Validation of the French Versions of the Flourish Index and the Secure Flourish Index

Abstract Individual well-being is generally thought of from the perspective of the risk factors which might compromise it. Assuming that positive dimensions of well-being are also worth considering, VanderWeele and Wȩziak-Białowolska et al. developed and tested the Flourish Index (FI) and the Secure Flourish Index (SFI). These 10- and 12-item questionnaires, respectively, measure 5 and 6 dimensions of human flourishing. This article presents the translation of these indexes into French (as FI–F and SFI–F) and the assessment of their psychometric properties on a sample of 2,376 French-speaking respondents. The validity and reliability indicators used by Wȩziak-Białowolska et al. were calculated for our French indexes and found to be very close to theirs. Item groupings were confirmed using correlation and factor analyses. The hierarchical structure of both French indexes matched the English indexes’ second-order and bi-factor model analyses. Reliability and construct validity were good. The brief time required to complete the FI–F or SFI–F and their excellent psychometric properties make them very promising tools for research on the general well-being of Francophone populations.


Introduction
Several academic disciplines (e.g., psychology, medicine, public health, management) have combined the notions of health and well-being.However, those concepts are usually considered in either a fairly narrow sense or in investigations of what might be detrimental to them (i.e. from a pathological perspective) rather than what might contribute positively to them (i.e. from a salutogenic perspective; VanderWeele, 2017).In the last 20 years, the positive psychology movement (Seligman & Csikszentmihalyi, 2000;Shankland, 2014;Al Taher, 2015) and the "rediscovery" of the works of Antonovsky (Mittelmark & Bauer, 2017) have encouraged researchers to broaden the range of dimensions associated with well-being and to explore the more positive ones by using, for example, the concept of human flourishing (VanderWeele et al., 2019).

Measuring well-being
Work on well-being carried out by Diener et al. (2010Diener et al. ( ), (2015) ) led to the development of two measurement tools.The first, the 8-item Flourishing Scale (FS), evaluates psychosocial well-being (Diener et al., 2010).The second, the 12-item Scale of Positive and Negative Experience (SPANE), measures subjective feelings of positive well-being (6 items) and negative well-being (6 items; Diener et al., 2010).The FS evaluates social relationships, feelings of leading a purposeful life, participation in activities with others and one's interest in one's own activities, the search for self-respect and optimism, and feelings of competence and capability (Diener et al., 2010).However, the FS does not address physical and mental health or financial and material stability (VanderWeele, 2017).The SPANE is centred on emotions and feelings and the frequencies with which they are experienced, whether positively or negatively (e.g., joy, sadness; Diener et al., 2010).Likewise, this scale is also unable to capture all of the dimensions of general well-being (VanderWeele, 2017).
Based on an empirical review of the literature examining longitudinal, experimental and quasiexperimental research, VanderWeele (2017) found that the concept of "psychological wellbeing"-frequently used in numerous research or clinical disciplines-was unable to affirm whether a person was flourishing.A more comprehensive concept was necessary.VanderWeele proposed the concept of "complete human well-being" encompassing domains other than human psychological well-being.In VanderWheele's opinion, the two necessary conditions for a domain to be considered part of complete human well-being were that it could (a) be seen as an end in itself (meaning that the domain can be a life goal in itself rather than a way to achieve certain other goals) and (b) be almost universally desired (meaning that most human beings, regardless of their culture or personal traits, would perceive this domain as desirable).The domains VanderWeele identified through these two postulates were: (D1) finding joy and satisfaction in life; (D2) mental and physical health; (D3) leading a purposeful and meaningful life (reflecting the subjective value attached to life); (D4) adopting good character and virtues, seen as behaviour matching the four cardinal Platonic virtues of wisdom, justice, courage and moderation; (D5) close social relationships (in sufficient quantity and quality); and (D6), added by VanderWeele himself, financial and material stability.Indeed, he saw this sixth domain as necessary for supporting domains one to five through time (VanderWeele, 2017).Built on this platform, human flourishing can be defined as "a state in which all aspects of a person's life are good" (VanderWeele, 2017, p. 8149).
Dimensions D1 to D5 make up the Flourish Index (FI); when D6 is added to them, we obtain the Secure Flourish Index (SFI; Wȩziak-Białowolska et al., 2017).The Secure Flourish Index is currently the only instrument capable of measuring general well-being or human flourishing, as defined above, using a single, rapidly completed scale composed of 12 questions.Indeed, other scales often leave out certain domains (e.g., physical health) or measure them indirectly using several subscales, sometimes significantly lengthening the time needed to complete them (Wȩziak-Białowolska et al., 2019a).Finally, VanderWeele (2017) determined that there was a multitude of very well-known measures of human well-being in each of the domains.Their degree of precision was often proportional to the number of questions asked.Thus, VanderWeele developed a short scale measuring what he considered to be the essential elements of the five (FI) or six (SFI) domains.To do this, he borrowed from existing, frequently-used instruments with robust psychometric properties and selected two questions per domain.The questions for domain D4 -"Character and Virtue"-are the exceptions, as he created these himself (VanderWeele, 2017).

Psychometric properties of the English-language Flourish Index
Validity, reliability and applicability were tested in different studies (Wȩziak-Białowolska et al., 2017, 2019b, 2019a).First, those authors tested whether empirical data justified how items were grouped into the domains of flourishing by assessing a correlation matrix and using exploratory factor analysis (EFA).Second, they analysed whether the data reflected the hierarchical structure of the indexes (items grouped into domains, domains grouped into the indexes) using confirmatory factor analysis (CFA) models.They chose a bi-factor model to confirm that grouping items into the five or six domains of FI and SFI, respectively, was relevant.Third, they checked whether both indexes could be calculated as a reflection of a single construct or should be separated into subscales.Fourth, correlations with existing measures of good health allowed them to test convergent validity, and measures of health behaviours were used to test discriminant validity (Campbell & Fiske, 1959).Fifth, they assessed the FI and SFI's properties of measurement invariance using multi-group confirmatory factor analysis (MG-CFA).
Because of our theoretical interest in the concepts of well-being, flourishing and salutogenic approaches, together with the empirical potential for applying the easy-to-understand and rapidly completed FI and SFI, and their robust psychometric properties in English, we decided to translate them into French and validate them.

Participants
In February and 2 March 2020, 376 French-speaking participants answered the FI-F plus the two additional items on the SFI-F as a part of a large-scale health study (manuscript in preparation).All were undergraduate students enrolled in different faculties of the University of Applied Sciences and Arts Western Switzerland (HES-SO).Three respondents' answers were removed from the dataset for double entries, and two for withdrawal of consent.
The mean participant age was 23.03 years old (sd = 4.06), 67.8% were women, 30.4% were men and 1.9% identified as non-binary or preferred to self-describe.More than half lived with their parents (54.5%), whereas others lived with roommates (17.9%), as part of a couple (11.8%) or alone (11.6%).The largest proportion was studying in the field of health (34.1%), with others enrolled in engineering and architecture (21.6%), economics and business (18.6%),social work (17.4%), design and visual arts (4.5%) and music and the performing arts (3.7%).Moreover, 51.5% earned money through a job held in parallel to their studies.

Flourish Index
This index is composed of ten questions-two for each domain.Each item is measured on an 11point scale running from 0-10: the more favourable the response, the higher the score.Domainspecific scores are obtained by averaging the scores for their two questions.The overall FI mean score is calculated by averaging the score for each domain; thus, the final score ranges from 0 (minimum) to 10.00 (maximum).Respondents with the highest scores are those who perceive themselves to be as "flourishing" as much as possible.

Secure Flourish Index
The two supplementary questions for the sixth domain of "Financial and Material Stability" complete the ten from the FI and give us the SFI.The two items allow us to measure the FI's sustainability over time because they examine two elements deemed necessary to its maintenance.The overall mean SFI score is calculated in the same manner as the FI.

Translation
We followed the ten steps of the translation process described by Wild et al. (2005; see, Table 1).The original English-language items and their French translation are available in Annex 1.

Ethics
This research was approved by the Human Research Ethics Committee of the Canton of Vaud (CER-VD, protocol no.2019-01379).All participants gave their informed consent to take part in the research.

Analyses
To ensure a comparable analysis of the items and their internal structure, we followed Wȩziak-Białowolska et al.'s (2017) validation procedure for the FI and SFI questionnaires.

Exploratory factor analyses
FI is supposed to reflect five dimensions (Happiness and Life Satisfaction, Mental and Physical Health, Meaning and Purpose, Character and Virtue, Close Social Relationships) and SFI six (the first five, plus Financial and Material Stability).We thus performed an Exploratory Factor Analysis (EFA) to test the questionnaires' dimensional structures.As the domains of Flourishing are correlated (Keyes, 2005), Wȩziak-Białowolska et al. ( 2017) advocated using an oblimin rotation in these analyses.We used the Principal Factor Method for our factor analyses.Invite the developer to be part of the process.
The principal investigator (PI) organised this process in collaboration with a research trainee and contacted Prof.
VanderWeele, inviting him to participate.He gave his permission for the translation.

2.Forward translation
Have more than one native speaker of the target language (French) translate the questionnaire.
Three members of the research team translated the questionnaire independently of one another.

Reconciliation
Discuss any differences between the existing translations.
Decide on a single forward translation.
The introduction and the 12 questions were thoroughly and thoughtfully discussed.The researchers could not reach a consensus on the precise meaning of the scale's sixth item ("purpose in life") or on how to translate this properly.Thus, the trio of translators (1) asked the developer to clarify the term's meanings and connotations in English, and (2) examined existing literature on this concept to understand the etymology, meanings and possible connotations of "purpose" with regard to the concept of "purpose in life".They were subsequently able to propose an item question best adapted to the context and settings in French.When added to the original author's explanations, this combined strategy helped the researchers establish a consensus.

Back translation
Translate the reconciled translation back into the source language (English).
One back translation was carried out by a native-English speaker, a professional translator and editor with whom the research team is used to working.

Back translation review and 6. Harmonisation
Review the back translation against the source language to ensure the translation's conceptual equivalence.
The three researchers examined all the discrepancies between the original scale and its French translation.Any statements or terms that were problematic were examined and fixed.No clarification was needed from the developer.As there were no other French translations of the FI or SFI, the harmonisation step, as originally described, was unnecessary.

Cognitive debriefing
Test the translated scale with members of the target population.
The scale was included in the main questionnaire's pre-test.French-speaking students and research trainees who had not participated in the translation process were asked to test the FI-F and the SFI-F.

Review of cognitive debriefing results and finalisation
Review the remarks made or questions asked by the pre-test population.
No remarks were made on the FI and SFI scales.No re-wording was necessary.

Proofreading
Check for minor errors.
The PI checked the final questionnaire with particular attention to the FI and SFI questions.
10. Final report Describe the procedure followed.
A research trainee wrote a full report on the translation process under the PI's supervision.

Confirmatory factor analyses
As per Wȩziak-Białowolska et al. ( 2017), we tested three models via a Confirmatory Factor Analysis (CFA).First, the unidimensional model assumes that a single latent construct underlies all ten items in the FI and all twelve in the SFI (Figure 1a).The second-order model assumes that a general latent construct underlies all five primary dimensions (six in the SFI).Each of these dimensions reflects one domain of flourishing that, in turn, influences its two respective items (Figure 1b).Finally, the bi-factor model assumes that both the general construct and the domainspecific constructs directly influence the items (Figure 1c).All the latent variables' covariances were constrained to be orthogonal, and their variances were fixed to unity.

Reliability
Like Wȩziak-Białowolska et al. ( 2017), we assessed the questionnaires' reliability by computing their Explained Common Variances (ECV), their Percentage of Uncontaminated Correlations (PUC, a measure of the construct's unidimensionality) and their omega hierarchical coefficients (ω h , the coefficient that reflects the amount of variance explained by a general construct while considering the model's multidimensional nature; Rodriguez et al., 2016;Zinbarg et al., 2005).

Construct validity
The Pearson correlation coefficient was used to evaluate our indexes' convergent validity with two variables: the first item of the World Health Organization Quality of Life questionnaire (WHOQOL-BREF Skevington et al., 2004), "How would you rate your quality of life?" was answered from 1-Very poor to 5-Very good; and the total score of the Perceived Stress Scale (PSS-14, Jiang et al., 2017;Lesage et al., 2012), which asks individuals to describe how often they have been exposed to stressors during the last month.We expected FI-F and SFI-F scores to be strongly associated with these measures-positively with the former and negatively with the latter.
Discriminant validity was evaluated through the Pearson's correlation of two questionnaire variables that should not necessarily be associated with flourishing.First, participants' willingness to retake the study one year from then ("Do you agree to be contacted in a year's time to complete the same questionnaire and thus explore the evolution of your situation?/Yes-No").The second variable was whether participants had a means of earning money in parallel to their studies ("Do you have any incidental gainful activity?/Yes-No").The correlation matrix in Table 3 shows that all the items were positively correlated and generally strongly associated (r > .45)with the other item in their dimension of measurement, except for items D4.1 and D4.2 (Character and Virtue, r = .34).It is of note that items D2.2 and D3.1 were highly correlated with each other (r = .62)and with items D1.1 (r = .72,r = .63,respectively) and D1.2 (r = .73,r = .60,respectively).Thus, overall, the internal structure of our items corresponded to what was expected.Items D1.1 and D1.2ʹs dominant role in flourishing was reported previously in validation studies for the FI and the SFI (Wȩziak-Białowolska et al., 2019b, 2019a).These items concern overall satisfaction with life and general happiness, respectively; it is thus not surprising that they are strongly associated with the other items.

Exploratory factor analyses
EFA revealed that the items could indeed be separated into five dimensions (six for the SFI-F; Tables 4 and 5).Yet, items D2.1 and D2.2 (Mental and Physical Health) were most closely associated with the same factor as items D1.1 and D1.2 (Happiness and Life Satisfaction), whereas items D4.1 (acting to promote good in all circumstances) and D4.2 (ability to give up some happiness now for greater happiness later) did not correlate with the same factor.
In Węziak-Białowolska et al.'s study (Wȩziak-Białowolska et al., 2017, 2019a), items D1.1 and D1.2 and D3.1 and D3.2 (Meaning and Purpose) correlated on the same factor.These findings confirmed the preceding section concerning items D1.1 and D1.2.Their very general nature makes it difficult to statistically discriminate them from other items.The same concern was observed for dimension 1 (combining those items), which is difficult to distinguish statistically from dimensions 2 and 3.However, this is not problematic if we consider that one general factor underpins the different domains measured in the questionnaire-a supposition that Węziak-Białowolska et al. verified using CFA.

Confirmatory factor analyses
We tested the unidimensional, second-order and bi-factor models (Figures 1a, 1b and 1c), described earlier, on the FI-F and the SFI-F, and evaluated three fit indices: the root mean square error of approximation (RMSEA), the standardised root mean square residual (SRMR) and the comparative fit index (CFI; Table 6).According to Schermelleh-Engel et al. (2003), to remain within acceptable limits, the RMSEA should not exceed 0.08 and the SRMR should not exceed 0.10.Moreover, the CFI should reach at least 0.90 to be considered acceptable (Kline, 2016).
The unidimensional model was unsatisfactory for both the FI-F and the SFI-F.On the contrary, the second-order and bi-factor models had very good fit indices, with a very slight advantage in favour of the former, although this has no practical implications.These results echoed Wȩziak-Białowolska et al.'s (2017) and indicate that each of the domains of the FI-F and SFI-F reflect different facets of flourishing that cannot be attributed to a unique source.

Reliability
The unidimensionality of the data was assessed using the ECV, the PUC and the ω h .According to Reise et al. (2013), for the data's multidimensionality not to be significant, the PUC must be lower than 80%, the ECV higher than 60% and ω h higher than 0.7.Here, our findings differed slightly from those of Wȩziak-Białowolska et al. (2017), who reported that the FI-F and SFI-F could be conceptualised as unidimensional.In our data, the ECV indicated that the general factor  0 = Worry all of the time, 10 = Do not ever worry 0.22 0.22 0.21 0.24 0.18 0.17 0.11 0.10 0.15 0.16 0.71 1 accounted for 65.5% of the variance in the items of the FI-F and 50.2% of the items in the SFI-F, whereas the PUC accounted for 88.9% and 90.9%, respectively.Finally, ω h levels were 0.84 for the FI-F and 0.79 for the SFI-F.These results suggested that the FI-F might be interpreted as unidimensional, but that the SFI-F was better when separated into subscales.This makes sense with regard to the low correlation between items D6.1 and D6.2 (Financial and Material Stability) and the others.This is also in line with that domains' definition.Indeed, whereas domains 1 to 5 represent facets of flourishing, domain 6 (Financial and Material Stability) reflects their stability over time rather than their level (see below for further discussion).

Construct validity
As expected, the FI-F and SFI-F were strongly associated with item 1 from the WHOQOL-BREF about evaluating one's quality of life (r = .52,p < .001for FI-F and r = .55,p < .001for SFI-F) and negatively associated with the PSS-14 score, assessing perceived stress (r = −.65,p < .001for FI-F and r = −.66,p < .001for SFI-F).It was only weakly associated with having a job to support one's studies (r = .07,p = .001for FI-F and r = .04,p = .085for SFI-F) and was not associated with the willingness to participate in the study's second phase (r = −.02,p = .430for FI-F and r = −.03,p = .179for SFI-F).This indicates that the FI-F and SFI-F do indeed measure a construct that is close to well-being and are not influenced by variables that are irrelevant to it.

General discussion
Over the past 40 years, a new approach to health theory, research and practice has been developed to focus on the resources individuals can use to improve their own health: salutogenesis (Mittelmark & Bauer, 2022).Although health has often been conceptualized as the binary opposite of illness, some approaches advocate conceptualizing health as a continuum (Antonovsky, 1987;Mittelmark & Bauer, 2022), which provides researchers with greater granularity when studying concepts such as well-being.Indeed, by shifting the focus to the positive end of the continuum, the FI and SFI enable us to assess not only whether individuals are healthy or sick but also how well they are.Indeed, one could be healthy yet not flourishing.Both scales also allow us to assess which of their dimensions contribute positively to individuals' well-being, as per VanderWeele's theoretical proposition.As French is a widely-used language, the translation and validation of the FI and SFI will enable the promotion, dissemination and use of these concepts in several new countries, thus generating more data for the international comparisons begun by Wȩziak-Białowolska et al. (2019b).
Our results showed that the FI-F can be used for both its overall score and the sub-scores it produces in each of its five dimensions.The sixth dimension, added to create the SFI-F-Material and Financial Stability, i.e. the ability to sustain human flourishing over time-should be used as an indicator of the stability of the scores in the other dimensions rather than be included in the overall score.This finding is somewhat different from the conclusions drawn by Wȩziak-Białowolska et al. (2017), who advocated using both the FI and the SFI as unidimensional scales.However, the respective study populations may have perceived the sixth dimension differently: Węziak-Białowolska et al.'s middle-aged employees in America and the present study's higher education students in Switzerland.Indeed, Swiss students' characteristics and life conditions (Fischer et al., 2021) probably differed significantly from those of the population primarily investigated by Wȩziak-Białowolska et al. (2017).
Despite our scales' very similar internal structures, our results also differed from Wȩziak-Białowolska et al. 's (2017) in that the responses observed for most items were approximatively one point lower.The exceptions were contentedness with friendships and relationships and worrying about safety, food or housing, for which our participants responded slightly higher than Węziak-Białowolska et al.'s.Again, this is not surprising considering the differences between the studies'populations: young populations, such as students, are living through a relatively unstable phase of their life during which their identity and relationships are still developing and their future can feel uncertain (Dunkel et al., 2011).This could explain their lower scores on most flourishing items.However, items on which our participants scored higher could be explained by the differences between US workers vs Swiss students).Swiss students' social relationships are generally well developed (Fischer et al., 2021), and the majority have easy access to many campus activities and services that contribute to developing their personal networks.Furthermore, Swiss students often live with and financially rely on their parents (Fischer et al., 2021).Finally, university tuition fees are quite modest in Switzerland compared to other countries (about CHF 1,500 per year).

Strengths and limitations
The present research's main strength was to have replicated Wȩziak-Białowolska et al.'s (2017) validation study findings to a satisfactory level of reliability on a large sample.On the other hand, the fact that this was not a general population sample could call into question that validation's generalisability.Nevertheless, obtaining results comparable to those of the indexes' original authors on a different population is a strong argument in favour of using the FI and SFI scores and their translated French versions.
Another potential limitation was the period during which our validation occurred.Data collection took place in February 2020, just before the first wave of the COVID-19 pandemic struck Switzerland.Although Switzerland's government had yet to institute any disease prevention measures, feelings of worry and incertitude may have been growing and may have weighed on participants' answers.

Conclusion
The present article described our work to translate into French and validate the Flourish Index and Secure Flourish Index developed by VanderWeele (2017) and Wȩziak-Białowolska et al. ( 2017).We closely followed the validation procedure applied by those authors so that our results would be comparable with theirs.Our results were in close agreement with Węziak-Białowolska et al.'s, which indicated that our translation is valid and reliable: it can be used for French-speaking populations and to conduct studies comparing the Flourish index score of people speaking different languages.Measuring human flourishing can help approach well-being from a salutogenic perspective-one oriented toward factors that sustain it.The scales could prove to be precious tools for assessing the effectiveness of interventions to improve individuals' well-being.Finally, because the FI-F or SFI-F are simple and rapidly completed instruments, they will be easy to integrate into studies using multiple scales.You are free to: Share -copy and redistribute the material in any medium or format.Adapt -remix, transform, and build upon the material for any purpose, even commercially.The licensor cannot revoke these freedoms as long as you follow the license terms.
Under the following terms: Attribution -You must give appropriate credit, provide a link to the license, and indicate if changes were made.You may do so in any reasonable manner, but not in any way that suggests the licensor endorses you or your use.

No additional restrictions
You may not apply legal terms or technological measures that legally restrict others from doing anything the license permits.

Figure
Figure 1.Models tested for the FI-F and SFI-F: (a) Unidimensional model, (b) Second-order model, (c) Bifactor model.Dotted lines are the model parameters fixed at value 1. "Gen" stands for a general factor.Remark: The FI-F models present the same structure without dimension 6 and items D6.1 and D6.2.
In general, how happy or unhappy do you usually feel?In general, how would you rate your physical health?How would you rate your overall mental health?Overall, to what extent do you feel the things you do in your life are worthwhile?0 = Not at all worthwhile, 10 = Completely worthwhile I always act to promote good in all circumstances, even in difficult and challenging situations.0 = Not true of me, 10 = Completely true of me I am always able to give up some happiness now for greater happiness later.0 = Not true of me, 10 = Completely true of me My relationships are as satisfying as I would want them to be.How often do you worry about being able to meet normal monthly living expenses?0 = Worry all of the time, 10 = Do not ever worry How often do you worry about safety, food or housing?
and dialog with, expert editors and editorial boards • Retention of full copyright of your article • Guaranteed legacy preservation of your article • Discounts and waivers for authors in developing regions Submit your manuscript to a Cogent OA journal at www.CogentOA.com

Table 3 . Correlation matrix of FI-F and SFI-F items
Overall, how satisfied are you with life as a whole these days?