Participants The study involved a total of 269 adolescents living in the Autonomous Community of Galicia (Spain), out of whom 142 (52.8%) were male and 127 (47.2%) were female. Their age ranged from 12 to 17 years (M=13.46; SD=.93). Instrument A scale elaborated by the authors was used to measure students’ self-image. The scale, which was validated in a previous macro-study (Santos Rego, 2014), was initially composed of 23 items with a Likert response format and five response options (from ‘strongly agree’ to ‘strongly disagree’). 5 items were removed due to the lack of relevance of the item-total correlation, the resulting scale consisting of 18 items grouped into four factors with acceptable validity and reliability indexes. Procedure The scale implementation took place collectively in the classroom, using tutors of each group specially trained, not only for this task, but also within the framework of a broader data collection effort for an educational research project.

Descriptive statistics of items Item 1: I want to complete Compulsory Secondary Education (CSE) to continue studying (M=4.49, SD=1.0366); Item 2: I think I have very good qualities (M=3.90, SD=.8506); Item 3: I do my school work correctly (M=3.84, SD=.8767); Item 4: Sometimes I feel really useless (M=3.57, SD=1.2682); Item 5: In general, I am satisfied with myself (M=3.83, SD=1.1425); Item 6: My teacher thinks I am a hardworking student (M=3.51, SD=1.1218); Item 7: I work hard in class (M=3.68, SD=1.0514); Item 8: My teacher appreciates me (M=3.62, SD=1.1481); Item 9: I'm a good student (M=3.72, SD=1.0412); Item 10: My teacher thinks I am intelligent and hardworking (M=3.60, SD=1.0442); Item 11: I have the ability to get what I want (M=4.40, SD=.7642); Item 12: When I do not understand something, I ask the teacher (M=3.89, SD=1.0467); Item 13: Working in groups with peers from other cultures makes my grades worse (M=4.06, SD=1.2051); Item 14: I would rather work by myself than in groups (M=3.49, SD=1.3346); Item 15: I would be willing to do class work with peers with lower grades than mine, if that helped them in their studies (M=3.82, SD=1.1366); Item 16: Collaborating with others is a waste of time and useless work (M=4.35, SD=1.1283); Item 17: I like sharing my time and work with my peers in class activities (M=4.03, SD=.9080); Item 18: My peers’ success in their studies, regardless of whether they are my friends or not, gives me satisfaction (M=3.74, SD=1.0289). Dimensionality. Exploratory Factor Analysis (EFA) and internal consistency The dimensionality of the scale was determined by an Exploratory Factor Analysis (EFA), following the Principal Components extraction method and a Varimax rotation. The KMO tests = .830 and χ2 (153) = 1519.73 p <.001 reported good sampling adequacy and a significant correlation between items, respectively. The initial extraction yielded four significant factors (companionship, self-efficacy, feedback and self-esteem), which account for 53.60% of the variance, with good internal consistency (with α values ranging from .67 to .79). The results of the rotated solution are shown below: - Companionship factor (α=.72): Item 16 (.710), Item 13 (.657), Item 17 (.652), Item 18 (.646), Item 15 (.613), Item 14 (.594). - Self-efficacy factor (α =.77): Item 1 (.709), Item 3 (.674), Item 9 (.653), Item 7 (.594), Item 15 (.461). - Feedback factor (α=.79): Item 8 (.807), Item 6 (.751), Item 10 (.734). - Self-esteem factor (α=.67): Item 4 (.767), Item 5 (.688), Item 2 (.616), Item 11 (.613). Proposal of an explanatory model, Confirmatory Factor Analysis (CFA) A CFA was performed using the maximum likelihood method and the bootstrapping procedure, starting from 500 different subsamples, to check the estimators which were not affected by normality. Mardia’s coefficient was 118.3 (Mardia, 1970), which, according to Bollen (1989), was lower than p (p + 2), where p is the number of observed variables. All Standardized Regression Weights are significant (p< .01). The covariance between feedback, self-esteem and self-efficacy (p< .01) is also significant. However, the covariance between companionship and the other three factors is not significant . According to these data, the proposed model confirms the existence of four factors which predict self-image in Secondary Education students. The values obtained were: χ2 / gl = 2.10 (value lower than 3; Bollen, 1989); GFI= .91 (interpreted as a coefficient of determination, whose value should be close to 1) (Jöreskog and Sörbom, 1993); CFI= .90 is a comparative adjustment index and values ranging between .90 and .95 indicate a good fit (Bentler, 1990); RMSEA = .064 reports the difference between the population correlation matrix and the matrix proposed in the used sample model, whereas values lower than .08 indicate a good fit (Steiger and Lind, 1980); SRMR= .050 reports the standardized residuals and, like in the previous case, values lower than .08 indicate a good fit (Hu and Bentler, 1999). CONCLUSIONS The purpose of this work was to develop a model of self-image measurement for Compulsory Secondary Education (CSE) students from Galicia, Spain. The proposal includes a scale, based on a theoretical model, with 18 items. The exploratory and confirmatory factor analyses allowed the authors to state that the solution was satisfactory, both in terms of structure (with four factors of the scale: companionship, self-efficacy, feedback and self-esteem) and internal consistency of the evaluated levels.