedness, and lower imposter phenomenon scores. In termso sel-ecacy, the ully recovered group had signicantly highersel-ecacy than the active eating disorder group, but did not diersignicantly rom the partially recovered group. Lastly, it is notablethat the partially recovered group and the active eating disordergroup looked similar in terms o global sel-esteem, sel-ecacy,sel-directedness, and the imposter phenomenon. Eect sizes or the

univariate analyses or these global sel-concept constructs rangedrom 2 = .21 to 2 = .48; or most constructs, eating disorder statusaccounted or almost hal o the variance in the global sel-conceptmeasures.

Analyses involving domain-specic sel-esteem ound signicantgroup dierences in sociability, intimate relationships, morality, in-telligence, sense o humor, physical appearance, job competence,and being an adequate provider. O note, these dierences weresignicant even when using a conservativep-value based on a Bon-erroni correction or the 11 comparisons ollowing the signicant

multivariate eect (.05/11 = .005). The ully recovered group was


15/27


comparable to the healthy controls and better o than the active eat-

ing disorder group in all o these domains o dierence. When com-pared with the partially recovered group, the ully recovered grouphad signicantly higher sel-esteem only in the area o appearance.The partially recovered group was not signicantly dierent romthe active eating disorder group, although there was a marginallysignicant dierence (p = .06) or intimate relationships. Eect sizesor the signicant domain-specic sel-esteem ndings ranged rom2 = .17 to 2 = .46; or most constructs, eating disorder status ac-counted or about one-ourth o the variance in the domain-specicsel-esteem subscales.3

disCussion

With ull eating disorder recovery dened comprehensively, usingpsychological, behavioral, and physical indices, very clear sel-con-cept dierences emerged between the ully recovered group andthose with an active eating disorder or those partially recoveredrom an eating disorder, with our hypotheses largely conrmed.Given the interrelatedness o these constructs, it may be that these

ndings refect the operation o a single underlying actor o sel-concept. In that sense, the ndings indicate that positive sel-con-cept is present among those recovered rom an eating disorder onphysical, behavioral, and psychological levels, to the degree that

3. While some would argue that controlling or depression in an ANCOVAramework in examining the relation between eating disorder status and sel-esteemis warranted, there is strong support against using depression as a covariate in thiscontext (Miller & Chapman, 2001). In particular, given that depression is generallynot seen as an independent, conounding actor in the eating disorders, but rather as

substantively related to eating disorders, statistical methods cannot remove the eecto depression rom eating disorder symptomatology. Using depression as a covariatein this context would produce a residualized eating disorder grouping that is a much-diminished representation o the construct that the eating disorder grouping measures(Miller & Chapman, 2001, p. 45). That said, when depressive symptomatology wasincluded as a covariate, a similar pattern o results emerged or sel-directedness andthe imposter phenomenon, but there were no longer group dierences in sel-ecacyand the pairwise comparisons or sel-esteem indicated that the healthy controlsand ully recovered groups were no longer signicantly dierent rom the partiallyrecovered group. Whereas there was still a multivariate eect or the domains osel-esteem ater controlling or depression, there were ewer signicant pairwisecomparisons.


16/27

836 BARDONE-CONE ET AL.

TABLE2.

ComparisonoSel-Concep

tAcrossEatingDisorderStatusGroups

ActiveED

Partially

Recovered

Fully

Recovered

Healthy

Controls

Signif

cance

Pair-W

ise

Comparisons

Global

Sel

-Concep

tScales

(Mu

ltivariateEect:F=

11

.37

,p

PRED

,AE

D

Se

l-E

cacy

56

.92(10.80)

63

.33(9

.53)

66

.25(5

.78

)

67

.81(9

.79)

F(3

,151)

=11

.63;

p

AED

FRED>

AED

Se

l-D

irec

tedness

60

.47(12.47)

68

.60(11

.03)

84

.10(7

.00

)

81

.00(10

.63)

F(3

,149)

=42

.13;

p

PRED

,AED

FRED>

PRED

,AE

D

Im

poster

Phenomenon

68

.98(11.71)

63

.27(16

.56)

47

.21(12.56)

45

.55(11

.77)

F(3

,146)

=38

.74;

p

AED

Se

nseo

Humor

2.8

2(.72

)

2.8

4(.71)

3.2

9(.54)

3.5

1(.54)

F(3

,141)

=13

.17;

p

PRED

,AED

FRED>

AED

Ph

ysical

Appearance

1.8

5(.62

)

2.2

3(.79)

3.0

9(.49)

3.1

3(.67)

F(3

,141)

=40

.10;

p

PRED

,AED

FRED>

PRED

,AE

D

Athletic

Abilities

2.2

4(.79

)

2.2

2(1

.05)

2.2

6(.60)

2.6

1(.81)

F(3

,141)=

2.4

8;

p=.0

64;

2=

.05

JobCompetence

2.6

5(.62

)

2.9

0(.61)

3.1

8(.46)

3.4

1(.57)

F(3

,141)

=16

.40;

p

PRED

,AED

FRED>

AED

House

ho

ldManage-

men

t

2.9

3(.75

)

2.9

0(.80)

3.1

2(.56)

3.1

8(.61)

F(3

,141)=

1.6

0;

p=.1

93;

2=

.03

Adequate

Provi

der

2.4

1(.80

)

2.6

8(.74)

3.2

5(.37)

3.3

1(.57)

F(3

,141)

=16

.77;

p

PRED

,AED

FRED>

AED

Note.

ED=eatingdisorder;AED=activeea

tingdisorder;PRED=partiallyrecoveredeatingdisorder;FRED=ullyrecovered

eatingdisorder;HC=healthycontrols.

For

allmeasures,higherscoresrefecthigherlevelsotheconstruct.

Forglobalsel-estee

m,sel-e

cacy,sociability,intimaterelationships,morality,andadequateprovider,

WelchsF-statisticisreportedandpairwisec

omparisonswereperormedusingtheG

ames-Howelltest,asrecommendedinc

aseswherethehomogeneityovariance

as-

sum

ptionhasbeenviolatedinthecontexto

unequalsamplesizes(Field,

2009).For

allothersel-conceptvariables,thehom

ogeneityovarianceassumptionwasmetand

the

standardF-statisticandresultsromTuke

ystestsarepresented.

Pairwisecomparisonslistedweresignicantatleastatp

Aspects of Self Concept

Documents

Transcript of Aspects of Self Concept