INTRODUCTION

Metacognition is a widely studied construct and different studies show its importance for success in learning (^{Flores and Pacheco, 2020}; ^{Harris and Graham 2017}; ^{Tobias and Everson (2009)}; ^{Embleton, 2023}) and for the development of critical thinking (^{Bernal et al., 2019}; ^{Lozano and Lozano, 2016}; ^{Suárez and Gonzalez, 2021}; ^{Aguilar-Perez, et al., 2023}; ^{Pérez, et al., 2023}). More recently, some research also shows its relationship with self-control conflicts and how to learn to resolve them successfully (^{Bürgler et al., 2022}), and with the use of social networks and disorders related to internet gaming (^{Cavalcante-Pimentel, et al., 2022}; ^{Akbari et al., 2023}).

The origin of metacognition dates to the 1970s, when ^{Flavell (1971)}, based on his research on cognitive development inspired by the work of Jean Piaget, introduced the concept of metamemory. Later ^{Flavell (1976)} defines metacognition as the knowledge that people have about their cognitive process and the way it regulates their learning. In this definition, two aspects of metacognition are appreciated: knowledge of cognition referred to the awareness of one’s own knowledge, metacognitive regulation, or self-regulation of learning (^{Flavell, 1979}; ^{Brown, 1983}; ^{Cerezo, et al., 2019}).

Flavell (1979) emphasizes that the metacognitive experiences on which he refers can be short or extensive, of a cognitive or affective nature, based on continuous observation and reflection as strategies to assess the processes in relation to the cognitive objects on which they are based, to become aware and establish new goals, so that theoretical understanding involves metacognitive knowledge, metacognitive experiences, strategies and goals or objectives, in the interweaving of the metacognitive regulation process.

^{Flavell (1985)} emphasizes in his model about the interaction and collaboration between internal cognitive factors and the environment for the construction of knowledge. He also points out that in the cognitive encounter with the world, assimilation and accommodation are present with equal importance, which allows us to appreciate the relevance given to previous learning or prior knowledge, and to learning environments, as activating and detonating environments of new learning.

On the other hand, metacognitive regulation refers to the ability of students to use information about their thinking to direct or regulate their learning, it implies the ability to think strategically and solve problems (^{Aizpurua et al., 2018}; ^{Solaz y Sanjosé, 2008}; ^{De Jesús, 2020}; ^{Ricardo-Fuentes et al., 2023}), plan, set goals, organize ideas, and evaluate what to know and not to know. For ^{Flavell (1992)}, a large part of cognitive development is self-motivated, so metacognition implies the ability to teach others and make the thought process visible.

Brown (1983) describes three ways in which we direct our own learning. First, planning approaches to tasks, which involves identifying the problem, choosing strategies, organizing our thoughts, and predicting outcomes. Second, exercise supervision of activities during learning, this implies testing, reviewing, and evaluating the effectiveness of our strategies. And finally, check the results, which requires evaluating the results based on specific criteria of efficiency and effectiveness.

For ^{Nelson and Narens (1990)}, monitoring and control are the primary processes of metacognition; in monitoring, strategies for the development of a task are established, and through control, the necessary adjustments and modifications are made. In this way, metacognition allows students to ask themselves about how to develop a task, carry out strategies, self-evaluate and based on the results make the necessary adjustments (^{Wilson and Bai, 2010}; ^{Kleitman and Narciss, 2019}; ^{Escalante, et al., 2023}), that is, it implies high self-regulation processes. Cerezo, et al. (2019) suggest that students self-regulate their learning when they adequately interpret tasks and the contexts, which allows them to activate a path to solving, thus making appropriate decisions, and selecting the required strategies. For ^{Boekaerts et al. (2000)} metacognition has three fundamental aspects, firstly, the subject’s awareness of his weaknesses and strengths in the learning process; secondly, the cognitive resources he has for the development of his academic activities and, finally, the regulation of his processes, which implies that he evaluates his performance with the aim of obtaining better performance in his tasks.

During the last decades, metacognition has been studied by various authors from psychology, pedagogy, and neurosciences, consolidating these understandings and catapulting the ideas previously exposed. ^{Carretero (2001)} refers to metacognition as “the knowledge that people build regarding their own cognitive functioning” (p.191), for ^{Díaz and Hernández (2010)} metacognition consists of “that knowledge we develop about our own processes and knowledge products ”(p.187) and for ^{Frade (2011)} metacognition“ is the thinking ability that consists in evaluating the knowledge that one has, what is known, but also what is thought, felt, and even the movement itself that is carried out or not, so as to reflect on what is missing to know, do or feel ”(p. II).

Complementing, the theory of ^{Schraw (2002)} that postulates the existence of a general metacognitive capacity that allows individuals to know and regulate their own learning process, regardless of the domain. Cognitive skills are necessary to perform a task while metacognition is necessary to understand how the task was performed. Initially, he considered metacognitive knowledge to be a specific domain or task and regulatory skills as those that allow selecting appropriate learning strategies (^{Barrero, 2001}). On the other hand, ^{Winne and Azevedo (2014)} define procedural knowledge as a set of skills that students have on how to perform cognitive work to perform tasks effectively.

Furthermore, ^{Ugartetxea (2001)}; ^{Carvallo (2001)} and ^{Efklides (2009)} refer that metacognition involves affect, knowledge, emotions, feelings and, attitudes and motivation, which are factors that contribute to long-term self-regulation, the motivation induced by metacognitive experiences activates existing cognitive knowledge and metacognitive skills, the control of feelings of difficulty and confidence regulate the efforts in the accomplishment and challenges that come with troublesome tasks and situations. Specifically, for ^{Efklides (2011)} metacognition is divided into two interdependent domains, the first refers to metacognitive knowledge understood as the explicit verbalizable beliefs that a subject has about how his or her mind works and the second is metacognitive experiences, which refer to the feelings that a person experiences when learning, remembering and thinking about his or her own learning. For Arango-Muñoz metacognitive experiences are inherently subjective, that is, they occur regardless of whether they are accurate or not.

Metacognitive control is said to be the ability to evaluate and make one’s own thought processes conscious. Metacognitive control is what enables individuals to regulate their thoughts and interact with their environment according to their desired goals. From a competency perspective, from the socioformative approach, metacognition is conceived as the process by which human beings reflect to become aware of our development in the face of different life activities and to self-regulate said performance in order to avoid mistakes and seek to have more clarity in what we do, beyond the cognitive field (knowing how to know) as it has traditionally been approached, but also in the socio-affective field (knowing how to be / to live together) and in the acting field (knowing how to do). For ^{Arguelles and Nagles (2010)} “metacognition in the perspective of competence development deals with aspects such as learning potential, cognitive processes, variables of cognitive activity, understanding of mental activity and self-regulation processes” (p.118).

Based on the postulates around the subject of metacognition, researchers around the world have developed studies from various perspectives to broaden their understanding. To start the characterization of metacognitive skills at undergraduate level is widely carried out around the world with the purpose of guiding students with respect to the responsibility of assuming the learning process as their own (^{Bortone & Sandoval, 2014}; ^{García et al., 2015}). This awareness of metacognition involves knowledge, control, and regulation of learning, that is, knowledge about one’s own activity of knowing or higherorder capacity, which allows directing and regulating cognitive, affective and motivational processes to achieve a specific objective. Therefore, there are as many metacognition modalities as there are cognitive processes (^{Brown, 1987}; ^{Otero, 1990}; ^{Pintrich et al. 1991}; ^{Ifenthaler, 2012}; ^{Hernández y Camargo, 2017}).

Several studies have explored the metacognitive skills of undergraduate students and the metacognitive strategies they use in different contexts and activities. ^{Maturano et al. (2002)} researched on the cognitive and metacognitive strategies used by undergraduate students to understand an expository text, they worked with a sample of 59 students from different degrees, and concluded that they can take more literal information easily, that they establish relations between different ideas, that their comprehension of the text is limited and their vocabulary is poor. The researchers point out that these results are not those expected for undergraduate students and, therefore, teachers are required to actively engage so that students develop cognitive and metacognitive strategies.

^{Ceniceros and Gutiérrez (2009)} undertook a study on metacognitive skills with the involvement of 218 undergraduate and graduate students from a Mexican university. They used a structured questionnaire to analyze two dimensions of metacognition: self-knowledge (10 items) and self-regulation (10 items) with a Likert scale between 1 and 5. In the first category, the average in each item was greater than 4, which reflects a systematic use of strategies related to the knowledge of their own cognition, students in general are aware of the cognitive processes they use to learn. In the second, the average of each item was close to 4, which shows that students are aware of their self-regulation process.

There are, in turn, studies on metacognitive monitoring in educational settings that are grouped into three models, the first one that is based on the use of Gamma Statistic, evaluating the precision of the judgments about correct performance, and the specificity that measures the precision of judgments about incorrect performance (^{Schraw et al., 2013}). The second, makes a judgment of two performance factors, referring to correct and incorrect and the general monitoring model, which refers to two incident processes in the error and precision process (^{Gutiérrez et al., 2016}; Schraw et al., 2013; ^{Schraw et al., 2014}).

On the other hand, Bortone and Sandoval in 2014 carried out a study with the objective of characterizing the metacognitive profile of first-semester students from different engineering programs. They used the Metacognition Inventory of ^{Sandoval and Franchi (2007)} with a Likert scale between 1 and 5. The results showed that new students had a metacognitive profile between Moderate and High, which the authors consider as an adequate metacognitive ability to manage and control the knowledge they own.

^{Arias et al. (2014)} as part of their study with 273 freshman-year psychology students from three Peruvian universities used Kolb’s Inventory of Metacognitive Strategies and found that a high percentage of students were located at low levels of metacognition, self-knowledge, self-regulation and evaluation. In this study, it is concluded that those students who have established goals and achievement motivation are the ones who have better performance and greater academic success, regardless of their learning style.

^{Campo et al. (2016)} in their study with 462 university students from France and Colombia identified, through a self-report questionnaire, that students use their metacognitive knowledge in text production processes and that planning strategies are used most often by French students. Some of the metacognitive strategies used were: information organization, goal setting, planning, seeking help with tutors, and self-reward. The results of the study also found that there is a positive correlation between the level of performance and metacognitive knowledge, for which the authors conclude that metacognitive knowledge is a requirement for controlling what is produced, and that self-regulation strategies contribute to the planning, editing and revision of texts.

^{Arias and Aparicio (2020)} analyzed the metacognitive awareness of 96 firstcycle university students at the beginning and end of the semester to determine if the training received at the university contributed to their development. They used the ^{Schraw and Dennison (1994)} metacognitive skills inventory adapted for use in Peru. The results show, in general, that participants are at an average level with significant differences in metacognitive awareness, and dimensions of planning and self-regulation, but not in evaluation, and that there are no significant differences by gender or correlation with academic achievement. The researchers suggest that the favorable change observed is due to the fact that teachers have provided students with opportunities for the appropriate use of metacognitive planning and self-regulation strategies related to setting goals, identifying the resources to be used, and supervising and monitoring during the development of the activities.

In a recent study, ^{Gutiérrez & Montoya (2021)} examined the generality / specificity of metacognitive skills (knowledge of cognition: declarative, procedural, and conditional; regulation of cognition: planning, information management, debugging, monitoring and evaluation), with a sample of 507 university students from the education, medicine and psychology programs. It was hypothesized that the existence of statistically significant differences between the domains is an indicator of the specificity of metacognitive skills to a domain, if on the contrary, they are general, that is, they are not a function of the domain of the task of learning, there will be no significant differences. It was found that there are significant differences between the domains for all skills, except for procedural knowledge and debugging. The authors expected metacognitive awareness to be different among medical and educational students, and among medical and psychology students. However, there was no significant difference.

Studies on the correlations between the metacognitive abilities of university students with academic performance, gender, the nature of the institution (public or private) and semester are still incipient, they are at the descriptive and exploratory levels. As an example, ^{Andriani & Mbato (2021)} in a study on academic reading from a gender perspective with a mixed method with open and closed questionnaires and a semi-structured interview. Through a t-test, it was established that only the use of the strategy of guessing the meaning during the planning phase is significantly different. However, the students used the reading strategies differently, with the men using the guessing strategy more during the monitoring phase, while the women taking regular breaks and using other strategies, the women activating their prior knowledge more often. The authors indicate that the only significantly different strategy use was guessing the meaning, used more frequently by men.

On the other hand, ^{Gutiérrez de Blume, et al. (2022)}, sought to establish the differences in metacognitive knowledge: declarative, procedural, and conditional, and regulation: planning, monitoring, and evaluation with a sample of 771 Colombian teachers based on the variables of gender and type of professional performance based on a Manova analysis, finding significant differences between teachers based on level of training and gender.

Finally, ^{Gutiérrez de Blume and Montoya (2023)} explored the relationship between objective and subjective measures of metacognition with a sample of 352 Colombian students. Based on descriptive statistics, zero-order bivariate correlations, simultaneous multiple regression, and multivariate analysis of covariance, conscientiousness and openness were found to be the personality traits that positively predicted metacognition. Objective and subjective measures of metacognition are weakly related, and subjective measures of metacognition are more strongly related to personality traits. Gender affected subjective and objective measures of metacognition.

The above shows the importance of knowing the metacognitive abilities of university students because the identification of strengths and needs allows the development of new learning strategies that contribute to the strengthening of knowledge and regulation of cognition, and this may impact other areas. Additionally, it is important to recognize the effect of variables such as gender, age and semester and the incidence of institutional methodologies in the development of students’ cognitive abilities during the first university semesters, to forge awareness about the learning process and the strategies that are useful individually, contributing to educational practice.

This research analyzes the metacognitive skills of university students from three Colombian universities and establishes whether there are significant differences according to the university of origin, gender, or semester of the participants. More specifically, the following questions were posed:

METHOD

Research Design

In this study, it is quantitative, descriptive and correlational, descriptive and inferential statistics are combined.

Participants

The convenience sampling technique was used to select the study participants, who had to be pursuing the first four terms of their academic programs. Informed consent was requested in accordance with the guidelines of the Colombian Data Protection Law of 2012, for the application of the instrument in three Colombian universities. A total of 1204 students responded to the instrument, however, when analyzing the data, 5 outliers were detected and excluded. Of the 1199 participants 40,6 % were men and 59,4 % were women, as shown in Table 1.

In relation to age, it stands out that 55 % of the participants were under 21 years of age, followed by 19 % between 21 and 25 years of age; approximately 10 % between 26 and 30 years, and the rest distributed in the other three age ranges established as shown in Table 2.

UNAD is a Public University, UAN and UNAB are Private Universities, UNAD participants belonged to distance learning programs and the other participants to face-to-face learning programs. The modality in Colombia, in accordance with decree 1330 of July 25, 2019 of the Ministry of National Education, is the means used integrating a set of organizational and / or curricular options that seek to respond to specific requirements of the level of training and attend to conceptual characteristics that facilitate access to students, in different conditions of time and space.

Instrument

The Metacognitive Awareness Inventory (MAI) designed by Schraw & Dennison in 1994 and validated by ^{Huertas et al. (2014)} and ^{Gutiérrez et al. (2021)} in Colombia, was applied. This instrument has 52 items in total distributed in two categories: Knowledge of Cognition and Regulation of Cognition. Knowledge of Cognition refers to knowledge that subjects have about their own knowledge and comprises three subcategories: Declarative Knowledge (8 items), Procedural Knowledge (4 items) and Conditional Knowledge (5 items). The Regulation of Cognition comprises activities that help the subject control learning and is made up of five sub-processes: Planning (7 items), Organization (10 items), Monitoring (7 items), Debugging (5 items), and Evaluation (6 items). Each item includes a five-point Likert-type scale whose score ranges from 1 (completely disagree) to 5 (completely agree). The overall internal consistency reliability was α = 0,94.

Procedure

The students were contacted directly, and their participation was voluntary. In one of the universities, the students answered the instrument on paper, then the data was entered. In the other two universities, the instrument was designed as a Google form and was sent to students by institutional mail. Students responded anonymously, only age, gender, and semester were collected. Students had to be pursuing the first half of their study programs, that is, be in semesters 1 to 4.

Data analysis

Data processing was done with SPSS v25. For the analysis and presentation of the data, descriptive and inferential techniques are combined. For the descriptive part, for each participant a global value was determined by averaging the 52 items of the instrument that allows knowing the metacognitive level in general, and a value for each of the two components and eight subcomponents averaging respectively the items that comprise them, this in turn allows knowing the level in each one of these. Taking into account that these averages are on a scale of 1 to 5, the interpretation of ^{Figueroa et al. (2012)} to describe the results obtained

This description was combined with inferential techniques. Since the data were not normal, nonparametric tests were used. To determine significant differences in the level of development of metacognitive skills of the students according to the institution of origin, a Kruskal Wallis test was performed. To determine if there were significant differences in the level of development of metacognitive skills of Colombian university students, considering only gender, a T-test was carried out. And to determine significant differences in the level of development of metacognitive skills of Colombian university students, considering only the semester studied, a Pearson correlation was made, the participants were between 1 and 4 semesters.

RESULTS

Within the framework of the research questions that guided this study, the results are divided into three parts. First, a description is made of the metacognitive abilities from all students at a general level and in each of the components and subcomponents evaluated through the MAI, question 1. In a second moment, it is analyzed if there are significant differences when considering the university of origin, question 2. And finally, it is analyzed if there are differences based on gender or semesters, questions 3 and 4.

General analysis

University students Colombians have a high level of metacognition with an average of 3,86 (SD = 0,46), with better knowledge of their cognition than of their regulation with general averages of 3,95 (SD = 0,48) and 3,81 (SD = 0,48) respectively, as observed in Figure 1.

When analyzing the subcomponents of metacognition, it is evident that the aspect that should be strengthened the most is the evaluation with an average of 3,52 (SD = 0,64), that is, university students must improve the analysis they make on the effectiveness that have the strategies they use, verify that what has been done allows them to meet the objective of the activity and has allowed them to learn as much as possible. In contrast, they are stronger in debugging, with an average of 4,08 (SD = 0,57), that is, students identify weaknesses in their learning process and adjust improve their performance, for example, they ask for help, reread, or change strategy when they do not understand or when they are confused.

Figure 1 Average in Each Component and Subcomponent of Metacognition 

Most university students are at the High Level, only 21,4 % are at the Very High Level, at general level, 29 % in knowledge of cognition, and only 17,8 % in the regulation component of cognition. It should be noted that no student is at a very low level and only the 1,4 % is at a low level., as shown in Table 3.

Analysis by Universities

When analyzing each institution, it is evident that in the three universities more than 70 % of students are located at a High or Very High Level both at a general level and in the two components of metacognition: knowledge and regulation of cognition, as shown in figure 2. It should be noted that at UNAD, more than 90 % of students are located at Higher Levels.

Figure 2 Percentage of Students in the Different Levels by University 

When doing a Kruskal Wallis analysis, it was possible to observe that there were significant differences in the level of metacognitive skills of the students according to the University of origin, as well as in each of the components and subcomponents, the p value was less than 0,05 in all cases. By doing a Post hoc analysis it was possible to establish that the differences are significant in favor of the results obtained by the students from UNAD, as shown in Table 4.

UNAB students have higher metacognitive levels than UAN students both at a general level and in each one of the categories and subcategories; all these differences are significant. With respect to UNAB students, UNAD students also have higher metacognitive levels at a general level, in conditional knowledge, in the category regulation of cognition and all its subcategories, except organization. There are also significant differences in favor of UNAB students, in the regulation of cognition and its planning, monitoring and evaluation subcategories, with respect to the results from UAN students.

The figure 3 shows that overall average of UNAD students is 0,21 higher than those from UAN, and 0,14 with respect to those from UNAB, these differences are significant. In the knowledge of cognition, although the average of UNAD students is 0,08 higher than those of UNAB, this difference is not significant, but it is compared to UAN students who are 0,13 below.

Figure 3 Average by University for Knowledge of cognition and its subcomponents 

It stands out that in the Conditional Knowledge subcomponent is where the greatest difference between UNAD and UAN and UNAB is, also in the only cognition subcomponent in which UAN students have a better average than UNAB students, with averages of 3,94 and 3,93 respectively, however this difference is not significant.

Regarding the Regulation of Cognition (figure 4), the students from UNAD have better averages than those from UAN and UNAB, both at the general level and in the subcomponents, and these differences are significant, except in organization, where the students from UNAD and UNAB have the same average. Students from the three universities have greater strength in Debugging, being Evaluation the one that should be strengthened the most.

Figure 4 Average by university for regulation of cognition and its subcomponents 

Analysis by Gender and Semester

Women have better averages both at the general level and in each component and subcomponent (see Tables 5 and 6), however, when doing the U de Mann-Whitney test it was found that the differences are not significant (p> 0,05). It was also found that there are no differences by gender when considering the University.

Using Spearman correlation coefficient, it was possible to establish that there are differences depending on the semester taken by the students (p< 0,05) in all categories and subcategories except procedural knowledge. And contrary to what would be expected, students have better metacognitive levels when they start their studies and as they progress the average decreases as evidenced in Table 7.

DISCUSSION

This study aimed to describe the metacognitive skills in its components Knowledge of Cognition and Regulation of Cognition, and its subcomponents in a group of university students and to establish differences in relation to the university of origin, gender and term of the participants. It was determined that 63,6 % of students are at a High Level and 21,3 % are a Very High Level, that is, more than 80 % of young university students have a good level in the development of their metacognitive skills. These findings are similar to those reported by Arias and Aparicio (2020), where 75 % of students were placed at an Average Level, which would be equivalent to a High Level in this study, and 11,5 % at a very High Level, since those reported by Bortone and Sandoval (2014) and by Ceniceros and Gutierrez (2009) with averages close to 4 in both categories, which shows that students have a reflective and systematic process towards learning processes and the evaluation of results.

The results show that there were differences depending on the Participants’ University of origin, the students from UNAD, a university that carries out all its learning programs based on Distance Modality, have better metacognitive skills, with significant differences in most of the subcomponents with respect to the other two universities. It is possible that this is explained because the distance modality requires autonomy as a capacity to respond independently to the responsibilities acquired in an educational process, additionally, it is an indicator of maturity for learning associated with the discipline to fulfill the acquired commitments or to manage techniques, tasks and time. The construction of knowledge is generated independently, of a place, a time or a teacher (^{Peters, 2003}: ^{Preti, 2005}; ^{Moore and Kearsley, 2007}; ^{Gottardi, 2015}).

Furthermore, although in this study an analysis of metacognitive skills by domain was not made, that is, in the three institutions the participants were from different academic programs but there is no information in this study by training area, the positive outcomes for students in distance modality could support the hypothesis of some researchers those metacognitive cognitive skills, especially in adulthood, are of general domain (Gutiérrez et al., 2016; Schraw et al., 2013) and that other factors can explain it.

It was also found that there is difference by semester, which not grees with the study carried out by ^{Cantillo et al. (2014)} nor with the research of ^{Hermosilla and Ossa (2022)}. These results also differ from the results presented by ^{Martínez-Fernández (2007)} which found that students who are finishing their degree have a greater use of metacognitive strategies with respect to students who are at the beginning or in the middle of their programs. The results in this study, at a general level, show that Colombian university students decrease their metacognitive levels as they progress in their programs, both at the level of cognition knowledge and its regulation.

Regarding gender, although women have higher averages, the differences are not statistically significant, which implies, as pointed out by Cantillo et al. (2014) that it is important to analyze other aspects that are decisive for the development or not of metacognitive skills. In contrast to the results reported by Gutiérrez de Blume et al. (2022), where it was found that women reported significantly higher regulation skills (monitoring and evaluation) compared to men. In subsequent studies, Gutiérrez de Blume and Montoya (2023) found, exploring the relationship between objective and subjective measures of metacognition with a sample of 352 Colombian students, that men obtain better results in monitoring, awareness of their knowledge, and regulation of cognition than women, and women reported overconfidence. Which suggests that it is necessary to deepen studies in this line because it is complex and dynamic.

LIMITATIONS AND RECOMMENDATIONS

The data from this study differ from others and therefore further investigation is required. It is recommended to design and implement strategies aimed at strengthening the development of metacognitive skills in university students and that allow determining if they can be generalized in different learning domains. Authors such as ^{Greene et al. (2015)}; Kleitman & Narciss (2019); ^{Olaya et al. (2023)} point out that general metacognition can be instructed simultaneously in different learning situations, thereby hopefully transferring to new learning situations, in contrast specific metacognition must be taught for each task or domain separately.

The leveling strategies, peer tutoring, time organization, establishing study habits and strategies, as well as the implementation of technological strategies, are part of the activities that promote the development of metacognitive skills (^{Castro, 2017}), in this sense, other studies may focus on using these skills and testing new strategies in general or specific domains. In this sense, it is important to take into account the recommendation of ^{Azevedo (2020)} who points out that a process aimed at strengthening metacognitive skills must meet three conditions: 1) integrate metacognitive instruction into the content to ensure connectivity, 2) inform learners about the usefulness of metacognitive activities to make them exert the initial additional effort, and 3) prolong the training to ensure the smooth and sustained application of metacognitive activity.

CONCLUSIONS

This study allowed to know the metacognitive skills of Colombian university students, through the analysis in three universities that include public and private institutions and with the use of different teaching modalities. It is evident that although students have a high level in the development of their skills, they still need to be strengthened in an important way. The results show that it should not be taken for granted that university students are aware of their ways of learning and the strategies that work best for them. It was also possible to determine that the mere fact of advancing in the career does not have a significant impact on the strengthening of their metacognitive skills, and contrary to what would be expected their metacognitive skills are decreasing. In other words, there are significant differences in the level of skills of Colombian university students according to the semester in which they are located, but as they progress in their programs, their skills decrease. It is important to take into account that in this study only students from the first to the fourth semester were considered.

This shows the importance of working with students, from the first semesters and in an explicit manner, in becoming aware of their study habits and not assuming that they already have these aspects developed, but rather that in any educational process they must continue to strengthen. It is necessary that students can assess the effectiveness of the strategies they implement to achieve their objectives and determine if they allow them to learn. It is necessary to work on this from the first semesters of University since it favors students to be aware of the learning process and what they require to make it meaningful, information that will also allow teachers to plan educational practice.

Although 70 % of university students in general, are located at high and very high levels in the two components of metacognition: knowledge and regulation of cognition. It stands out that more than 90 % of the students of the public university, who attend distance education programs, are located in the high levels of metacognition. It was established that there are significant differences between the level of development of the skills and the university of origin. It is possible that this can be explained by the educational models that the institutions have. The UNAB, which was the University where the students had the best levels, is a University that offers distance programs, for which the autonomy of the students in their learning process is worked from the beginning, this requires studying in greater depth.

On the other hand, it is determined that university women have better averages than men, at a general level as well as in each component and subcomponent of metacognition. However, the Mann-Whitney U test allowed us to determine that the differences are not significant.

It is important to delve into the correlational studies between the metacognitive abilities of university students and variables such as: academic performance, gender, nature of the institution (public or private) since the results do not show conclusive trends. It is necessary to combine subjective and objective measures of metacognition in research, together with the control of variables and the selection of samples to reduce bias and achieve greater clarity about predispositions.