Translation and Construct Validity of the Feeling Scale and the Felt Arousal Scale in Portuguese Recreational Exercisers

Traducción y Validez de Constructo de la Feeling Scale y la Felt Arousal Scale en Ejercitadores Recreativos Portugueses

Tradução e Validade de Constructo da Feeling Scale e Felt Arousal Scale em Praticantes Portugueses de Exercício Recreacional

Affective responses to physical exercise are variable and sensitive to individual, environment, and task constraints (Box et al., 2020; Ekkekakis & Brand, 2019). There is evidence that exercising in nature environments induces greater positive affect than exercising indoors (Focht, 2009; Niedermeier et al., 2017), listening to music while exercising may also lead to increased positive feelings, and high intensity exercise typically affects mood in a negative way (Andrade et al., 2011; Jones et al., 2014). Furthermore, arousal is an important aspect of performance in sport (Córdoba et al., 2020; Moreno-Fernández et al., 2019). In short, one individual may experience increases as well as decreases in affect and arousal during an exercise session (Ekkekakis et al., 2011). To measure such fluctuating affective states, sport and exercise psychologists have at their disposal a large number of valid questionnaires and scales which in many instances have similarities and distinctions between them (Boyle et al., 2015). Some examples are the Exercise-induced Feeling Inventory (Gauvin & Rejeski, 1993), the Profile of Mood States (McNair et al., 1971), and the Positive and Negative Affect Scale (Watson et al., 1988). Such questionnaires assess self-perceptions usually verbalized in a sheet of paper or simply as a question from someone else (i.e. researcher, health professional, coach), and thus, are highly sensitive to language. Notable examples of extensively used affective scales in exercise and physical activity research are Hardy & Rejeski's (1989) Feeling Scale (FS), and Svebak & Murgatroyd's (1985) Felt Arousal Scale (FAS) (Evmenenko & Teixeira, 2020). Despite this prominence in affect/hedonism research, the FS and FAS have not yet been translated to Portuguese (Portugal) and have not had their psychometric properties tested in Portuguese participants, which limits their usage in this sociolinguistic context. During the process of creation and validation of self-reported instruments, questionnaires and scales are designed and shaped by the designer’s language and related culture. Thus, to make these useful instruments available to more cultural and linguistic contexts, rigorous language adaptations are needed, if evidence-based theories and interventions are to be generalizable to how humans behave in sport and exercise (Badenes-Ribera et al., 2020; Tenenbaum et al., 2012).

There is ongoing discussion regarding the ability of such short scales as the FS and FAS to capture complex mental phenomena (Hoeppner et al., 2011; Sarstedt & Wilczynski, 2009). However, single-item scales have been endorsed in the literature because they show strong correlations to multiple-item questionnaires measuring the same constructs. The one-item Fatigue Scale showed convergent validity with other more complex screening instruments (Temel et al., 2006). The Rating of Perceived Exertion (Borg, 1998) is a single-item perceived exertion scale which consistently shows strong correlations to physiological exertion markers (Arney et al., 2019; Canário-Lemos et al., 2020). Svebak and Murgatroyd (1985) showed that arousal states from the FAS were correlated with physiologic arousal of the body of participants, and the FS showed strong correlations to the Multiple Affective Adjective Check List (Zuckerman & Lubin, 1985). During the development and validation of the FS, Hardy and Rejeski (1989) discussed that the bipolarization of affective states during exercise may be governed by the central aspect of the pleasure/displeasure emotion, thus, the single-item FS was able to measure positive and negative affect accurately.

Exercise psychologists, coaches, health professionals and researchers would benefit from theory-supported and practical questionnaires that are able to assess mental states accurately and repeatedly during exercise interventions, without language barriers. To this end, the Feeling Scale and the Felt Arousal Scale are single-item scales which are suited to be used across cultures to assess exercise-related affective states, benefiting from theoretical support and practicality, provided they are demonstrably capable of capturing the same constructs across language and cultural barriers (Evmenenko & Teixeira, 2020), since both the FS and FAS have been used extensively in sport and exercise psychology research (Lacharite-Lemieux et al., 2015; Turner & Stevinson, 2017; White et al., 2015).

In the cross-cultural adaptation of questionnaires, construct validity, sensitivity, and internal consistency is important (Andrade et al., 2013). In the case of single-item scales such as the FS and the FAS, internal structure is not applicable and factor analysis is not possible (Cid et al., 2022). Consequently, establishing convergent construct validity by correlating the scores of the scale with validated questionnaires is an appropriate approach to test construct integrity (Tenenbaum et al., 2012). Additionally, regression analysis is recommended to test if the scores of the scale that is to be translated and adapted to a new language can be predicted by the scores of a scale of reference already measuring the construct of interest (Boateng et al., 2018). In the past, validations following this method were performed for single-item scales (Konrath et al., 2018; Yohannes et al., 2011).

To adapt the new instruments, a reference instrument that can demonstrably capture the construct of interest must be selected. In this case, the Self-Assessment Manikin (SAM) (Bradley & Lang, 1994; Lang, 1980) is a suitable scale that can be used as a reference instrument to measure affect and arousal. The SAM is a distinct instrument because its three items are displayed as pictures (pleasure, arousal, and dominance), relative to an object or event. Assuming that the meaning of positive and negative pictures is common and shared among cultures (at least western cultures), then the SAM is not limited to a particular country or language, and can thus be applied in wider contexts than language-based questionnaires and scales (Bynion & Feldner, 2017). Bradley and Lang (1994) conducted a tested the content validity of the SAM with participants that rated the SAM and the Semantic Differential Scale (Mehrabian & Russel, 1974), an affective scale of 18 bipolar adjectives dispersed in three dimensions (pleasure, arousal and dominance). By using principal components analysis, they reported that the three factors of pleasure, arousal and dominance accounted for 24%, 23% and 12% of the variation, respectively. Additionally, the SAM scores were strongly correlated to the adjectives in the Semantic Differential Scale (above 0.90 in the affective and arousal dimensions). Thus, the SAM is valid to be used in multiple cultural settings, including in Portuguese participants, as has been used before (Monteiro et al., 2011; Soares et al., 2015).

To increase the availability of the FS and FAS for research and field use by exercise and health psychologists and researchers in Portuguese contexts, the aim of this study was to translate and test the construct validity of these scales in Portuguese exercisers. The FS, FAS, and SAM protocols and forms are freely available in repositories easily accessible from an online search. Nonetheless, permission for using these instruments was requested from the original developers (Bradley & Lang, 1994; Hardy & Rejeski, 1989; Svebak & Murgatroyd, 1985), which was kindly granted by all.

This research is a two-phase design. First, a translation of the FS and FAS was conducted, which is described in the procedure section, followed by a construct validation study. The translation occurred in January 2021, and the data collection for the validation was collected during the months of March and April of 2021. This study was not preregistered. Approval was received from the Ethics Committee of the University (Grant no. 30/2020), in accordance with the Declaration of Helsinki.

One participant was excluded from the FS analyses due to incorrect filling of the scales. The linear regression scatter plots for the pre-exercise (Figure 1) and post-exercise (Figure 2) assessments, and the respective regression equations are presented below. Pearson Correlations, slopes, and intercept values are shown in table 1, including the 95% confidence intervals. The predictions of the FS and FAS based on correspondence with the SAM reference is displayed in table 2.

Pre-exercise assessment

The FS showed a strong correlation with the SAM affect item (r = .697, p < 0.001). The SAM affect item explained 49% of the variance in FS scores (R2 = .486). The lower, middle, and higher ends of the SAM affect item (1, 5, and 9 values) corresponded respectively to -1.6, +1.6, and +4.8 in the FS. The FAS showed a strong correlation with the SAM arousal item (r = .651, p < 0.001). The SAM arousal item explained 42% of the variance in FAS scores (R2 = .424). The lower and higher ends of the SAM arousal item (1 and 9 values) corresponded respectively to 2.3 and 5.5 in the FAS.

Post-exercise assessment

The FS showed a moderate correlation with the SAM affect item (r = .556, p < 0.001). The SAM affect item explained 31% of the variance in FS scores (R2 = .309). The lower, middle, and higher ends of the SAM affect item corresponded respectively to -1.3, +1.6, and +4.5 in the FS. The FAS showed a high correlation with the SAM arousal item (r = .719, p < 0.001). The SAM arousal item explained 52% of the variance in FS scores (R2 = .517). The lower and higher ends of the SAM arousal item corresponded respectively to 2.6 and 5.7 in the FAS.

The aim of this study was to translate and test the construct validity of the Portuguese versions of the FS and FAS, using the SAM as the reference instrument, in Portuguese recreational exercisers. Correlations and linear regressions were conducted and predictions of the FS and FAS ratings according to the reference scale were calculated. The hypothesis of this study was that the FS and the affect item of the SAM would measure the same constructs, thus, it was expected that the same individual would score these homologous scales identically, regardless of the value range of the scales. The same was hypothesized to be true about the FAS and the arousal scale of the SAM.

The FS and FAS are two important scales in the study of affective valence during exercise that have been widely used in international exercise psychology studies (Evmenenko & Teixeira, 2020). The translation and validation of these instruments and others into multiple cultural settings allows their dissemination and contributes to an increasing body of knowledge on the topic of affective response in the exercise context (Batista et al., 2019; Rodrigues et al., 2020). This study used a method similar to that of other authors who translated and validated the contents of FS and FAS to the German language by correlation with the SAM, which, similarly to this study, showed moderate-to-strong correlations (r = 0.72 to 0.73 for FS and SAM, and r = 0.50 to 0.62 for FAS and SAM) (Maibach et al., 2020). In that study, the authors employed the scales at various time-points during exercise, whereas in this study only the assessments before and after the exercise were recorded. Despite this difference in design between the two studies, the results here do not suffer from not screening the subjects during exercise, as the aim of this study was not to measure affective valence, but to test the construct validity of the FS and FAS as opposed to a homologous scale of reference. Furthermore, in this study, the correlation values between the FS/FAS and SAM remained strong in two distinct moments in terms of physical exertion (i.e. before and after exercise). Furthermore, the SAM scales revealed a substantial predictive power over the FS and FAS, suggesting that the scales share the same constructs.

In spite of the potential utility of these instruments, there is some discussion that their single-item structure prevents the identification of factors that may influence the exercisers’ affective valence. Engaging in physical activity may also cause negative experiences such as fear, breathlessness, and pain, despite it being widely associated to greater wellbeing (Ekkekakis et al., 2011). However, they offer a general measure of pleasure/displeasure and arousal that are the result of a set of positive and negative feelings related to engaging in physical activities and exercise. Most researchers of affective phenomena include at least the two dimensions of affective valence and arousal in their models (Mehrabian and Russel, 1974; Thayer, 1989) and according to Ekkekakis and Petruzzello (2002), “different affective states are considered combinations of varying degrees of these two constituent dimensions” (p. 38). Thus, these instruments gain other advantages in comparison to more complex affective scales and questionnaires, since they are quicker to rate, are able to capture affective state, and thus are easier to employ during exercise, as opposed to complex, time-consuming questionnaires which may impact the exercise interventions.

Another particular advantage of having the FS and FAS as available instruments for research and field use, is that the scores of both scales in an individual may be plotted as an affect-arousal (x,y) ‘affective space’, which is known as the circumplex model, which has been applied in several contexts, including physical exercise (Evmenenko & Teixeira, 2020; Niedermeier et al., 2017). Thus, increasing the availability of practical and valid instruments like the FS and FAS will allow for the application of the circumplex model more widely.

As limitations in this study, the SAM did not predict the lower ends of the FS and FAS accurately. This is probably because there were no observations of extremely low affect or arousal in the sample, since the participants were probably engaging in their favourite past-time activities. It is inferred that if there had been more negative scorings in the scales it would have helped the linear prediction even further, since the predictions were positive, nonetheless. Another limitation were the successive lockdowns during the Covid-19 pandemic, which imposed limitations on the researchers, ultimately denying reliability analyses (test-retest). However, one could infer that if construct validity has been obtained in correspondence to the original English versions of the scales, then reliability of the newly translated version ought to be similar to the original versions as well, although it is acknowledged that reliability testing is needed to reach definitive validation.

Further validation of these scales is warranted and should focus on recruiting other samples, such as the elderly or sedentary individuals, as well as designing trials where the scales are scored during high intensity activities to produce more negative scorings in the scales. Alternatively, researchers can manipulate the participants towards negative affect and arousal prior to engaging in exercise, to improve the predictive power of the SAM over the FS and the FAS. Also, reliability tests should be conducted by replicating the exercise with the same participants, although these scales capture extremely variable behavior, i.e. affect is known to rise and fall substantially within individuals in short periods of time (Brose et al., 2020).

This study shows that the Portuguese translated versions of FS and FAS capture the constructs of affect and arousal and may be used to assess these psychological states in Portuguese recreational exercisers. The translated forms are available as supplemental material.

HB and DA were partly funded by the Fundação para a Ciência e a Tecnologia [grant numbers SFRH/BD/147966/2019, and UIDB/00447/2020].

The authors report there are no competing interests to declare

Data and materials from this study are available as supplemental material, or by contacting the main author