Several tools have been identified in the scientific literature to measure social skills, such as the Teenage Inventory of Social Skills (TISS) is the only self-report designed exclusively to reflect behaviors functionally related to peer acceptance in adolescence (Inderbitzen and Foster, 1992; Inglés et al., 2005) the Matson Evaluation of Social Skills with Youngsters Scale (MESSY), describes an analysis of aggressive and antisocial behavior in girls and boys aged 7 to 15 years (Matson et al., 1983) the Multidimensional Social Expression Scale (EMES-M), a scale that addresses cognitive aspects of subjects that may play basic roles in the expression of socially skilled behavior (Caballo, 1993) the Interpersonal Difficulties Questionnaire (DCI), describes the development and psychometric properties of a new self-report measure aimed at detecting the level of difficulty that adolescents present in their interpersonal relationships (Eceiza et al., 2008) the Inventory of Social Skills (IHS), this study analyzed research that develops psychometric instruments to assess social skills and other related constructs, such as assertiveness, social competence and social self-efficacy (Del Prette et al., 1998; Orlando et al., 2009) the Social Skills Scale (EHS), which reviews the general background of instruments commonly used to assess social skills and presents the general characteristics of the instrument studied, reviewing its psychometric measurement properties (Gismero, 2000; Miranda-Zapata et al., 2014) and the Interpersonal Difficulties Assessment Questionnaire for Adolescents, which detects the level of difficulty that adolescents have in their interpersonal relationships (Inglés et al., 2009). However, most of these instrument’s focus on the adolescent stage, so it is essential to develop a tool that considers the specific characteristics of primary school students. During this stage, children begin to interact with their peers, which is crucial for their personal and academic development (Kerklaan et al., 2020).

One of the limitations observed in the use of existing instruments for self-assessment of social skills is that they are outdated. In addition, it is difficult to find culturally appropriate instruments focused on the educational level to be addressed (Pareja de Vicente et al., 2021). This hinders an adequate intervention by teachers, since the diagnosis obtained does not allow identifying the main difficulties encountered in these educational contexts (Civitillo et al., 2021). At present, research carried out in previous years in this area is still being used. For this reason, and after analyzing the scientific literature, it is considered essential to provide a questionnaire with excellent psychometric properties that analyzes the construct. For this purpose, the questionnaire “Social Skills Scale for Young Immigrants” (SSSYI) by Tomé-Fernández et al. (2020) has been used as a reference. This questionnaire consists of 33 items, organized in different dimensions: ability to say “no” and cut interactions, self-expression in social situations, defense of rights as a consumer, expression of anger or disagreement, ability to make requests to others and ability to initiate positive interactions with people of the opposite sex, related to those applied in previous studies (Bunford et al., 2015a,b; Hintermair et al., 2016; Caballo and Salazar, 2017). This instrument was designed following the parameters and conditions stipulated to obtain valid and reliable results, since an instrument with adequate psychometric properties allows the assessment of specific social skills involved in adaptive and non-adaptive behaviors (Tomé-Fernández et al., 2020).

To avoid problems related to overfitting, the samples for each type of factor analysis corresponded to two data collection processes in the target population (Strayhorn et al., 2022). At first, the aim is to find the structure and dimensions in which the questionnaire items are organized according to the validation using EFA. For its procedure, data collection was carried out within a selection of eight Infant and Primary schools from different provinces of Andalusia from which students were surveyed (Table 1). Participants at this stage reported ages ranging from 6 to 14 years (M = 9.65 years; SD = 1.395). Of these, 460 (60.4%) were boys and 301 (39.60%) girls (Table 2). In addition, 651 (85.5%) were of Christian religion, 29 (3.8%) were Islamic, 81 (10.60%) did not know or did not answer. In terms of ethnicity, 84 (11%) belonged to the Roma ethnic group and 677 (89%) to the Castilian ethnic group. Finally, with regard to race, 29 (3.8%) students were of African origin and 732 (96.2%) were white. The second data collection process (AFC) allowed us to retrieve an additional set of 807 student surveys, within the selection of the eight Infant and Primary schools in the different provinces of Andalusia previously selected in the previous procedure (AFE) from which students who wished to participate were re-surveyed (Table 1). Participants at this stage reported ages ranging from 6 to 14 years (M = 9.65 years; SD = 1.365). At this stage, 479 (59.7%) were boys and 324 (40.3%) were female (Table 2). In addition, 690 (85.9%) were of Christian religion, 29 (3.6%) were Islamic, 84 (10.5%) did not know or did not answer. In terms of ethnicity, 85 (10.6%) belonged to the Roma ethnic group and 718 (89.4%) to the Castilian ethnic group. Finally, with regard to race, 29 (3.6%) students were of African origin and 774 (96.4%) of white origin.

Initially, the scale of social skills in Primary Education students in diverse contexts was carried out by reviewing the scientific literature and analyzing the different instruments related to the topic, such as, for example, the Social Skills Learning Team Questionnaire (CHSEA; Mendo-Lázaro et al., 2018), Social Skills Questionnaire (CHASO-III; Salavera et al., 2017), the Social Skills Inventory (Oldmeadow et al., 2013), the Social Skills Questionnaire for University Students (SSQ-U; Morán et al., 2015), Social Skills Scale for Young Immigrant (Tomé-Fernández et al., 2020), the Social Skills Improvement System-RS (Bunford et al., 2015a,b). Subsequently, we opted for the adaptation and validation on which this research is based, which considered all previous instruments more appropriate in terms of the simplicity of the language of the constructs assessed, and fulfilling excellent psychometric properties of validity and reliability (Klimeckil et al., 2018), as well as being novel in the definition and diagnosis of social skills. The initial instrument was structured into 33 items that are scored along a Likert scale with four response options. Responses range from never (1) to always (4). The scale is relevant to the study context, with the items having been literally translated into Spanish from the items provided by Tomé-Fernández et al. (2020). Once the first version of the questionnaire had been designed, the content validation process was set in motion to analyze the degree to which the instrument was capable of measuring the different dimensions and items to be analyzed, taking into account both its clarity and relevance. To this end, in the preliminary phase of the study, the questionnaire was submitted to the judgment of experts who made it possible to refine and add modifications to the initial questionnaire. These were professionals related to the development of social skills of students of different ages, nationalities and cultures working in educational institutions and researchers from Spanish universities, whose lines of work are linked to the diagnosis and evaluation of social aspects in immigrants (Shelanee et al., 2020). These consisted of 25 experts from different universities, of which 10 were researchers related to diverse contexts, 9 were associated with social skills in adolescents in the field of early childhood and primary education and 6 had lines of work related to interculturality. The questionnaires were sent by email to each of the experts. The scaling method was used to rate from 1 to 4 (1 = never, 2 = sometimes, 3 = quite a lot, 4 = always) the importance of each item. The experts were consulted on the face validity and readability of the questionnaire in order to assess the appropriateness of the instrument’s information for primary school and immigrant youth, as well as the comprehension and wording of the items (Mérida et al., 2015). The experts provided qualitative responses to open-ended questions. Based on the experts’ responses, modifications were made to several items of the questionnaire, both significant and non-significant. To carry out these modifications, the indications of Barbero (2006) for inter-judge agreement were followed. The mean value of each item was analyzed, being equal to or greater than 2.5, and the median value was considered as the item value. In addition, an ambiguity coefficient was established to measure the dispersion in the judges’ agreement, using the interquartile range as the criterion. If the difference between the 75th percentile (P75) and the 25th percentile (P25) was equal to 0.0 or 1, the item was accepted and/or slightly modified; if the difference was between 1 and 2, the item was revised and reformulated; and if it was greater than 2, the item was rejected due to high dispersion among expert judgments. The final questionnaire was obtained and its hierarchical structure was demonstrated through factor analysis.

The purpose of the present study was the validation of an instrument, which was administered individually to students from different schools who expressed their willingness to participate in the research. To this end, the corresponding informed consent was requested from the management team of each institution and from the parents or legal guardians of the minors, in compliance with the ethical and legal requirements established for carrying out research with subjects under the age of majority. The questionnaire was administered in school classrooms, in the presence of teachers, and the response time for its completion varied between 12 and 20 min approximately. It is important to mention that the students did not require assistance to understand the items in the questionnaire, as most of them are fluent in the official language of the host country, which is the language that predominates in their social interaction outside the family environment. Throughout this procedure, the ethical criteria established in the Declaration of Helsinki in 1975 and then updated in Brazil in 2013 and the protocol approved by the Ethics Committee of the University of Granada (reference code: 742/CEIH/2018) were followed.

All collected data were imported into IBM SPSS version 28 software (Wanna, 2022) in order to conduct a preliminary descriptive analysis and examine the normality of the data. Following this, an exploratory factor analysis (EFA) was conducted, previously it was decided to employ the principal axis factorization method with Oblimin rotation, due to the fact that the factors are correlated (Yong and Pearce, 2013). The overall reliability of the scale was determined using Cronbach’s alpha coefficient, setting the index reliability at 95%. In the second phase, a confirmatory factor analysis (CFA) was conducted using the latest version of the IBM Amos package. For this purpose, the quantitative data acquired were analyzed in two periods: the first, through which the EFA was carried out using the latest version of SPSS statistical software; the second, in which the CFA was carried out using the AMOS statistical programmed, confirming the hierarchical structure of the previous analysis. This took into account the goodness-of-fit criteria established by Kock (2014) and Hu and Bentler (1999). To study model fit, the Chi-square statistic was calculated, together with the Comparative Fit Index (CFI), Normalized Fit Index (NFI), Incremental Fit Index (IFI), and The Root Mean Square Error of Approximation (RMSEA). These incremental fit indices assess the improvement of the proposed model relative to a previous baseline model (Wanna, 2022).

An EFA of the 33-item scale was conducted using the principal axis method of factor extraction, using Oblimin rotation, since the factors are correlated (Yong et al., 2021). The Bartlett’s statistic produced in relation to this revealed an acceptable fit, as this factor analysis was validated using the Kaiser-Meyer-Olkin sampling adequacy (KMO) with a value of 0.831 with a Bartlett’s sphericity of 3477.120 (p = 0.000) which was equally acceptable. Since the closer the value obtained from the KMO test is to 1, it implies that the relationship between the variables is excellent (Siembida et al., 2018), as is the case in this study. The results of the factor analysis are presented in Table 5.

According to Siembida et al. (2018) after extraction, the communalities must be ≤0.30 to conjecture that the measure assumes excellent validity (Table 6). After extraction, the instrument meets this criterion in all of its items, so they were not eliminated, maintaining the 33 items that make up the questionnaire. In addition, the study (AFE) grouped the 33 initial items into five factors or dimensions with factor loadings ranging from 0.77 to 0.511 (Table 6).

Five different dimensions of the Social Skills in Primary School Students in Diverse Contexts scale (SPSDC) have been identified. The first of these is called “Denial” and deals with situations where the young person considers that saying “no” and cutting off interactions is not a difficulty, but something opportune. It has an optimal internal scale consistency, with a Cronbach’s alpha value of 0.70 and consists of 8 items (22, 23, 19, 24, 20, 21, 25, 18). The second dimension identified is called “Self-Expression,” and the items included refer to self-expression in social situations (33, 7, 8, 6, 9, 2). The items referring to this dimension also present a reliability above the minimum required values (Cronbach’s alpha of 0.76). The third-dimension deals with young people’s expression of emotions “Emotional expression” (32, 3, 5, 1, 4, 15, 14) where once again the reliability of the scale is optimal (Cronbach’s alpha of 0.60). The fourth dimension identified refers to “No opinion,” where seven items address not daring to express one’s opinion to others (11, 17, 12, 31, 10, 16, 13). This dimension represents a reliability of 0.62. Another dimension is called “Relationship with peer group.” This dimension consists of 5 items dealing with relationships with friends (29, 27, 28, 26, 30). Furthermore, the component analysis showed that the five dimensions have 62.03% of the total variance explained. The distribution of the items in five factors coincides with the hierarchical structure of the instrument reference (Maggiori et al., 2017), which bases the dimensions of social skills on important theoretical references (Sánchez-Bolívar et al., 2019), which specify that when a subject presents social skills, they are able to express their emotions in different situations, defend their opinion, show denial and express themselves clearly, as well as initiate interactions with peers. These aspects are developed in the five dimensions established in the scale called Social Skills in Primary School Students in Diverse Contexts (Burchinal et al., 2020).

After conducting the FEA, the CFA is carried out to corroborate the suitability of the indicators to assess the latent variables (Anderson-Butcher et al., 2016). This test was conducted on a sample of N = 803 and analyzed with AMOS software, version 24. At this stage the data (CFA) are presented in path diagrams, where circles represent latent variables and squares represent observed variables. Single-headed arrows are used to indicate a direction of presumed influence, and double-headed arrows represent the covariance between the five latent variables. This type of analysis, according to authors such as Salavera Bordás et al. (2019), makes it possible to assess the validity and reliability of each item of the questionnaire and performs an individual hypothesis test, where it measures whether the instrument is valid and subsequently measures its reliability. Taking into account this procedure, it can be deduced that the proposed model presents a reasonable contiguity of the data, and ratifies the hypothesis of the multidimensionality of the analyzed construct. The latent variables for the SPSDC were presented in 8, 6, 7, 7, 7, and 5 items (Figure 1). In the first factor, the items are adequately explained by the hypothesized construct, showing a minimum saturation of 0.65 and a maximum of 1.00. The second factor presents saturations ranging from 0.80 to 1.40. The third factor ranges between 1.00 and 1.55. The fourth factor ranges between 1.00 and 1.85. Finally, the fifth factor presents a minimum saturation of 1 and a maximum of 1.36, which shows that the items show a satisfactory correlation with the total test (Figure 1).

To carry out the CFA, the CMIN (Cb, minimum value of the discrepancy) was observed with a x2 distribution (Table 7).

Establishing the criteria of other studies (Bentler, 2007; Byrne, 2010), the Chi-square goodness-of-fit test (x2) tends to be insufficient for the sample size used; Therefore, it is recommended to take into account the comparative fit index (CFI), the Tucker-Lewis index (TLI) and the root mean square error of approximation (RMSEA) with the intention of knowing the degree of fit between the covariance matrix of the observed data and the covariance matrix predicted by the model, established by the goodness-of-fit index. The CFI, TLI and RMSEA obtained at this stage show good internal consistency of the latent factors (Salavera Bordás et al., 2019). Furthermore, according to Aljaberi et al. (2022), the CFI and TLI are considered to have an acceptable fit the closer they are to 1, a characteristic that is fulfilled in the results of this research, whose CFI = 0.772 and TLI = 0.737 (Table 8). As for the RMSEA value, it is considered an excellent fit, since its value is 0.42 (Table 9). It can be deduced that the proposed model exhibits reasonable contiguity of the data and ratifies the hypothesis of the multidimensionality of the construct. The latent variables for the KABIP-S were presented in 33 items (Figure 1). Once the CFA was completed and the 33-item instrument structure was confirmed, the internal consistency test (Cronbach’s alpha), obtaining a reliability factor of α = 0.667 for the first dimension, α = 0.613 for the second dimension, α = 0.663 for the third dimension, α = 0.638 for the fourth dimension and, finally, α = 0.668 for the last dimension. This indicates a good reliability of the scale (Laurencelle, 2021).