Abstract:

The concern of the knowledge society for the quality of university education has stimulated the use of indicators for diagnosing and predicting the results of the academic activity of university students in the social sciences.

In this context, the increasing generation of nested data by the university system of a hierarchical nature should not be ignored, otherwise errors are made when assigning the same value of variables to macro (context) units to micro units, or when making inferences about individual dynamics on the basis of aggregated data.

Therefore, our hypothesis is that the academic performance of university students is influenced by the context in which it develops, that is, by the interaction between the individual characteristics of students and the variables describing the different activities and contexts in which university institutions develop, within the framework of a hierarchical data structure.

The methodology used to estimate and validate our hypothesis is based on a two-level multilevel model in which students, level 1 units, are nested within larger units such as groups, courses, grades, etc., which make up level 2 units. These hierarchical models include a wide range of individual and contextual variables and provide details of the different environments in which the phenomena of university academic performance unfold, allowing us to overcome the problems arising from the atomistic and ecological fallacies.

The model has made it possible to study the academic reality of the university at different levels and between levels. That is, to differentiate the variance explained by each predictor at the 2 selected levels of aggregation and to make inferences with the variables that act at the student and institutional levels of the Bachelor of Commerce at the Complutense University of Madrid.

Moreover, we find that the nesting effect modifies the classical interpretation of the coefficients accompanying nested predictor variables, since aggregation levels have significant effects on student level through global, relational, analytical or structural predictor variables, which act above the basic level.

Keywords:

Nesting, hierarchy, academic achievement, interaction, context, hierarchical model.