Is there a reliable factorial structure in the 20-item Toronto Alexithymia Scale?: A comparison of factor models in clinical and normal adult samples

https://doi.org/10.1016/S0022-3999(03)00033-3Get rights and content

Abstract

Objective

The 20-item Toronto Alexithymia Scale (TAS-20) is the most widely used instrument for measuring alexithymia. However, different studies did not always yield identical factor structures of this scale. The present study aims at clarifying some discrepant results.

Method

Maximum likelihood confirmatory factor analyses of a German version of the TAS-20 were conducted on data from a clinical sample (N=204) and a sample of normal adults (N=224). Five different models with one to four factors were compared.

Results

A four-factor model with factors (F1) “Difficulty identifying feelings” (F2), “Difficulty describing feelings” (F3), “Low importance of emotion” and (F4) “Pragmatic thinking” and a three-factor model with the combined factor “Difficulties in identifying and describing feelings” described the data best. Factors related to “externally oriented thinking” provided no acceptable level of reliability.

Conclusion

Results from the present and other studies indicate that the factorial structure of the TAS-20 may vary across samples. Whether factor structures different from the common three-factor structure are an exception in some mainly clinical populations or a common phenomenon outside student populations has still to be determined. For a further exploration of the factor structure of the TAS-20 in different populations, it would be important not only to test the fit of the common three-factor model, but also to consider other competing solutions like the models of the present study.

Introduction

The term ‘alexithymia’, coined by Sifneos [1], [2], describes a set of affective and cognitive characteristics. They are believed to reflect deficits in the processing or regulation of emotions through mental processes [3]. As indicated by the Greek term, its most prominent feature is a marked difficulty in describing one's own feelings to others. Alexithymia was defined as a multifacet construct composed of the following conceptually distinct, yet logically related, salient features: (a) “difficulty identifying feelings and distinguishing between feelings and the bodily sensations of emotional arousal”, (b) “difficulty describing feelings to other people”, (c) “constricted imaginal processes as evidenced by a paucity of fantasies” and (d) “a stimulus bound, externally orientated cognitive style” [3, p29]. Yet still, there is the problem of an adequate factorial structure of the construct in self-report measures.

The Toronto Alexithymia Scale (TAS-26) [4] was introduced as a reliable and valid measure of the construct. After several revisions, the actual 20-item version (TAS-20) [5], [6] resulted. For this scale, a three-factor structure was proposed [5]: (F1) “Difficulty identifying feelings”, (F2) “Difficulty describing feelings” and (F3) “Externally oriented thinking”. Although the majority of the factor analytic studies about the TAS-20 seem to support the original three-factor solution, a closer review shows that there might be alternative factor structures. The results of studies using exploratory factor analyses (EFAs) varied. In three studies, the common three-factor solution was preferred [5], [7], [8]. In one study, a one-factor model was found [9], and as had already occurred with the two previous versions of the TAS, some researchers found a two-factor solution in various language versions where the items of F1 and F2 formed a common factor [10], [11], [12]. However, two of these studies [10], [11] were criticized by Taylor et al. [13] because of sample characteristics or divergent results of exploratory and confirmatory factor analyses (CFAs) (cf. Ref. [14]). Other researchers found a factor structure where “Externally oriented thinking” decomposed into two factors “Pragmatic thinking” and “Lack of subjective significance or importance of emotions” [10], [15], [16], [17], [18], [19] (cf. also Ref. [20]).

In addition to EFA, in a large number of studies, CFAs of the TAS-20 with structural equation modeling were conducted [5], [9], [10], [14], [18], [19], [21], [22], [23], [24], [25], [26], [27], [28], [29]. Most of these studies yielded an acceptable fit of the three-factor model established by Bagby et al. [5]. Only in 5 out of these 15 studies, the common three-factor model showed no acceptable fit [9], [10], [18], [27], [29]. However, the studies reported above had some limitations. The fit indices (Goodness of Fit Index [GFI], Adjusted Goodness of Fit Index [AGFI] and RMR) or the criteria for acceptable fit of these indices used in most of the cited studies with CFA had been criticized [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41], [42]. Furthermore, competing models with two or three factors were only compared in few studies. Alternative models like those mentioned above [10], [16], [17], [18], [19] were tested in none of the studies, although they are empirically supported and might be equally plausible.

Taking all this together, many of the studies reviewed here support the original three-factor model [5] of the TAS-20. However, there are hints that the factor structure of the TAS-20 may vary across various studies, samples and language versions (cf. also Ref. [43]). The factor structure, however, is clearly an important issue regarding the multidimensionality of the alexithymia construct. The current study was intended to contribute to the clarification of the inconsistent findings concerning the factor structure of the TAS-20 reported above and to address the limitations of many of these studies. Thus, in this study, five competing factor models of the TAS-20 with one to four factors were tested for their fit with data from a clinical and a nonclinical sample. The models were compared concerning (a) the global fit, (b) the relationships among the TAS factors, (c) parameter estimates among the factors and (d) the reliability of the factors.

Section snippets

Subjects and procedure

The clinical sample consisted of 204 inpatients (59.3% women) of a large hospital for psychosomatic disorders and a clinic for substance abuse. Their diagnoses were primarily anxiety disorders, somatoform disorders, somatic diseases traditionally regarded as psychosomatic and alcohol abuse. The mean age of the patients was 47.1 years (S.D.=8.0) with a range from 23 to 63 years. Regarding the educational level, 56.1% of the patients had finished secondary school (Hauptschule), 31.2% had finished

Global fit of models

The global model fit indices for both samples are presented in Table 2. Using the combinational rule of SRMR≤0.11 and RMSEA≤0.06, the two-factor model (b) (ID/DE–ET), the three-factor model (c) (ID–DE–ET), the three-factor model (d) (ID/DE–PR–IM) and the four-factor model (e) (ID–DE–PR–IM) in the patient sample showed a good fit with the data. The SRMR met the predefined criterion in these models. Regarding the lower boundary of the 90% confidence interval of the RMSEA, the possibility cannot

Factor structure and model fit

The two-factor model (b) (ID/DE-ET), the three-factor models (c) (ID–DE–ET) and (d) (ID/DE–PR–IM) and the four-factor model (e) (ID–DE–PR–IM) showed a good global goodness-of-fit in the patient sample. In the normal adult sample, none of the models fit the data particularly well. Though the difference between the tested models was small, the four-factor model yielded the best fit in both samples and should be preferred on the basis of this criterion. This was confirmed by a significant χ2

Conclusions

When the TAS-20 is applied for individual diagnosis of alexithymia, the reliability and the conceptual clarity of the factor “externally oriented thinking” should be increased. Further, there is evidence that the factor structure of the TAS-20 may vary across samples. As a consequence, the factor structure may emerge as more or less differentiated. Whether factor structures different from the common three-factor structure are an exception in some mainly clinical populations or a common

Acknowledgements

The authors wish to thank Johanna Schams and Benedikt Reuter for their help in collecting parts of the data reported in this article. The support of the Rhön-Rehabilitationsklinik Bad Kissingen and the Saaletalklinik Bad Neustadt is greatly appreciated. This paper was prepared with the assistance of the members of the CERE Group and the support of the Maison des Sciences de l'Homme, Paris.

References (55)

  • M Bagby et al.

    Affect dysregulation and alexithymia

  • GJ Taylor et al.

    Toward the development of a new self-report alexithymia scale

    Psychother Psychosom

    (1985)
  • M Bach et al.

    Validierung der deutschen Version der 20-Item Toronto-Alexithymie-Skala bei Normalpersonen und psychiatrischen Patienten

    Psychother Psychosom Med Psychol

    (1996)
  • I Fukunishi et al.

    Is alexithymia a culture-bound construct? Validity and reliability of the Japanese versions of the 20-item Toronto Alexithymia Scale and Modified Beth Israel Hospital Psychsomatic Questionnaire

    Psychol Rep

    (1997)
  • MC Lambert et al.

    Emotions and psychopathology

    West Indian Med J

    (1999)
  • MG Haviland et al.

    Structure of the twenty-item Toronto Alexithymia Scale

    J Pers Assess

    (1996)
  • G Loas et al.

    Factor analysis of the French version of the 20-item Toronto Alexithymia Scale (TAS-20)

    Psychopathology

    (1996)
  • T Erni et al.

    Two factor solution of the 20-item Toronto Alexithymia Scale confirmed

    Psychopathology

    (1997)
  • GJ Taylor et al.

    Assessment of alexithymia

  • G Loas et al.

    Confirmatory factor analysis of the French translation of the 20-item Toronto Alexithymia Scale

    Percept Mot Skills

    (1997)
  • T Ritz

    Alexithyme Merkmale und affektive Intensität: adaptation und Zusammenhang zweier Fragebogeninstrumente

    Z Differ Diagn Psychol

    (1994)
  • T Ritz et al.

    Zur Konstruktvalidität einer deutschen Fassung der Toronto Alexithymia Scale

    Z Differ Diagn Psychol

    (2000)
  • M Franz et al.

    Faktorenstruktur und Testgütekriterien der deutschen Version der Toronto-Alexithymie-Skala (TAS-20) bei psychosomatischen Patienten

    Psychother Psychosom Med Psychol

    (2001)
  • M Joukamaa et al.

    Psychometric properties of the Finnish 20-item Toronto Alexithymia Scale

    Nord J Psychiatry

    (2001)
  • TN Wise et al.

    Comparison of 26-item and 20-item versions of the Toronto Alexithymia Scale for psychiatric outpatients

    Psychol Rep

    (2000)
  • JDA Parker et al.

    Factorial validity of the 20-item Toronto Alexithymia Scale

    Eur J Pers

    (1993)
  • R Pandey et al.

    Cross-cultural alexithymia: development and validation of a Hindi translation of the 20-item Toronto Alexithymia Scale

    J Clin Psychol

    (1996)
  • Cited by (94)

    • Factors associated with alexithymia in adults with epilepsy

      2021, Epilepsy and Behavior
      Citation Excerpt :

      Finally, the TAS-20 is the most commonly used method to measure alexithymia, but this metric has an inherent limitation in that it is a self-rating scale, and patients with alexithymia often have difficulties with self-reflection. In addition, the EOT subscale had low internal consistency, which is an important limitation to our categorical approach to alexithymia [25,26]. In summary, alexithymia was more severe in degree, but not more prevalent, in PWE than in healthy controls.

    View all citing articles on Scopus
    View full text