The scale was administered to 1507 university students from four cities in Mexico: Toluca, Estado de México; Ciudad Juárez, Chihuahua; Ciudad Victoria, Tamaulipas and Zacatecas, Zacatecas; with a mean age of 20.85 (S.D. = 3.41).The sample was divided into two groups; group one (n = 754) was used for the exploratory factor analysis and group two (n = 753) was used for the confirmatory factor analysis. There were no statistically significant differences between the two groups in age, gender, marital status, and city.

Demographics

Participants were asked about their age, gender, marital status, and city they lived in (see Table 2).

Psychological Well-Being Scales (PWBS)

The original instrument was developed by Ryff (1989b) and it consisted of six factor with 20 items per factor. van Dierendonck (2004) analyzed a shorter version of the scale with 39 items that was validated in a sample from Spain (Díaz et al., 2006) . This last version of the scale was validated in this study with a structure of three factors: Self-Actualization Tendency (19 items), Interpersonal and Intrapersonal Relationships (eight items), and Autonomy (four items). The internal consistencies, measured with Cronbach’s alpha, for the total of the scale and factors ranged from .63 to .92 in group one, and from .64 to .92 in group two (see Table 3).

Note: EFA = exploratory factor analysis; CFA = confirmatory factor analysis.

University students were approached in four universities from different cities in Mexico. They were asked to participate, then they were given the consent form that explained the study, benefits and risks, rights and they were given the opportunity to ask questions. They were given a set of scales that took approximately 20 minutes to complete. Once they turned in the scales, they were explained the study with more details. Participants’ data was captured and analyzed in SPSS 23 and AMOS 23. The sample was divided into two groups to perform an exploratory factor analysis (n = 754) and a confirmatory factor analysis (n = 753). The internal reliability was analyzed for the factors and the total scale in both groups.

The first step was to explore the factor structure of the PWBS. Exploratory factor analyses were performed with the 39 items, in a sample size of 754, using the unweighted least squares method with a direct oblimin rotation. Several factor structures with different number of factors were analyzed. For the first result, there were seven eigenvalues greater than one, so the first factor structure analyzed included seven factors. Even though the seven factors were analyzed, many items had shared loadings and there were factors with only two items. Other exploratory factor analyses were performed forcing the structure to six, five, four, three, two and one factors. After revising all the structures, the best fit was the analysis with three factors (see Table 4). The KMO index was .94 and the Barlett’s sphericity test (Snedecor & Cochran, 1989) indicated an adequate and normally distributed sample (p < .01). For each factor, an item was retained if the highest loading had a value of .30 or greater and if the difference of the two highest loadings was at least .10;items were excluded if the highest loading of an item was less than .30 or if an item had shared factor loadings in two or more factors.

The three factor solution had 31 items with unique factor loadings from .42 to .75 (see Table 5), six items were excluded because of shared factor loadings

Note: Items with bolded loadings are items retained in that factor; F1 = Personal Orientation; F2 = Positive Relationships; F3 = Autonomy.

(items 3, 5, 13, 21, 27, and 36) and two items were excluded because of low factor loading values (items 10 and 34). The first factor explained 26.96% of the total variance, and it was composed of 19 items with factor loadings ranging from .45 to .75. According to the theme of these items, this factor was named “Self-Actualization Tendency”. An item for this factor is “In general, I feel confident and positive about myself”. The second factor explained 8.31% of the total variance and it consisted of eight items with a range of factor loadings from .49 to .58. The theme for these items was “Interpersonal and Intrapersonal Relationships” and an item from this factor was “I often feel lonely because I have few close friends with whom to share my concerns”. The third factor explained 4.05% of the total variance and it included four items with factor loading values between .42 to .67; the theme for these items was “Autonomy” and an item from this factor was “I tend to worry about what other people think of me”.

The Cronbach’s alpha was used to evaluate the internal reliability of the sum of all of the items of the scale and for each factor (see Table 3). The internal reliability for all of the items was α = .77, for the first factor was α = .92, for the second factor was α =.79 and for the third factor was α = .63.

A confirmatory factor analysis, with a sample of 753 participants, was used to cross validate the factor structure of the PWBS. The model for this analysis included the three factors from the previous analysis and with a total of 31 items. To calculate the model, the variance of an item per factor had to be constrained to one. According to Hu and Bentler (1999) , to evaluate the model fit, the following cut-off points per index describe a good model fit: RMSEA ≤ .06, CFI ≥ .90, NFI ≥ .90, and GFI ≥ .90 (Hu & Bentler, 1999) . The model fit for the scale with a structure of three factors was evaluated, and it had the following fit indices: RMSEA = .05, CFI = .90, NFI = .86, GFI = .89, χ²(427) = 1356.15 (p < .01), and χ²/df = 3.18. A good model fit should not have a significant χ² value, but it is important to note that the χ² index is sensitive to large sample sizes (Cheung & Rensvold, 2002) .

Factor loadings for all of the items are reported in Table 4. The range of the factor loadings for the first factor was from .43 to .81, for the second factor was from .51 to .70, and for the third factor was from .35 to .74. The internal reliability of all of the items was α = .73, for the first factor was α = .92, for the second factor was α = .80, and for the third factor was α = .64.

Finally, we correlated the factors of the adapted version in this study, finding values ranging between r = −.67 and r = .68, both positive or negative according to the content of each factor (see Table 6).

The different versions (of 29, 39, 54 and 84 items) of the Ryff’s PWBS in Spanish (Table 1), have been tested in studies with Spanish and Latin American samples (including Dominican Republic, Mexico, Colombia and Puerto Rico) with samples of adolescents, adults, elderly, or general samples. The results of these studies indicate dimensions of two, four, five and six factors without second order factors, as well as six factors with one or two second order factors. As discussed in the introduction of this work and as observed in the studies in Table 1, while some studies obtain a structure similar to that of Ryff’s original proposal (Gallardo & Moyano-Díaz, 2012; Ryff & Singer, 2006; Rosa-Rodríguez et al., 2015) , others have obtained somewhat different results different from the factorial structure of origin (Abbot et al., 2006; Springer, Hauser, & Freese, 2006; Valenzuela, 2015) .

**p < .01.

Based on these inconsistencies and in accordance with the literature reviewed, there was a need to evaluate the factorial structure of the test in its version with 39 items, which was the purpose of this research. Since the estimation of a model begins with the formulation of a theory that can be tested with real data (Ruiz et al., 2010) , the exploratory and confirmatory analyses, revealed that the factor structure with the best model fit (Hu & Bentler, 1999; Ruiz et al., 2010) was the one that grouped the items into three factors: Tendency to Self-actualization, Interpersonal and Intrapersonal Relations, and Autonomy.

Regarding the first factor, Tendency to Self-actualization, items were grouped together to conform a new dimension that included items that in the original version of the scales belonged to the factors of Purpose in Life, Self-Acceptance, Environmental Mastery and Personal Growth.

One of the basic theories of psychological well-being is positive psychology and further backward are the humanist theories commanded by authors such as Abraham Maslow (as cited in Lafarga Corona & Gómez del Campo, 1989 ), whose contribution was the hierarchy of human needs. At its highest level, Maslow presents self-actualization, also called personal guidance. In his studies, he indicated that these needs for self-actualization are needs of growth or being, which are not born of the lack of something, but of the desire for personal growth and that individuals who have good psychological functioning are characterized by a tendency to self-actualize (competence, strength, and awareness of weaknesses).

It was later when Everett Shostrom (1964) operationalized the construct of self-actualization by indicating that self-actualizing people experience themselves as being capable of becoming angry (acceptance of aggression); experience and accept their weaknesses (self-acceptance); they are able to be loving with others (ability to intimate contact) and with themselves (self-esteem); they accept the values of others and are flexible in applying their own values; they attend to their own needs and feelings and are sensitive to the needs of others; they manage to find meaning in their experiences (purpose of life); are interested in reaching their maximum potential, are concerned about their growth (personal) and can see the difficulties of life as challenges (environmental mastery).

The 14 items of the tendency to self-actualization refer to positive aspects related to self-learning, to finding direction and goals in life, to the ability to handle responsibilities in life and to make changes if necessary, to feel safe and positive with oneself, and these items represent the most powerful domain of the scales. Findings that are not similar, but consistent with the way these factors were configured in this new dimension are those of Chitgian-Urzúa et al. (2013) who found that the dimensions of environmental mastery and self-acceptance have high correlations with each other, as well as purpose inlife and self-acceptance, so they indicate that they seem to be measuring the same construct, and in this study, these four dimensions merge in a new factor, revisiting the postulates of humanistic psychology as a background that supports psychological well-being.

The second factor, called interpersonal and intrapersonal relations, groups eight items that in the original version of Diaz et al. (2006) belong to the factors positive relations with others (four reactants), environmental mastery, purpose in life, personal growth and autonomy (with an item from each factor, respectively) and that in this study evaluate the difficulty of maintaining close relationships and of trust with others, of being disappointed by one’s own achievements, of difficulties in accepting oneself, or of feeling that the demands of daily life depress and overwhelm the person.

As Ryff points out, one of the components of psychological well-being is the ability to engage and maintain close and warm relationships with other people, to care about the well-being of others, and strong empathy. This component in the Ryff model, called positive relations with others, would also have its negatively written items, deficits that the person experiences in affection, intimacy and empathy towards the others, since the loadings of the four items have a negative sign, and for this reason it was decided to call it interpersonal relations.

In the same way, this second factor integrates four other items whose content refers to negative attitudes towards oneself, not accepting positive aspects of oneself, not knowing clearly what one wants in life and a negative assessment of the person’s past, which when integrated would reveal a deficit in the acceptance and view of oneself and it corresponds to that relationship with oneself (intrapersonal from the humanistic psychology view).

The third factor, autonomy, includes four items that in the version of Diaz et al. (2006) belonged to the dimension denominated autonomy, although it retains less items than the version translated into Spanish and contains those of negative type. The items for this factor reflect the concern for the evaluation that others make of the person, of his vital choices, of the possible influence that others exert on the person and of changing the own decisions if others do not agree with them. Unlike the other two factors, which group items of different dimensions, the autonomy factor groups only four items that in the original scale were included in the dimension of the same name.

The analysis of the different Spanish versions of the Ryff’s PWBS presented in Table 1 indicates that in two studies the factor structure was not analyzed and in five of them the factor analyses carried out was of confirmatory type, forcing the structure to six dimensions. The rest of the studies presented in Table 1, performed an exploratory or exploratory and confirmatory analyzes in the same study. In our study, we performed an exploratory and confirmatory factor analyses in order to ensure that the scale’s dimensionality was the most adequate.

In this sense, the contribution of this study is that according to the contradictory findings of previous studies, several models were tried in our analyses, in the understanding that the structural equation models “facilitate the adequate statistical framework to evaluate the validity and reliability of each item rather than a global analysis, helping the researcher to optimize both the construction of a measurement instrument and the analysis of results” (Escobedo Portillo, Hernández Gómez, Estebané Ortega, & Martínez Moreno, 2016: p. 19) , which guarantees a greater precision of the findings regarding the dimensional structure of the PWBS.

After explaining each of the factors from this study, our findings support a three-dimensional structure as the one obtained by Kafka and Kozma (2002) who found a three-factorstructure because the one proposed by Ryff was not replicated in their study even though they found the same face validity as Ryff. Springer and Hauser (2006) found a similar overlap between the factors of personal growth, purpose in life, self-acceptance and environmental mastery, which in this research were grouped in the first factor we named tendency to self-actualization. These authors do not criticize the theoretical model generated by Ryff, but criticize the way psychological well-being is operationalized, since there is an interdependence of the dimensions, revealed by the high correlations among the factors, so it could be presumed that the items of these dimensions are actually measuring the same construct. Later, Springer, Hauser and Freese (2006) reveal in a new study that there are high correlations among four of the six dimensions of the Ryff’s model, which means that the items do not discriminate sufficiently the construct. Findings similar to those of Springer and Hauser are reported later by Valenzuela (2015) , who with a sample of university students reported that the six-factor structure has inappropriate model fit indices and a low internal consistency; this author found that a two-factor model (personal growth and self-acceptance) has a better model fit and he also concludes that there is a one-dimensional overlap of the factors of personal growth, purpose in life, self-acceptance and environmental mastery.

As Gallardo and Moyano-Díaz (2012) argued in an analysis of the PWBS, that the model proposed by Ryff has had an important weight in the literature on psychological well-being; however, it would be necessary to review in detail the dimensions originally proposed, in order to inquire whether a new version or a reconfiguration of the dimensions is needed, because in some cases the results of the replications are different from the original model. So far, a first finding in our study on the Ryff’s PWBS is that the factor analysis does not indicate a good theoretical fit of the six-factor structure initially proposed by Ryff (1989a) and subsequently translated, reformulated, and adapted by Díaz et al. (2006) , but in our study the model with the best model fit is that of three factors.

With regard to the reliability, the internal reliability of the total scale (α = .73) as well as the second factor (α = .80), are acceptable; in the case of the alpha value of the first factor is high (.92) and that of the third factor has a low value (α = .64). The low value of the third factor and the acceptable value of the total scale are considered adequate in exploratory studies (Nunnally, 1980) but not desirable in confirmatory studies since the Ryff measure, has shown higher alpha values several other studies.

The internal reliabilities obtained in this research share similarities with the findings of Gallardo and Moyano-Díaz (2012) , who found between acceptable and bad consistencies in three scales and good consistencies in the other three; Medina-Calvillo et al. (2003) found an acceptable total consistency in the subscales; Rosa Rodríguez et al. (2015) obtained low internal consistencies in the environmental mastery and in personal growth factors; Valenzuela (2015) indicated that internal consistencies are far from the recommended standards; Véliz (2012) found lower consistencies in comparison to the results of previous studies; Vera-Villarroel et al. (2013) reveal that four of the six scales do not have good internal consistencies, and Vivaldi and Barra (2012) concluded that two of the six factors have reliability problems. These findings, on internal consistencies, are different from those reported by Díaz et al. (2006) , Díaz, Stavraki, Blanco and Gandarillas (2015) , Chitgián-Urzúa et al. (2013), Loera Malvaez et al. (2008), Molina Sena and Meléndez Moral (2006) and van Dierendonck et al. (2008) , who report good or high consistency values with different samples.

Although the reliability is a necessary, but not sufficient, condition to guarantee the validity of an instrument (Campo-Arias & Oviedo, 2008) , it is desirable that this property, that reveals the magnitude in which the items measure the same construct, has values in a certain range to demonstrate the degree of homogeneity of the measure.

In addition to the findings that support the difficulties of the internal consistencies of the Ryff’s PWBS, there is another explanation supported by Campos-Arias and Oviedo (2008) who state that better reliability indices “are achieved when the instrument is applied in a sufficiently heterogeneous population in relation to the attribute or characteristic that is intended to be measured” (p.837), so as a suggestion, it would be interesting to see how the internal consistencies behave when performed with different age groups in the same study, since in this research, we worked with samples of young people.

Based on the reliability data of the scales and the total test in this study, a second important point is that the internal consistencies observe some difficulties and would support the findings of authors who have found similar results, and that these values are far from the high values reported in other studies.

The confirmatory analysis of the Ryff’s PWBS included the use of fit indices that allowed verifying the suitability of the model. If we compare the model fit indices proposed by Hu and Bentler (1999) and by Ruiz et al. (2010) , with those obtained in the confirmatory factor analysis of this study, it is clear that in this sample of university students, from four Mexican states, the model that showed the best fit was the one that allowed to keep all the original items of the scale and that simplified the factorial structure in three factors.

The discriminant validity of the scale was contrasted by the traditional way of calculating the correlation matrix among the factors of the scales. The correlation between the factors of tendency to self-actualization (whose items are written in a positive sense) and interpersonal and intrapersonal relations (with negatively worded items) reveal a moderate and negative correlation; between the factor of tendency to self-actualization and the factor of autonomy (the latter, with negatively worded items) the correlation is negative and low; The correlation between the factor of interpersonal and intrapersonal relations and the factor of Autonomy (both, with negatively worded items) was moderate and positive. The results of these correlations in the validated version of the scale in this study indicate that the regrouping of the factors of the original scale in a three-dimensional version reveals better values than those obtained in other studies, which are detailed below.

In some studies, (Chitgian-Urzúa et al., 2013; Gallardo & Moyano-Díaz, 2012; Springer, Hauser, & Freese, 2006; Véliz, 2012) inter-item correlations or between scales reveal high values, which are indicative that there are items or scales that are assumed to be independent but they might actually be evaluating similar aspects. In some cases, there is a model with six dimensions but with a high intercorrelations among factors, which undermines the discriminant validity of the scales, and perhaps this overlap also undermines the construct validity of the test.

For Gallardo and Moyano-Díaz (2012) , the proposed model by Ryff has had a very important weight in the literature of positive psychology, nevertheless, based on these findings, it is convenient to think that the raised dimensions are subject to revision to see if a new version arises, since the results of the high correlations among factors would presume an overlap among them, this would also explain that some of the dimensions coalesce with each other.

Véliz (2012) indicates that one of the problems of the Ryff’s PWBS in its Spanish version is that a six-factor model is difficult to replicate because there are high intercorrelations among the factors, which does not contribute to the discriminant validity that the authors of the scale defend. Véliz defends the idea that in order to improve the results of the scales, studies with larger samples should be carried out, an action that is presented in this study; although we have a large sample, the scales still present difficulties to fit a model of six factors, but problems with the internal consistencies in this modified version are no longer present.

As an important information on Ryff’s PWBS, the analyzes of the correlations among factors are indicative of a greater interdependence, which allows to appreciate that these are factors or dimensions more or less independent although part of a common construct, and that in this study there is no overlap between scales as it is reflected in other reviewed studies.