In order to achieve our stated objective, data collection was conducted twice. A pilot test was carried out to enable the final selection of items, followed by a final test. The population of the pilot test comprised 300 participants, 203 from Catalonia and 97 from Venezuela, attending years one to three of primary school. The age range was 5.92 to 10.08 years (M = 7.98; SD = 0.90).

To obtain the sample for the final test, we proceeded as follows. In Catalonia, a list of state schools was compiled from the province of Girona (including 6 counties of the province, and excluding 3, for reasons of geographical proximity) and five schools were randomly selected. For the Venezuela sample, the selection was based on non-probabilistic purposive sampling. In both cases, we received consent from the schools’ governing bodies. The test was not administered to some children for one of the following reasons: a) they would not understand, in the teacher’s opinion, the language in which the test was administered; b) they did not receive parental consent; c) they did not attend school on the day of the test.

The final test sample consisted of 636 schoolchildren aged 6 to 9 years old (range: 6 to 9.83 years; M = 7.89, SD = 0.90). Table 1 provides data for the age of participants by school year, gender, and country. Participants were from Venezuela (267) and Catalonia (369), where they attended years one to three of Basic Education (Venezuela) or primary school (Catalonia).

Table 1 Table 1 Age of participants at the final test (in years) by gender, school year and country

The self-reported test used was designed specifically for this research. It consisted of a group administration Likert scale, comprising 53 items for the pilot test and 32 for the final test. The items consisted of positively-worded statements and referred to positive aspects of children’s lives, with the exception of some items related to emotions.

For each item, participants had to mark the most appropriate response with a cross. There were 3 possible answers, based on the frequency with which the child felt in the way described: “always,” “sometimes,” or “never”. Each item had a value of 0-2 points. In some items, the response “always” received 2 points and “never” 0 points, while in others it was the reverse (0 points for “always” and 2 points for “never”). “Sometimes” always scored one point. Thus, the possible range of scores for the final test was 0 to 64 points.

The following tests were reviewed for the wording of items: the self-esteem assessment scale (De Tejada, 2009), the self-description questionnaire (Marsh, 1988), the self-perception profile for children (Harter, 1985), the pictorial scale of perceived competence and social acceptance for young children (Harter & Pike, 1984), the self-esteem inventory (Coopersmith, 1981), the Piers-Harris children's self-concept scale, 2nd edition (Piers, Harris, & Herzberg, 2002), and finally, the multimedia and multilingual self-esteem assessment questionnaire (Ramos & Santamaría, 2010). Three of the four authors of the present study wrote the items of the pilot study by grouping the items of these tests according to their content (grouping the items with similar content). These items were in turn classified into 7 dimensions.

The 53 items included in the pilot test were initially written in Spanish and then translated into Catalan. Once the pilot had been administered, a principal component analysis was carried out with 7 fixed factors, for both the Catalan and Venezuelan samples. On the basis of these analyses, items were chosen for the final test which weighed a score superior to 0.3 on the same factor. In the pilot sample, the Kaiser-Meyer-Olkin measure of sampling adequacy was 0.734 in Catalonia and 0.493 in Venezuela, and Bartlett’s test of sphericity was statistically significant in both samples (Catalonia: x² = 3295.21, p < 0.001; Venezuela: x² = 1827.32, p < 0.001).

Thus, a total of 32 items were selected to comprise the final test. The following factors were found: a) social acceptance (10 items); b) negative emotions (7 items); c) physical self-assessment (3 items); d) school satisfaction (3 items); e) family relationships (3 items); f) ability to learn (3 items); g) physical activity and sport (3 items). Apart from these 32 items, 2 example items not included in the final test analysis were also included in order for participants to understand how to respond to the scale.

School heads were contacted prior to administering the pilot and final tests to explain the aims of the study.

Both tests were administered to the students as a group. The items were read aloud by the test adminstrator, who checked the children were paying attention and marking their responses in the appropriate row.

The instructions given to participants were as follows: “We’re going to do a different activity to the ones you normally do in the classroom. We’re going to read out some questions about how you feel”. The questionnaires were then handed out and participants were asked to use a pencil to mark their answer. They were first asked to fill in the personal data section of the questionnaire, and then told, “We’re going to read a few sentences about how we feel sometimes, always, or never, in relation to the activities we do with our parents, teachers, and friends. As we read, mark one of the three options in the column: I always feel like this, I sometimes feel like this, or I never feel like this, according to how you feel about the sentence. Think carefully before you answer”.

Exploratory factor analysis was used in the Catalan sample to explore the factorial structure of the Self Esteem Scale, and then this model was fitted in the Venezuelan sample using Confirmatory factor analysis.

Version 19 of the SPSS statistics package was used for the principal component factor analysis and version 18.0 of the AMOS program for the confirmatory factor analysis. A total of 63 responses had missing values from a total of 46 participants. Thus, 590 participants responded to all questions correctly. In the factor analysis, missing values were replaced by the mean of the item.

The goodness of fit indices used included the Satorra–Bentler index and the Non-Normed Fit Index (NNFI) and Comparative Fit Index (CFI). NNFI and CFI values around 0.90 are acceptable. The Standarized Root Mean Squared (SMRM) and the Root-Mean Squared Error of Aprroximation Measures Residuals (RMSEA) were also used. SMRM and RMSEA values lower than 0.05 are acceptable.

First, the psychometric properties of the scale of 32 items in the Catalan sample (n = 369) were analysed by means of exploratory factor analysis with varimax rotation. The Kaiser-Meyer-Olkin measure of sampling adequacy was 0.851 and Bartlett’s sphericity test was statistically significant (x² = 3150.01; p < 0.001), indicating that it was appropriate for the factor analysis. The scree plot suggested an initial solution of 5 factors. Only those items loading more than 0.4 on a certain factor and less than 0.3 on the rest were retained (see Table 2). Following this criterion, items 3, 6, 10, 13, 16, 22, 24, 25, 28, 30 and 31 were removed.

Table 2 Table 2 Factor loadings of the 32-item 5-factor scale Note:

Items loading more than 0.3 appear in bold.

Only factors with over 0.5 internal consistency were retained for the analysis, leading to the elimination of the fifth factor (alpha = 0.44), which included items 7 and 14, and the analysis was repeated using the remaining 19 items and four factors. This new analysis revealed that item 9 did not reach a loading value of 0.4 on any of the factors and had a factor loading greater than 0.3 on factors 2 and 3, leading to the item being removed and the analysis being repeated with the remaining 18 items and 4 factors. All items met the criteria with this analysis. The psychometric properties of the selected items are presented in Table 3. The seven items that loaded on the first factor are related to social acceptance by friends. The four items that loaded on the second factor express negative emotions. The three items that loaded on the third factor refer to school satisfaction, and finally, the four items that loaded on the fourth factor were related to self-assessment. The variance explained by the four factors was 50.55%.

Table 3 Table 3 Description, descriptive indices, and factor loadings of the items and Cronbach’s alpha, eigenvalues, and explained variance for rotated factors for each factor *

Cronbach’s alpha if item is removed

In order to test the goodness of fit of the factor solution for the four factors identified in the Catalan sample by means of explotatory factor analysis, we conducted a confirmatory factor analysis on the sample of Venezuelan students (n = 267). Version 18.0.0 of the AMOS software was used and the adjustment was made using the method of maximum likelihood estimation. Missing data were processed using the EM (Expectation Maximization Method). The goodness of fit indices used were the Satorra-Bentler index and the NNFI (Non-normed Fit Index) because data did not follow normality. The SMRM (Standardized Root Mean Squared) and RMSEA (Root-Mean Square Error of Approximation Measures Residuals) were also used. The results show that the fit is good (x² = 253.8, p < 0.001; NNFI = 0.89; CFI = 0.94; SRMR = 0.043, RMSEA = 0.039, 95% CI (0.032:0.046)). Table 4 shows the standarized factor loadings of the scale EVA 2015. The Cronbach’s alpha values were as follows: social acceptance 0.81 (7 items); negative emotions 0.61 (4 items); school satisfaction 0.61 (3 items); and self-assessment 0.47 (4 items).

Table 4 Table 4 Factor loadings EVA 2015