Assessing alexithymia: Psychometric properties of the Perth Alexithymia Questionnaire and 20-item Toronto Alexithymia Scale in United States adults

Highlights

• Overall, the PAQ and TAS-20 both displayed good factorial and convergent validity.

• However, the TAS-20 assessed alexithymia only for negative (not positive) emotions.

• Our PAQ analyses highlighted the importance of assessing both emotional valences.

• All PAQ scores had good reliability, but the TAS-20 EOT score did not.

• The PAQ therefore appeared to provide a more comprehensive facet-level profile.

Abstract

Alexithymia is a trait involving difficulty identifying feelings, difficulty describing feelings, and externally orientated thinking. Over the past two decades, it has usually been assessed using the 20-item Toronto Alexithymia Scale (TAS-20). Recently, the Perth Alexithymia Questionnaire (PAQ) was developed to provide more detailed facet-level (and valence-specific) analysis. To date, no studies have compared these two measures, and few data are available on the PAQ. We examined the psychometric properties of the PAQ and TAS-20 in United States adults (N = 675). Confirmatory factor analysis indicated that both measures had a theoretically congruent factor structure. While the PAQ assessed alexithymia for both negative and positive emotions, the TAS-20 assessed alexithymia only for negative emotions. All PAQ and TAS-20 subscales had good reliability, except for the TAS-20 externally orientated thinking subscale. Pearson correlations supported the convergent validity of both measures. We conclude that the PAQ and TAS-20 both have strong qualities and assess a similar alexithymia construct, however, the PAQ appears to provide a more comprehensive facet-level profile.

Introduction

Alexithymia is a multidimensional trait comprised of a cluster of emotion processing deficits: difficulties identifying one's own feelings (DIF), difficulties describing feelings (DDF), and an externally orientated thinking style (EOT) characterised by an excessive focus on external stimuli rather than internal emotional states (Bagby, Parker, & Taylor, 1994; Lane, Weihs, Herring, Hishaw, & Smith, 2015; Preece, Becerra, Allan, Robinson, & Dandy, 2017; Sifneos, 1973). Alexithymia is a risk factor for psychopathology symptoms and poor well-being (Taylor, Bagby, & Parker, 1999), and thus is a key assessment target for researchers and clinicians. One of the earliest measures of alexithymia was the 20-item Toronto Alexithymia Scale (TAS-20; Bagby et al., 1994), which remains the most widely used measure in the field. Preece, Becerra, Robinson, Dandy, and Allan (2018), however, recently introduced the Perth Alexithymia Questionnaire (PAQ) to try to enable more detailed facet-level (and valence-specific) analyses of the construct. This is in line with recent trends in the alexithymia field, whereby authors are increasingly interested in examining alexithymia at the facet level (e.g., Demers et al., 2019; Kajanoja, Scheinin, Karukivi, Karlsson, & Karlsson, 2019; Leweke, Leichsenring, Kruse, & Hermes, 2012; Lyvers, McCann, Coundouris, Edwards, & Thorberg, 2018). To date, no existing studies have compared the psychometric performance of both the PAQ and TAS-20. Few data are available on the PAQ and, with respect to valence-specific measurement, it is also unclear whether TAS-20 assesses alexithymia for negatively valenced emotions, positively valenced emotions, or some aggregate of both. Our purpose here was therefore to examine the psychometric properties of the PAQ and TAS-20 in adults from the United States.

The TAS-20 (Bagby et al., 1994) is a 20-item self-report measure. It was originally designed to provide only a total scale score as an overall index of alexithymia (i.e., no facet level scores; Bagby, Taylor, Quilty, & Parker, 2008), but it is now standard practice for separate subscale scores to also be derived: DIF (e.g., “I am often confused about what emotion I am feeling”), DDF (e.g., “It is difficult for me to find the right words for my feelings”), and EOT (e.g., “I prefer talking to people about their daily activities rather than their feelings”) (Taylor, Bagby, & Parker, 2003). Two of the DIF items ask about negatively valenced emotions (e.g., “When I am upset, I don't know if I am sad, frightened, or angry”) and all other items do not specify an emotional valence (e.g., “I have feelings that I can't quite identify”). All items are answered on a 5-point Likert scale, with higher scores indicating higher levels of alexithymia (five items are reverse-scored).

The psychometric properties of the TAS-20 have been tested in a range of sample types, with fairly well-replicated patterns (for a review, see Sekely, Bagby, & Porcelli, 2018). In terms of factor structure, three theoretically meaningful first-order factors usually emerge (DIF, DDF, EOT), alongside a reverse-scored item method factor¹ (e.g., Meganck, Vanheule, & Desmet, 2008; Preece, Becerra, Robinson, & Dandy, 2018; Watters, Taylor, Ayearst, & Bagby, 2016). The DIF and DDF items consistently have strong factor loadings, though some of the EOT items usually have poor loadings (a pattern that has been ascribed, at least in part, to the high concentration of reverse-scored items in the EOT subscale; Meganck et al., 2008). The TAS-20 total scale score and DIF and DDF subscales usually have sound internal consistency (Cronbach's α ≥ 0.70), but the EOT subscale does not (e.g., Taylor et al., 2003). Convergent validity has been evidenced by correlations with markers of psychopathology and emotion dysregulation (e.g., Leweke et al., 2012), though the low reliability of the EOT subscale score may reduce the utility of the TAS-20 for studies specifically interested in that alexithymia facet (Demers et al., 2019).

The PAQ (Preece, Becerra, Robinson, Dandy, & Allan, 2018) is a 24-item self-report measure. Items are answered on a 7-point Likert scale, with higher scores indicating higher levels of alexithymia. The PAQ is designed to provide separate scores for DIF, DDF and EOT. For the DIF and DDF facets, separate valence-specific subscale scores can also be derived for the processing of negative and positive emotions. This valence-specific approach was in response to recent findings that valence is an important consideration when assessing emotional constructs (e.g., Becerra, Preece, Campitelli, & Scott-Pillow, 2019; van der Velde et al., 2013). A five subscale structure was therefore intended: Negative-Difficulty Identifying Feelings (N-DIF; “When I'm feeling bad, I get confused about what emotion it is”), Positive-Difficulty Identifying Feelings (P-DIF; “When I'm feeling good, I can't tell whether I'm happy, excited, or amused”), Negative-Difficulty Describing Feelings (N-DDF; “When something bad happens, it's hard for me to put into words how I'm feeling”), Positive-Difficulty Describing Feelings (P-DDF; “When I'm feeling good, I can't talk about those feelings in much depth or detail”), and General-Externally Orientated Thinking (G-EOT, “I prefer to focus on things I can actually see or touch, rather than my emotions”). Preece, Becerra, Robinson, Dandy, and Allan (2018) suggest that these subscales can also be combined to yield several theoretically meaningful composite scores, including a total scale score.

Two papers have examined the psychometric properties of the PAQ, both using Australian samples (Preece et al., 2020; Preece, Becerra, Robinson, Dandy, & Allan, 2018). In Preece, Becerra, Robinson, Dandy, and Allan's (2018) original development study, they found good support for a five-factor structure corresponding to the five intended subscales (N-DIF, P-DIF, N-DDF, P-DDF, G-EOT). In support of the importance of considering emotional valence in alexithymia assessments, this five-factor model was superior to simpler solutions that did not have valence-specific factors, and participants tended to report significantly more difficulties identifying and describing their negative emotions than their positive emotions. The scale authors (Preece et al., 2020; Preece, Becerra, Robinson, Dandy, & Allan, 2018) reported high levels of internal consistency for all PAQ subscale and composite scores across four samples (α ≥ 0.80), and these scores had good convergent and discriminant validity against markers of psychopathology symptoms or emotion dysregulation.

We examined the factorial validity, internal consistency, and convergent validity of the PAQ and TAS-20 in a sample from the United States. In comparing these measures, a key aim was to test whether valence was an important consideration in alexithymia assessments, and if so, whether the non-valenced DIF and DDF items from the TAS-20 were measuring the processing of negative emotions, positive emotions, or some aggregate of both. Descriptions from Bagby and colleagues (e.g., Bagby et al., 1994; Bagby, Parker, & Taylor, 2020) indicate that these TAS-20 subscales were intended to capture both valence types, but this has not yet been tested empirically by factorial modelling.