6. Discussion
p. 81-85
Texte intégral
1The purpose of this study was to employ different multivariate techniques to analyze the relationship between the saber11 test, socioeconomic variables and the saber pro test in 22,202 students from 29 universities in Bogotá, Colombia. Regarding the first research question, what is the best way to combine seven socioeconomic variables for producing a single ses index?, the mca model allowed the use of variables measured at categorical and lower level (nominal) for explaining 62.44 % of the variance of the socioeconomic factors. It would be interesting to ask for more information related to the students’socioeconomic condition in order to improve the quality of the index as a predictor value. The parents’educational level, the students’strata and the family income were the best variables represented in the mca solution space.
2Nevertheless, it is highly recommended to review the way in which information related to parents’occupation is collected, considering that those variables showed the lowest correlation with the dimension 1 of the mca. In the case of this study, the information given by the student strata added value to the explanation of the ses index. Thus, there is a partial agreement with the available literature, particularly with Sirin (2005) who states that “… Duncan, Featherman, and Duncan’s (1972) definition of the tripartite nature of ses (…) incorporates parental income, parental education, and parental occupation as the three main indicators of ses” 81 Concerning the second research question, what is the relation between the ses index,saber11 test outcomes andsaber pro test outcomes?, a moderate relationship between ses index and saber11 (symmetric measures closer to 0.3) was observed, but this relation was greater than the relation existing between the ses index and saber pro (symmetric measures closer to 0.2). A test like saber11 is more closely related to the socioeconomic conditions of the students, because it might be measuring the degree to which a student has access to better information or better preparation for the test. As reported by Zwick (2012), there are two hypotheses related with this point:
The content hypothesis (…) is that the test questions are not well tied to the high school curriculum and focus on material that is more familiar to students from wealthier families. Therefore (…) test-takers do not have equal opportunities to learn the material (…) The coaching hypothesis says that coaching, or test preparation, is more likely to be available to wealthier test takers, creating an association between the test scores and socioeconomic status. (p. 25)
3Apparently, this finding represents some evidence for thinking about redesigning the saber11 test, leading to the proposal of making it a competence test instead of an achievement test. Moreover, a considerable relationship was found between saber11 and saber pro, despite each test’s different levels of measurement. There is consistency with the literature, considering that saber11 could represent a measure of previous knowledge of pre-university students. Li, Chen and Duanmu (2010) state that: “prior academic achievement is a key academic predictor of the students’further achievements at higher levels of study. A number of studies have shown that it plays a dominant role in predicting students’learning outcomes.” (p. 391). As a consequence, the saber11 test represents a predictor of academic achievement. It is undoubtedly imperative to discuss more deeply the importance of standardized tests in order to assess not only the level of performance but also the quality of the Colombian educational system.
4In relation to the third research question, is it possible to use socioeconomic variables to classifysaber11 andsaber pro performance group’s membership?, the mda and Logistic Regression did not report a correct percentage of classification larger than 50 % for both cases, although more correctly classified cases were found in the saber11 performance. An increase in ses conditions, i. e., an increase in the ses index, implies an increase in academic performance. Specifically, it is more likely to improve in the saber11 performance than in the saber pro performance when there is an improvement in the ses conditions. This finding is aligned with the results reported in the second research question of this study and thus leads to the conclusion that students’academic performance is not solely determined by their socioeconomic conditions, but also by academic, psychosocial, cognitive and demographic factors (McKenzie & Schweitzer, 2001).
5 Which are the most significant predictors in these classifications? According to the mda, the mother’s educational level and family income were the best predictors for saber pro performance and the variables that discriminate saber11 performance the most were student stratum and the father’s educational level. The sisben level had a low discriminating power in both cases. Of all factors examined in the literature, a family’s ses is one of the strongest correlates of a student’s academic performance (Sirin, 2005). In summary, the ses index does a better job for classifying the saber11 performance than the saber pro performance.
6With respect to the fourth research question, are there significant differences in the Critical Reading and Quantitative Reasoning centroids between: a) thesaber11 performance groups?, there are significant differences in the Critical Reading and Quantitative Reasoning centroid among the saber11 performance groups. In fact, the highest scores for the saber pro competences were estimated for the saber11 high performance group. Particularly, more homogeneity of variance was determined for Critical Reading among the saber11 groups than for the Quantitative Reasoning scores. In contrast, there are no significant differences in the Critical Reading and Quantitative Reasoning centroid among the ses index groups. In relation with the fifth research question, the effect of belonging to a different saber11 performance group on saber pro outcomes does not differ for students of high, middle and low ses.
7Relative to the sixth research question, how universities affect the progress of the students in Quantitative Reasoning and Critical Reading?, there are differences among universities in achievement in Quantitative Reasoning; apparently, universities give added value to the progress in Quantitative Reasoning for the most able students. In fact, the most able students vary more than the least able students at the university level, which could be explained whether the student’s major is considered. It is expected that students with prior high level of performance in area such as Mathematics (i. e., a higher saber11 score) choose programs such as Engineering, Physics, Mathematics, Economy. If this is the case, students must take more courses oriented at developing elements related to mathematical thinking. Thus, a measure of their Quantitative Reasoning at the end of their university career could be higher than the measure of a student who has graduated in another discipline. In the same way, there are differences among universities in achievement in Critical Reading; possibly, universities give added value to the progress in Critical Reading for the least able students. In fact, the least able students vary more than the most able students at the university level. Apparently, students are developing elements related to their Critical Reading competence regardless of the program in which they are enrolled. Consequently, it would be appropriate to propose a three level model, where information about the students’academic program can be included.
8 Are there different university effects for saber pro students’outcomes according to their ses index? In relation to the ses index, students from different socioeconomic conditions tend to vary the same at the university level in both Quantitative Reasoning and Critical Reading. As a matter of fact, the least wealthy students have more variance in Critical Reading and the wealthier students vary more in Quantitative Reasoning. It is possible that the findings described above would remain the same without distinctions made by the ses index.
9Regarding the methodology employed in this study, the multilevel modeling allowed for the use of all the data to obtain inferences about universities. Some advantages of this modeling methodology are reported in literature such as Goldstein (2011). However, there are some unconsidered variables at level two (universities) that could be explaining the results. For example, the faculty to student ratio, the educational level of faculty members and the number of research groups at the university. Undoubtedly, however, student achievement can be explained due to individual predictors at student, teacher, and school levels. Given this, a better understanding of predictors at school level is necessary, through the fit of new models which also allow for cross-level interactions among all possible predictors at student, teacher, and university levels.
Le texte seul est utilisable sous licence Licence OpenEdition Books. Les autres éléments (illustrations, fichiers annexes importés) sont « Tous droits réservés », sauf mention contraire.
Socioeconomic Factors and Outcomes in Higher Education
A Multivariate Analysis
Carlos Felipe Rodríguez Hernández
2016