Validation of the Italian Translation of the Philadelphia Mindfulness Scale

Assessing mindfulness skills is an important topic for the mindfulness research community around the world. Having a variety of mindfulness measures validated in different languages would increase the quality of research on the topic. Thus, we propose a contribution to the validation of the Italian version of the Philadelphia Mindfulness Scale (PHLMS) and its relatively short form, including only 10 out of the original 20 items. We also study its relationships with other psychological measures, and how it relates to meditation experience. We translated the original scale into Italian and then assessed its psychometric properties in two convenience samples of healthy participants from the general population (total n = 429). We analyzed the validity and the dimensionality of the scale (study 1 and 2), its construct validity and relationships with other psychological measures (study 3), and then how it relates to meditation experience (study 4). An exploratory factor analysis (study 1) on sample 1 confirmed the original PHLMS structure, indicating two orthogonal dimensions named awareness and acceptance. A successive confirmatory factor analysis (study 2) on sample 2 also revealed a good fit of the model for the two-factor structure with correlated error. The short form also revealed a good model fit. In the successive studies conducted on a pooled sample including both sample 1 and 2, we confirmed the predominant role of acceptance in determining psychological well-being (study 3) and that meditation experience was related to increased mindfulness skills (study 4). The results support both the long and short forms of the Italian PHMLS (PHLMS-I) as valid and reliable instruments for measuring mindfulness skills in non-meditative and meditative samples.

Defining how to measure mindfulness is one of the main tasks of psychological research on mindfulness and its effects on psychological well-being. Measurement of mindfulness is important to understand the relationships of dispositional mindfulness with personality and well-being (Rau & Williams, 2016), and to collect evidence on the effectiveness of mindfulness-based interventions (Baer, 2019). A reliable and valid measurement of mindfulness is indeed crucial for empirical investigation and the assessment of the effects of mindfulness-based interventions. However, finding such a measurement is challenging, as it implicated conducting statistical analyses on validity and reliability, and solving conceptual issues related to the content of the self-report measures (Bergomi et al., 2013;Gherardi-Donato et al., 2020;Park et al., 2013). Moreover, mindfulness was both defined (Baer, 2019;Bishop et al., 2004;Kabat-Zinn, 1991) and operationalized (Baer et al., 2006;Brown & Ryan, 2003;Feldman et al., 2007) in different ways. Thus, working on a compelling and broadly accepted mindfulness measure is a crucial task for mindfulness-related research in order to make it possible to share and compare easily study and intervention results.
To date, several questionnaires have been developed for this aim. One of the most widely used questionnaires is the Mindful Attention Awareness Scale (MAAS; Brown & Ryan, 2003), which includes only a general measure of mental presence, considering mindfulness as a unidimensional construct. Other influential mindfulness questionnaires are the Kentucky Inventory of Mindfulness Skills (KIMS; Baer et al., 2004), the Freiburg Mindfulness Inventory (FMI; Walach et al., 2006), the Five Facet Mindfulness Questionnaire (FFMQ; Baer et al., 2006), and the Cognitive and Affective Mindfulness Scale-Revised (CAMS-R; Feldman et al., 2007). All of these questionnaires include multiple factors, thus considering mindfulness as a multicomponent set of capabilities or skills. In particular, the FFMQ was developed by combining all items of the previous questionnaires and then performing an overall factor analysis. With this method, Baer et al. (2006) identified five mindfulness factors or facets: observing experiences and sensations, describing/labeling sensations, acting with awareness, nonreactivity to inner experience, and nonjudging of inner experience. This scale has been used in hundreds of studies, although it was empirically but not theoretically founded, as it was constructed in a bottom-up or datadriven way. Developing a measure that was more grounded in the theory of mindfulness was thus useful to bridge the gap with research on defining mindfulness, with a conceptual foundation of the proposed measures grounded on both Buddhist teachings and secular conceptualizations of mindfulness (Bishop et al., 2004;Brown et al., 2007;Chiesa, 2013). Moreover, a theory-driven approach to the measurement of mindfulness may enable a more accurate association with cognitive and affective processes and mechanisms (Malinowski, 2013).
To this end, Cardaciotto et al. (2008) developed a new mindfulness scale, called the Philadelphia Mindfulness Scale (PHLMS). The PHLMS was based on the consensus definition proposed by Bishop et al. (2004), in which mindfulness includes two main components: "The first component involves the self-regulation of attention so that it is maintained on immediate experience, thereby allowing for increased recognition of mental events in the present moment. The second component involves adopting a particular orientation toward one's experiences in the present moment, an orientation that is characterized by curiosity, openness, and acceptance" (Bishop et al., 2004, p. 232). Based on this definition, Cardaciotto et al. (2008) designed an instrument with two components. The first component was called awareness and the second acceptance. Awareness measures the ability of individuals to pay attention to the present moment, i.e., to physical sensations (e.g., "When I shower, I am aware of how the water flows over my body."), bodily experiences ("I notice changes inside my body, like my heart beating faster or my muscles getting tense."), and feelings and thoughts while they are happening (e.g., "I am aware of thoughts I am having when my mood changes."). Acceptance measures the disposition to not judging such experiences and to let go of thoughts and feelings without avoiding them. In order to present this construct in a more accessible way to participants not familiar with meditation and mindfulness, it is measured with reverse-scored items measuring non-acceptance (e.g., "I tell myself that I shouldn't feel sad." and "I try to stay busy to keep thoughts or feelings from coming to mind."). Thus, awareness refers to the "what" part of mindfulness, that is, the focus of present-moment attention, and acceptance to the "how," that is, the attitude and quality of this attention (Baer, 2019;Bishop et al., 2004).
This bipartite model of mindfulness was further elaborated in the Monitor and Acceptance Theory (MAT; Lindsay & Creswell, 2017, 2019, in which it was proposed that present-moment awareness increases emotional reactivity while acceptance could modulate such response. While this model seems not to be fully supported by data available to date (see Simione et al., 2021), it shows how having a theory-grounded operationalization of mindfulness can lead to the formulation of clear and testable hypotheses. Despite its unresolved issues, the proposal of MAT pointed out the importance of having competing or alternative theories of mindfulness that could be directly tested and falsified. In order to test such theory-driven hypotheses, instruments grounded in the same theoretical background could greatly help. As PHLMS was specifically designed to follow the bipartite model of mindfulness (Bishop et al., 2004), it represents a valid instrument for testing mindfulness theories such as MAT (Lindsay & Creswell, 2017).
Compared to the most used mindfulness questionnaires, PHLMS offers a more reliable theoretical background and potentially wider use. MAAS (Brown & Ryan, 2003) measures only the present-moment awareness as the reverse of absent-mindedness, and thus, it completely lacks measures of the "how" part of mindfulness, while PHLMS includes both. The FMI (Walach et al., 2006) is based on the same theoretical model of PHLMS, thus including attention and acceptance as mindfulness facets, but it was designed for people already experienced with mindfulness meditation. Then, it could not be used in experiments designed to study dispositional mindfulness in the general population, while PHLMS was validated with both non-meditators and meditators (Cardaciotto et al., 2008;Morgan et al., 2020).
Also the FFMQ (Baer et al., 2006) has been used with people naïve to meditation, but it presents three main limitations as compared to the PHLMS. First, it was constructed in a data-driven manner, without any specific theoretical model of reference. This could lead to confusion about the expected relationships with other variables. For example, both nonreacting and nonjudging scales could be related to the acceptance or "how" factor of mindfulness (Lindsay & Creswell, 2017). However, these mindfulness facets showed inconsistent patterns of correlations with mental health measures (Rau & Williams, 2016;Simione et al., 2021). Furthermore, the describing score included in both FFMQ and KIMS was not credited as part of a core component of mindfulness (Bergomi et al., 2013). Moreover, FFMQ includes twice as many items (39 items) with respect to the PHLMS (20 items), thus demanding a longer compilation time and a possible decrease in compliance. Furthermore, FFMQ showed psychometric problems for the observing and nonreacting scales (Baer et al., 2006;Tran et al., 2013). Finally, several results have suggested that the five-facet structure of the FFMQ could be reduced to the two main aspects of mindfulness proposed in the literature Lilja et al., 2013;Rau & Williams, 2016;Tran et al., 2013). Overall, these findings have confirmed the need for a simpler but more reliable measure of mindfulness for both dispositional and experimental studies, such as PHLMS. However, PHLMS has been validated only in English (Cardaciotto et al., 2008;Morgan et al., 2020), Spanish (Tejedor et al., 2014), Chinese (Zeng et al., 2015), and Portuguese (Teixeira et al., 2017). To date, a validated Italian version of PHLMS is missing.
The main objective of this study was to provide a validated measure of mindfulness in Italian that was an alternative to the FFMQ (Giovannini et al., 2014) or the MAAS (Veneziani & Voci, 2015). In particular, we aimed to assess if the original two-factor structure of the PHLMS was confirmed for the Italian version and how such version correlated with the FFMQ and with measures of ill-and well-being. To this end, we conducted a set of successive studies on two non-clinical samples of the general population recruited at different times (sample 1, n = 226; sample 2, n = 203). Study 1 focused on the translation process of the Italian version of the PHLMS and on a first exploratory factor analysis carried out in sample 1. Study 2 involved a confirmatory factor analysis conducted in another sample (sample 2) in order to assess the final structure of the scale in the Italian population. We also tested a short version of the PHLMS that included only 10 items (as described by Zeng et al., 2015). In the successive studies, we combined the two samples in a pooled sample (total n = 429). With this pooled sample, we first assessed construct and convergent validity of the PHLMS (study 3), and then the relationship between PHLMS scores and meditation expertise (study 4), by comparing meditators and non-meditators as well as by conducting correlation analyses between meditation expertise and mindfulness scores.

Participants
The first sample (sample 1) was a convenience sample of 226 Italian healthy volunteers from the general population (females = 150, males = 76). The average age in years was 29.07 (SD = 8.85), and the average education level in years was 14.97 (SD = 2.63). About their job, 108 participants reported to be students, 101 worked either full time (61) or partial (40) time, and 17 reported to be unemployed. About the personal relationship, 80 were single and 146 in a relationship. Regarding religious faith, 118 were Catholic and 4 reported other religions; 104 reported being agnostic or atheist. Fourteen participants reported the presence of a psychological condition. About mindfulness experience, 40 participants reported some experience with mindfulness or similar meditation practice, with 42.89 h (SD = 206.87) of average lifetime meditation experience, while 186 reported no experience. Sample 1 was recruited during October 2019.

Procedure
Participants gave their informed consent before completing the battery of questionnaires, which was administered through a series of successive online forms or a printed fascicle. The questionnaires were presented in the order reported in the Materials section. Afterwards, participants also compiled a demographic form, reporting sex, age, education level, presence of a psychological condition, and their meditation experience in terms of estimated lifetime experience in hours.
The PHLMS was translated into Italian with the typical forward-backward translation process (Brislin, 1980). Three independent translators produced three versions of the PHLMS items in Italian. One of the translators was also an expert in mindfulness. Then, they compared the items that showed different translations and proposed a final version of each item by resolving the discrepancies between the independent translations. The 20 Italian items produced were back-translated into English again by a bilingual translator not included in the first set of translators. The new set of items were very close to the original ones. In light of the differences between the original items and the back-translated items, a final Italian version of the 20 items was prepared.
The scale was then submitted to an expert committee for the evaluation of its content validity. The experts were six researchers with expertise in mindfulness, psychology, or both. The experts evaluated the proposed item translation and suggested modifications to improve the scale at both conceptual and semantic levels. They raised minimal to no problems with the proposed items, and thus, we produced the final 20 items by incorporating such last modifications.
As the last step, we administered this final version of the scale to 10 students who were debriefed after filling the questionnaire about the understandability of the 20 items and their interpretation by people naïve to mindfulness. They confirmed that the items were clear and that the five-point Likert scale proposed for responding was meaningful. We then administered this final version to sample 1 for the subsequent analysis.

Measures
To validate the Italian translation of the PHLMS, we used a range of scales investigating mindfulness and mental health. Mindfulness was assessed using the FFMQ and the translated version of the PHLMS. FFMQ (Baer et al., 2006;Giovannini et al., 2014) is a 39-item questionnaire evaluating five domains or facets of mindfulness, i.e., observing, describing, acting with awareness, nonjudging, and nonreacting (to inner experience). Each item was evaluated on a 5-point Likert scale ranging from 1 ("never or very rarely true") to 5 ("very often or always true"). Facets were computed as the sum of the corresponding items. Higher scores indicate a higher degree of mindfulness. In our study sample, FFMQ scales revealed a good to excellent Cronbach's alpha, with values of 0.75, 0.87, 0.88, and 0.89, respectively, for the observing, describing, acting with awareness, nonjudging, and nonreacting scales, while the nonreacting scale revealed an acceptable alpha of 0.68. McDonald's omega for these scales was 0.76, 0.88, 0.88, 0.89, and 0.68.
PHLMS (Cardaciotto et al., 2008) is a 20-item inventory measuring two main aspects of mindfulness, i.e., awareness and acceptance. Each score was calculated as the sum of 10 items, and each item was evaluated on a 5-point Likert scale ranging from 1 ("never") to 5 ("very often") according to the frequency of subjective experience of the content of the item. Higher scores reflect higher mindfulness ability.
Mental health was assessed with the Symptoms Checklist Revised or SCL-90R (Prunas et al., 2012). SCL-90R is an inventory of psychological symptoms ranging from psychoticism to somatization. Each item was rated on a 5-point Likert scale from 0 ("never") to 4 ("a lot") to assess the extent to which participants experienced the listed symptoms. We computed an overall distress index as the Global Severity Index or GSI, which revealed an excellent Cronbach's alpha score of 0.98 and a McDonald's omega score of 0.98. A higher GSI score indicated greater presence of psychological symptoms.

Data Analyses
To validate the Italian translation of the PHLMS, we conducted an exploratory factor analysis (EFA) in order to assess its factor structure in the Italian sample. Before conducting this analysis, we checked its assumptions by performing the Kaiser-Meyer-Olkin test (KMO) and the Bartlett's test of sphericity. The EFA was conducted with a varimax rotation, as we expected that the two factors should be orthogonal (as in Cardaciotto et al., 2008). Then, the factors individuated in the EFA were inspected at item level and their internal validity assessed by means of Cronbach's alpha, McDonald's omega, inter-item correlation, and item-score correlation.

Exploratory Factor Analysis
The PHLMS items in sample 1 were checked for EFA assumptions. The KMO test revealed an acceptable value of 0.85. The Bartlett's test of sphericity was significant, χ 2 (190) = 1683.14, p < 0.001, indicating that between-item correlations were sufficiently large to perform an EFA.
We then inspected the scree plot, which suggested the presence of two factors to be retained. Also, Kaiser-Guttman's criterion supported the adequacy of a two-factor solution, with only factor 1 and factor 2 showing an eigenvalue greater than one, respectively, of 4.67 and 2.54. We decided to test a two-factor model. The result of the EFA is reported in Table 1, along with the original items and measures of internal consistency. The translated version of the scale is reported in the Supplementary materials, Table 1S.
The evaluated two-factor solution explained in total 38% of the variance. Items clustered in the factors as expected based on the original PHLMS structure. Only item 7 and 15 showed a relatively high complexity, respectively, of 1.92 and 1.81, as they showed cross-loading on both factors. However, they both loaded higher on factor 2 than on factor 1 in the expected direction, so we assigned them to factor 2 accordingly. Then, all the items of the original acceptance scale loaded onto factor 1 and all the items of the original awareness scale loaded onto factor 2. Thus, this analysis suggested the presence of the same two factors as the original model. Following such model, we called factor 1 "acceptance" and factor 2 "awareness". The two factors showed a weak negative correlation in sample 1, with r = − 0.25, p < 0.01.

Internal Consistency
We conducted reliability analysis on the two subscales. Both factors showed good internal reliability, with Cronbach's alpha of 0.80 and 0.87, respectively, for the awareness and acceptance scores. McDonald's omega was 0.81 and 0.87 for the two scales. For the awareness scale, inter-item correlations ranged from 0.19 to 0.37, mean value = 0.29, and for the acceptance scale, inter-item correlations ranged from 0.32 to 0.46, mean value = 0.39. No inter-item correlation fell outside the recommended parameters of 0.15 and 0.50. We also computed item-score correlations. For the awareness scale, they ranged from 0.39 to 0.70, mean value = 0.60, and for the acceptance scale, they ranged from 0.55 to 0.78, mean value = 0.67. These analyses confirmed the good reliability of the two factors of the Italian version of the PHLMS.

Discussion
This first study confirmed the two-factor structure of PHLMS in the proposed Italian translation. This result showed that the participants interpreted our translation of the scale in the same way as the original English version. We found that the factorial structure was exactly as expected, with items loaded onto two factors representing the awareness and acceptance facets of mindfulness.
All items reported high factor loadings, i.e., higher than 0.40, except items 3 and 9, which had a more modest loading of, respectively, 0.34 and 0.36. The English versions of these items are the following: "When I shower, I am aware of how the water is running over my body" and "When I walk outside, I am aware of smells or how the air feels against my face." Thus, it seems that the items referring to awareness of physical sensations reported lower loadings. In line with this, items referring to awareness of emotions and thoughts showed instead higher factor loadings. This pattern of results seems different from the findings reported by Cardaciotto et al. (2008), showing that all items reported similar loadings on the awareness factor. Instead, the factor analysis reported by Zeng et al. (2015) showed a similar configuration, with physically related items as less related to awareness than emotionally oriented items. Moreover, the high-loading items corresponding to the items they proposed to be related to a stricter Vipassana interpretation of awareness, i.e., less related to awareness toward the external world (as in items 3, 5, and 9). We tested the same proposed short form of the scale in the next study, and then we assessed whether the same pattern also applied to our data.
We found that the two factors negatively correlated. This result was in contrast with previous validation studies of the PHLMS (Cardaciotto et al., 2008;Morgan et al., 2020), although it was also found in other studies reporting a negative correlation between the two factors (Tejedor et al., 2014;Zeng et al., 2015). A possible explanation of this finding is that the two scales, although mainly independent, could also be weakly interrelated. Also, study 3 in Cardaciotto et al. (2008) reported a negative correlation between the two scales significant at p < 0.05, with r = − 0.10. This could be further supported by the fact that all acceptance items were reverse-coded, leading to possible misinterpretation, depending on the sample. A more compelling explanation could be found in some latent profile studies conducted with the FFMQ questionnaire. In these analyses, usually a judgmentally observing profile was found (Bravo et al., 2018;Pearson et al., 2015), with a high score on observing but a low score on nonjudging, suggesting that these two factors, roughly corresponding to the awareness and acceptance scales of PHLMS, could be negatively related in the general, non-meditative population. Future studies can be conducted to disentangle this pattern of result, assessing if and under which conditions these two mindfulness skills act as interrelated or orthogonal.

Study 2
In study 2, we performed a confirmatory factor analysis (CFA) on the factor structure that emerged from study 1 to assess whether such structure was further confirmed. We further assessed if both the original two-factor model and the short version of the PHLMS as proposed in Zeng et al. (2015) were reliable instruments in their Italian version.

Participants
Sample 2 was a convenience sample including 203 Italian volunteers from the general population (F = 151, M = 52). The average age in years was 28.60 (SD = 11.13), and the average level of education in years was 15.68 (SD = 3.27). One hundred and three participants were students, 87 were occupied full time (54)

Procedure
The procedure for this study was identical to the one of study 1 described above. Therefore, participants gave their informed consent, then completed the battery of questionnaires and the demographic form.

Measures
Sample 2 completed a range of scales investigating mindfulness and mental health in terms of ill-being, well-being, and emotional regulation.  (Bottesi et al., 2015;Henry & Crawford, 2005); DASS-21 measures three domains of psychopathology, each of which with 7 items, with each item evaluated on a 4-point Likert scale ranging from 0 ("never") to 3 ("often") based on the experienced symptom reported in the item. A total score could be computed as the sum of the three scores, indicating the overall level of psychological distress. In this sample, the DASS-21 revealed excellent internal reliability, with Cronbach's alphas of 0.88, 0.88, 0.89, and 0.94, respectively, for depression, anxiety, stress, and total score. McDonald's omega for the same scales was 0.89, 0.88, 0.89, and 0.94.
We assessed emotional regulation problems with the Difficulties in Emotion Regulation, 18-item version, or DERS-18 (Sighinolfi et al., 2010;Victor & Klonsky, 2016), evaluating five aspects of emotional dysregulation on an inventory of 18 emotional experiences evaluated with a 5-point Likert scale ranging from 1 ("never") to 5 ("very often"). We computed an overall score, with a higher level of problems corresponding to a higher total score. In this sample, DERS-18 showed an excellent Cronbach' alpha value of 0.90 and a McDonald's omega of 0.90.
Lastly, well-being was assessed through three questionnaires: the Satisfaction With Life Scale or SWL (Di Fabio & Gori, 2016), measuring perceived satisfaction in life on 5 items; the Subjective Happiness Scale or SHS (Iani et al., 2014), evaluating the perceived level of happiness on 4 items; and the General Health Questionnaire, 12 item version, or GHQ-12 (Giorgi et al., 2014), measuring a positive factor, i.e., proactivity and self-efficacy, and a negative factor, i.e., depressive symptoms and distress. All these scales reported good reliability, with Cronbach's alphas, respectively, of 0.90, 0.82, 0.84, and 0.82, in line with McDonald's omega for the same scales that were 0.90, 0.84, 0.84, and 0.82.

Data Analyses
We conducted a confirmatory factor analysis (CFA) on the PHLMS items using maximum likelihood estimation with robust (Huber-White) standard errors. To assess whether the two factors were considered as components of an overarching mindfulness construct or as separate components, we tested both a two-factor model as it was identified via EFA in the previous sample and a single-factor model as control analysis. For each model, we controlled for the presence of an acquiescence by modeling it as a common method bias (Podsakoff et al., 2003). Following the suggestions of Welkenhuysen-Gybels et al. (2003), we added a latent factor that loaded onto all items with a fixed loading of 1, and that was orthogonal to the target factors, i.e., its covariance with other factors in the model was fixed to 0. Each model was evaluated using four fit indices: relative chi-square (χ 2 / df), comparative fit index (CFI), root mean square error of approximation (RMSEA), and standardized root mean square residual (SRMR). We assumed as a good fit relative χ 2 ≤ 3.00, CFI ≥ 0.90, RMSEA ≤ 0.05, and SRMR ≤ 0.08 (Hu & Bentler, 1999;Kline, 2011). The internal validity of the individuated factors was then inspected. The same pattern of analysis was applied to the PHSLM short version, as proposed in Zeng et al. (2015).

Confirmatory Factor Analysis
CFA results and statistics about the tested models are reported in Table 2. First, we tested a single-factor model. This model showed a poor fit, relative χ 2 = 7.41, CFI = 0.24, RMSEA = 0.18 (90% confidence interval: 0.17 to 0.19), and SRMR = 0.22, suggesting that it was not an adequate solution. Then, we tested a two-factor model based on the EFA conducted on study 1. This model showed marginally adequate fit indices, relative χ 2 = 2.17, CFI = 0.89, RMSEA = 0.07 (90% confidence interval 0.06 to 0.08), and SRMR = 0.06. However, this model outperformed the onefactor model, Δχ 2 = 937.37, indicating that this solution was better than the former. In this two-factor model, awareness and acceptance did not correlate, r = 0.09, p = 0.34, and the loadings of all items with the two factors were significant.
We then inspected the modification indices to determine the best-fitting model, based on the two-factor solution. The error was correlated for the following pairs of items: items 5 and 9 (both measuring awareness of external sensations), items 4 and 12 (both related to avoidance of thoughts), items 13 and 17 (both referring to the ability to notice inner changes relative to emotional states), items 10 and 20 (both referring to avoiding thoughts), and 11 and 19 (both referring to the ability to notice emotional states of self or others). This model with correlated error showed an overall good fit, relative χ 2 = 1.66, CFI = 0.94, RMSEA = 0.05 (90% confidence interval: 0.04 to 0.07), and SRMR = 0.06. Items 1, 3, and 8 reported relatively small loadings on the respective factor, b < 0.40; however, these paths were significant. Also, in this model, awareness and acceptance did not correlate, b = 0.07, p = 0.43, and all the items' loadings to the two factors were significant. Compared to the model with uncorrelated error terms, this model showed a significant better fit, Δχ 2 = 82.55. Overall, this analysis suggested that the two-factor model with correlated error terms was the best-fitting model (Fig. 1).

Internal Consistency
The awareness scale revealed good internal consistency, with Cronbach's alpha = 0.81, inter-item correlations ranged from 0.20 to 0.38, mean value = 0.30, and item-score correlations ranged from 0.44 to 0.71, mean value = 0.61. The acceptance scale also revealed excellent internal consistency, with Cronbach's alpha = 0.88, inter-item correlations ranged from 0.34 to 0.51, mean value = 0.43, and item-score correlations ranged from 0.58 to 0.80, mean value = 0.70.

Confirmatory Factor Analysis on PHLMS Short Version
In this analysis, we assessed whether a short version (Zeng et al., 2015) of the PHLMS could also be fitted for our Italian translation. To this aim, we conducted a further CFA analysis by means of maximum likelihood estimation with robust standard errors. The model included only the short-version items, that is, items 1, 11, 13, 17, and 19 for the awareness factor and items 2, 6, 14, 16, and 20 for the acceptance factor. This 10-item model (see Fig. 2) showed an excellent fit to the data, relative χ 2 = 1.59, CFI = 0.98, RMSEA = 0.05 (90% confidence interval 0.01 to 0.08), and SRMR = 0.03. As expected, also in this model, the two factors were not correlated, b = 0.11, p = 0.30, and the loadings of all items to the two factors were significant. Only item 1 displayed a small loading on factor "awareness," b = 0.35. Compared to the 20-item model, the short version showed a significantly better fit χ 2 , Δχ 2 = 201.42, but comparable relative χ 2 , Δχ 2 /df = 0.07, which indicated that there is no significant difference between the fit level of these two versions. The two short-version mindfulness scores were highly correlated to their corresponding long-version scores, with r = 0.84 and r = 0.92, respectively, for awareness and acceptance scores. PHLMS short form is shown in the Supplementary materials, Table 2S.

Internal Consistency of the Short Version
The awareness scale revealed good internal consistency, with Cronbach's alpha = 0.77, inter-item correlations ranged from 0.34 to 0.48, mean value = 0.41, and item-score correlations ranged from 0.62 to 0.80, mean value = 0.72. The acceptance scale also revealed good internal consistency, with Cronbach's alpha = 0.81, inter-item correlations ranged Fig. 1 Confirmatory factor analysis for the two-factor model with correlated errors in sample 2. The figure reports the unstandardized coefficients that describe the loadings of the two factors "Awareness" and "Acceptance" and the covariance paths between latent and observed variables. For ease of presentation, error terms and common method bias latent factor are omitted Fig. 2 Confirmatory factor analysis for the short version of the twofactor model in sample 2. The figure reports the unstandardized coefficients that describe the loadings of the two factors "awareness" and "acceptance" and the covariance paths between latent variables. For ease of presentation, error terms and common method bias latent factor are omitted from 0.41 to 0.52, mean value = 0.46, and item-score correlations ranged from 0.68 to 0.82, mean value = 0.75. Overall, the PHLMS short version revealed an acceptable reliability.

Discussion
In study 2, we reported that a two-factor model fitted adequately our data based on the PHLMS. Thus, the Italian translation of the PHLMS showed the same factor structure of the original scale (Cardaciotto et al., 2008), including the two orthogonal factors of awareness and acceptance. This structure seems to be remarkably stable, as it was reported in a number of validation studies in different cultures (Teixeira et al., 2017;Tejedor et al., 2014;Zeng et al., 2015). We then also confirmed this structure in our Italian sample, as already shown in study 1. We further showed that the Italian translation of the PHLMS could also be used in its short form, including only five items per factor. This short version could be useful in case of limited time for the administration of the scale, or if a research focuses on Buddhist meditation, as the short version provided a mindfulness measurement less confusable in the context of Buddhism (Zeng et al., 2015).

Study 3
In this study, we tested the PHLMS for discriminant and convergent validity. To this end, we pooled the two tested samples, thus obtaining a larger sample for analysis.

Data Analyses
First, we assessed if the models for the 20-and the 10-item version of the PHLMS also fitted the large sample (n = 429). Thus, we conducted two CFA analyses with maximum likelihood estimation and robust standard error. We then analyzed their internal consistency. Correlation analyses were then conducted between the PHLMS scores and the measured demographic and psychological variables, such as mindfulness and mental health scores. In particular, convergent validity was assessed by means of correlation with the FFMQ scores.
In the pooled sample, the two subscales of awareness and acceptance from the 20-item version revealed excellent internal consistency, with Cronbach's alpha respectively of 0.80 and 0.87, inter-item mean correlation respectively of 0.30 and 0.41, and item-score mean correlation respectively of 0.61 and 0.69. For the short version, again, both the awareness and acceptance scores revealed good internal consistency, with Cronbach's alpha respectively of 0.78 and 0.79, inter-item mean correlation respectively of 0.41 and 0.43, and item-score mean correlation respectively of 0.73 and 0.74.

Construct Validity: Relationships Between Mindfulness Scores and Other Variables
As the correlations between the normal and short version scores of the PHLMS were very high, all further analyses were based on the 20-item version. Furthermore, since we conducted a large number of comparisons calculated using this sample, we considered significant an alpha level of p < 0.01. All correlation coefficients are reported in Table 3.
Regarding demographic variables, female sex was related to higher awareness and lower acceptance scores; age was positively significantly correlated with both scores; that is, mindfulness skills increased with age; education was positively correlated with acceptance, while the presence of a psychological condition was negatively correlated with it.
Concerning the convergence validity, we correlated the two PHLMS scores with the FFMQ facets for the pooled sample. We found that awareness was mainly related to observe and describe facets of mindfulness as measured through the FFMQ, while acceptance was mainly related to act with awareness and nonjudge. Acceptance was also negatively correlated with the FFMQ observe score.
Lastly, we correlated the two PHLMS scores with measures of mental health and well-being. On sample 1, we collected the SCL-90R GSI score that was negatively correlated with acceptance. On sample 2, we found that awareness was related to an increased subjective happiness rating and decreased emotional regulation's problems, while acceptance positively correlated with all measures of well-being, i.e., SHS and SWLS, and negatively with all measures of ill-being, i.e., DASS subscales and DERS.

Relationships with Psychological Symptoms
We calculated the regression coefficient of the two mindfulness scores on psychological symptoms using regression analyses. Each model included the two PHLMS scores as predictors and a measure of psychological distress as outcome, i.e., SCL-90R GSI for sample 1 and total DASS score for sample 2. As expected, both analyses revealed that acceptance significantly predicted psychological symptoms, with β = − 0.56, p < 0.01, and β = − 0.44, p < 0.01, respectively, for the model on GSI and DASS. In contrast, the awareness score significantly predict neither GSI, β = − 0.10, p = 0.08, nor DASS, β = − 0.01, p = 0.84.

Discussion
In study 3, we showed that the two PHLMS scores have good convergent validity. The awareness score correlated positively with the observe and describe scores of the FFMQ, i.e., to the abilities to notice sensations and feelings and to report such feelings in words. The last result is interesting, as awareness items do not refer directly to the describe capacity but only to the capacity of being aware of experiences. Then, this finding suggests that the describe capacity does not simply rely on the awareness capacity, but could also be part of the same capacity in non-meditators (de Bruin et al., 2012). The acceptance score was related to both the facets of acting with awareness and nonjudging. Interestingly, it correlated weakly with the nonreacting ability, which has been considered as strongly related to the acceptance construct (see Lindsay & Creswell, 2017). We should note that no acceptance items of PHLMS included a nonreactivity statement, such as item 29 of FFMQ "When I have distressing thoughts or images, I am able just to notice them without reacting." Thus, the acceptance measure of PHLMS could slightly differ from the related facets measured in FFMQ, which include a "letting go" aspect (Chadwick et al., 2008) which is not considered in PHLMS. This controversy should be addressed in future studies on mindfulness components or in a revision of the PHLMS.
Our analyses confirmed that acceptance was related to a better psychological condition whereas awareness was not (see Simione et al., 2021). Previous validations of the PHLMS reported the same pattern of results (Cardaciotto et al., 2008;Morgan et al., 2020;Zeng et al., 2015), also in line with theoretical perspectives pointing out the central role of acceptance in explaining the beneficial effects of mindfulness on mental health (Bishop et al., 2004;Lindsay & Creswell, 2017, 2019. Baer et al. (2006) reported that awareness was not related or positively related to psychological symptoms in non-meditators, whereas in meditators, it was negatively related to psychological symptoms. However, a recent validation study of PHLMS on meditators and nonmeditators (Morgan et al., 2020) showed that awareness was not related to any psychological symptoms' outcome in both samples, whereas acceptance was strongly (in non-meditators) or weakly (in meditators) related to reduced symptoms.

Study 4
As our pooled sample included a number of meditators, we conducted this further study in order to clarify the relationship between PHLMS scores and psychological well-being in meditators and non-meditators.

Participants
For this study, we divided the pooled sample of study 3 in two groups, i.e., a group of non-meditators and a group of meditators. The first group included 350 participants who reported no experience with meditation (246 females, 104 males; mean age = 27.91 years, SD = 9.42; mean level of education = 15.05 years, SD = 2.84). The second group included instead 79 participants who reported at least some experience with meditation (55 females, 24 males; mean age = 33.01, SD = 11.35; mean education level = 16.44 years, SD = 3.28).
In the group of meditators, the average meditation experience was 241.66 h, SD = 447.34, from a minimum of 10 h to a maximum of 2500 h of practice, median = 50 h. On average, meditators reported to have spent about 6 h meditating in the last month. The two groups differed in age, t(427) = 4.20, p < 0.01, and education, t(427) = 3.80, p < 0.01, with meditators reporting both higher age and education level. Instead, the two groups did not differ for the proportion of males/ females, χ 2 (1) = 0.01, p = 0.99, and for the presence of psychological conditions, χ 2 (1) = 1.10, p = 0.30.

Data Analyses
First of all, we compared the factor structure of the PHLMS in the two groups. We tested hypotheses about measurement invariance over groups in three successive steps (as indicated in Kline, 2011). Then, we tested in sequence configural invariance, i.e., equal number of factors and factor-indicator correspondence; metric invariance, i.e., unstandardized factor loadings of each indicator are equal across the groups; and strict invariance, i.e., equivalence of residual variances and covariances. We then conducted correlation analysis between meditation expertise and the two PHLMS scores. Furthermore, we conducted a regression analysis with meditation expertise as a regressor and PHLMS scores as outcomes to assess their relationship while controlling for demographic variables, i.e., sex, age, education, and psychological condition. Then, we compared meditators and non-meditators along both PHLMS and FFMQ scores by means of independent samples t tests.
In the pooled sample (sample 1 and sample 2 combined), meditation expertise was significantly correlated with awareness, r = 0.13, p < 0.01, but not with acceptance, r = 0.08, p = 0.08. The result of regression analysis confirmed that meditation expertise predicted the level of awareness, β = 0.10, p < 0.05, but not the level of acceptance, β = 0.05, p = 0.26. Considering only the group of meditators, we obtained the same result: meditation experience predicted the awareness score, β = 0.24, p < 0.05, but not the acceptance score, β = 0.01, p = 0.94. However, as compared to nonmeditators, participants in the group of meditators reported a significantly higher acceptance, t(425) = 2.65, p < 0.01, whereas awareness did not differ between the two groups, t(425) = 1.20, p = 0.20 (see Fig. 3).
We also compared the two groups along the FFMQ dimensions. Meditators reported a higher score than nonmeditators on observing, t(425) = 3.70, p < 0.01, and describing, t(425) = 3.01, p < 0.01, but similar scores on the other facets (see Fig. 4). About the correlation in the meditators' sample, this analysis revealed a significant positive correlation between meditation experience and the mindfulness facets observing, r = 0.22, p < 0.05, describing, r = 0.29, p < 0.01, and nonreacting, r = 0.40, p < 0.01, with a non-significant correlation for the facets acting with awareness, r = 0.13, p = 0.20, and nonjudging, r = 0.17, p = 0.10.

Discussion
This study investigated the relationships of PHLMS-I scores with meditation experience. The results showed that meditators reported a higher score than non-meditators on acceptance, while meditation experience was positively correlated with awareness. Therefore, practicing mindfulness meditation could lead to a higher acceptance capacity, although the degree of awareness better discriminated between meditators with a lower experience and meditators with a higher experience.
The positive correlation between mindfulness awareness and meditation practice has already been reported in the literature (e.g., Baer et al., 2008). Morgan et al. (2020) reported that mindfulness experience was related to both higher awareness and acceptance scores of PHLMS, although the effect on awareness showed a larger effect size. Baer et al. (2006) showed that the observe facet of mindfulness was related to a worse psychological condition in a sample of non-meditators and, in contrast, to a better psychological condition in a sample of meditators, thus suggesting a difference in both quality and quantity of awareness in the two groups. Combined with our findings, these results support the role of awareness as a fundamental mindfulness ability (Bishop et al., 2004).
Even if acceptance was not correlated with meditation expertise, it was higher on average in meditators as compared to non-meditators. Thus, people practicing meditation could have developed the acceptance skill more than non-meditators, although this skill might have grown more slowly in time as compared to awareness, or non-linearly. Another possible explanation is that the acceptance skill developed more slowly, but then became more stable. The results reported in Easterlin and Cardeña (1998) support this hypothesis: the acceptance score was more stable against fluctuation of perceived stress in experienced meditators than in novices. Alternatively, participants who meditated could have a higher acceptance level from the beginning with respect to the non-meditators. The cross-sectional design of this study did not allow disentangling or addressing this point directly. Moreover, in our study as well as in other reported investigations (Baer et al., 2006(Baer et al., , 2008Morgan et al., 2020), the sample of meditators also included participants with a limited meditation experience. Future studies may thus need to involve meditators with a higher meditation experience, e.g., Buddhist monks or very-longterm lay meditators.

General Discussion
We aimed to present and test an Italian translation of the Philadelphia Mindfulness Scale (PHLMS; Cardaciotto et al., 2008). Our main objective was thus to investigate the validity and reliability of the Italian version of the PHLMS, i.e., the PHLMS-I. To this end, we conducted a series of studies on two samples from the general population (total n = 429). Overall, the results support the adaptability of the PHLMS to the Italian context and show that the two-dimensional structure of the PHLMS also held for our Italian sample. In the first study, we presented the translated version of the scale and the results of an exploratory factor analysis. The analysis confirmed the two-dimensional factor structure of the scale, with two main components, i.e., awareness and acceptance. As expected, the awareness scale includes all the items referring to the ability to notice present-moment experiences, while the acceptance scale includes all the items referring to the capability of not judging or avoiding the experiences. Therefore, the factor structure of the Italian version replicated those presented in previous attempts of validation of PHLMS (Cardaciotto et al., 2008;Morgan et al., 2020) or translation (Tejedor et al., 2014;Zeng et al., 2015). The PHLMS-I confirmed a solid factorial structure that also reflects its theoretical grounding (Bishop et al., 2004).
In the second study, we confirmed the factorial structure of the scale by means of a confirmatory factor analysis on a different sample. Again, the model including the two scales of awareness and acceptance showed good fit indexes, while the model with a single mindfulness factor showed a poor fit. Our study confirmed the multi-faceted structure of the mindfulness construct, as already highlighted in the main mindfulness questionnaires (Baer et al., 2004(Baer et al., , 2006Cardaciotto et al., 2008;Walach et al., 2006) and theories (Baer, 2019;Bishop et al., 2004;Lindsay & Creswell, 2017). Furthermore, the PHLMS-I did not show a positive or negative correlation between the two mindfulness components in our samples, suggesting that the two main mindfulness abilities are interconnected but distinct constructs. This also suggests that it could be possible to train either awareness or acceptance, or both (as in Lindsay et al., 2018, and that the effect of awareness and acceptance could be more independent than hypothesized (see Simione et al., 2021).
The differential effects of the two mindfulness components were further shown in successive studies. In study 3, the dimensions of the PHLMS-I were associated with other constructs selected for the evaluation of the convergent validity as expected. In fact, study 3 confirmed that acceptance skill was related to reduced psychological symptoms and distress (Carpenter et al., 2019;Simione et al., 2021). Awareness was instead only involved in increased subjective happiness and reduced emotional dysregulation, with an effect, however, smaller than acceptance on the same variables. This effect was in accordance with that reported by Morgan et al., (2020; see also Simione et al., 2021), with acceptance more than awareness as mainly related with reduced mental problems in both meditators and non-meditators.
Study 4 showed instead how experience with mindfulness meditation was differently related to the two mindfulness components. Awareness was linearly and positively correlated with meditation experience, while acceptance was not. However, acceptance was higher on average in participants with meditation experience than in the others. The result of this study suggests that mindfulness skills could increase differently after a mindfulness training. This hypothesis could be further supported by classical mindfulness teachings and modern mindfulness theory, in which awareness is considered as the base of acceptance (Lindsay & Creswell, 2017). As we did not include a specific sample of long-term meditators, an alternative explanation is that in our sample individuals with higher acceptance were more prone to enter a mindfulness training or just meditate by themselves. A future study could address this issue by involving a larger and more specific sample of meditators.
We also showed that a short form of the PHLMS-I could be used, including only five items for each subscale. This version showed a high positive correlation with the complete form of the scales and the expected pattern of results with other measured variables linked to mental health and well-being. Following Zeng et al. (2015), this short version would be particularly useful in the context of studies on Buddhists, as it was based on a theoretical analysis of the original items. As it also performed well with non-Buddhists and non-meditator participants in our sample, we suggest using this short version of the scale even with such general population participants, in particular in the context of long cross-sectional data collection. Future studies would assess the usefulness of this short version with the general population.

Limitations and Future Research
It has to be noted that our study was not free from limitations. First, we used only self-report measures of both mindfulness and mental health status. For example, correlating mindfulness skills to measures such as heart rate variability (HRV; Kim et al., 2018) or other biomarkers linked to stress (Yang & Jiang, 2020) could provide a better test of the mindfulness capability of reducing distress. Moreover, a comparison between the general population and a clinical group, such as individuals with anxiety or post-traumatic stress disorder, could even lead to a more affordable evaluation of the effectiveness of dispositional mindfulness in reducing mental problems. As a second limitation, in study 4, we compared meditators and nonmeditators but we did not collect data from expert meditators; moreover, our group of meditators was smaller than the group of non-meditators, and it included both shortterm meditators (10 h lifetime) and long-term meditators (2500 h lifetime). We found different results as compared to previous attempts to compare meditators and non-meditators with both PHLMS (Morgan et al., 2020) and FFMQ (Baer et al., 2008;de Bruin et al., 2012), in which meditators reported a higher score than non-meditators for all the investigated mindfulness components. Instead, we found a difference only for the acceptance score. This difference could be due to a sampling bias in our study. More refined comparisons between meditators and non-meditators with the PHLMS-I should be conducted in future. As a third limitation, we tested the short version of the PHLMS-I with the same sample used for the long version, and then its stability should be further assessed in future studies. Lastly, we collected our measures in a cross-sectional design, and thus, causal relationships between PHLMS-I scores and other variables should be further evaluated in future studies, i.e., by conducting meditation training and measuring mindfulness before and after such intervention, as well as its effect on mental health.
Author Contribution LS designed and executed the study, analyzed the data, and wrote the first paper draft. CDB collaborated with the study design and execution. LC collaborated in the writing and editing of the manuscript. AR collaborated with the design of the study and editing of the final manuscript.
Funding Luca Simione has been supported by the grant from BIAL Foundation (Portugal) on the project "Assessing static and dynamic effects of mindfulness meditation on peripersonal space."

Declarations
Ethics Approval All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Ethical approval for this study was granted by the Research Ethics Board of Sapienza, University of Rome.

Conflict of Interest
The authors declare no competing interests.

Informed Consent Informed consent was obtained from all individual participants included in the study.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.