Vol. 16: e37-e44, March 2011
ORIGINAL RESEARCH PAPER
Perfectionism as a mediator between perceived criticism and eating disorders S. Sassaroli1,2, M. Apparigliato1, S. Bertelli3, L. Boccalari3, F. Fiore1,2, C. Lamela4, S. Scarone3,5, and G.M. Ruggiero1,2 1“Studi Cognitivi”, Post-graduate Cognitive Psychotherapy School, Milano, 2“Psicoterapia Cognitiva e Ricerca”, Post-graduate Cognitive Psychotherapy School, Milano, 3Eating Disorders Unit, Ospedale San Paolo, Milano, 4“Scuola Cognitiva Firenze” Post-graduate Cognitive Psychotherapy School, Firenze, 5Psychiatric Branch, Department of Medicine, Surgery and Dentistry, University of Milan, Milano, Italy
ABSTRACT. AIM: In this work we aimed to test the hypothesis that perfectionism plays a third variable role in the psychological process leading from perceived criticism to eating disorders (ED). METHOD: Forty-nine individuals with ED and 49 controls completed the Concern over Mistakes subscale of the Multidimensional Perfectionism Scale, the Perceived Criticism Inventory, and the Drive for Thinness, Bulimia, and Body Dissatisfaction subscales of the Eating Disorders Inventory. Mediational and moderational models were tested. RESULTS: Analyses revealed that perfectionism mediates between perceived criticism and drive for thinness. Results for bulimia and body dissatisfaction were controversial. Moderational models were rejected. DISCUSSION: Results suggest that restrictive dieting is related to a process in which perceived criticism is the initial factor and perfectionism is an intervening mediator. (Eating Weight Disord. 16: e37-e44, 2011). ©2011, Editrice Kurtis
INTRODUCTION
Key words: Anorexia, bulimia, criticism, mediation, perfectionism, regression analysis. Correspondence to: Sandra Sassaroli, “Studi Cognitivi”, Foro Buonaparte 57, 20121, Milano, Italy. E-mail: grupporicerca@studicognitivi. net Received: June 11, 2010 Accepted: September 28, 2010
Perfectionism is considered one of the most important psychological factors related to the development of eating disorders (ED) (1-6). In turn, perfectionism interacts strongly with an intense proneness to perceive criticism from significant others (711). This association between perceived criticism and perfectionism is also possible in individuals with ED (3, 12). However, while this association is a wellestablished finding, it is not clear how these two factors would interact with each other. In order to explore this interaction, this work used a regression methodology, which assessed whether a variable played a third variable (mediational or moderational) role between the independent and the dependent variables. While our first hypothesis was that perceived criticism was the independent variable and perfectionism played the third variable role, it is important to stress that the methodology also permitted us to test rival hypotheses. Of course, measures of ED were the dependent variables in both cases. Clinical and developmental reasons support the hypothesis that perceived criticism
is the independent variable placed at the beginning of the process. Perceived criticism should precede perfectionism in a developmental process which leads to ED. In fact, episodes of perceived criticism related to harsh parenting styles are probably experienced during childhood or adolescence and should temporally precede the emergence of perfectionism in an adult individual. Kawamura et al. (7) have reported evidence in support of this hypothesis. Data from the clinical field of mood and anxiety disorders also support the association between open verbal abuse from parents during childhood and negative cognitive self-schemata in adulthood (13-15). However, the inverse scenario is less intuitive but nonetheless possible: perfectionism could precede the perception of criticism. In fact, present-time perceived criticism could be influenced not only by past events, but also by current experiences and personality. In addition, empirical results regarding the real behavior of parents, as remembered by individuals, are controversial. For example, Brewin et al. (16) found that self-criticism is related to perceived parental criticism, but not to parents’ own reports of criticism. It could be possible that e37
S. Sassaroli, M. Apparigliato, S. Bertelli, et al.
a perfectionist personality would increase vulnerability to perceive criticism and to overrate past episodes of criticism. For this reason, it is correct to test both criticism and perfectionism as independent or third variables. In addition, we also ignore whether the process leading from either criticism or perfectionism to the development of eating disorders is mediational or moderational. Actually, from a clinical viewpoint testing whether the effect of a variable is mediational or moderational is not only a sophisticated but a blank statistical game. In fact, a mediator is a variable embedded within the psychological process, given that it either is influenced by or influences the other variables. Thus, the mediator seems to measure a psychological state. On the other hand, a moderator operates outside the relationship between independent and dependent variables. In other words, the moderator is a contextual factor which influences the relationship but it is not influenced by the relationship between independent and dependent variables. We could say that it describes a frame or even a trait. Given that perfectionism and perceived criticism can be conceived either as a thought, a belief related to an on-the-spot cognitive state or as a long-term personality trait, thus, we tested both the mediation and the moderation model.
METHODS Participants Clinical participants were recruited from a population of 80 individuals requesting clinical help from the Eating Disorders Unit of the Ospedale San Paolo of Milano (Italy). Fortynine Italian individuals with ED participated in the study (47 females, 2 males, with a mean age of 32.36±14.89 yrs), while 8 received no diagnosis of ED and 23 abandoned the process before engaging in the clinical treatment. The recruitment was carried out during the initial assessment phase of treatment for ED. All the clinical participants were treated as outpatients and received pharmacological treatment and cognitive psychotherapy (1 session per week for 3 months at least). We classified the participants as individuals with anorexia nervosa (N=13), bulimia nervosa (N=12), or an eating disorder not otherwise specified (EDNOS) (N=24) according to the Structured Clinical Interview axis I (SCID-I) for the IV version of the Diagnostic and Statistical Manual of Mental Disorders (DSM). The SCID was administered as part of the routine diagnostic procedure of the Unit. e38
Eating Weight Disord., Vol. 16: N. 1 - 2011
Forty-two of the participants were high school graduates and 7 were university graduates. Eight individuals were married. All the non-student participants were full- or part-time employed during the six months prior to the study. Clinical participants were informed about the procedures and aims of the study and knew that the results of their assessments would be discussed during the initial sessions of treatment. Each clinical participant provided written consent. The Ethical Committee of the “Studi Cognitivi” cognitive psychotherapy school of Milano and of the S. Paolo Hospital of Milano approved the study. Forty-nine Italian individuals without ED were recruited as controls. The recruitment was carried out in order to obtain a control sample that was matched accurately with the clinical sample for gender distribution, age range, level of education, marital status, and employment. This meant that the control sample included 47 females, 2 males, had a mean age of 34.12±11.43 yrs; 42 of the controls were high school graduates, and 7 were university graduates; 8 individuals were married; and all the non-student participants were full- or parttime employed during the six months prior to the study. Regarding the place of recruitment of the control sample, 32 individuals were recruited among students of the “Studi Cognitivi” postgraduate program at the Cognitive Psychotherapy School of Milano, Italy, and 17 individuals were recruited among the population of employees of a firm. Demographic data, height, and weight were collected on a paper questionnaire. Individuals in the control group were not remunerated for their participation. Psychologists trained in cognitive therapy (four years of training, according to the criteria of the Italian Ministero dell’Istruzione, dell’Università e della Ricerca) collected data and interviewed individuals of the control sample in order to detect possible diagnostic criteria of ED. None of them revealed a symptom of an ED. Given that perfectionism and perceived criticism are known to be present in anxiety and mood disorders, we excluded individuals with these disorders from the control group. Given that the majority of individuals with ED have a mood or an anxiety disorder in addition to ED (17, 18), we did not exclude individuals with ED combined with an anxiety or mood disorder in the clinical sample. Measuring instruments The Italian version of SCID-I (19, 20) is a structured interview based on DSM criteria for
Criticism and perfectionism in eating disorders
diagnosis of axis I psychiatric disorders. The Drive for Thinness, Bulimia, and Body Dissatisfaction subscales of the Eating Disorders Inventory version 3 (EDI-3, 21) were chosen to measure ED. Drive for Thinness assesses a cardinal feature of ED: Excessive concern with dieting and fear of weight gain, and it is useful in screening for ED (21-24). The definition of the subscale derives from clinical conceptualizations of Bruch (25) and Russell (26). However, Drive for Thinness does not exhaust the complete spectrum of the psychopathology of ED. In the description of the EDI, the Bulimia and the Body Dissatisfaction subscales are also considered important psychopathological measures of ED (21, 24). The Bulimia subscale assesses the tendency to worry about and engage in uncontrollable bingeing, which is one of the defining features of bulimia and bingeing-purging subtype of anorexia (21, 24). Body dissatisfaction, although present in non-clinical young women, is also a central feature of ED when present in extreme degrees (21, 24). In sum, the author of the EDI describes the ensemble of the abovementioned three subscales together as an exhaustive set of dependent variables for measuring the symptomatology of ED (21, 24). The Perceived Criticism Inventory (PCI; 27) is an Italian instrument designed to evaluate many dimensions of parental criticism. In this study, we used the amount of perceived criticism subscale (hereafter PC) suffered from parents during childhood and adolescence. Psychometric analyses show that PC has adequate reliability (Cronbach’s alphas higher than 0.7) and concurrent validity with another instrument measuring criticism: The Parental Criticism subscale of the Multidimensional Perfectionism Scale (Pearson’s r higher than 0.7) (27). The Multidimensional Perfectionism Scale (MPS) is a 35-item self-reporting questionnaire based on Frost’s theoretical model of perfectionism (28). In this study, we used the Concern over Mistakes subscale (hereafter CM) of MPS as a measure of perfectionism. Psychometric studies show that the MPS and the subscale CM have adequate reliability (Cronbach’s alphas higher than 0.7) (28, 29). G.M.R. translated the EDI and the MPS into Italian. The Italian version of the scales and the questionnaires were then back-translated into English by a native English speaker who was not familiar with the questionnaire. The author of the EDI compared the original version and the back-translated version and did not identify any differences (D.M. Garner, January 15, 1997, personal communication to B. VanAntwerp). Although G.M.R. translated
the previous version of the EDI (the so-called EDI-2), the version-2 translation is still valid because the new version (EDI-3) does not contain any new items that were not included in the EDI-2. In particular, the Drive for Thinness scale was not changed (which means that it is totally identical in EDI-2 and EDI-3), while the only changes of the Bulimia and Body Dissatisfaction subscales were that each of them included an additional item, which was not actually new, but had shifted from another subscale of the EDI-2: The Interoceptive Awareness subscale (21). One of the authors of MPS compared the original version and the back-translated version of the MPS and did not note any meaningful differences (R. Frost, December 29, 2004, personal communication). The English and Italian translations of the EDI and the MPS were also reviewed and approved by a native bilingual Italian and English speaker who grew up in an Italian family that had emigrated to the USA and now lives in Italy and works as an English language teacher (R. Denaro, June 23, 2004, personal communication). Reliability based on internal consistency was confirmed with a Cronbach’s alpha coefficient >0.7 for each instrument and each sample (0.85 for Drive for Thinness of the EDI in the clinical sample; 0.87 for Drive for Thinness of the EDI in the non-clinical sample; 0.71 for Bulimia of the EDI in the clinical sample; 0.73 for Bulimia of the EDI in the non-clinical sample; 0.83 for Body Dissatisfaction of the EDI in the clinical sample; 0.84 for Body Dissatisfaction of the EDI in the non-clinical sample; 0.81 for the CM scale of the MPS in the clinical sample; 0.88 for the CM scale of the MPS in the non-clinical sample; 0.83 for the PC scale of the PCI in the clinical sample; 0.83 for the PC scale of the PCI in the non-clinical sample). Statistical procedures Mediation moderation analysis We implemented the regression analyses in the clinical and non-clinical samples separately and used centered predictor values in order to reduce multicollinearity (30). In addition, given that the sample size was insufficient for statistical inference and that from power analysis it resulted that we needed a sample of 500 individuals at least, we implemented analyses using a number of bootstrap resample equal to 1000. In order to test which variable acted as a third variable and whether the third variable acted as mediator or moderator, we implemented two sets of product-term regression analyses, according to the guidelines devised by Baron and Kenny (31). The hypothesis Eating Weight Disord., Vol. 16: N. 1 - 2011
e39
S. Sassaroli, M. Apparigliato, S. Bertelli, et al.
assumed in this work was that perfectionism was the third variable between perceived criticism (independent variable) and symptoms of ED (dependent variables). Of course, the method also permitted different results: first, that perfectionism was the independent variable and perceived criticism was the third variable. In both the first and the second regression analyses, a measure of ED was used as a dependent variable. In the first regression analysis, the independent variable (perceived criticism) was entered in the first step; the independent variable (perceived criticism) and the third variable (perfectionism) were then entered jointly in the second step; and the product of the independent variable (perceived criticism) and the third variable (perfectionism) was entered in the third step. The second regression analysis was different from the first only in step one, where the third variable was entered in place of the independent variable. The three-step procedure of the productterm regression analysis allowed for four different possible results through which a third variable can influence the relationship between the dependent and the independent variable: 1) In the first step of the second regression, a significant effect of the third variable means a direct effect of the third variable on the dependent variable (31); 2) In the second step of the first regression, a significant effect of the third variable and a non-significant effect of the independent variable suggest that the third variable could play a complete mediator role and intervene in the process between the other two variables (31, 32); in addition, a significant effect of both the third variable and the independent variable is still compatible with a partial mediational model; 3) In the second step of the second regression, a significant effect of the independent variable and a non-significant effect of the third variable suggest that the third variable indirectly influences the dependent variable via the independent variable. If confirmed, this result would correspond to a rival model in which the independent variable is the complete mediator that mediates the impact of the third variable on the dependent variable (32). In addition, in this case, a significant effect of both the third variable and the independent variable is still compatible with a partial mediational model; 4) In the third step, a significant effect of the product of the independent variable and the third variable is indicative of an interaction effect, which is a good approximation of a e40
Eating Weight Disord., Vol. 16: N. 1 - 2011
moderating effect in which the moderator influences the strength of the relation between the independent and the dependent variable (31, 32). The rationale of implementing the interaction analysis as a third step was that it was desirable to test first that the moderator variable was not correlated with both the independent and the dependent variable (31). The above described procedure missed the regression of the independent variable on the mediator. For this reason, we implemented this last regression and completed a rigorous four steps mediation analysis as described by Baron and Kenny (31) to further test the mediational model suggested in the previous analyses.
RESULTS Preliminary analyses All of the tests gave evidence that the distribution of residuals met the requirements for normality (Kolmogorov-Smirnov, ShapiroWilk, Normal Q-Q plot, Detrented Normal Q-Q plot). Neither the histogram nor the probability-probability plot indicated that the assumption of normality of residuals was violated. Descriptive analyses Results confirmed that on all the considered variables, the patients with ED showed significantly higher scores than individuals from the control sample: The Drive for Thinness, Bulimia, and Body Dissatisfaction subscales of EDI, the CM subscale of MPS and the PC scale of PCI (Table 1). Results testing the hypotheses Three steps regression analyses Table 2 shows the product-term regression analyses for patients with ED with perfectionism as third variable and perceived criticism as independent variable. Given that the CM scale had a significant effect and the perceived criticism scale did not have a significant effect on Drive for Thinness on the second step of the regression, this result suggests that perfectionism plays a perfect mediating role between perceived criticism and Drive for Thinness. On the other hand, the reverse result was derived on the other two subscales of the EDI (i.e., the CM scale did not have a significant effect and the perceived criticism scale had a significant effect on both Bulimia and Body Dissatisfaction on the second step of the regression), which suggested that, actually, perceived criticism mediates the impact of perfectionism on bulimia and body dissatisfaction of the EDI. Finally, the
Criticism and perfectionism in eating disorders
TABLE 1 Descriptive statistics and ANOVA differences.
Drive for Thinness of EDI
Bulimia of EDI
Body Dissatisfaction of EDI
Concern over Mistakes of MPS
Perceived Criticism of PCI
N
Mean
SD
Patients with ED
49
12.88
6.57
Non-clinical individuals
49
1.63
2.95
Patients with ED
49
7.45
6.46
Non-clinical individuals
49
0.22
0.69
Patients with ED
49
15.08
7.31
Nonclinical individuals
49
8.82
7.15
Patients with ED
49
27.55
8.48
Non-clinical individuals
49
21.45
6.39
Patients with ED
49
33.04
9.45
Non-clinical individuals
49
28.08
7.52
third step of regression analysis revealed that the product CM * PC was not significantly correlated with Bulimia or Body Dissatisfaction of the EDI and showed an inverse significant correlation with Drive for Thinness of the EDI (Table 2). This inverse correlation is not easy to interpret, because it implies that the interaction between perceived criticism and perfectionism would predict a decrease of drive for thinness. We will discuss how to interpret this counterintuitive result in the following section (see section 4, Discussion). In short, three steps regression analyses implemented in individuals with an ED suggested that: a) perfectionism has a mediating effect between perceived criticism (PC) and drive for thinness; b) perceived criticism has a mediating effect between MP and bulimia/body dissatisfaction; c) the interaction of MP and PC showed a controversial result of difficult interpretation: an inverse correlation (opposite of expected) with drive for thinness. Table 3 shows the product-term regression analyses for non-clinical individuals with perfectionism as a third variable and perceived criticism as an independent variable. Even though five of the p-values were significant, according to the rules of the three-step procedure the results did not confirm any possible mediation or moderation model. Four-steps mediation analysis Rigorous mediation analysis was implemented using the four steps methodology of Baron and Kenny (31) and confirmed the mediating role of CM between PC and Drive for Thinness (Fig. 1). On the other hand, the four steps mediation analysis did not confirm the model fore-
F
Sig. (2-tailed)
10.936
p<0.001
7.779
p<0.001
4.289
p<0.001
4.021
p<0.001
2.875
p<0.01
TABLE 2 Three steps product-term regression analyses in patients with ED (N=49). Step
Dependent variable: Drive for Thinness
Beta
t
p-value
1
Concern over Mistakes (CM)
2
Perceived Criticism (PC)
0.388
2.883
<0.01
-0.007
-0.043
NS
3
Product CM * PC
-0.331
-2.267
<0.05
Step
Dependent variable: Drive for Thinness
Beta
t
p-value
1 2 3
Perceived Criticism (PC) Concern over Mistakes (CM) Product CM * PC
0.297 0.320 -0.331
2.129 2.023 -2.267
<0.05 <0.05 <0.05
Step
Dependent variable: Bulimia
Beta
t
p-value
1 2 3
Concern over Mistakes (CM) Perceived Criticism (PC) Product CM * PC
0.291 0.408 -0.213
2.088 2.648 -1.453
<0.05 <0.05 NS
Step
Dependent variable: Bulimia
Beta
t
p-value
1 2 3
Perceived Criticism (PC) Concern over Mistakes (CM) Product CM * PC
0.449 0.408 -0.213
3.443 2.648 -1.453
<0.01 <0.05 NS
Step
Dependent variable: Body Dissatisfaction
Beta
t
p-value
1 2 3
Concern over Mistakes (CM) Perceived Criticism (PC) Product CM * PC
0.328 0.339 -0.146
2.379 2.118 -0.971
<0.05 <0.05 NS
Step
Dependent variable: Body Dissatisfaction
Beta
t
p-value
1 2 3
Perceived Criticism (PC) Concern over Mistakes (CM) Product CM * PC
0.411 0.156 -0.146
3.093 0.996 -0.971
<0.01 NS NS
NS: not significant.
Eating Weight Disord., Vol. 16: N. 1 - 2011
e41
S. Sassaroli, M. Apparigliato, S. Bertelli, et al.
TABLE 3 Three steps product-term regression analyses in non-clinical individuals (N=49). Step
Dependent variable: Drive for Thinness
1 2 3
Concern over Mistakes (CM) Perceived Criticism (PC) Product CM * PC
Step
Dependent variable: Drive for Thinness
1 2 3
Perceived Criticism (PC) Concern over Mistakes (CM) Product CM * PC
Step
Dependent variable: Bulimia
1 2 3
Concern over Mistakes (CM) Perceived Criticism (PC) Product CM * PC
Step
Dependent variable: Bulimia
1 2 3
Perceived Criticism (PC) Concern over Mistakes (CM) Product CM * PC
Step
Dependent variable: Body Dissatisfaction
1 2 3
Concern over Mistakes (CM) Perceived Criticism (PC) Product CM * PC
Step
Dependent variable: Body Dissatisfaction
1 2 3
Perceived Criticism (PC) Concern over Mistakes (CM) Product CM * PC
Perceived Criticism (Independent variable)
Beta
t
p-value
0.232 0.213 0.211
1.636 1.466 0.992
NS NS NS
a
Beta
t
p-value
Perceived Criticism (Independent variable)
0.295 0.299 0.211
2.113 2.902 0.992
<0.05 NS NS
Beta
t
p-value
0.181 0.328 0.374
1.262 2.309 1.849
NS <0.05 NS
Beta
t
p-value
0.353 0.095 0.374
2.590 0.667 1.849
<0.05 NS NS
Beta
t
p-value
0.119 0.269 -0.010
0.818 2.554 -0.102
NS <0.05 NS
Beta
t
p-value
0.219 0.227 -0.010
1.537 2.158 -0.102
NS <0.05 NS
c
Drive for Thinness (Dependent variable)
Concern over Mistakes (Mediator variable) b
c’
Drive for Thinness (Dependent variable)
a path (direct effect of independent on mediator): coeff. = 0.47; S.E. = 0.11; t = 4.19; p<0.001 b path (effect of mediator on dependent controlling for the independent): coeff. = 0.24; S.E. = 0.12; t = 2.02; p<0.05 c path (direct effect of independent on dependent): coeff. = 0.21; S.E. = 0.10; t = 2.13; p<0.05 c’ path (effect of independent on dependent controlling for the mediator): coeff. = 0.09; S.E. = 0.11; t = 0.82; NS.
FIGURE 1 Mediational analysis from Perceived Criticism to Drive for Thinness in patients with ED.
by the CM scale of MPS) and perceived criticism is plausibly mediational (Table 2 and Fig. 1) with respect to a reliable measure of presence of an ED: the subscale Drive for Thinness of the EDI (21, 24). To our knowledge, these results are the first confirmation that the two abovementioned factors are interwoven with each other in the psychopathology of ED. In fact, until now, previ-
NS: not significant.
Concern over Mistakes (Independent variable)
seeing perfectionism as an independent variable, perceived criticism as mediator, or either bulimia or body dissatisfaction as dependent variables (Figs. 2 and 3).
c
Perceived Criticism (Mediator variable) a
DISCUSSION While the importance of the role of perfectionism in the psychopathology of ED is a wellestablished finding, the role of criticism is less supported from an empirical viewpoint. This work helps to focus attention of the scientific community on the variable of perceived criticism and enlightens the nature of the relationship between perfectionism and perceived criticism in the cognitive process leading to ED. The regression and mediation analyses of this study seem to support the idea that the relationship between perfectionism (as measured e42
Eating Weight Disord., Vol. 16: N. 1 - 2011
Bulimia (Dependent variable)
Concern over Mistakes (Independent variable)
b
c’
Bulimia (Dependent variable)
a path (direct effect of independent on mediator): coeff. = 0.61; S.E. = 0.15; t = 4.08; p<0.001 b path (effect of mediator on dependent controlling for the independent): coeff. = 0.18; S.E. = 0.10; t = 1.77; NS c path (direct effect of independent on dependent): coeff. = 0.22; S.E. = 0.11; t = 2.09; p<0.05 c’ path (effect of independent on dependent controlling for the mediator): coeff. = 0.11; S.E. = 0.12; t = 0.93; NS
FIGURE 2 Mediational analysis from Maladaptive Perfectionism to Bulimia in patients with ED.
Criticism and perfectionism in eating disorders
Concern over Mistakes (Independent variable)
c
Body Dissatisfaction (Dependent variable)
Perceived Criticism (Mediator variable) a Concern over Mistakes (Independent variable)
b
c’
Body Dissatisfaction (Dependent variable)
a path (direct effect of independent on mediator): coeff. = 0.61; S.E. = 0.15; t = 4.08; p<0.001 b path (effect of mediator on dependent controlling for the independent): coeff. = 0.21; S.E. = 0.11; t = 1.91; NS c path (direct effect of independent on dependent): coeff. = 0.28; S.E. = 0.12; t = 2.38; p<0.05 c’ path (effect of independent on dependent controlling for the mediator): coeff. = 0.15; S.E. = 0.13; t = 1.12; NS
FIGURE 3 Mediational analysis from Maladaptive Perfectionism to Body Dissatisfaction in patients with ED.
ous studies had shown the connection between perceived criticism and perfectionism in disorders other than ED (7, 9, 11, 33-35). From these results it might be cautiously inferred that perceived criticism cognitively precedes perfectionism in the psychological process that leads to the highly motivated decision to obsessively control weight and fat, in the mind of individuals affected by ED. Thus, we could imagine that perfectionism is a sort of reaction to painful experiences of perceived criticism, which in turn would generate the ED when the person decides to focus his or her perfectionism on the narrow domain of body aspect, weight, and fat. From a cognitive viewpoint, we could infer that severely criticized individuals may learn to develop a perfectionistic attitude, which helps them to cope with the emotional suffering generated by criticism. On the other hand, the results of the non-clinical control sample (Table 3) show some significant relationship but do not reveal any complex relationship involving both perfectionism and perceived criticism. This supports the hypothesis that the interaction between perceived criticism and perfectionism is a psychopathological process, absent in non-clinical individuals. In conclusion, the results could be interpreted as suggesting that perceived criticism is the relational factor facilitating the intergenerational transmission of perfectionism in individuals with ED. This is in agreement with Kawamura et al. (7) and Soenens et al. (36). However, we cannot forget that the inference of causative or develop-
mental relationships from correlations is always questionable and only longitudinal studies could definitively confirm this interpretation. The analyses confirmed the mediation model for the Drive for Thinness subscale only. In fact, the results were different with respect to the variables of Bulimia and Body Dissatisfaction of the EDI (Figs. 2 and 3). The results shown in Table 2 suggest that perceived criticism is the mediator intervening between perfectionism and measures of ED. The clinical inference would be that, in the case of body dissatisfaction and bulimia, a condition of subjective vulnerability to criticism would depend on a preexisting perfectionistic attitude. However, the mediational analyses did not confirm this model (Figs. 2 and 3). Thus, the results for bulimia and body dissatisfaction are controversial and need further research. The clinical implications of these results are that, when treating drive for thinness, the therapist may presume that perfectionism is the cognitive mediator from criticism to restrictive behaviors. In this case, the intervention should be aimed at assessing the sufferance related to the conflicting parental relationships and helping the patient to recognize that they would use perfectionism to cope with the emotional pain generated by criticizing relationships. This intervention could be particularly useful when the patient does not recognize or, even worse, denies the influence of the suffered criticism. On the other hand, when treating bulimic behaviors and body dissatisfaction, the therapist may presume that perfectionism is the initial psychological attitude and criticism is the intervening mediator. In this case, the therapist should encourage the patient to recognize their vulnerability to criticism as a consequence of a propensity to perfectionism. The limitations of the study include the small size of the sample, the non-longitudinal design, and the result (the significant inverse correlation between the product of perceived criticism and perfectionism and drive for thinness). This significant inverse correlation is very difficult to interpret from a clinical viewpoint, because it would mean that that the interaction between two risk factors for ED (like perfectionism and criticism) would play a protective role against ED, which we fear is a clinical absurd (Table 2). At the present moment, we are inclined to interpret this result as simply a statistical artifact, probably due to the small size of the sample.
REFERENCES 1. Fairburn CG, Cooper Z, Shafran R. Cognitive behavEating Weight Disord., Vol. 16: N. 1 - 2011
e43
S. Sassaroli, M. Apparigliato, S. Bertelli, et al.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
e44
iour therapy for eating disorders: a “transdiagnostic” theory and treatment. Behav Res Ther 2003; 41: 509-28. Halmi KA, Sunday SR, Strober M, et al. Perfectionism in anorexia nervosa: Variation by clinical subtype, obsessionality, and pathological eating behavior. Am J Psychiatry 2000; 157: 1799-805. Hewitt PL, Flett GL, Ediger E. Perfectionism traits and perfectionistic self-presentation in eating disorder attitudes, characteristics, and symptoms. J Pers Soc Psychol 1991; 60: 456-70. McLaren L, Gauvin L, White D. The role of perfectionism and excessive commitment to exercise in explaining dietary restraint: Replication and extension. Int J Eat Disord 2001; 29: 307-13. Sassaroli S, Romero L, Ruggiero GM, et al. Perfectionism in depression, obsessive compulsive disorder and eating disorders. Behav Res Ther 2008; 46: 757-65. Vohs KD, Bardone AM, Joiner TE Jr, et al. Perfectionism, perceived weight status, and selfesteem interact to predict bulimic symptoms: A model of bulimic symptom development. J Abnorm Psychol 1999; 108: 695-700. Kawamura KY, Frost RO, Harmatz MG. The relationship of perceived parenting styles to perfectionism. Pers Indiv Differ 2001; 32: 317-27. Frost RO, Lahart CM, Rosenblate R. The development of perfectionism: A study of daughters and their parents. Cogn Ther Res 1991; 15: 469-89. Huprich SK. Depressive personality and its relationship to depressed mood, interpersonal loss, negative parental perceptions, and perfectionism. J Nerv Ment Dis 2003; 191: 73-9. Irons C, Gilbert P, Baldwin MW, et al. Parental recall, attachment relating and self-attacking/self-reassurance: Their relationship with depression. Br J Med Psychol 2006; 45: 297-308. Vieth AZ, Trull TJ. Family patterns of perfectionism: An examination of college students and their parents. J Pers Assess 1999; 72: 49-67. Dunkley DM, Blankstein KR, Masheb RM, et al. Personal standards and evaluative concerns dimensions of “clinical” perfectionism: A reply to Shafran et al. (2002, 2003) and Hewitt et al. (2003). Behav Res Ther 2006; 44: 63-84. Gibb BE. Childhood maltreatment and negative cognitive styles. A quantitative and qualitative review. Clin Psychol Rev 2002; 22. 223-46. Rose DT, Abramson LY. Developmental predictors of depressive cognitive styles: Developmental perspectives on depression. In Cicchetti D, Toth SL (Eds) Developmental Perspectives on Depression. Rochester symposium on developmental psychopathology. Rochester, University of Rochester Press, 1992, pp 32349. Sachs-Ericsson N, Verona E, Joiner T, et al. Parental verbal abuse and the mediating role of self-criticism in adult internalizing disorders. J Affect Disord 2006; 93: 71-8. Brewin CR, Andrews B, Furnham A. Self-critical attitudes and parental criticism in young women. Br J Med Psychol 1996; 69: 69-78. Braun DL, Sunday SR, Halmi KA. Psychiatric comorbidity in patients with eating disorders. Psychol Med 1994; 24: 859-67. Grilo CM, Levy KN, Becker DF, et al. Comorbidity of DSM-III-R axis I and II disorders among female inpa-
Eating Weight Disord., Vol. 16: N. 1 - 2011
19.
20.
21.
22.
23.
24.
25. 26.
27.
28. 29.
30.
31.
32.
33.
34.
35.
36.
tients with eating disorders. Psychiatr Ser 1996; 47: 426-9. First MB, Spitzer RL, Gibbon M, et al. Structured Clinical Interview for DSM-IV Axis I Disorders – Patient Edition (SCID-I/P, version 2.0, 4/97 revision). New York, NY, Biometrics Research Department, New York State Psychiatric Institute, 1996. Mazzi F, Morosini P, De Girolamo G, et al. SCID, Intervista Clinica Strutturata per il DSM-IV [Structured Clinical Interview for DSM-IV]. Firenze, Organizzazioni Speciali, 2000. Garner DM. The Eating Disorder Inventory - 3 (EDI-3) Professional Manual. Odessa, FL, Psychological Assessment Resources, 1991. Abood DA, Black DR. Health education prevention for eating disorders among college female athletes. Am J Health Behav 2000; 24: 209-19. Engström I, Kroon M, Arvidsson CG, et al. Eating disorders in adolescent girls with insulin-dependent diabetes mellitus: A population-based case-control study. Acta Paediatr 1999; 88: 175-80. Garner DM. The Eating Disorder Inventory - 2 (EDI-2) Professional Manual. Odessa, FL, Psychological Assessment Resources, 1991. Bruch H. Eating Disorders: Obesity, Anorexia Nervosa and the Person within. New York, NY, Basic Books, 1973. Russell GFM. Anorexia nervosa: Its identity as an illness and its treatment. In Price JH (Ed) Modern Trends in Psychological Medicine. London, Butterworths, 1970, pp 131-64. Apparigliato M, Ruggiero GM, Sassaroli S. Il Perceived Criticism Inventory (PCI): Un nuovo strumento di valutazione del criticismo genitoriale [The Perceived Criticism Inventory (PCI): A new instrument for the evaluation of parental criticism]. Psic Cogn Comp 2007; 13: 275-300. Frost RO, Marten P, Lahart C, et al. The dimensions of perfectionism. Cognitive Ther Res 1990; 14: 449-68. Parker WD, Adkins KK. A psychometric examination of the multidimensional perfectionism scale. J Psychopathol Behav 1995; 17: 305-21. Aiken LS, West SG. Multiple Regression: Testing and Interpreting Interactions. Thousand Oaks, CA, Sage, 1991. Baron RM, Kenny DA. The moderator-mediator variable distinction in social psychological research: Conceptual, strategic and statistical considerations. J Pers Soc Psychol 1986; 51; 1173-82. Cohen J, Cohen P, West SG, et al. Applied Multiple Regression/Correlation: Analysis for the Behavioral Sciences. New Jersey, Lawrence Erlbaum, 2003. Enns MW, Cox BJ, Clara I. Adaptive and maladaptive perfectionism: Developmental origins and association with depression proneness. Pers Indiv Differ 2003; 33: 921-35. McArdle S, Duda JL. Exploring the etiology of perfectionism and perceptions of self-worth in young athletes. Soc Dev 2008. Available at: [http://www3. interscience.wiley.com/journal/120174519/abstract? CRETRY=1&SRETRY=0]. Rice KG, Lopez FG, Vergara D. Parental/social influences on perfectionism and adult attachment orientations. J Soc Clin Psychol 2005; 24: 580-605. Soenens B, Elliot AJ, Goossens L, et al. The intergenerational transmission of perfectionism: Parents’ psychological control as an intervening variable. J Fam Psychol 2005; 19: 358-66.
