Psychological flexibility has been conceptualized as the capacity to alter the function of internal experiences by responding flexibly to negative thoughts, feelings, and events, improving well-being, and helping people open up to these experiences, while continuing to make behavioral decisions in the service of what is valuable in their lives (Hayes et al., 2011; Rolffs et al., 2018). It is linked to a wide range of personal and social skills: adaptation to situational demands, modification of dysfunctional behaviors, maintenance of balance between the important domains of life and a conscious, open and committed attitude through behaviors that express congruence with the values held by the individual (Kashdan et al., 2020; Kashdan & Rottenberg, 2010).
The hexaflex model identifies six dimensions of psychological flexibility considered fundamental to well-being (Doorley et al., 2020; Rolffs et al., 2018; Stabbe et al., 2019): Acceptance - implies fully accepting one's own experience, it is the ability to remain in contact with painful intimate experiences without trying to alter their form or frequency; Contact with the Present Moment Awareness - consists of paying attention, intentionally and without judgment, counteracting the tendency to avoid or withdraw from painful thoughts, feelings and circumstances; Self as Context - fostering a perspective of the self as an observer; Defusion - a process that makes it possible to observe products and thought processes; Committed Action - implies acting in the direction of one's own values considered important; Values - directions for intentional action, which can give new meaning, purpose, and vitality to a range of behaviors (Forsyth & Eifert, 2008; Kashdan et al., 2020).
The model proposes, in turn, six dimensions that comprise Psychological Inflexibility (Rolffs et al., 2018): Experiential avoidance - unwillingness to come into contact with unpleasant inner experiences; Lack of contact with the present moment - not paying attention to present experiences; fusion with a particular concept of itself; Fusion - implies the inability to detect the current thought process as different from the product of thinking; Lack of contact with values - difficulty connecting with what is valuable and meaningful in life; Inaction - a tendency to avoid painful events that can lead to losing contact with values, with effective action, and the present moment - (Hayes et al., 2006; Kashdan et al., 2020).
Measuring Psychological Flexibility
The most widely used measures of psychological flexibility have been the Acceptance and Action Questionnaire (AAQ; Hayes et al., 2004), the Acceptance and Action Questionnaire-II (AAQ-II; Bond et al., 2011), and the Avoidance and Fusion Questionnaire for Youth (AFQ-Y; Greco et al., 2008). Nevertheless, these scales consider psychological flexibility as a single dimension, despite the fact that the theory postulates up to twelve different dimensions. Scales have also been developed using some of the individual components, for example, the Conscious Attention Awareness Scale (MAAS; Brown & Ryan, 2003).
However, because these scales have never been integrated into a complete measure with a stable twelve-factor structure, making it directly mapped to the hexaflex model, Rolfs et al., (2018) developed the Multidimensional Inventory of Psychological Flexibility (MPFI), a self-administered 60-item questionnaire that assesses its twelve dimensions. A shorter version of the MPFI also includes 24 items assessing the twelve dimensions of the original scale.
Psychological flexibility has been explored in its different aspects and contexts. Numerous studies indicate that it is related to broad benefits in the physical and psychological functioning of people, and interventions aimed at increasing the level of psychological flexibility have shown valuable results (see Hayes et al., 2006 for a review). Studies carried out with the oncological population consistently show that psychological flexibility is associated with high levels of well-being in cancer patients (Aguirre-Camacho & Moreno-Jiménez, 2017) and acceptance of pain in patients with chronic pain (Flores, 2016). Likewise, young people who present destructive or dysfunctional experiential avoidance (Cobos-Sánchez et al., 2017).
The MPFI demonstrated adequate psychometric properties in different countries such as Canada (Grégoire et al., 2020), China (Lin et al., 2020), France (Grégoire et al., 2020), Italy (Landi et al., 2021), Japan (Lin et al., 2020), Taiwan (Lin et al., 2020) and the United States (Seidler et al., 2020).
Grégoire et al. (2020) translated the MPFI-24 into French and administered to the population of French-speaking students in Canada and France and to the population of French-speaking employees in Canada, Belgium, France, Switzerland and Luxembourg, demonstrating validity and reliability. The results of the confirmatory factor analysis indicated that a second order structure with six flexibility factors and six inflexibility factors constitutes the best fit model. The findings suggest that the MPFI-24 scale is a short, reliable, and valid measure of flexibility and psychological inflexibility (Grégoire et al., 2020).
Seidler et al. (2020) performed a confirmatory factor analysis of the short scale composed of 24 items, supporting the proposed factorial structure for both the short scale and the extensive scale of the MPFI. The authors suggest potential areas of improvement for the elements of the “defusion” dimension. They point out that the pattern of the findings indicates that although flexibility and psychological inflexibility are related, they correspond to different repertoires of responses.
This suggests that flexibility and inflexibility are unlikely to simply belong to a single common dimension, since they show distinct patterns of results, can change independently of one another over time, and only demonstrate weak to modest negative correlations, that is, a low score in psychological flexibility does not necessarily correspond to a high level of inflexibility. In fact, the results with the MPFI suggest that examining each of the 12 processes that make up the scale could offer critical information for the planning and evaluation of patients' treatments (Rogge et al., 2019).
Lin et al. (2020) developed the MPFI translation in three Asian languages (Traditional Mandarin, Simplified Mandarin, and Japanese). The MPFI subscales demonstrated excellent internal consistency in all languages and subscales continued to show convergent patterns of correlation with indices of well-being and psychological distress in all languages.
Landi et al., (2021) demonstrated that the Italian MPFI reproduced the original factorial structure of the MPFI, demonstrating an excellent construct validity, demonstrating a measurement invariance for sex, age and clinical status. The Italian MPFI showed good internal consistency and convergent and concurrent validity.
The Present Study
Given the lack of studies in Spanish-speaking populations this study aims to explore evidence of internal validity and factorial Invariance of the Multidimensional Psychological Flexibility Inventory.
METHOD
Participants
Main Spanish-speaking Sample
An incidental non-probabilistic sample was used that included 866 participants from the general population of Argentina with ages between 18 and 84 years (M = 47,33; SD = 17,26) and of both sexes (Men = 40,9%; Women = 59,1%).
Comparison English-speaking Sample
A convenience sample of 1630 online, English-speaking respondents from the United States was recruited primarily (88.5%) using the ResearchMatch system. Respondent ages ranged from 18 to 85 years old (M = 50.1; SD = 18.7), they were predominantly White (86.6%) and identified as women (77.0) with 20.6% identifying as men and 2.4% identifying as gender minorities.
Measures
Multidimensional Inventory of Psychological Flexibility
Multidimensional Inventory of Psychological Flexibility (MPFI) (Rolffs et al., 2018) is a self-administered 60-item questionnaire with a Likert-type format with six response anchors, 1 being "Never" and 6 "Always". Although the English-speaking comparison sample used the original 6-point response scale, in the survey for the main Spanish-speaking sample, those responses were simplified slightly to a 5 point scale. The MPFI assesses the twelve dimensions of the hexaflex model. Psychological Flexibility inlcudes six dimensions (1) Acceptance (5 items; e.g. "I allowed myself to have negative thoughts and emotions accepting them instead of rejecting them" / “Me permití tener pensamientos y emociones negativas aceptándolos en lugar de rechazarlos”; (2) Contact with the Present Moment Awareness (5 items; e.g. “I paid close attention to what I was thinking and feeling”/ “Presté mucha atención a lo que estaba pensando y sintiendo”); (3) Self as Context (5 items; e.g. “I tried to stay centered even when life collapsed me” / “traté de mantenerme centrado incluso cuando la vida me derrumbó”); (4) Cognitive Defusion (5 items; e.g. "I was able to let negative emotions come and go without being trapped in them" / “pude dejar que las emociones negativas vayan y vengan sin quedar atrapado en ellas”); (5) Contact with Values (5 items; e.g., "I was very connected with what is important to me and to my life" / “estuve muy conectado con lo que es importante para mi y para mi vida”); (6) Committed Action (5 items; e.g., “even in the face of failure I did not stop pursuing what is important to me” / “Incluso ante el fracaso no dejé de perseguir aquello que es importante para mi”).
For its part, Psychological Inflexibility is made up of six components: (7) Experiential avoidance (5 items; e.g. “when I had a bad memory, I tried to distract myself to disappear ” / “cuando tuve un mal recuerdo, traté de distraerme para que desapareciera”); (8) Lack of contact with the present moment (5 items; e.g., “I did most things automatically without paying much attention to what he was doing" / “hice la mayoría de las cosas de forma automática sin prestar mucha atención a lo que estaba haciendo”); (9) Self as Content (5 items; e.g., “I thought that some of my emotions were bad or inappropriate and that I shouldn't have them” / “pensé que algunas de mis emociones eran malas o inapropiadas y que no debería tenerlas”); (10) Cognitive Fusion (5 items; e.g., "my negative thoughts and emotions accompanied me for a long time" / ““mis pensamientos y emociones negativas me acompañaron por mucho tiempo”); (11) Lack of Contact with Values (5 items; e.g., “often my priorities and values were left behind in my daily life” / ““a menudo mis prioridades y valores quedaron postergados en mi vida cotidiana”); (12) Inaction (5 items; e.g., "often negative emotions paralyzed me and prevented me from acting" / “a menudo las emociones negativas me paralizaron y me impidieron actuar”).
The questionnaire of the abbreviated version of the MPFI respects the twelve dimensions of the original scale, and is made up of 12 items that are grouped into the six factors Psychological Flexibility (items 3 and 4 from the "Acceptance" subcale; items 7 and 9 representing "Contact with the Present Moment"; 13 and 14 "Self as Context"; 16 and 17 "Cognitive Defusion"; 21 and 23 "Values"; 26 and 27 "Action Committed ") and six factors of Psychological Inflexibility (31 and 33" Experiential Avoidance; 36 and 39 "Lack of Contact with the Present Moment"; 41 and 45 "I as Content"; 46 and 47 "Fusion"; 51 and 53 " Lack of Contact with the Values ”; 56 and 58“ Inaction”).
Procedure
For the adaptation and validation of the MPFI, the team approach procedure was used, two independent experts were asked to translate the instrument and subsequently compare the translation to locate the problem areas. In a second instance, they discuss both versions and between the two they agree on a single final version (Bolaños-Medina & González-Ruiz, 2013; Douglas & Craig, 2007), from which an attempt was made to maintain the psychological meaning of each term.
The main study was carried out on a population of individuals living in Argentina. The questionnaires were administered through a digital form, guaranteeing the anonymity of the participants. The execution time of the questionnaire took approximately 20 minutes. Participation was voluntary and none of the participants received any financial compensation for collaborating with the study. Simiarly, the survey within the comparison sample was conducted entirely online, took approximately 20-25 minutes, and none of the participants received any financial compensation for collaborating with the study.
Ethics Statement
Main Spanish-speaking Study
This study adheres to ethical standards aimed at protecting the rights of individuals who participate in the reported research, in accordance with the Declaration of Helsinki. Specifically, measures have been taken to ensure the anonymity of the participants and to obtain their informed consent for their involvement in the research. The research project in which the study is framed was, in the first place, subjected to evaluation by an ethics committee. Once approved, the participants were invited to participate voluntarily in the study requesting their informed consent, complying with the codes of ethical conduct established by the National Council for Scientific and Technical Research (CONICET) (Res. D N ° 2857/06). Finally, participants were informed that the results would be used for exclusively academic-scientific purposes in accordance with National Law 25,326 on the protection of personal data.
Comparison English-speaking Sample
The study conducted in the United States was evaluated and approved as a minimal risk study by a university IRB. The study was conducted adhering to ethical guidelines for human subject research and the first webpage of the survey presented participants with an information sheet to obtain informed consent prior to any survey responses being collected. The data was collected in a completely anonymous manner to protect the privacy of respondents and only aggregate findings are presented.
Data analysis
In the adaptation and validation of the MPFI, descriptive and inferential statistics were used with the support of the statistical software SPSS 22 and Lisrel 22 For data analysis, a polychoric correlation matrix was applied, considering the ordinal nature of the items. First, construct validity was analyzed using confirmatory factor analysis. This type of analysis makes it possible to verify whether the data collected in the field conform to the factorial structure of the technique and to the theoretical model proposed by the author (Medrano & Muñoz-Navarro, 2017; Weston & Gore, 2006). For this purpose, the Robust Maximum Likelihood Estimation method was used (Holgado-Tello et al., 2018; Ruiz et al., 2010).
To evaluate the goodness of fit of the model, the χ2, the Incremental or Incremental Adjustment Index or Fit Index (IFI), the Normed Fit Index (NFI), the Non-Normed Adjustment Index or Non-Normed Fit Index (NNFI), the Comparative Fit Index (CFI) and the RMSEA were considered because they tend to be among the most robust indicators to estimate factorial models (Bandalos & Finney, 2010; Medrano & Muñoz-Navarro, 2017). The IFI, NNFI and CFI indices values higher than .90 and the RMSEA lower than .05 figures were considered as indicators of a good fit (Jordán Muiños, 2021; Kline, 2005).
The reliability of the MPFI was subsequently evaluated from the internal consistency analysis taking into account the McDonald's Omega coefficient, following the recommendations of the literature (Zhang & Yuan, 2016). Values above .70 were considered as indicators of a good fit (Martínez Arias et al., 2006; Raykov & Marcoulides, 2011).
Finally, the measurement invariance of the Spanish MPFI in comparison to the original English MPFI was evaluated using multi-group CFA analyses in Mplus (version 7.11). Using procedures outlined by van de Schoot and colleagues (2012), we specified the scale of our latent variables by setting their intercepts (i.e., means) to zero and their variances to 1, as this allowed the factor loadings of all items to be estimated. Consistent with the factor structure of the MPFI, we ran models in which 12 subscales loaded onto two higher order factors.
Our first multi-group model allowed the item loadings to freely vary across the Spanish-speaking and English-speaking samples. Obtaining adequate fit from that model would demonstrate configural invariance as it suggests that both forms of the MPFI have the same basic factor structure. Our second multi-group model then constrained the item loadings to be identical across the two forms of the MPFI. Obtaining adequate fit from that model would demonstrate metric as it suggests that the items load on their respective factors to the same degree across the two forms of the MPFI.
RESULTS
Scale translation.
First, the experts independently translated the items of the original version of the MPFI. Then, these direct translations were reviewed and after discussing the differences between the two versions, an agreement was reached between the experts. To evaluate the preliminary Spanish version of the MPFI, and to ensure its clarity and comprehension, it was administered to a group of 30 respondents from the general population and additional modifications were subsequently introduced.
Confirmatory Factor Analysis.
In order to adapt the MPFI, a confirmatory factor analysis was first performed. In Table 1 it can be seen that the analysis has contributed to confirm that the proposed model fits the data collected in the field (Hair et al., 2009) (see Table 2).
X 2 (gl) | NFI | NNFI | CFI | IFI | RMSEA | |
MPFIlf | 4033.096 (1697) | .97 | .98 | .98 | .98 | .041 |
MPFIsf | 630.945 (239) | .97 | .98 | .98 | .98 | .044 |
Note. MPFIlf= Multidimensional Psychological Flexibility Inventory Long Form; MPFIsf= Multidimensional Psychological Flexibility Inventory Short Form; IFI= Índice de Ajuste Incremental, NFI= Índice de Ajuste Normado, NNFI= Índice de Ajuste No Normado, CFI= Índice de Ajuste Comparado SRMR= Residuo Estandarizado Cuadrático Medio, IFI= Incremental Fit Index; NNFI = Non-Normed Fit Index; CFI= Comparative fit index.
Determination coefficients
Determination coefficients were tested for MPFI long form (see Table 3).
ITEM | R2 | P | ITEM | R2 | P | ITEM | R2 | P |
MPFI1-A | .352 | .594 | MPFI21-V | .664 | .815 | MPFI41-T | .837 | .915 |
MPFI2-A | .535 | .732 | MPFI22-V | .698 | .836 | MPFI42-T | .877 | .937 |
MPFI3-A | .408 | .639 | MPFI23-V | .702 | .838 | MPFI43-T | .574 | .758 |
MPFI4-A | .585 | .765 | MPFI24-V | .609 | .781 | MPFI44-T | .605 | .778 |
MPFI5-A | .407 | .638 | MPFI25-V | .491 | .701 | MPFI45-T | .494 | .703 |
MPFI6-P | .712 | .844 | MPFI26-C | .755 | .869 | MPFI46-F | .736 | .858 |
MPFI7-P | .786 | .887 | MPFI27-C | .806 | .898 | MPFI47-F | .839 | .916 |
MPFI8-P | .743 | .862 | MPFI28-C | .806 | .898 | MPFI48-F | .866 | .931 |
MPFI9-P | .606 | .779 | MPFI29-C | .758 | .871 | MPFI49-F | .715 | .846 |
MPFI10-P | .430 | .656 | MPFI30-C | .649 | .806 | MPFI50-F | .582 | .763 |
MPFI11-S | .627 | .792 | MPFI31-E | .755 | .869 | MPFI51-W | .580 | .762 |
MPFI12-S | .819 | .905 | MPFI32-E | .802 | .896 | MPFI52-W | .724 | .851 |
MPFI13-S | .822 | .907 | MPFI33-E | .813 | .902 | MPFI53-W | .790 | .889 |
MPFI14-S | .685 | .828 | MPFI34-E | .628 | .793 | MPFI54-W | .700 | .837 |
MPFI15-S | .648 | .805 | MPFI35-E | .574 | .758 | MPFI55-W | .737 | .859 |
MPFI16-D | .672 | .820 | MPFI36-L | .595 | .772 | MPFI56-I | .797 | .893 |
MPFI17-D | .669 | .818 | MPFI37-L | .784 | .886 | MPFI57-I | .762 | .873 |
MPFI18-D | .641 | .801 | MPFI38-L | .891 | .944 | MPFI58-I | .826 | .909 |
MPFI19-D | .519 | .721 | MPFI39-L | .894 | .946 | MPFI59-I | .813 | .902 |
MPFI20-D | .627 | .792 | MPFI40-L | .893 | .945 | MPFI60-I | .707 | .841 |
Note: A=Acceptance; P=Present Moment Awareness; S=Self as Context; D=Defusion; V=Values; C=Committed Action; E=Experiential Avoidance; L=Lack of Contact with the Moment Present; T=Self as Content; F=Fusion; W=Lack of Contact with Values I=Inaction
Determinant coefficients were then tested then for MPFI Short Form (see Table 4).
ITEM | R 2 | λ | ITEM | R 2 | λ |
MPFI3-A | .221 | 0.471 | MPFI31-E | .436 | 0.661 |
MPFI4-A | .839 | 0.916 | MPFI33-E | .902 | .950 |
MPFI7-P | .624 | 0.790 | MPFI36-L | .521 | 0.722 |
MPFI9-P | .731 | 0.855 | MPFI39-L | .952 | 0.979 |
MPFI13-S | .674 | 0.821 | MPFI41-T | .546 | 0.739 |
MPFI15-S | .749 | 0.866 | MPFI45-T | .700 | 0.837 |
MPFI16-D | .783 | 0.885 | MPFI46-F | .826 | 0.909 |
MPFI17-D | .654 | 0.809 | MPFI47-F | .813 | 0.902 |
MPFI21-V | .678 | 0.824 | MPFI51-W | .532 | 0.730 |
MPFI23-V | .624 | 0.790 | MPFI53-W | .797 | 0.893 |
MPFI26-C | .774 | 0.880 | MPFI56-I | .767 | 0.876 |
MPFI27-C | .883 | 0.940 | MPFI58-I | .874 | 0.935 |
Note: A=Acceptance; P=Present Moment Awareness; S=Self as Context; D=Defusion; V=Values; C=Committed Action; E=Experiential Avoidance; L=Lack of Contact with the Moment Present; T=Self as Content; F=Fusion; W=Lack of Contact with Values I=Inaction
Internal Consistency
The internal consistency was then tested using McDonald's Omega (Ventura León & Caycho Rodriguez, 2017), obtaining acceptable values (Martínez Arias et al., 2006; Raykov & Marcoulides, 2011) (see Table 5).
Long [ IC 95%] | Short [IC 95%] | X 2 | df | p | ||
A | ꞷ | .807 [.786-.826] | .672 [.626-.712] | 29.109 | 1 | .000 |
P | ꞷ | .904 [.894-.914] | .808 [.781-.832] | 50.206. | 1 | .000 |
S | ꞷ | .928 [.920-.935] | .832 [.808-.853] | 75.536 | 1 | .000 |
D | ꞷ | .893 [.881-.904] | .836 [.813-.856] | 18.737 | 1 | .000 |
V | ꞷ | .896 [.885-.906] | .789 [.759-.815] | 52.340 | 1 | .000 |
C | ꞷ | .939 [.932-.945] | .906 [.893-.918] | 19.220 | 1 | .000 |
E | ꞷ | .926 [.918-.933] | .797 [.768-.822] | 107.666 | 1 | .000 |
L | ꞷ | .956 [.951-.960] | .848 [.827-.867] | 162.918 | 1 | .000 |
T | ꞷ | .912 [.903-.921] | .767 [.734-.796] | 100.152 | 1 | .000 |
F | ꞷ | .948 [.942-.953] | .901 [.887-.913] | 43.208 | 1 | .000 |
W | ꞷ | .923 [.915-.931] | .797 [.768-.822] | 99.258 | 1 | .000 |
I | ꞷ | .947 [.941-.952] | .901 [.887-.913] | 40.645 | 1 | .000 |
Note: A=Acceptance; P=Present Moment Awareness; S=Self as Context; D=Defusion; V=Values; C=Committed Action; E=Experiential Avoidance; L=Lack of Contact with the Moment Present; T=Self as Content; F=Fusion; W=Lack of Contact with Values; I=Inaction
Measurement Invariance
The invariance of the MPFI factor structure was tested across the main Spanish-speaking sample and the English-speaking comparison sample using multi-group CFA models. The configural model allowing the item loadings to freely vary across the two groups demonstrated adequate fit (χ2(3290) = 9758.4; CFI = .946; SRMR = .044; RMSEA = .039, 90% CI: LL = .039, UL = .040). Moreover, the metric model constraining the item loadings to be identical across groups also demonstrated adequate fit (χ2(3350) = 10531.9; CFI = .941; SRMR = .070; RMSEA = .041, 90% CI: LL = .040, UL = .042) with only nominal shifts in the CFI and RMSEA fit indices, well below the thresholds indicating a lack of invariance (see Chen, 2007), thereby supporting strong factorial invariance (i.e., scalar invariance). Taken together, these results suggest that the Spanish translation of the MPFI has a highly similar factor structure to the original English version of the scale. The Spanish MPFI was shown to have both weak (metric) and strong (scalar) factor invariance compared to the English MPFI given to a sample from the United States.
DISCUSSION
The present study has intended to provide a Spanish version of the Multidimensional Psychological Flexibility Inventory (MPFI). The 60 items in the adapted version are similar to those proposed by Rolffs et al. (2018) in the original version. The results of the confirmatory factor analysis allow us to observe an adequate fit of the data to the model proposed by the authors, similar that those reported in Italy (Landi et al., 2021) the United States (Seidler et al., 2020) Canada (Grégoire et al., 2020), China (Lin et al., 2020), France (Grégoire et al., 2020), Japan (Lin et al., 2020), Taiwan (Lin et al., 2020) (RMSEA ≤ .08 ; CFI ≥ .95). Measurement invariance analyses suggested metric factorial invariance of the Spanish MPFI when contrasted with the English MPFI in multigroup analyses.
On the other hand, it is important to note that the values of the estimated parameters are adequate (≥ .50), considering that the factor loadings tend to oscillate around .50 in most studies psychological (Beauducel & Herzberg, 2006).
Regarding internal consistency, McDonald's Omega reached adequate values in both versions -long and short-. When comparing the internal consistency indices of both versions of the MPFI -long vs. short-, statistically significant differences were observed in all dimensions. This means that the homogeneity of each dimension was affected by the reduction of items. Although it is expected to obtain lower consistency indices by reducing the number of items (Tavakol & Dennick, 2011), it is important to ensure that item selection in the design of a short version does not affect its homogeneity (Robinson, 2018). The item pool of the long version could be reviewed in the future to make a new selection of items that will result in a new short version that guarantees equivalent homogeneity indices.
It should be noted, regarding its limitations, that the results of the study reported refer exclusively to the factorial structure -factor validity- and the internal consistency -reliability- of the MPFI, which only implies a first approach in the task of adapting the instrument to the local environment. In this sense, future research should continue exploring the psychometric properties of the MPFI in Argentina. It would be convenient to evaluate the convergent validity of the MPFI by studying the correlations with related psychological constructs in the specialized literature, such as the Acceptance and Action Questionnaire-II (AAQ-II) and the Comprehensive Assessment of Acceptance and Commitment Therapy (CompACT).