Student Profiles and Academic Achievement in Mexican Middle Schools

The aim of the study was to describe student profiles associated with educational achievement from four sets of personal variables: psychological, academic, health, and technology use. A representative sample of middle school students from Baja California, Mexico, was given an ad hoc questionnaire and academic results were obtained from a standardized academic achievement test. Cluster analysis k-means allowed us to define two student profiles. Equal numbers of students were found in both groups, with Cluster 1 grouping high-achieving students and Cluster 2 those with a lower level of educational achievement. All of the personal variables help to describe and differentiate between the two groups, except for the number of times they have changed schools.

Although such analytical procedures contribute to configuring various types of student profiles, entailing an understanding of the variables related to educational achievement, this type of technique has not been used frequently and the variables involved tend to be linked to, and restricted to, specific fields, which suggests the use of more comprehensive analysis schemes (De Groot et al., 2015).Existing studies that focus on student profile characterization tend to include personal variables, either of a psychological or affective-motivational nature.

Variables Analyzed in Studies on Student Profiles
Motivation.This variable has received most attention in studies on student profiles (Hayenga & Corpus, 2010).Valle et al. (2009) found that students who were motivated to learn and achieve better results than others, and who had high levels of perception of their own academic skills, displayed better academic performance.Corpus and Wormington (2014) observed that students included in the intrinsic motivation cluster displayed higher academic performance.activities, which are in turn linked to the selection and use of learning strategies (Bipp, Kleingeld, Van Den, & Schinkel, 2015;Bjørnebekk, Diseth, & Ulriksen, 2013;Dull et al., 2015;King & Mcnerney, 2016;Rogers, 2013).In a study conducted by Gargallo and Suárez (2014) with high-performing students, it was found that these students are characterized by better learning strategies (cognitive, metacognitive and support strategies) and motivation toward mastery.In another study, Valle et al. (2008) concluded that students who self-regulate their learning had better educational results.Linnakylä and Malin (2008) showed that students with better academic performance display a higher degree of school engagement, commitment, and organization in schoolwork, and respond to various academic demands in a timelier manner.A study by Bravo et al. (2015) revealed that students' educational expectations were one factor that explained performance in mathematics.
Self-esteem and self-concept.The way in which a student perceives and describes him or herself has proven to be a variable associated with educational achievement (Hassan, Jami, & Aqeel, 2016;Moradi-Sheykhjan, Jabari, & Rajeswari, 2014).Imhof and Spaeth-Hilbert (2013) observed that students with positive academic self-concept and high levels of invested effort showed higher learning and lower dropout rates.
Physical exercise.Physical exercise is a factor associated with academic achievement that has recently been included in studies on student profiles.Pellicer et al. (2015) found that students from the cluster with the highest level of physical exercise displayed higher academic performance.
Internet Use.Torres, Duart, Gómez, Marín, and Segarra (2016) found that students who downloaded software and audio and video content, and who made use of all entertainment possibilities, were less likely to fail courses than students who only used the Internet minimally.

Other Variables in Psychological and Educational Literature
Although the findings from these studies that focus on student profile characterization have served to expand the bodies of knowledge on factors associated with educational success, the exclusion of variables explored in the most recent findings in psychological and educational literature limits the explanatory power of the models proposed and our understanding of them as an object of study in a broader context.

Academic performance variables.
Various studies have produced results that point to previous academic performance as a predictor of academic success (Gómez, Oviedo, & Martínez, 2011;Ocaña, 2011;Regueiro, Suárez, Valle, Núñez, & Rosário, 2015).Another variable commonly associated with academic achievement is attendance.Ning and Downing (2015) showed that motivation to attend class was strongly linked to school achievement, while attendance and punctuality were associated with better grades in school.Study expectations are also positively associated with educational achievement (Figueroa, Padilla, & Guzmán, 2015;García-Castro & Bartolucci, 2007;Shaw, 2013).
Authors like Griffin, MacKewn, Moser, and VanVuren (2012) have discussed the importance of concentration and attention in improving learning while studying.Furthermore, authors such as Bove, Marella, and Vitale (2016) have shown that students who perceive a better school environment tend to show better academic results.Additionally, there is a relationship between preschool attendance and academic achievement.Children who attend preschool perform better in academic tasks (Aslan & Arnas, 2014).
Finally, there is some evidence that the age of entry into elementary education is a determining factor in educational achievement (Sakic, Brusic, & Barbatovic, 2013).In addition, Anderson (2017) stated that changes of schools are a variable associated with educational achievement in middle school students.
Health variables.A relationship exists between health-oriented behavior and academic achievement.Kristjánsson, Sigfúsdóttir, and Allegrante (2010) showed that personal healthcare variables such as a healthy diet and physical activity are associated with higher educational achievement in university students.Substance use is one of the factors most strongly associated with low performance and school dropout rates.Hill and Mrug (2015) found that consumption of tobacco, alcohol, and marijuana, and a combination of these, was associated with low academic performance.Sung, So, and Jeong (2016) showed that junior high school students who often consumed alcohol had, on average, lower school grades.Specialized literature has documented the repercussions of sexual activity on adolescent development.Price and Hyde (2009) demonstrated that early sexual activity in adolescents was associated with other risk behaviors such as substance use, low academic performance, and dropping out of school.
Computer skills.A study by Ziya, Dogan, and Kelecioglu (2010) showed the order of importance of predictive variables for PISA achievement scores in mathematics: (a) self-reliance in performing Internet-related operations; (b) using computers for program and software purposes; (c) self-reliance in performing operations requiring a high level of computer skills (using the word processor, using an electronic tabulating program to draw graphs, prepare presentations, prepare multimedia presentations and design web pages); and (d) using computers for Internet and entertainment purposes.However, variables (b), (c), and (d) had a negative effect on scores in mathematics.
In sum, broader sets of variables, which provide a more detailed explanation of student profiles, have yet to be incorporated.Therefore, the aim of this study was to characterize middle school student profiles associated with educational achievement by including four sets of variables, namely psychological variables (analyzed in studies on student profiles) and three other sets of variables: academic, health-related, and those concerning technology use.All of these have been reported as explanatory variables in the literature.

Participants
The information analyzed in this study was taken from the database of a research project (Contreras, Rodríguez, Caso, Díaz, & Urias, 2012).The sample was made up of 21,724 students (50% male and 50% female) aged between 11 and 16, from 88 schools in the state of Baja California, Mexico, and made up as follows: general (57.8%), technical (26.3%), private (9.7%), and distance-learning junior high program (6.2%).

Instruments
Student questionnaire.The questionnaire comprised 26 variables divided into four aspects: psychological, academic, health, and technology use variables (see Table 1).The levels used to measure each variable are given in the Appendix (Table A1).The instrument includes acceptable reliability and validity values [the name of the study in which the validity was analyzed was omitted to respect the double-blind review].

Data Analysis
In order to sort the high number of subjects in the sample into two groups, a k-means cluster analysis was performed based on the set of variables mentioned.Prior to that, it was necessary to recode and standardize the values that are assignable to these variables to homogenize the measurement scales so that they all had a mean of 0 and a typical deviation of 1.

Results
After 20 iteration steps, the final solution was reached with 12,202 students.However, the distances between the centroids were minimal from the first five iterations (see Table A2 in the Appendix).In cluster 1 there were 6,360 students, and 5,842 students were grouped in cluster 2. The final mean values for each variable in both clusters are shown in Table 2, and reflect the characteristics of the prototypical case for each cluster.Table 3 shows which variables contributed to the cluster solution.Variables with high F values provide greater separation between clusters.The variables are arranged in descending order by their F value, and all are significant except for the number of times the student changed school.Thus, cognitive strategies for study were the highest contributor to the final solution, and the age of entry into elementary education had the least impact.(Hill & Mrug, 2015;Sung et al., 2016), and considered these substances harmful to health.In addition, their attitude toward personal healthcare and the environment is positive, which leads to a healthy diet and more regular exercise (Pellicer et al., 2015).It has been confirmed that many factors contribute to educational achievement, and there is a diverse set of characteristics that may influence students' academic results.As a result, explanatory models of achievement must acknowledge the multi-factorial nature of this construct.It follows from all the above that using k-means clusters was a useful data mining technique in describing student profiles (DeFreitas & Bernard, 2015;De Groot et al., 2015).
Finally, it is worth noting that one limitation of this study is the fact that the variables used were not planned in advance but were taken from a preexisting database.This restricted the ability to obtain a better-configured student profile.

Likert scales)
• Academic self-concept (score based on 1-4 Likert scales) • Academic expectations (junior high school, high school or technical course, bachelor's degree, master's degree or doctoral degree) • Causal attribution of low grades (I never get low grades, financial problems, family problems, problems understanding teachers, health problems, laziness) • Problems concentrating while studying (score based on 1-4 Likert scales) (can't remember, 5 years old or less, 6 years old, 7 years old, 8 years old or older) • Number of times he/she has changed school (never, once, twice, 3 times or more) • Lack of punctuality in attending class (never, once or twice, 3 or 4 times, 5 or more times) • Perception of school environment (score based on 1-4 Likert scales) • Exam preparation activities (score based on 1-4 Likert scales) • Enjoyment of reading (I don't like it at all, I like it a little, I somewhat like it, I like it a lot) • Learning strategies (score based on 1-4 Likert scales) years old or more) • Age at which alcohol was first consumed (I have never drunk alcohol, 10 years old or younger, 11 years old, 12 years old, 13 years old, 14 years old or more) • View of the use of harmful substances (score based on 1-4 Likert scales) • Personal healthcare (score based on 1-4 Likert scales) • Healthy diet (score based on 1-4 Likert scales) • Physical exercise (never, sometimes, very often, always)

Copyrights
Copyright for this article is retained by the author(s), with first publication rights granted to the journal.
This is an open-access article distributed under the terms and conditions of the Creative Commons Attribution license (http://creativecommons.org/licenses/by/4.0/).

•
Use of the Internet to read e-mails • Use of the Internet to chat • Use of the Internet to read the news • Use of the Internet to consult a dictionary or encyclopedia • Use of the Internet to look for information on a specific topic • Use of the Internet to take part in debates or forums Figure 1 sh the one ha favorable p low-achiev Figure 2 proposed b excellent p insufficien

•
Use of the Internet to take part in social networksNational Assessment of Academic Achievement in Schools(ENLACE, in Spanish).ENLACE scores in Spanish and mathematics was used.ENLACE is an assessment carried out by the Secretariat of Public Education for informational purposes and taken by students from public and private schools.(http://www.enlace.sep.gob.mx/).

Table 2 .
Central values for the final solution

Table 3 .
ANOVA values for each variable grouped across both clusters