This is a quantitative, explanatory study that encompasses research aimed at testing models of relationships between a set of variables derived from an underlying theory (Ato et al., 2013). Data collection took place during the months of May and June 2022. After obtaining the necessary permissions from the institutions and teachers contacted through convenience sampling, they were provided with the questionnaire to share with the students, ensuring voluntary, anonymous, and confidential participation.

Through convenience sampling (Cohen et al., 2018), a total of 276 students participated in the study. They were either enrolled in vocational education (n = 87, 65.5% in science and technology programs) or baccalaureate (n = 189, 64.6% in science and technology track) in the cities of Burgos (86.2%), Madrid (11.6%), or Salamanca (2.2%). The mean age of the participants was 18.05 years (SD = 4.56; MD = 17). Slightly more than half of the sample identified as female (51.3%), while the rest identified as male.

The instrument used for measuring expectancies of success was designed and validated by Thomas and Strunk (2017) and was available in three versions suitable for students, parents, and teachers. The student version consisted of seven items (e.g., I can get good grades in science and technology; Science and technology are challenging for me). In the parent version, the same items were used, but the wording was modified to reflect the student’s beliefs about their parents’ expectancies of success. For example, the item “I can get good grades in science and technology” was modified to “During secondary school, I perceived that my parents believed I could get good grades in science and technology.” Finally, the teacher version included similar wording changes (e.g., During secondary school, I perceived that my teachers believed I could get good grades in science and technology).

Firstly, the construct validity and internal reliability of each version of the instrument were examined, as the original version was published in English. The items were adapted to Spanish following a cross-cultural translation procedure (Beaton et al., 2000). Subsequently, an exploratory factor analysis was conducted to determine to what extent the Spanish versions of the instrument have the same factorial structure as the original versions. Specifically, the items were subjected to a robust exploratory factor analysis using the Maximum Likelihood method, as this procedure provides more accurate results and is theoretically superior to other extraction methods (Ferrando et al., 2022). The number of factors to extract was determined using parallel analysis (Lloret-Segura et al., 2014). Consistent with the original instruments and Eccles and Wigfield’s Expectancy-Value Theory (1995, 2020), a one-dimensional structure is expected for each version of the instrument (students, parents, teachers). The reliability of internal consistency was established using Cronbach’s alpha (α) and McDonald’s omega (ω), with appropriate minimum values above .70.

The Kaiser-Meyer-Olkin (KMO) index and Bartlett’s test of sphericity indicated that the data were suitable for exploratory factor analysis. Specifically, the KMO values were .86, .88, and .87, respectively for the student, parent, and teacher versions, and Bartlett’s test of sphericity was statistically significant (p < .001) in all three cases. The exploratory factor analysis revealed the presence of a two-dimensional structure in all three questionnaires, which was supported by the results of the parallel analysis (see the scree plots and parallel analysis charts in the Appendix).

However, it is important to note that the second factor is composed of items that are phrased negatively; this would seem to indicate that it is a statistical artifact rather than an appropriate factorial structure (see Tables 1-3). Thus, even though two out of the three versions of the instrument (students and teachers) reveal two factors with eigenvalues greater than 1, this criterion is not recommended in the literature as it may result in an excessive extraction of factors (Lloret-Segura et al., 2014). Furthermore, specialized literature indicates that negatively phrased items often load onto another factor, as is the case here (Zhang et al., 2016). Therefore, in instruments where the structure is unidimensional, the use of negative items can distort the dimensionality of the scale. Consequently, given these limitations of negative items, and because the two-dimensional structure is not in line with the theoretical foundation of the construct of expectancies of success (Eccles and Wigfield, 1995, 2020), the decision has been made to retain unidimensional structures, which exhibit appropriate factor loadings and optimal internal consistency reliability.

Therefore, the questionnaire for students explained 47.9% of the variance, and in the present sample, the scores exhibited good reliability (α = .85, ωt = .83). Regarding the questionnaire assessing parental expectancies of success, the one-dimensional structure explained a total of 46.7% of the variance and also yielded scores with adequate reliability in the current sample (α = .87, ωt = .84). Finally, the one-dimensional version of the instrument measuring teacher expectancies of success explained a total of 44.4% of the variance and also demonstrated good score reliability in the current sample (α = .87, ωt = .87). These values suggest that the instrument in its adapted Spanish version has evidence of validity and reliability that meets current psychometric standards (Ferrando et al., 2022; Lloret-Segura et al., 2014).

Table 1
Exploratory Factor Analysis for the Student Version

Table 2
Exploratory Factor Analysis for the Parents version

Table 3
Exploratory Factor Analysis for the Teacher version

Various independent samples t-tests were conducted to examine potential differences in levels of expectancies based on the type of studies (baccalaureate or vocational education) and field (scientific-technological or non-scientific-technological). The statistical assumptions were met (Knapp, 2018): (i) normal distribution of variables (kurtosis and skewness ±2); and (ii) no violation of the assumption of variance homogeneity (p values > .05). Cohen’s d was used to determine the effect size: small (.2), moderate (.5), and large (.8).

Additionally, a logistic regression model was employed with three continuous variables (student expectancies, parental expectancies, and teacher expectancies) after meeting the required statistical assumptions (Knapp, 2018): (i) inspection of histograms and kurtosis and skewness levels (±2) indicated a normal distribution of variables; and (ii) Pearson correlation between variables ranged from .62 to .70, confirming the absence of multicollinearity issues in the data. Given the sufficient sample size, both analyses were conducted independently for girls and boys.

The independent samples t-tests (Figure 1) revealed that, in the case of boys, vocational education students have significantly higher expectancies of success than their baccalaureate counterparts t (133) = 2.04, p = .04, d = .35, with a medium effect size. Conversely, for girls, statistically significant differences were found, with a medium effect size, in favor of those enrolled in baccalaureate in the expectancies of success of their parents t (139) = -2.51, p = .01, d = .48, and those of their teachers (139) = -2.45, p = .02, d = .49.

Figure 1
Expectancies of success in vocational education and baccalaureate

Regarding the logistic regression analysis (Table 4), the model with the three predictor variables was statistically significant for boys c2 (3, n = 135) = 14.72, p < .01, but not for girls c2 (3, n = 141) = 7.94, p = .05). This indicates that, in the case of girls, their expectancies of success, those of their parents, and those of their teachers do not influence or predict their choice of studies. Conversely, in the case of boys, the model explained 14% (Nagelkerke R2) of the variance in their choice of study type.

The strongest predictor was the expectancies of success from their parents, with an odds ratio of 2.61 (p = .002). This suggests that boy students whose parents have high expectancies of success in science and technology for them are more than twice as likely to enroll in baccalaureate studies compared to students whose parents have low expectancies of success.

The second statistically significant predictor was their own expectancies of success, with an odds ratio of 0.36 (p = .001). This indicates that, surprisingly, students with high expectancies of success in science and technology are less likely to enroll in baccalaureate studies than their peers with low expectancies of success. Finally, the expectancies of the teachers did not reach statistical significance (p = .498).

Table 4
Predictive model for baccalaureate choice

The independent samples t-tests (Figure 2) revealed that, in the case of boys, students in scientific and technological fields have significantly higher expectancies of success in science and technology than their peers in other fields of study, with a moderate effect size, t (133) = 2.51, p = .01, d = .57. For girls, statistically significant differences were found with moderate to large effect sizes in favor of those enrolled in scientific and technological studies in all three variables: their own expectancies of success t (139) = 7.22, p < .01, d = 1.23, those of their parents t (139) = 4.67, p < .01, d = 79, and those of their teachers t (139) = 3.28, p < .01, d = .56.

Figure 2
Expectancies of success in scientific-technological and other type of studies

Regarding the logistic regression analysis, the model was statistically significant for both boys c2 (3, n = 135) = 12.85, p < .01, and girls c2 (3, n = 141) = 52.54, p < .01, explaining 15.4% and 41.5% (Nagelkerke R2) of the variance in the chosen study mode (scientific and technological or not), respectively. Table 5 summarizes the results of the logistic analysis.

For boys, the variable with the highest statistical significance was their expectancies of success, with an odds ratio of 3.13 (p = .005), suggesting that students with high expectancies of success in science and technology are three times more likely to enroll in scientific and technological studies. The second significant variable was parental expectancies, with an odds ratio of 0.32, indicating that students whose parents express high expectancies of success are more likely to enroll in non-scientific and technological studies. Finally, teacher expectancies did not make a significant contribution to the model (p = .165).

On the other hand, for girls, the variable with the highest statistical significance was also their own expectancies of success, with an odds ratio of 9.22 (p < .001). These values indicate that girls with high levels of expectancies of success in science and technology are nine times more likely to choose scientific and technological studies. Additionally, teacher expectancies also influence their choice of studies, but in an inverse manner, as indicated by the odds ratio of 0.32 (p = .011). Therefore, female students whose teachers have high levels of expectancies of success are less likely to enroll in scientific and technological studies. Finally, parental expectancies do not influence their decisions (p = .154).

Table 5
Predictive model for scientific-technological type of studies

Regarding the choice between vocational education and baccalaureate studies, the implications differ for boys and girls. Surprisingly, in the case of boys, those with higher expectancies of success chose vocational education over baccalaureate, and there were no differences in parental and teacher expectancies of success between the two groups. However, in the case of girls, those with higher expectancies of success from their parents and teachers chose baccalaureate studies. Therefore, these results suggest that educational measures aimed at encouraging girls to pursue baccalaureate studies should focus on the development, implementation, and practice of tools and methodologies that improve the perception of teacher and parental expectancies of success in girls.

Regarding the choice of scientific and technological studies, boys with higher expectancies of success in science and technology chose an educational option related to this field. Additionally, the results show that girls pursuing studies related to science or technology outperform their peers in terms of high levels of their own expectancies of success, those of their parents, and those of their teachers. Therefore, educational initiatives aimed at increasing interest in scientific and technological studies should provide experiences that enhance students’ expectancies of success, such as the use of inquiry-based teaching methods (Muñoz-Domínguez et al., 2022). Furthermore, in the case of girls, these educational measures should include specific strategies to effectively convey parental and teacher expectancies of success to students.

Despite their importance and interest, it is important to consider the following limitations when interpreting these results. Firstly, since the sample for this study was collected using convenience sampling techniques, the results cannot be generalized to all Spanish students. This underscores the need for further research with a representative sample. Secondly, the expectancies of success from parents and teachers were collected based on the beliefs and perceptions of the students. It would be interesting to replicate this study by also surveying parents and teachers and analyzing whether their self-reported expectancies of success, compared to the students’ beliefs, provide any additional explanatory value.