ARTICLE ORIGINAL
Psychological risk factors for colorectal cancer?
Facteurs de risque psychologiques du cancer colorectal ?
S. Kreitler
M.-M. Kreitler A. Len Y. Alkalay F. BarakAbstract: The paper deals with personality correlates of colorectal cancer patients in the framework of the cognitive orientation theory. The cognitive-motivational approach and the construction and testing of a reliable and valid questionnaire for assessing the personality correlates of colorectal cancer are reviewed in the first part. In the second part in a new sample of 230 colorectal cancer patients the themes in the questionnaire are clustered and their structure is tested in a confirmatory factor analysis.
Further, following the expectation that colorectal cancer is gender bound, the differences in the themes and belief types are applied to testing differences between men and women corresponding to the medical differences. Finally the questionnaire was applied to identifying the detected personality correlates in an attenuated form in a sample of Crohn’s disease patients who are known to be at risk for colorectal cancer. Discriminant analysis showed that the
questionnaire provided a highly significant correct identifi- cation of cases of the three groups (165 healthy controls, 90 patients with Crohn’s disease and 230 colorectal cancer patients). The thematic clusters that constitute the persona- lity correlates of colorectal cancer were found to be tendencies for compulsiveness, control of oneself and especially of anger, self effacement, pleasing others, self assertion, distancing oneself from others, keeping regula- tions, and performing to perfection all ones obligations. The three major foci of these tendencies are perfect duty performance, and two contradictory pairs: self effacement versus self assertion, and closeness to others versus distancing from others. The clusters and the contrasts constitute potentially sources of tension. It is suggested that the identified personality correlates be considered as psychological risk factors for colorectal cancer.
Keywords:Colorectal cancer – Crohn’s disease – Cognitive orientation – Risk factors
Re´sume´ : Cet article traite des caracte´ristiques de personna- lite´ des patients atteints de cancer colorectal (CCR) dans le cadre d’une the´orie de la personnalite´ d’orientation cognitive.
Cette approche dite motivationnelle-cognitive, ainsi que la construction d’un questionnaire valide´ et fiable pour mesurer les caracte´ristiques de personnalite´ lie´es au cancer colorectal, sont de´veloppe´es dans la premie`re partie. Dans la seconde partie, a` partir d’un nouvel e´chantillon de 230 patients atteints de CCR, les the`mes du premier questionnaire ont e´te´
range´s en clusters et leur structure a e´te´ valide´e par une nouvelle analyse factorielle. Ensuite, partant du fait que le CCR se manifeste sur un plan me´dical de manie`re diffe´rente en fonction du sexe, les diffe´rences dans les the`mes et les types de croyances, ont e´te´ mises en application pour tester des diffe´rences hommes/femmes correspondant a` ces diffe´- rences me´dicales. Enfin, le questionnaire a e´te´ applique´ pour identifier les meˆmes caracte´ristiques de personnalite´ mises en e´vidence, dans une moindre mesure, chez des patients
Le cancer colorectal
S. Kreitler (*)
Department of Psychology,
Tel Aviv University and Psycho-Oncology Research Center, Sheba Medical Center, Tel Hashomer
E-mail: Krit@netvision.net.il M.-M. Kreitler
Psycho-Oncology Research Center, Sheba Medical Center, Tel Hashomer A. Len
Surgery ward, Tel-Aviv
Sourasky Medical Center, Tel-Aviv Y. Alkalay
Social Sciences faculty, Tel Aviv University F. Barak
Oncology Department, Barzilai Medical Center, Ashkelon, Israel
DOI 10.1007/s11839-008-0094-9
atteints de maladie de Crohn, connus pour eˆtre a` risque de cancer colorectal. L’analyse statistique a montre´ que le questionnaire permettait d’identifier de manie`re significative les caracte´ristiques de personnalite´ de chaque groupe. Le mode`le en structure the´matique du questionnaire a e´te´
pertinent pour discriminer les diffe´rences entre les trois groupes ainsi qu’entre les hommes et les femmes (groupe controˆle, n = 165 ; patients atteints de maladie de Crohn, n = 90 ; patients atteints de cancer colorectal, n = 230). Les clusters the´matiques trouve´s ( ou variables de personnalite´
du CCR), font ressortir les tendances a` la compulsivite´, au controˆle de soi et spe´cialement de la cole`re, l’auto- effacement, le fait de se rendre agre´able aux autres, l’auto- affirmation, la distanciation des autres, l’attachement aux re`gles et le besoin de perfection dans les taˆches. On peut dire que ces tendances se manifestent en trois tendances majeures : L’accomplissement d’un travail parfait et deux tendances contradictoires, l’auto-effacement et l’auto- affirmation d’une part, et d’autre part, la proximite´ des autres et la distanciation. Ces clusters et oppositions peuvent constituer des sources de tension. Il est propose´ que les traits de personnalite´ identifie´s puissent eˆtre conside´re´s comme des facteurs de risque psychologique du CCR.
Mots cle´s :Cancer colorectal – Maladie de Crohn – Orientation cognitive – Personnalite´ facteurs de risques
Introduction
Search for risk factors of colorectal cancer yields in the Medline file over one thousand (n=1618 for August 20, 2008) and in the PsychNet file about one hundred (n=98) scientific papers listing a great variety of factors, such as genetics, age, nutrition, obesity, ethnic background, alcohol consumption, lack of physical exercise, or ingestion of specific medications [30-33], but very few if at all refer to psychological or for that matter psychiatric risk factors. Even an extended search for the relation of colorectal cancer with life events or traumata does not yield any additional information. The tendency to overlook potential psychological risk factors for colorectal cancer cannot be attributed to lack of interest in psycholo- gical aspects of cancer because there are practically hundreds of papers that deal with other psychological aspects in regard to cancer, such as the impact of colorectal cancer on the patients’ quality of life and life style or the psychological effects of genetic testing for colorectal cancer [3,8,28]. The paucity of scientific interest in psychological risk factors of cancer is understandable in view of contradictory or unclear findings in this domain that have been yielded by previous research concerning psychological correlates of different kinds of cancer.
Cancer and personality: Previous research
Most of the research focused on specific psychological variables, sometimes referred to as the Type C personality
type [44], including mainly extraversion, rationality/anti- emotionality, low or high emotional expressivity, repressive- ness, submissiveness, depression, anxiety, hostility and facade of pleasantness [10,12,15]. However, even the apparently better supported tendencies, such as anti-emotionality, repressive- ness or submissiveness were not found in recent replications [1,29,42,43]. In some cases the obverse was found [27] or the psychological trait was shown to be the result of the diagnosis of cancer [24]. Moreover, some of the personality correlates were found to apply to different chronic diseases (e.g., CHD, diabetes) rather than specifically to cancer as had been previously maintained [2,7,41].
There are several reasons for the failure to identify conclusively personality correlates of cancer. First, the subjects selected for participation in these studies vary in medical and psychosocial characteristics that could seriously affect the results, in particular, cancer diagnoses, stages of disease, duration of disease, state (viz. undergoing treatment or not), degree of information about the disease, and quality of life.
Secondly, the psychological variables examined in most studies did not constitute a comprehensive profile, and were not grounded in a theory or methodology that relates them to the disease. Rather, each study focused on one or more variables selected each for a different rationale, representing a variety of psychosocial dimensions, ranging from personality traits through socioeconomic state to life event stresses.
Most importantly, the various studies devoted to personality correlates of cancer did not deal specifically with colorectal cancer. It is however unjustified to assume that personality correlates apply equally to all types of cancer. The obverse seems both more justified empirically and theoretically [19].
Colorectal cancer and personality:
Previous research
Few studies were devoted specifically to personality correlates of colorectal cancer (CRC), despite the fact that it is the third most common form of cancer and the second leading cause of cancer-related death in the Western world [39]. There seem to be three notable studies in this domain. In an interview study with 637 colorectal patients and 714 controls it was found that references to denial and repression of anger and other negative emotions, commitment to social norms resulting in the impression of a nice person, suppression of reactions that may potentially be offensive to others, and conflict avoidance were significantly more frequent in the CRC patients than the healthy controls [26]. The authors consider the mentioned variables as representing potentially risk for CRC and note that these variables were independent of other risk factors, such as diet, beer intake, and family history of colorectal cancer, as well as of demographic factors, including socioe- conomic level, marital status, religion and country of birth.
In another study [16] the subjects were 61 male veterans diagnosed with CRC. The psychological variables were based
on the responses of theses subjects to the MMPI 2 to 41 years earlier. Significant differences were found by a stepwise conditional regression analysis between the patients and controls on the Aggressive Hostility variable (p<0.018).
The third study [11] differs in scope, methodology and theoretical approach from the other two described studies, that focused on variables based either on informal observations and impressions of CRC patients or on data that were available by chance. In contrast to these studies, the study by Figer et al. [11] was rooted in the rationale that the attempt to identify personality correlates of CRC must be based on a coherent and theoretically viable theory that would ensure the detection of a set of personality characteristics, that would be (a) comprehensive (at least on the level on which the study variables are defined), (b) relevant for CRC, and (c) reliable.
It was expected that this approach would yield a set of characteristics of CRC patients that would form the basis for subsequently examining the role of the identified variables for the incidence and progression of the disease and for their integration as adjuvant psychological treatment in the standard treatment of CRC patients.
First Phase of the Studies on the Cognitive Orientation of CRC
The theoretical and methodological characteristics of the third mentioned study [11] will be described in some detail because it constituted the first step of the project whose further step will be presented in the current paper.
The major component of the new approach consisted in applying to the study of the personality correlates of colorectal cancer the theory of cognitive orientation (CO).
The personality characteristics examined in the study were defined in the framework of the cognitive orientation (CO) theory which provides a well-validated and empirically- based approach to the impact of psychological variables on disease-relevant physiological processes [18,19-21,23]. The main tenet of the theory is that specific kinds of cognitive contents affect a person’s psychosomatic state in a way that may potentially enhance the likelihood of specific diseases.
This in no way indicates that the psychological factors cause the disease but merely that they constitute a part of the matrix of background organismic factors which consti- tute the context within which the pathogen (e.g., virus or microbe) operates and is made more or less likely to bring about the disease. Thus, if a person is exposed to a particular pathogen, the likelihood of contracting the disease depends to some extent on whether that person has the psychological makeup that was found to be characteristic for individuals who suffer from the specific disease. The variables that define the relevant psychological makeup can be considered as potentially constituting the psychological risk factors for the specific disease. They may be considered as actual risk factors if empirical studies provide sufficient evidence for
their reliability and independent contribution to predicting the state of the disease.
The CO theory has been applied successfully to a range of medical diagnoses [9,22,25,34]. Its major advantage for the study of psychological factors that play a predisposing role in regard to colorectal cancer is that it provides a methodology for identifying the adequate kind of cognitive contents. In the present context this means that the cognitive contents have the following characteristics: they are relevant for a particular disease; they constitute theoretically a homoge- neous and comprehensive set of factors; they are not under the conscious or voluntary control of the participant; they may be empirically validated; and they are amenable for change in a well-controlled clinical intervention.
The psychological factors identified by the CO theory are cognitive contents, i.e., they express how the individual conceptualizes or experiences external and internal states and objects. The cognitive contents are expressed in the form of statements about some theme that may or may not be veridical, such as ‘Colorectal cancer is a common disease’,
‘I love my neighbors’, ‘People often lie’. These statements are cognitive units that are called ‘beliefs’ although they have nothing to do with faith or veridicality. The beliefs relevant for a disease are characterized by formal and semantic characteristics. Formally the beliefs are of four types: one type refers to the self (e.g., ‘I like spending money freely’,
‘I try not to accept help from anyone’); a second type refers to others and reality (e.g., ‘People who tend to be late are irresponsible’, ‘It is possible to reach perfection in everything one does’); a third type refers to goals and wishes (e.g., ‘I would like to avoid fights at any cost’) and a fourth type refers to rules and standards (e.g., ‘A person should do for oneself only the bare minimum’, ‘One should try to have as much fun in life as possible’).
In terms of semantics, the relevant cognitive contents refer to themes that represent deeper underlying meanings related to the particular disease. These meanings are called
‘themes’ and they are identified by interviewing pretest participants according to a standard procedure. The beliefs referred to themes that were extracted from interviews conducted with pretest participants (15 colon cancer patients and 15 healthy controls) according to a standard procedure [21]. The procedure consists of asking the participants to express the lexical interpersonally-shared meanings of the disease and then repeatedly the personal meanings of their responses. The responses that recur in the last phase of the interview in at least 50% of the patients and in fewer than 25% of the controls were identified as themes for the CO questionnaire of colorectal cancer. In sum, the cognitive contents that represent the psychological factors relevant for a disease may be presented in the form of a matrix whose columns are defined by the four types of beliefs and whose rows are formed by particular themes identified by the interviewing procedure (e.g., concern for cleanliness or for saving money).
The CO questionnaire for colorectal cancer (CO-CLC) was prepared by phrasing beliefs for each theme in each belief type. The final draft of the questionnaire was obtained after the completion of standard psychometric analyses and tests, and included four parts referring to each of the four types of beliefs: about self, about norms, about goals, and general
beliefs (total number of items was 181), and 44 themes. Table 1 presents a list of all themes with examples of items of the four belief types illustrating them. The items were presented in the form of brief statements with four response alternatives, scored 1-4: ‘‘very true’’, ‘‘true’’, ‘‘not true’’ and ‘‘not at all true’’
(in the case of goal beliefs the alternatives were ‘‘want very
Table 1.Complete list of themes and examples of beliefs of the CO questionnaire of colorectal cancer
Themes Examples of Beliefs
1. Promoting peace and quiet in one’s environment I would like to behave so that there would be no tension and fights among the people close to me (Go)
2. Avoid fighting with others Fights lead only to losses (GB)
3. Cleanliness It is impossible to live in an environment that is not perfectly clean (GB)
4. Orderliness It disturbs me a lot if there is around me any kind of disorder (BS)
5. Punctuality I want to have contact only with people who are on time (Go)
6. Avoid spending money I don’t like spending money on things that are not absolutely vital (BS) 7. Control over others I want others always to do exactly what I want and in the
manner I want it done (Go)
8. Controlling events It is important for me to feel that I control the situation down to the smallest details (BS)
9. Control of emotional expression One should not let one’s feelings affect one’s behavior (N)
10. Control of emotional experiencing It is possible to have a wonderful life also without any emotions (GB)
11. Control over one’s body I don’t succumb to physical limitations (BS)
12. Heavy load of responsibility I undertake many commitments and take care of many things which actually I did not have to do (BS)
13. Keeping promises and commitments Keeping promises means doing what is required and on time without considering circumstances & obstacles (GB)
14. Work as the most important thing in life One should consider ones work as the most important thing in life (N) 15. Pressuring oneself for maximum performance I want to demand from myself the maximum (Go)
16. Giving all to others I want to give to others all I can without any limitations or considerations (Go)
17. Self effacement: setting others before oneself Others are always more important than the person herself (GB) 18. Doing for oneself only the bare minimum I do very little for myself (BS)
19. Need for signs of love and affection from others One should not depend on signs of affection from others (N–)
20. Pleasing others It is impossible to behave so that would be pleased with you
and what you do (GB–)
21. Being nice & accommodating Almost everyone I know loves to be in my company (BS) 22. No demands for oneself I would like not to have any demands from life (Go)
23. Refraining from getting help from others I want never and under no circumstances to ask others for help (Go) 24. Need to specify demands and expectations I like to know exactly what is expected of me and what I can
expect from others (BS)
25. Avoid getting orders from others A person ought to be able to get orders from others (N–) 26. Complying with routines, rules and regulations I keep strictly all the socially accepted rules concerning proper
conduct among friends, at work, etc. (BS)
27. Giving up success for other things in life I want to be able to give up success in order to keep other things which are important to me (Go)
28. Avoiding anger One should avoid anger at all cost (N)
29. Avoiding the expression of anger I want to hide my anger deep inside me (Go)
30. Expressing anger in a fast explosive way & be done with it Letting one’s anger burst all at once brings immediate relief (GB) 31. Huge amounts of anger inside oneself One can contain within oneself a huge amount of anger (GB) 32. Being calm externally even if upset internally Even when I am all upset inside me, nothing can be noticed
on the outside (BS)
33. Not sharing any negative emotions and weaknesses I want to share with others everything I experience (Go–) 34. Insists on getting one’s due and rights It irritates me if I do not get precisely what I deserve (BS)
35. Getting everything exposed out in the open It is important to me that everything should always be open and clear, without any secrets or things done behind one’s back (BS)
36. Telling others one’s opinion of them even if they don’t like it I want to state my views even if it will not please others (Go)
37. Having to prove oneself (as in an exam) I feel as if I were constantly standing up for an exam, trying to prove myself (BS) 38. Ignoring problems and avoiding solving them Unpleasant things should simply be overlooked (N)
39. Trying to correct injustice I am upset by each instance of injustice (BS)
40. Setting duty prior to pleasure Entertainment should not be an option for anyone who has not completed all one’s duties (N)
41. No fun in life I like very much pleasure and entertainment of all kinds (BS–)
42. Not letting oneself make any mistakes It is forbidden to make any mistakes in any domain of one’s life (N) 43. Striving for perfection I expect from myself to perform up to perfection in every domain (BS) 44. Keeping constantly active Anyone who is not constantly busy is unhappy (GB)
Note.BS=Beliefs about self, GB=General beliefs (about others and reality), N=Beliefs about rules and norms, Go=Beliefs about goals. A minus sign following any item indicates that the scoring of the item is to be reversed. All themes are phrased in the direction supporting colon cancer.
much’’, ‘‘want’’, ‘‘don’t want’’, ‘‘don’t want at all’’). The reliability was satisfactory for each part separately and for the whole questionnaire in each of the samples.
The CO questionnaire yielded for each participant 44 scores, one for each theme representing the sum of the scores for all responses referring to the theme across belief types, and 4 scores, one for each belief type representing the sum of the scores for all responses referring to the type of belief across all themes.
A factor analysis of the themes in the CO of CLC questionnaire yielded eight factors (presented in descend- ing order of relative importance):The first factor defined by 10 themes represents focusing on attaining peace and quiet in one’s environment through restraint and self- regulation, both positively (by striving for orderliness and peace, ignoring problems, emphasizing work, and controll- ing one’s body, one’s emotional expressions and even one’s ambitions for success) and negatively (by avoiding fights, cutting down on expenses, and renouncing any demands concerning oneself).The secondfactor represents focusing on fulfilling one’s duties and commitments, including concern with cleanliness, punctuality, compliance with all rules, keeping promises and commitments, and pressuring oneself to the extreme for maximum performance and attainment of perfection. The third factor represents suppressing anger; the fourth – toughness in regard to oneself (not letting oneself have fun or commit mistakes);
the fifth– fighting injustice;the sixth– controlling others;
the seventh– avoiding being controlled by others; andthe eighth– pleasing others.
The CO of CLC questionnaire was administered to 106 patients with CLC and 99 healthy controls. Comparing the scores showed that the patients and controls differed significantly on all four belief types, on seven of the eight factors, and in the majority of the single themes as well (88%), including even themes that contradict each other, such as Keeping peace and quiet vs. antagonizing others or self-abasement vs. insisting on one’s rights. In regard to all variables, the patients scored higher, as expected in line with the CO theory. Discriminant analyses showed that on the basis of these psychological variables alone, the participants were classified correctly as patients or healthy in 77% (lowest rate) to 86.3% (highest rate) of the cases.
These percents constitute an improvement of 27% to 36.3%
over the 50% of correct classification expected by chance.
Among the belief types the best predictors were beliefs about self and general beliefs, and among the factors the best predictors were Factor 1(attaining peace and quiet in one’s environment by self-regulation), factor 4 (toughness in regard to oneself) and factor 6 (controlling others). Control analyses showed that the scores of the CO variables do not depend on demographic properties or medical characteristics, such as disease duration, disease site and stages, quality of life and age. Hence, the observed differences in the CO variables between patients and
healthy controls may be considered as characterizing the tested groups of participants per se.
The study showed that there exists a set of psychological variables, identified and defined in line with the CO theory, which characterize colon cancer patients and enable a highly significant differentiation between them and healthy controls.
These variables consist of specific cognitive contents that are related to the disease both theoretically (in terms of their derivation and identification) and empirically (in terms of the findings of this study). The identified psychological variables provide the blueprint for conceptualizing the personality of individuals diagnosed with colon cancer. The major features are manifested in the way these individuals view themselves and reality, including others (self-beliefs and general beliefs). The main characteristic is the striving to maintain peace and quiet around oneself (factor 1). This desired state is promoted both by doing certain things, such as taking care of orderliness, focusing on work, and controlling one’s body and emotional expressions, as well as by refraining from doing certain things, such as avoiding fights and renouncing all demands that concern oneself. The hallmark of this factor is self-restraint, applied in regard to fulfilling one’s own needs, spending money, striving for success, and emotional expression. Self-restraint is further elaborated by focusing on fulfilling to perfection all of one’s duties and commitments (factor 2), in regard to cleanliness and punctuality, living up to the explicit and implicit demands and expectations of others in regard to oneself, keeping one’s promises and commitments, and complying with rules and regulations. These two major foci – attaining peace and quiet around oneself and exercising self-restraint as a means for attaining this objective – allow for integrating also most of the other identified themes and factors, such as suppressing anger, and toughness in treating oneself, or pleasing others. Notably, keeping peace and quiet around oneself and attaining self-restraint, as well as most of the specific identified themes and factors are clearly sources of tension, both singly, each separately, and in clusters that readily denote conflicts. The tension derives both from the difficulty of conforming to the themes as well as of the impossibility of ever doing it right or to a satisfactory degree, e.g. suppressing anger, controlling one’s bodily needs or doing things to perfection. In addition, the findings showed further sources of tension in the patients, dependent on the CO variables – inconsistencies between the view of oneself, of reality, the norms and the goals. It may seem justified to conclude that it may difficult to be a CRC patient not only physically and medically but also psychologically.
In sum, the findings of this study provide support for the hypothesis that there is a set of comprehensive, disease-relevant, reliable, specific psychological varia- bles that may be considered as characterizing the colon cancer patient.
Second phase of the Studies on the CO of CRC
The second phase of the studies on the CO of CRC was designed to clarify three specific issues that may increase the validity and applicability of the CO of CRC and deepen the insight it provides into the psychological dynamics of the personality dispositions of CRC.
The Thematic Structure of the CO of CRC Questionnaire
Objective and Background.The first question we dealt with concerned the interrelations between the various themes constituting the CO of CRC. Table 1 presents the list of the 44 themes that were derived from interviewing CRC patients and served as basis for constructing the CO of CRC. In the previous study [11] a factor analysis of the themes yielded 8 factors, that were used for predicting group classification of the participants and for interpreting the personality dynamics of CRC patients. However, the factor solution did not seem to us fully satisfactory for the following reasons. First, all eight factors together accounted only for 66.88% of the variance.
Second, only the first two factors were strong enough according to the Scree test. Third, factors 4-8 were defined each by one or two themes only. Fourth, the factors allowed for different forms of integration in interpreting their interrelations. The preferred interpretation offered in the paper consisted in organizing all the factors around the two major related foci of ‘attaining peace and quiet around oneself’
and ‘exercising self-restraint’. However, as noted in the paper itself, alternate organizations are possible that focus on other anchors, such as control of oneself and others, or maintaining the proper order.
Therefore in the present study we adopted a different approach, based on testing a measurement model. For this purpose it was first necessary to organize the 44 themes into meaningful clusters. The task was performed by inviting five focus groups of CRC patients to cluster the themes in line with their meanings. Each group was asked to come up with 5 to 10 clusters of themes and provide a name for each cluster. The groups worked independently.
Each group included 3 to 6 members and met for 1-2 hours.
The clusterings by the five groups turned out to be highly similar. All patients met together for one session for resolving various inconsistencies. The final clustering into eight clusters is presented in Table 2. Notably, the clustering resembles the results of the exploratory factor analysis but differs from it in major respects.
The eight clusters were considered as an hypothetical set that needs to be tested. For this purpose the set was tested as a measurement model by means of a confirma- tory factor analysis.
Method.The participants were 230 CRC patients (101 men and 129 women), whose mean age was 61.54 yrs, range of disease duration .5 to 7.3 yrs, with different diagnoses (47% cancer of the rectum, 53% colon cancer), and in different disease stages (I 13.04%, II 38.7%, III 30%, IV 18.26%). All subjects responded to the CO of CRC questionnaire, used in the previous study [11].
Results.The results are presented in Table 3 (first part) and Fig. 1. The indices presented in Table 3 (CMIN, GFI, NFI, NNFI, CFI and RAMSEA) confirm that the model whereby all eight clusters represent one latent factor is fully supported by the data. Further, the standardized regression weights of all eight clusters are highly significant which indicates that all eight clusters are related significantly
Table 2.Definition of the theme clusters
Cluster Defining themes Suggested label
1 3. Cleanliness; 4. Orderliness; 5. Punctuality; 6. Avoid spending money Compulsiveness 2 2. Avoid fighting with others; 7. Control over others; 8. Controlling events; 9. Control of
emotional expression; 10. Control of emotional experiencing; 11. Control over one’s body;
28. Avoiding anger; 29. Avoiding the expression of anger; 32. Being calm externally even if upset internally; 33. Not sharing any negative emotions and weaknesses; 41. No fun in life
Control, especially over ones’
emotions and anger
3 17. Self effacement: setting others before oneself; 18. Doing for oneself only the
bare minimum; 22. No demands for oneself; 27. Giving up success for other things in life;
38. Ignoring problems and avoiding solving them
Self effacement
4 1. Promoting peace and quiet in one’s environment; 16. Giving all to others; 19. Need for signs of love and affection from others; 20. Pleasing others; 21. Being nice and accommodating
Pleasing others 5 34. Insists on getting one’s due and rights; 35. Getting everything exposed out in the open;
36. Telling others one’s opinion of them even if they don’t like it; 37. Having to prove oneself (as in an exam)
Asserting oneself
6 23. Refraining from getting help from others; 25. Avoid getting orders from others;
30. Expressing anger in a fast explosive way & be done with it; 31. Huge amounts of anger inside oneself
Distancing oneself from others
7 24. Need to specify demands and expectations; 26. Complying with routines, rules and regulations; 39. Trying to correct injustice
Keeping regulations 8 12. Heavy load of responsibility; 13. Keeping promises and commitments; 14. Work as the most
important thing in life; 15. Pressuring oneself for maximum performance; 40. Setting duty prior to pleasure; 42. Not letting oneself make any mistakes; 43. Striving for perfection;
44. Keeping constantly active
Performing commitments to perfection
Note.The numbers in the middle column indicate the numbering of the themes (see Table 1).
to the latent variable. In other words, all eight clusters belong to the same conceptual sphere, which is the motivational understructure of CRC. The squared multiple correlations show that clusters 8 and 1 (which represent compulsiveness and perfectionism, respectively) have the relatively strongest relations to the latent variable and clusters 5 and 6 (which represent self-assertion and distancing from others, respecti- vely) have the weakest relations to the latent variable. The strongest clusters share the theoretically related emphases on striving for perfection and normative behavior, whereas the weakest clusters share the theoretically related emphases on withdrawing from others, mainly due to the pent up anger one feels. Further, the model shows five interrelations between the error terms of the clusters that are not accounted for by the measurement model.
In terms of the thematic contents of the eight clusters, the dynamic image of the CRC personality that emerges is focused around three central issues. The first is the concern with behaving according to the requirements. This concern is evident in the four clusters (Nos. 1,2, 7, and 8) that refer to compulsivenss in regard to the classical issues of cleanliness, orderliness, punctuality and saving money; controlling oneself especially in regard to anger; complying with all regulations; and performing all one’s duties to perfection.
The second issue is the conflict in regard to the self, evident in clusters 3 and 5. This contrasting pair of clusters indicates tendencies for both self-effacement and self-assertion.
Although each as such is warranted, upholding both at the same time is impossible and constitutes a constant source of tension. Similar conclusions attend the third issue which
Table 3.The measurement model in the three diagnostic groups and the two gender groups
Group Clusters Standardized Reg. Weights Squared Multiple Correlations Linked Error Terms Cor. of error terms
CRC Cluster 1 .883*** .780*** e6 – e7 –.219***
Cluster 2 .864*** .747 *** e5 – e7 .295***
Cluster 3 .759*** .576 *** e2 – e6 .250***
Cluster 4 .729*** .531*** e2 – e5 –.123.054
Cluster 5 .719*** .517*** e2 – e3 .301***
Cluster 6 .678*** .460***
Cluster 7 .763*** .582***
Cluster 8 .915*** .836***
CMIN=28.450 DF=15 CMIN/DF=1.897* GFI=.970 NFI=.980 NNFI=.982 CFI=.990 RAMSEA=.063
CRC: Cluster 1 .876*** .767*** e5 – e7 .445***
Women Cluster 2 .843*** .711*** e2 3 e6 .436***
Cluster 3 .803*** .644*** e6 3 e7 –.146*
Cluster 4 .722*** .522***
Cluster 5 .633*** .401 ***
Cluster 6 .686*** .471***
Cluster 7 .690*** .476***
Cluster 8 .905*** .819***
CMIN=24.147 DF=17 CMIN/DF=1.420 GFI=.955 NFI=.968 NNFI=.984 CFI=.990 RAMSEA=.057
CRC Cluster 1 .883*** .780*** e2 3 e3 .522***
Men Cluster 2 .878*** .771*** e5 3 e8 .321*
Cluster 3 .732*** .536*** e6 3 e8 –.446***
Cluster 4 .752*** .565*** e6 3 e7 –.430***
Cluster 5 .776*** .603***
Cluster 6 .731*** .534***
Cluster 7 .838*** .703***
Cluster 8 .916*** .840***
CMIN=17.476 DF=16 CMIN/DF=1.092 GFI=.961 NFI=.975 NNFI=.996 CFI=.998 RAMSEA=.030
Healthy Cluster 1 .795*** .633*** e5 – e7 .197*
Cluster 2 .847*** .717*** e4 3 e5 .309***
Cluster 3 .757*** .573*** e3 3 e8 –.620***
Cluster 4 .603*** .364*** e3 – e5 –.389***
Cluster 5 .528*** .279*** e2 – e6 .398***
Cluster 6 .646*** .418*** e2 – e5 –.283***
Cluster 7 .655*** .429***
Cluster 8 .893*** .797***
CMIN=27.761 DF=14 CMIN/DF=1.983* GFI=.959 NFI=.964 TLI=.963 CFI=.982 RAMSEA=.077
Crohn Cluster 1 .707*** .499*** e2 3 e6 .352**
Cluster 2 .776*** .602*** e5 – e7 .261*
Cluster 3 .529*** .280*** e1 3 e4 –.372**
Cluster 4 .663*** .440*** e6 3 e7 –.147
Cluster 5 .536*** .287***
Cluster 6 .606*** .367***
Cluster 7 .633*** .401***
Cluster 8 .906*** .821***
CMIN=21.755 DF=16 CMIN/DF=1.360 GFI=.945 NFI=.935 NNFI=.967 CFI=.981 RAMSEA=.064
*p<.05; **p<.01; ***p.
Note:For the definition of the clusters see Table 2.
represents conflict in regard to the relations to others, evident in clusters 4 and 6. This contrasting pair of clusters indicates tendencies for both getting close to others, giving to them, being nice and accommodating and at the same time distancing oneself from others and refraining from contacts with others. Again, a case could be made for each form of interrelating separately, but upholding both at the same time is impossible and constitutes a constant source of tension.
The CO of CRC: The Gender Perspective
Objective and Background. The second question we dealt with concerned possible gender differences in the CO of CRC. The special interest in this question was evoked by the emerging evidence in the medical literature that possibly men and women may differ in regard to CRC.
Thus, Gao et al. [13] who studied trends of CRC in Canada reported that there is a lower incidence of CRC in women than in men. Although overall incidence and mortality rates for CRC are decreasing, but remain
substantially lower for women. For men the highest subsite was rectal cancer, which was third highest for women, whereas right colon cancer was highest for women.
Male/female ratios for incidence and surgeries were highest for distal cancer and are increasing with time.
Over the years, CRC show an increase in proximal subsites, but a greater decrease for the more distal subsites in females. Kotake et al. [17] who studied trends of CRC in relation to age, gender, site, and survival during the previous 20 years in Japan reported that there was a 2.5-fold increase in the number of cases, with consistent male predominance confined to the distal colon and the rectum. For females there was preferential localization of the disease in the proximal colon and a definite survival benefit. The reduced hazard ratio for women was the largest for proximal CRC in Stage I or II. Further, it seems that men and women are affected differently by known risk factors for CRC. Thus, obesity and large body size increases particularly exceeding 10 kg/m2 may increase the risk for CRC, but particularly among women [33].
Moradi et al. [31], who studied the effects of physical inactivity on the risk for CRC, found that the physical activity-related variation in risk among women is greatest in the proximal and middle parts of the colon, whereas the corresponding peak in men seems to be more distal parts of the colon. Further, there is growing evidence about genetic gender differences in CRC. For example, one study [37] found gender-related survival differences associated with the HER-1 R497K polymorphism. Women who had one variant (the HER-1 497 Arg/Arg variant) had better survival when compared with the other variants (Lys/Lys and/or Lys/Arg). But in men the opposite was true. Park et al. [35] found that the frequency of the –2578CA + AA (A allele-bearing) genotype had a protective effect for colon cancer in women, as well as the–2578CA + AA genotype associated with reduced risk in patients with proximal colon cancer, which is more frequent in women. A further example refers to the epsilon 2/epsilon 3 genotype that was shown to have an increased risk of colon cancer. But the association between the genotype and cancer was found to be significant only in men and only in men the proportion of patients with more advanced tumors (Dukes’ C&D) was significantly increased among those with the ApoE epsilon 2/epsilon 3 genotype [46].
clusters
clstr1
e1
e1
1
1
clstr2
e2
e2
a
1
clstr3
e3
e3
b
1
clstr4
e4
c 1 e4
clstr5
e5
e5
d
1
clstr6
e6
e6
e
1
clstr7
e7
e7
f
1
clstr8
e8
e8
g
1
cv5 cv4
cv2cv3 cv1
Fig. 1.The 8 cluster model in the group of the subjects with colorectal cancer
In sum, the thematic clusters define a dynamic understructure for CRC which is wrought with psychological hardships and pitfall. It is very difficult striving for perfect behavior, but it is no less difficult to try to be both self-effacing and self-asserting, and to be promote both closeness to others and withdrawal from others.
In sum, findings are accumulating which show that CRC tends to differ between men and women in incidence, site, effect of risk factors and genetic understructure.
Therefore, it seems justified to check whether there are corresponding gender differences in the CO of CRC. In the previous study [11] there were 10 significant differences in CO variables between the genders but only one (i.e. the theme of punctuality passed the Bonferroni criteria).
Hence we expected clearer findings with a larger sample, as in the present study.
Method. The same sample and the same instrument (CO of CRC questionnaire) were used as described above (seeThematic Structure of the CO of CRC Questionnaire).
The number of men was 101 and the number of women was 129. There were no significant differences between the gender groups in age, diagnoses, disease duration
or disease stages. The gender groups were compared in terms of means and in terms of the adequacy of the measurement model based on the eight thematic clusters.
Results.Table 4 presents the significant mean differences between the genders. These include differences in 13 themes, one belief type and three clusters. Only three themes (nos. 7, 8, 14), norm beliefs and two cluster (Nos. 2, 8) pass the Bonferroni criteria (for the significance level of .05, p<.001 for 44 themes, p=.0125 for four belief types and p<.01 for eight clusters), so that the other mean differences need to be interpreted with caution. Notably, norm beliefs play the most important role in identifying correctly group mem- bership (see next section The CO of CRC: In the Service of Identifying Risk for CRC). However, the major point that needs to be emphasized is that in all cases of the presented results, the means of the women were lower than those of the men. This observation clearly indicates that there may be a tendency in the female sample for lower disease severity. This tentative conclusion seems to correspond to some of the findings that emerge in the medical research of gender differences in CRC.
Table 3 (second and third sections) and Figs 2 and 3 present the results of testing the measurement model in terms of confirmatory factor analyses in the samples of women and men. Comparing the models shows that the fit is good in both samples and all eight clusters are related significantly to the latent factor in each of the two samples. Further, in both
clusters
clstr1
e1
e1
1
1
clstr2
e2
e2
a
1
clstr3
e3
e3
b
1
clstr4
e4
c 1 e4
clstr5
e5
e5
d
1
clstr6
e6
e6
e
1
clstr7
e7
e7
f
1
clstr8
e8
e8
g
1
Fig. 2.The 8 cluster model in the group of women subjects with colorectal cancer
clusters
clstr1
e1
e1
1
1
clstr2
e2
e2
a
1
clstr3
e3
e3
b
1
clstr4
e4
c 1 e4
clstr5
e5
e5
d
1
clstr6
e6
e6
e
1
clstr7
e7
e7
f
1
clstr8
e8
e8
g
1
Fig. 3.The 8 cluster model in the group of men with colorectal cancer Table 4. Significant differences between men and women in variables
based on the CO of CRC questionnaire
Variablea Mean and SD t-test
Men Women
Theme 5 14.44 [2.99] 13.25 [2.71] 3.18**
Theme 6 10.74 [2.64] 9.98 [2.45] 2.25*
Theme 7 9.05 [2.73] 7.69 [2.28] 4.08***
Theme 8 13.63 [2.84] 12.38 [2.70] 3.38***
Theme 10 8.72 [2.45] 7.74 [2.18] 3.19**
Theme 13 12.32 [2.27] 11.57 [2.03] 2.61**
Theme 14 10.76 [2.86] 9.31 [2.87] 3.81***
Theme 26 12.85 [2.12] 12.23 [2.06] 2.25*
Theme 29 9.85 [2.13] 9.07 [2.12] 2.77**
Theme 33 9.62 [2.12] 8.79 [1.87] 3.13**
Theme 37 10.85 [2.06] 10.36 [1.66] 1.98*
Theme 42 8.85 [1.72] 8.35 [1.60] 2.26*
Theme 43 10.67 [1.99] 10.05 [2.28] 2.16*
Norm Beliefs 2.64 [ .41] 2.47 [ .36] 3.42***
Cluster 1 11.84 [2.39] 11.15 [2.13] 2.33*
Cluster 2 11.06 [1.57] 10.37 [1.45] 3.42***
Cluster 8 11.59 [1.71] 10.87 [1.66] 3.24**
*p<.05; **p<.01; ***p<.001.
aFor the definition of the themes see Table 1, and for the definition of the clusters see Table 2.
samples clusters 8 and 1 (which represent compulsiveness and perfectionism, respectively) have the strongest relations to the latent variable. In the overall fit and in regard to the two strongest relations the results of the two gender samples correspond to those of the sample as a whole (see Table 3).
Yet, there are also differences between the results for men and women. First, the weakest relations to the latent variable are evidenced in the male sample by clusters 6 and 3 (representing distancing from others and self-effacement, respectively), and in the female sample by clusters 6 and 5 (representing distancing from others and self assertion, respectively).
Notably, clusters 3 (self-effacement) and 5 (self-assertion) are contradictory. This suggests that men and women tend to resolve in different ways the previously mentioned conflicts in regard to oneself and to relating to others. It may be assumed that men would tend to emphasize more self-assertion and women self-effacement. Secondly, the models in men and women differ also in regard to the number and kind of interrelated error terms that are not accounted for by the model, for example, in women the error terms for control and distancing from others (e2 and e6), whereas in men the error terms for control and self-effacement (e2 and e3).
The CO of CRC : In the Service of Identifying Risk for CRC
Objective and Background. One of the most important applications of the CO of CRC questionnaire is to identify risk for CRC. In order to test this potentiality of the questionnaire it was decided to apply it in regard to a group that does not have CRC but is known to be at risk for CRC. We chose for this purpose the group of patients with Crohn’s disease.
Crohn’s disease is an ongoing disorder that causes inflammation of the gastrointestinal tract, affecting espe- cially the lower part of the small intestine. The exact cause of the disease is unknown. It is however likely that both genetic and environmental factors are involved in the pathogenesis of Crohn’s disease. The disease has been linked to genes (e.g., mutation in CARD 15, ATG16L1) [6,36], autoimmunity [4], Ashkenazi Jewish background, living in northern countries especially in urban areas in the Western word, nutrition (e.g., high consumption of fatty, refined foods and low consumption of potassium, magne- sium etc.), and smoking [5,38].
It has been often shown that patients with Crohn’s disease have increased cancer rates for both the small and large bowel [40]. More recent studies confirmed the same
observation [47]. For example, in patients with Crohn’s disease, there was a significantly high risk of cancer of the colon (1.64) [14]. As compared with the baseline popula- tion of patients without Crohn’s disease, the relative risk of Crohn patients to get small bowel and of colorectal cancers was 28.4 and 2.4, respectively [45].
Since patients with Crohn’s disease do not have CRC but are only at risk for CRC, we expected that patients with Crohn’s disease would score on the CO of CRC question- naire lower than patients with CRC but higher than healthy controls. Accordingly, the hypothesis was that the CO of CRC questionnaire would enable a correct identification of the patients at risk for CRC but would differentiate between them and both the CRC patients and the healthy subjects.
Method.There were 485 participants in the study. These included 230 patients with CRC (see description of sample above in the sectionThe CO of CRC: The Gender Perspective), 90 patients with a confirmed diagnosis of Crohn’s disease and 165 healthy controls. The per cents of men and women in the three groups were 33.3% vs 66.7%, 50.6% vs 49.4%, and 43.9% vs 56.1%, respectively. All participants were adminis- tered the CO of CRC questionnaire described above (First Phase of the Studies on the Cognitive Orientation of CRC).
The three groups of subjects were compared in terms of the means of their responses to the questionnaire and the fit of the measurement model based on the eight thematic clusters in each group.
Results.Table 5 presents the oneway analyses of variance for the scores of the four belief types and the eight thematic clusters with control over age (because of age differences between the groups). The results show that in all variables without exception the differences between the three diagnostic groups were highly significant (the main effect for the group factor was significant at the p<.0001 level). Further, the differences in the scores of the three groups were always in the expected direction, with the scores of the CRC patients being highest, the scores of the healthy group being lowest, and the scores of the patients with Crohn’s disease being in the middle between the cancer patients and the healthy controls.
Further, in view of the significant differences between the groups, discriminant analysis was applied for examining whether it would be possible to identify correctly the group membership of the subjects on the basis of the CO of CRC questionnaire alone. Table 6 presents the findings for the four belief types. It shows that all four belief types have a significant contribution to the group identification, as predicted in line with the CO theory. The predictors formed two functions with significant contributions to the differen- tiation between the groups. The first and main one was defined primarily by general beliefs and beliefs about self (namely, the beliefs referring to facts), and the second was defined mainly by norm beliefs and goal beliefs (see the discriminant function coefficients). The first discriminated mainly between the groups of the healthy and CRC, the In sum, the CO of CRC questionnaire fits both
samples of men and women and in addition is sufficiently sensitive to detect differences between the genders that warrant further study insofar as the seem to correspond to gender differences detected on the medical level.
second mainly between the patients with Crohn’s disease and the CRC patients (see the distances between the functions at group centroids). The overall per cent of correctly identified cases (57.9%) deviated with high significance from that expected by chance (33.3%).
Similarly, Table 7 presents the findings for the eight thematic clusters. It shows that all eight clusters have a significant contribution to the group identification. The predictors formed two functions, with significant contribu- tions to the differentiation between the groups. In line with the discriminant function coefficients, the first and major function was defined mainly by clusters 1 (compulsiveness) and 3 (self-effacement), and the second was defined mainly by clusters 8 (perfectionism) and 2 (control). The first function discriminated mainly between the groups of the healthy and
CRC (as in the case of the four beliefs types as predictors), the second mainly between the healthy and the patients with Crohn’s disease (whereas the second function in the case of beliefs discriminated mainly between the two groups of patients) (see the distances between the functions at group centroids). The overall per cent of correctly identified cases (54.8%) deviated with high significance from that expected by chance (33.3%).
A final test of the CO approach in this context was done by applying the measurement model of the eight thematic clusters in each of the three groups separately. This test was expected to show whether the CO of CRC question- naire is assessing the same conceptual structure in all three groups despite the manifested and expected differences in level of the scores (see Table 5). Table 3 (first, fourth and
Table 5.Means, SDs and oneway analyses of variance the four types of beliefs and the eight thematic clusters with control for age
Variables Means & Sds Source of Variation Mean Square F Partial Eta Squared Power Observed
Self beliefs HL 2.52 (.26) Age 1.31 14.86*** .031 .970
CR 2.69 (.27) Group 1.97 22.86*** .086 1.000
CL 2.79 (.34) Error .09
Norm beliefs HL 2.28 (.31) Age 3.72 33.62*** .066 1.000
CR 2.39 (.27) Group .99 8.92*** .036 .973
CL 2.55 (.39) Error .11
Gen. beliefs HL 2.38 (.27) Age 2.31 25.53*** .051 .999
CR 2.54 (.24) Group 2.01 22.12*** .086 1.000
CL 2.69 (.35) Error .09
Goal beliefs HL 2.49 (.29) Age 2.37 22.88*** .046 .998
CR 2.75 (.30) Group 2.59 25.07*** .096 1.000
CL 2.75 (.36) Error .10
Cluster 1 HL 9.58 (1.75) Age 139.12 37.56*** .074 1.000
CR 10.69 (1.59) Group 76.86 20.75*** .081 1.000
CL 11.46 (2.28) Error 3.70
Cluster 2 HL 9.56 (1.35) Age 37.71 18.87*** .038 .991
CR 10.13 (1.30) Group 25.63 12.83*** .052 .997
CL 10.68 (1.55) Error 1.99
Cluster 3 HL 9.38 (1.22) Age 34.03 20.68*** .042 .995
CR 9.67 (.95) Group 2 4.41 14.84*** .059 .999
CL 10.58 (1.48) Error 1.64
Cluster 4 HL 11.44 (1.07) Age .004 .002 .000 .050
CR 11.49 (1.23) Group 18.19 10.87*** .044 .991
CL 12.16 (1.45) Error 1.67
Cluster 5 HL 11.28 (1.17) Age .63 .37 .036 .093
CR 11.42 (1.22) Group 14.87 8.74*** .036 .970
CL 11.89 (1.42) Error 1.70
Cluster 6 HL 8.93 (1.13) Age 15.42 9.13** .019 .854
CR 9.01 (1.34) Group 8.68 5.14** .021 .824
CL 9.69 (1.41) Error 1.69
Cluster 7 HL 11.10 (1.25) Age 11.48 5.33* .011 .635
CR 11.62 (1.21) Group 24.62 11.44*** .046 .993
CL 12.10 (1.70) Error 2.15
Cluster 8 HL 10.17 (1.24) Age 35.49 16.16*** .033 .980
CR 10.27 (1.34) Group 13.39 6.10** .025 .886
CL 11.188 (1.73) Error 2.20
*p<.05; **p<.01; ***p<.001.
Note:HL = Healthy, CR=Crohn’ disease, CL=Colorectal cancer; Group=Diagnostic group. In the second column, the first number is the mean and the number in parentheses is the SD. For the definition of the clusters see Table 2.
Significant group comparisons:
Self beliefs HL<CR***; HL<CL***; Norm beliefs HL<CR**; HL<CL**; General beliefs HL<CR***; HL<CL***; Goal beliefs HL<CR***; HL<CL***;
CR<CL*; Cluster 1 HL<CR***; HL<CL***; Cluster 2 HL<CR***; HL<CL***; Cluster 3 HL<CR*; HL<CL***; Cluster 4 HL<CL***; CR<CL**; Cluster 5 HL<CL***; CR<CL**; Cluster 6 HL<CL**; Cluster 7 HL<CR***; HL<CL**; Cluster 8 HL<CL**.
fifth sections) and Fig. 1, 4 and 5 show the results of applying the measurement model in all three groups. The results confirm clearly that the model fit is satisfactory in the three groups. The indices presented in Table 3 (CMIN, GFI, NFI, NNFI, CFI and RAMSEA) indicate that the model
whereby all eight clusters represent one latent factor is fully supported by the data in each group. Further, the highly significant standardized regression weights of all eight clusters imply that all eight clusters are related significantly to the latent variable, namely, they belong to
Table 6.Results of discriminant analysis with the four belief types as predictors and diagnosis (healthy, Crohn patients, colorectal cancer patients) as dependent variable
Predictors Wilks’ F Function 1 Function 2
Lambda Standard. Canon. Coeff.
Self beliefs .858 39.724*** .391 .018
Norm beliefs .894 28.558*** –.248 –1.392
General beliefs .833 48.463*** .866 –.377
Goal beliefs .879 33.143*** .012 1.799
Eigenvalues .217 .103
% of variance 67.9 32.1
Canon. Corr. .423 .305
Wilks’ Lambda .745 .907
Chi Square 141.495*** 46.975***
Functions at Group Centroids
Healthy –.614 –.142
Crohn .020 .669
Colorectal .430 –.160
Classification of cases:
Healthy –.614 –.142
Crohn .020 .669
Colorectal .430 –.160
Classification of cases:
Healthy Healthy 66.1% [Crohn 18.2%, Colorec. 15.8%]
Crohn Crohn 56.7% [Healthy 25.6%, Colorec. 17.8%]
Colorectal Colorec. 52.6% [Health 28.3%, Crohn 19.1%]
Total per cent of correct classifications 57.9% (Critical Ratio=7.809, p<.0001).
Table 7.Results of discriminant analysis with the 8 theme factors as predictors and diagnosis (healthy, Crohn patients, colorectal cancer patients) as dependent variable
Predictors Wilks’ F Function 1 Function 2
Lambda Standard. Canon. Coeff.
Cluster 1 .847 43.402*** .657 –.838
Cluster 2 .894 28.647*** –.072 –1.113
Cluster 3 .843 44.886*** .596 .634
Cluster 4 .931 17.859*** .000 .213
Cluster 5 .958 10.694*** –.037 –.044
Cluster 6 .927 18.948*** .017 .490
Cluster 7 .916 22.241*** .152 –.375
Cluster 8 .904 25.492*** –.232 1.298
Functions:
Eigenvalues .227 .083
% of variance 73.2 26.8
Canon. Correlation .430 .277
Wilks’ Lambda .753 .923
Chi Square 136.027*** 38.122***
Functions at Group Centroids
Healthy –.598 .171
Crohn –.092 –.599
Colorectal .465 –.112
Classification of cases:
Healthy Healthy 60.6% [Crohn 22.4%, Colorec. 17.0%]
Crohn Crohn 53.3% [Healthy 24.4%, Colorec. 22.2%]
Colorectal Colorec. 51.3% [Healthy 23.5%, Crohn 25.2%]
Total per cent of correct classifications 54.8 % (Critical Ratio=6.849, p<.0001).
