Sociocultural and demographic factors that influence academic performance: The pre-university case of the Universidad Politécnica Estatal del Carchi


Eduardo Javier Pozo-Burgos1 , Marco Rubén Burbano-Pulles1 , Jack Iván Vidal-Chica2 , Gabriela Elizabeth Revelo-Salgado1

1Universidad Politécnica Estatal del Carchi (Ecuador)
2Instituto Superior Tecnológico Sucre (Ecuador)

Received June 2021

Accepted November 2021


The pre-university course established at the Universidad Politécnica Estatal del Carchi (UPEC) after the disappearance of the Sistema Nacional de Nivelación y Admisión (National Equalization and Admissions System, known by its Spanish acronym as the SNNA) posed new challenges for the organizational and academic structure of this university. To start with, the challenge began with the adaptation of a curricular structure according to the educational needs of the nine undergraduate degrees offered at the university. It was followed by the implementation of the program, and finally its evaluation. The present research was designed and carried out as part of the latter goal. Its main objective was to determine the sociocultural and demographic factors that influence the academic performance of the students in the course. For the purposes of this investigation, from a methodological perspective, academic performance was associated with the number of subjects passed by the student. Using multivariate dependency techniques, a logistic regression was performed, which determined the degree of incidence of sociocultural and demographics factors on academic success. After applying the model, the results indicated that the demographic factors in this case do not influence academic success, while variables such as the educational level of the student’s parents, the student’s grade on university entrance exams and work status are significant when it comes to establishing a cause and effect relationship in this case study.


Keywords – Academic achievement, Regression analysis, Higher education, Entrance examination.

To cite this article:

Pozo-Burgos, E.J., Burbano-Pulles, M.R., Vidal-Chica, J.I., & Revelo-Salgado, G.E. (2022). Sociocultural and demographic factors that influence academic performance: The pre-university case of the Universidad Politécnica Estatal del Carchi. Journal of Technology and Science Education, 12(1), 147‑156.



    1. 1. Introduction

A student’s academic performance (AP) covers a conglomerate of variables that are not based solely on a grade. In order to obtain a deep understanding of what is involved in this construct in an educational context, it is necessary to know and evaluate the factors that have an influence on it, as well as the variables that allow us to predict it. The idea focuses on meeting the various needs of educational institutions, which would allow us to solve problems associated with this setting (Nájera-Saucedo, Salazar‑Garza, Vacío-Muro & Morales-Chainé, 2020)

For the purposes of establishing a weighted mean value, each institution of higher education established its own evaluation criteria, using benchmarks such as the number of subjects, the number of hours and the grade earned in each of the subjects. In the latter case, if we wish to establish it as a measure of the learning results, it must be considered that these are the product of certain conditions, such as the personal characteristics of the student and the instructor’s teaching strategies, as well as institutional and contextual factors; all these aspects have a direct or indirect effect on academic performance (Garbanzo, 2007).

With regard to research focusing on the analysis of academic performance or retention rates, different analytical methods have been used with different approaches to their analysis, as indicated by Alvarez, Callejas and Griol (2020), emphasizing parameters such as family background, socioeconomic status of the family and results on entrance exams.

Sánchez, Naranjo, Vidal, Salazar, Pérez and Jaramillo (2021) indicate that academic performance at the university level is associated with different factors, such as prior academic preparation, access to technology and scores on entrance exams. For this research, other additional variables were considered, such as those of a socioeconomic nature, including monthly family income, residence in rural areas, type of dwelling and gender.

Study habits and learning styles are very closely related to academic performance. This was demonstrated by Magulod (2019) in a study of 75 science students at Cagayan University in the Philippines. This same study also found a relationship between the father’s occupation and the type of school of origin and learning styles; it also detected a relationship in the students’ study habits when grouped according to academic level in secondary school, writing skills, the educational level of the mothers and test anxiety. The academic offerings of the universities also strongly influence student satisfaction. In a study conducted between 2014 and 2018, at the Universidad Politécnica de Cartagena, in Spain, with subjects in the field of Electronics, it was shown that by implementing new teaching methodologies, such as flipped classrooms, project-based learning, discussion sections and proprietary methodologies, student satisfaction with the classes increased by 15% (Hinojosa, Martínez-Viviente, Garcerán-Hernández & Ruiz-Merino, 2020).

Furthermore, the current use of ICT – and even more so as the result of the COVID-19 pandemic – is a key point in university academic activities. A study on 76 public universities in Thailand sought to identify the basic digital competences required in students by interviewing 1126 specialists; it identified seven components related to digital information: fundamentals, access, use, creation, communication, management and evaluation (Suwanroj, Leekitchwatana & Pimdee, 2019). But these competences are rendered useless when the student does not have access to the Internet or a computer. As a matter of fact, Suana (2018) conducted a study on 10 schools in urban areas in the province of Lampung, in Indonesia, focused on gender-related topics, with a group of 798 students in their final years of study. This study indicated that there were no gender gaps in this regard, and that women even had greater access to the Internet and computers and were the ones who made the most academic use of these resources.

Research on a global level has demonstrated that there are other factors that influence academic achievement. For example, in a public university in Colombia, the performance of 460 students in an undergraduate program was analyzed, identifying twenty-two attributes grouped into four factors: students, family, community and university, which allowed a correct classification of 91.7% to be obtained when associating academic performance with human factors, as opposed to material resources (Castrillón, Sarache & Ruiz-Herrera, 2020). Likewise, at the Universidad Peruana de Ciencias Aplicadas, a study was conducted with 28,586 students from the College of Information Systems and Computing between 2009 and 2014, using Knowledge Discovery in Databases (KDD) and decision tree methodology, obtaining a correct classification rate of 90.69% (Apolaya, Espinosa & Barrientos, 2015).

In a study conducted at the Universidad Agraria La Molina in Peru, analyzing 914 students during the second semester of 2013 and the first semester of 2014, four models were generated to predict passing the General Statistics course; the logistic regression method obtained an accuracy rate of 68.4%; the decision trees with J48 showed a rate of 68.3%; neuronal networks were 67.9% accurate; and the Naive Bayes classifier had an accuracy rate of 71.0% (Menacho-Chiok, 2017).

When researching online education, the influence of social networks on the students’ daily activities must also be considered. In this regard, Lau (2017) conducted a study at the University of Hong Kong in which 348 students from eight different colleges participated. Using hierarchical regression analysis, this researcher reached the conclusion that if social networks were used for academic purposes, there was no significant influence on performance, but if they were used for non-academic purposes, particularly in the case of video games, they had a direct relationship with low academic performance.

With regard to the above, the fundamental objective of this study is to determine the sociocultural and demographic factors that influence academic performance in the lower levels of the Universidad Politécnica Estatal del Carchi, as well as to propose a model estimating the probability that students fail at least one subject as opposed to academic success.

2. Methodology

2.1. Participants

The study population consisted of 444 students who entered the equalization system at UPEC, the list of which was provided by the national Secretariat for Education, Science and Technology and Innovation (known by its Spanish acronym SENESCYT) for the second semester of 2020, which was held online between November 2020 and March 2021. The students aspired to nine different degrees offered at the institution in the different programs. During the equalization period, all the students studied the generic subjects of Language and Communication and Mathematics; in addition, they also took a course called Introduction to the Profession, which is specific for aspiring students of each degree.

Two surveys were administered during the academic semester: the first was administered at the start of the period in order to identify the sociocultural characteristics and technological aspects of the students; the second was given at the end of the equalization period, in order to determine the level of academic demand, motivation, instructor attitude, instructional procedures and group management. Consent was obtained from 444 students, of whom 37% were male and 63% were female. The average age was 20, and most lived in an urban area (62%). With regard to parental education, the largest percentage corresponded to basic education (47%). 7% of the students worked and studied at the same time, and 93% of the students had Internet access at home. Consent was obtained from all participants to publish the results on a general level.

It should be mentioned that 506 students were originally admitted, of whom 62 dropped out; for this reason, they were not considered for the purposes of this research study.

2.2. Data

The data on student grades in Mathematics, Language and Communication and Introduction to the Profession were obtained from the academic records of the UPEC for the semester that ran from November 2020 to March 2021.

Table 1 shows a summary with the measures of central tendency for Mathematics, Language and Communication and Introduction to the Profession, as well as the final average corresponding to the non‑weighted mean of the grades earned by the students over the course of the semester in the aforementioned courses.

The grades in Mathematics, Language and Communication and Introduction to the Profession and the final average are calculated with a maximum grade of 10.

It must be added that the present study is focused on academic success. Table 2 presents the proportion of students who passed the different subjects, taking into account that the minimum passing grade is a 7.



Language and Communication

Introduction to the Profession

Final Average











Standard deviation
















Coefficient of variation









Table 1. Descriptive summary of the data




Language and Communication


Introduction to the Profession























Table 2. Pass and fail frequency

2.3. Materials and Methods

For the analysis of academic performance, the use of multivariate dependence methods through logistic regression is proposed, since it is a robust methodology in case the strict assumptions of normalcy and homoscedasticity are not met.

In this case, there is one single categorical dependent variable, represented by academic performance, and several independent variables (sociocultural and demographic factors). The dependent variable is related to academic success/failure based on passing the courses and academic performance; to determine this, two categories were assigned, according to the number of courses passed; if the student has passed all three subjects, they are assigned to category “1”, and if they have passed one or two courses, they are assigned to the category “0”, as indicated below:


It should be clarified that these nominal categories were constructed based on the general student regulations at UPEC, in Art. 5 of the admissions guidelines, where it indicates that the equalization phase for the degree program must be successfully completed (UPEC, 2019).

The logit model fit was evaluated by means of a Confusion Matrix and the Receiver Operating Curve (ROC). To supplement this, a Hosmer and Lemeshow test was applied, in which the null hypothesis (Ho) states that there are no differences between the real and predicted values.

3. Results

The independent variables associated with the logit model in this study – and the degree of incidence on the performance of which we intend to determine – were classified as follows:

Sociocultural factors: labor aspect, educational level of the parents, prior academic training and performance (grade on the university entrance exam), number of children.

Demographic factors: gender, age, income level in the home, area of residence.

Table 3 shows the study variables and their characterization.


Type of variable

Assigned values

Labor aspect


1 = Works

2 = Does not work

Father’s educational level


1 = None

2 = Basic education

3 = Mid-level education/Baccalaureate

4 = Higher education/Graduate degree

Mother’s educational level


1 = None

2 = Basic education

3 = Mid-level education/Baccalaureate

4 = Higher education/Graduate degree



1= Female

2= Male

Area of residence


1= Urban

2 = Rural

Household income


1 = Quintile 1

2 = Quintile 2

3 = Quintile 3

4 = Quintile 4

5 = Quintile 5



Age in years

Number of children


Number of children

Grade on university entrance exam


Student’s score on the university entrance exam, with a maximum score of 1000

Table 3. Characterization of the study variables

The model makes it possible to correctly estimate 92.8% of the cases (Table 4), discarding some variables that are not statistically significant, which is to say that in statistical terms, their effects are not relevant on the academic performance in the lower levels at UPEC during the second semester of 2020.

Generally speaking, the model indicators allow us to conclude that the selected factors (entrance exam grade, father’s level of education, work status) help explain the likelihood of whether the students of the equalization course at UPEC will pass all three subjects in the course (academic success), which is reflected by the percentage of correct predictions obtained by the model.

Independent variables

β coefficient

Wald statistic

Level of significance (p)


Odds Ratio

Entrance exam grade (EG)





Father’s level of education (None)





Father’s level of education (Basic education) (FEBE)





Father’s level of education (Mid-level education/Baccalaureate) (FEMB)





Father’s level of education (Higher education/Graduate degree) (FEHG)





Does not work (NW)










Table 4. Results of the logistic regression model

2LL Initial

-2LL Final

Correct Percentage

Hosmer-Lemeshow Test




Chi square = 7.803 p = 0.453

Table 5. Statistical summary of the logistic regression model

For this model, the three predicting variables were significant that allow us to estimate academic success using the following equation:





After analyzing the results of the model, it can be stated that the entrance exam grade had a positive coefficient and the odds ratio is greater than one; therefore, as the entrance exam grade increases by one unit, the student is 1018 times more likely to pass all the subjects.

The variable referring to the father’s level of education has a positive coefficient for higher levels of education; this indicates that students whose fathers have a basic level of education have a 12050 times greater chance of achieving academic success than those whose fathers had no level of formal education, those with Baccalaureate have a 9012 times greater chance and those with a higher education and/or graduate degree have a 6391 times greater chance.

The variable referring to the work aspect (Does not work) has a positive coefficient (β = 1.459) and an odds ratio greater than 1, which means it can be stated that students who do not work are 4304 times more likely to achieve academic success as compared to students who work and study at the same time.

Upon analyzing the confusion matrix, in which both the correct and incorrect estimations are quantified using the original data, it can be concluded that the percentage of correct predictions by the model is 92.8%.



Passing 1-2 subjects

Passing 3 subjects

% Correct cases

Passing 1-2 subjects




Passing 3 subjects




Overall percentage




Table 6. Confusion matrix of the logistic regression model

The area under the ROC curve (Figure 1) allows us to conclude that the model distinguishes students who achieved academic success from those who failed at least one subject with a probability of 0.7819.