The study of happiness has been one of the most reiterated topics in the history of human thought (Vázquez and Hervás, 2009), nevertheless it wasn’t until the 21st Century that Psychology began placing increased emphasis on it, apart from its focus on symptoms and disorders. Positive approach of Psychology is considered an antidote to psychological (depression) and physical (cardiovascular risk) discomfort and promotes learning, creativity and social cohesion (Kern et al, 2015; Seligman, 2005). It includes the study of individual character strengths that was pioneered by Peterson and Seligman (2004), who proposed the Values in Action Model (VIA) as a theoretical classification of 24 personal strengths grouped into six human virtues. These are: desirable, universal character traits with a clear moral value, and aim to foster wellbeing and success at an individual and a group level (Dahlsgaard et al, 2005). In this model the identification, use and development of strengths has been related to higher levels of wellbeing, engagement as well as work satisfaction and performance (Dubreuil et al, 2014; Forest et al, 2012; Gander et al, 2012; Kong and Ho, 2016; Linley and Harrington, 2006; Littman-Ovadia and Steger, 2010; Peterson et al, 2010).

The main questionnaire used to measure strengths is the Values in Action Inventory of Strengths (VIA-IS, Peterson et al, 2005) that is a 240-item self-report questionnaire and have shown a satisfactory internal consistency (Cronbach‘s Alpha $>$0.70) and stability over time with test-retest correlations above 0.70 (Ruch et al, 2010). The internal structure of the VIA-IS has been studied in different countries via principal component analysis with varimax orthogonal rotation. Nevertheless, up to now, no empirical evidence has been found to support the conceptual grouping of the 24 strengths in 6 virtues proposed by Peterson and Seligman (2004). To explain these heterogeneous results, some authors have focused on the definition of the scales and the items used to cover disparate concepts within a single strength (McGrath, 2014), the fact that there are scales that do not load on the factors that would be expected in theory (Ovejero, 2014; Ruch and Proyer, 2015), and the extension of the scale (Littman-Ovadia and McGrath, 2015). In addition, numerous studies used samples of university students to validate the measurement, adding the restricted range effect which reduces the variability of the results (Azañedo et al, 2014; Brdar and Kashdan, 2010; Khumalo et al, 2008; Macdonald et al, 2008; McGrath, 2014; Ovejero, 2014; Ruch et al, 2010; Singh and Choubisa, 2010).

Another criticism made of the VIA-IS is to assume that a high score on the different strengths is optimal, which implies that a greater presence of strengths would lead to a higher sense of psychological wellbeing, positive emotions, flourishing, higher achievement and better relationships (Peterson and Seligman, 2004). There is data that suggests that an excessive use of social intelligence and humility, and the underuse of enthusiasm, humor, self-regulation and social intelligence are associated with poor functioning and mental health problems, such as social anxiety (Freidlin et al, 2017). Moreover, the reduced use of strengths has been associated with reduced vitality and higher stress, constructs that relate to depression (Wood et al, 2011). These results indicate that it is a balanced use of strengths (scoring at an optimal point) that leads to good psychological functioning (Freidlin et al, 2017). In line with this, it has been found that people with balanced strengths report higher life satisfaction and a greater access to their individual resources (Littman-Ovadia and Steger, 2010).

There are also other questionnaires designed to assess strengths such as the Strength Profile questionnaire (Linley, 2008) and the Clifton Strengths (Asplund et al, 2007). The Strength Profile questionnaire measures 60 strengths and introduces concepts as the energy when strengths are used or the difference between conscious and unconscious strengths. The internal validity scores indicate that the 60 strengths present a low level of correlation, and therefore a relative independence from one another and the internal consistency is an average of 0.82, and the test-retest reliability lies between 0.63 and 0.80 within a week (Linley and Dovey, 2012). This test’s main limitations are: (1) the scant empirical support provided; (2) framing 60 independent strengths, which is difficult to manage, and (3) comprising 120 items, which takes much time to apply.

The Clifton Strengths is a measurement of personal talent that identifies themes in which individuals have a higher potential to build strengths. It is a self-report online questionnaire of 177 ipsative items that measures 34 themes (scales). The internal consistency of the different scales is below 0.70 for 24 of the scales (ranging between 0.54 and 0.76). The test-retest reliability at 8 to 12 weeks is between 0.52 and 0.81 for the 34 scales, and the internal structure of 34 scales has been verified through hierarchical cluster analysis (Schreiner, 2006). This model’s main limitation is not based on the concept of strength, but rather of talent, even though when perfectly used, it converts to a strength. Unlike talent, strengths are more changing over time. Their use can be trained and perfected and have a purpose in and of themselves, with no need to obtain later benefit (Duckworth et al, 2005). Furthermore, this test only shows the user the 5 main strengths, with no possibility of accessing the rest of information. Additionally, the test is mainly contextualized within the business world.

With the aim of integrating the different viewpoints on strengths (Asplund et al, 2007; Linley, 2008; Peterson and Seligman, 2004), we will consider strengths as a combination of adaptative thoughts, emotions and behaviors that can be developed and that a person experiences naturally and to different degrees. Strengths make a person feel more authentic, more vital and with higher energy levels (Csikszentmihalyi and Csikszentmihalyi, 1990; Linley, 2008) when they are displayed in a balanced way and contribute to the flourishing or optimal functioning of people, groups and institutions (Seligman and Csikszentmihalyi, 2000).

As other authors have already stated, some of the consequences related to the use of strengths might be experiencing greater degrees of wellbeing and satisfaction with life (Azañedo et al, 2014; Park et al, 2004; Ruch et al, 2010) and an increased presence of positive emotions (Güsewell and Ruch, 2012). All this can contribute to better adaptation to one’s environment.

The general goal of this research study is to develop and validate a short scale of balanced strengths, grouped by a general factor of optimal functioning. Unlike previous measures such as the VIA-IS, this model considers that it is the balanced score of each strength that will foster optimal psychological functioning.

Confirmatory factor analysis will be used to validate the internal structure of the test by contrasting three alternative models: (1) A model with a general factor of optimal functioning, in which all strengths are grouped, (2) a model with four independent factors (Emotional strengths, Interpersonal strengths, Transcendence strengths and Intellectual strengths) and (3) a hierarchical model with four first order factors (according to model 2), and a general factor of optimal functioning. The previous literature (Asplund et al, 2007; Linley, 2008; Peterson and Seligman, 2004; Seligman, 2005; Seligman, 2011) suggests that the different strengths will cluster in four factors:

- Emotional strengths: being made up of positivity, gratitude, forgiveness, humor, acceptance, managing emotions and emotional awareness.

- Interpersonal strengths: comprised by strengths such as cooperation, charisma, courage, harmony and honesty.

- Transcendence strengths: including the strengths called commitment and appreciation of beauty: quality, innovation, equality, legacy, persistence, universal love and curiosity.

- Intellectual strengths: consisting of the strengths of leadership, vitality, analysis, growth and organization.

The models will be contrasted with two independent samples to obtain cross-validation of the scale. In addition, the factorial invariance of the scale in terms of sex and age will be verified. Finally, we will examine the evidence of convergent validity of the conducted test with other related measurements and criterion validity with regards to a measurement of life satisfaction.

The research study was carried out with two independent samples, employing the second sample to replicate the results of the first sample and obtain cross-validation.

Sample 1. Consisted of 851 participants, 72.7% of which were women, with an average age of 37.66 years (SD = 11.11). At the time of the study 13.3% were unemployed, 10.6% students, 2.8% retirees and 73.3% working population.

Sample 2. Evaluated a total of 909 individuals with a mean age of 44.15 years (SD = 8.27), with 74% being women, 0.3% unemployed, 4.8% students, 0.3% retirees and 94.6% working population.

Balanced Strengths Questionnaire (FORTE). Measures 26 strengths that represent behavioural, emotional and cognitive aspects of individual behaviour (see Appendix 1). Each strength is evaluated with three ipsative items that refer to the use of strengths that the individual must arrange according to the frequency with which the experiences them in life. Scores 5 and 6 on a scale are for Overused, scores 3 and 4 are for a Balanced level, and, scores 1 and 2 are for Underused. Average scores (balanced strengths) fall under greater optimum functioning. For each strength, three phrases are presented and the person assessed must rank them from highest to lowest preference according to how he feels identified with each phrase, assigning scores from 1 to 3. For example, for Acceptance, the three phrases would be the following:

• When nothing can be changed, it is better to let go and not resist/struggle uselessly.

• I am not easily satisfied with the circumstances I live in I feel good when I do what.

• I can to improve and accept what I cannot change.

A person who uses strengths in a balanced way should have assigned the following number sequences: 2, 1, 3 or 2, 3, 1. That is, the most balanced answer is the second phrase. There are items that have more balanced responses in each of the three types of categories: cognitive, behavioural and emotional.

Social Desirability Scale by Crowne and Marlowe (Crowne and Marlowe, 1960), adapted by Ferrando and Chico (2000). Ten items taken from the Spanish adaptation of the scale were included to measure the social desirability in the sample and detect participants that attempt to provide an overly positive image of themselves. The items of highest internal discrimination (between 0.609 and 0.908) were selected, that is, items 6, 12, 13, 15, 16, 19, 20, 26, 30 and 33.

The Satisfaction with Life Scale (SWLS, Diener et al, 1985). The scale measures the cognitive components of subjective wellbeing on a Likert scale of 5. The adaptation by Vázquez et al. (2013) was employed. In the present study Cronbach’s Alpha of this scale was 0.854.

The construction of the strengths scale (FORTE) followed these steps. First, three judges’ experts in Positive Psychology aimed to unify the descriptions found in other existing strengths models (Asplund et al, 2007; Linley, 2008; Peterson and Seligman, 2004; Seligman, 2005; Seligman, 2011) and their corresponding measurements (VIA-IS, Strengths Profile and Clifton Strengths). A total of 26 strengths were identified. Then, each judge provided statements referring to a thought, an emotion and a behavior that characterizes each strength into three possible levels of expression: highly present (overused), balanced and absent (underused). A total of 234 items were proposed: 9 for each of the 26 strengths (9 $\times$ 26 = 234). Finally, of the 9 items for each strength (3 for behavior, 3 for thoughts and another 3 for emotions), 3 were selected: the one evaluating behavior for an overused level, the one evaluating emotion for the balanced level, and the one evaluating thought for the underused level.

Second, a first version of the instrument was administered to 29 experts on Positive Psychology, with the aim of evaluating the appropriateness of the items in measuring the 26 strengths on a conceptual level. The suggestions obtained from this second groups of experts were incorporated, and the final version of the scale was designed.

Finally, the three questionnaires were administered through the online platform http://www.iepp.es. In order to obtain a higher level of commitment and participation in the study a raffle was held, with one tablet being drawn and allotted amongst all participants that filled in the different questionnaires. Participants were recruited through social networks, such as Facebook, by posting messages on the website of the European Institute of Positive Psychology (IEPP). Mails were also sent to people who had requested information on the IEPP website and who had taken a course in the past. The Ethics Committee of the Cardenal Cisneros Center for Higher Education approved the present investigation (approval number: 20-09-2020).

First, the distribution of the items on the FORTE scale were analyzed to assess their normal distribution. Second, the presence of social desirability in the participants’ answers was calculated by correlating the FORTE items with the Social Desirability Scale. These calculations were carried out using SPSS (Version 28.0, IBM Corp., Armonk, NY, USA) (Arbuckle, 2006).

Third, the evidence of internal structure of the FORTE scale was analysed, as well as the factorial invariance in terms of sex and age using the program AMOS (Version 20.0, IBM Corp., Chicago, IL, USA) (Arbuckle, 2006). This type of analysis requires at least three indicators of each latent factor with a minimum of 100 participants and 10 times the number of observed variables (Byrne, 2001). In both samples, there criteria were widely met: Sample 1 with 851/26 $\approx$ 33 participants per indicator, Sample 2 with 909/26 $\approx$ 35 participants per indicator. The procedure of maximum likelihood was used to adjust the models. The goodness of fit statistics used to evaluate the adequacy of the models were: (1) Absolute fit of the model to empirical data with the statistic $\chi$². The null hypothesis, the matrix theoretical and empirical data being equal, is commonly rejected with large samples, so the ratio $\chi$²/df is often used (Bentler and Bonett, 1980), indicating a good fit with values ​​less than 3. Another absolute fit index is Root Mean Square Error of Approximation (RMSEA) (Steiger, 1990), whose values, when below 0.05 indicate good fit. Additionally, another absolute fit index is to examine the standardized residuals matrix: in the event that there are few values below $|$$\pm$1.96$|$, it can be considered that there is little discrepancy between the matrix of observed covariance and the matrix estimated by the model (Byrne, 2001). (2) Incremental fit measures compare the resulting model with the null model. The Normed Fit Index (NFI) (Bentler and Bonett, 1980), Tucker Lewis Index (TLI) (Hu and Bentler, 1999) and Comparative Fit Index (CFI) (Bentler, 1990) are the most frequently used. Values above 0.95 indicate good fit and then the empirical model is significantly different from the null model. (3) Parsimony fit measures evaluate the model fit versus the estimated number, taking into account the complexity of the hypothesized model in the assessment of overall model fit. The Parsimony Goodness of Fit Index (PGFI) (Jöreskog and Sörbom, 1993) and Parsimony Normed Fit Index (PNFI) (James et al, 1982) are more representative coefficients, and values above 0.50 indicate good fit.

Fourth, the reliability of the final scale was calculated. Finally, we verified the evidence of convergent validity of FORTE with the SWLS scale by calculating the Pearson correlation of both scales with SPSS V 28.0 (Arbuckle, 2006).

Table 1 display the results of the descriptive statistics (mean and standard deviation) of both samples, as well as the indicators of asymmetry and kurtosis for the FORTE scale. According to West et al. (1995), the values of asymmetry and kurtosis are not excessively elevated (asymmetry below 2 in absolute value, kurtosis below 7 in absolute value). With these values, the estimation procedure for maximum likelihood can be used, as it is a robust measurement of small deviations in normality. Following this criterion, all items present normal distribution in both samples.

To verify if the FORTE scale is influenced by social desirability, the items from both samples were correlated with the Social Desirability Scale. Following Cohen’s (1992) criteria, a correlation of 0.10 presents a low effect size, 0.30 medium and 0.80 high. The correlations between the different items are very low (r $\le$ 0.28) which indicates that the scales present low social desirability (see Table 1).

Three confirmatory models were compared for both samples separately: a model of a single general factor of optimal functioning, a model of 4 independent factors and a hierarchical model made up of one secondary order factor and 4 first order factors. Table 2 represents the adjustment indices. The hierarchical model clearly offers the best goodness of fit. The majority of indices are within the parameters of good fit, showing better adjustment in Sample 1 that in Sample 2. Fig. 1 displays this model with its corresponding factor loading in both samples. The majority of factor loadings are quite high, specially between the second order and first order factor (between 0.75 and 0.97).

Once the fit of the hierarchical factor structure was confirmed the reliability of the scale was calculated in each sample. Table 1 show the correlations of each item with the total of the scale deleting that item. In Sample 1 none of the items shows correlations below. 20 therefore it would not be advisable to eliminate any of them (Abad et al, 2011). The reliability of the scale according to Cronbach’s alpha ($\alpha$) is quite high: 0.853. In Sample 2 the reliability is adequate but slightly lower ($\alpha$ = 0.755) and some items show correlation values below 0.20. Nevertheless, given that this effect was not observed in Sample 1 and the overall reliability of the scale is above 0.70 it was agreed to not eliminate any item. And the reliabilities for the different subscales were as follows for sample 1 and 2: 0.556 and 0.620 for Emotional Strengths, 0.562 and 0.600 for Transcendental Strengths, 0.732 and 0.758 for Intellectual Strengths, and 0.674 and 0.743 for Interpersonal Strengths. Some subscales have low reliability because the number of items is lower than that of the total scale. In any case, the results obtained with the subscales with low reliability should be interpreted with caution.

The factorial invariance of the scale was study in terms of sex and ages of the participants in Sample 1 and 2 for the hierarchical model. The samples were divided in three age groups by $\pm$1 standard deviation: group 1 up to an age of 32.11 (37.66 – 5.55 = 32.11 years), group 2 with ages between 32.12 and 43.20 years and group 3 containing the ages above 43.21 years (37.66 + 5.55 = 43.21 years). Table 3 displays the results of these contrasts by sex and age. In the case of sex, we can see that even though the $\chi$² results are statistically significant at 5% the incremental value of CFI does not exceed 0.01, therefore we can assume strong metrical invariance between both sexes. The $\chi$² results for age are not statistically significant when comparing the four models that were compared and the CFI increment is below 0.01, therefore strict metrical invariance can be assumed (Cheung and Rensvold, 2002). Hence, the hierarchical model is the same for both sexes and the three age groups.

The evidence of convergent validity was analysed in Sample 1. Table 1 shows the correlations of the different strengths of the FORTE test and the SWLS, as well as the factors extracted through the confirmatory factor analysis. The results are statistically significant for most variables and positive, with values between 0.19 and 0.31 for the latent factors. This indicates the presence of convergent validity evidence with the SWLS.

It will be essential to replicate the results in other populations, e.g., clinical, and increase the evidence of the questionnaire’s validity, and apply it to different cultures in order to evaluate the possible differences in expressing strengths. Furthermore, the FORTE questionnaire could be studied in relation to other strengths measures, such as the VIA-IS (Peterson et al, 2005), the Strength Profile questionnaire (Linley and Dovey, 2012) and the Clifton Strengths (Asplund et al, 2007) to analyze the evidence of convergent validity. Moreover, recent studies have found a positive relationship between strengths and PERMA dimensions (Wagner et al, 2020). These findings open new doors in studying wellbeing that could guide the development of strength-based interventions adapted to individual needs. Finally, remaining pending is a study, based on the degree of balance, on the relationship between strengths and psychological malaise, in order to analyze their relationship with this type of construct.