There are different SRL models available in the literature to define this construct and its components (for an in-depth review of each model, it is suggested to consult the works of Panadero, 2017; Puustinen and Pulkkinen, 2001). However, all of them agree that the student is the primary agent of his/her learning, has the responsibility, and can learn in interaction with the context. The present study is based on Zimmerman’s theoretical model (Zimmerman, 2000), chosen mainly for its outstanding recognition at the international level (Wong et al., 2019) and for its ease of operationalization. This approach considers SRL as a competence developed through different tasks and environments.

The model proposed by Zimmerman (2000) considers a socio-cognitive perspective and emphasizes on strategies with complex and dynamic interactions influencing the students’ engagement and effort toward optimal learning and performance, which are enhanced in highly autonomous learning environments. This author described SRL processes as triadic, referring to the dynamic influence of self-regulatory processes on environmental events, personal, behavioral factors. Thus, he distinguished three phases of self-regulation that function in a cyclical, interrelated, and recursive manner during the learning process: disposition, performance, and self-evaluation (Zimmerman, 2013).

In the first phase, disposition, self-regulated learners are expected to actively engage in task analysis processes (i.e., goal setting and planning, designing a strategy for their learning). Then, the performance phase is activated when learners employ strategies to process learning material, seek help when needed, manage their time, structure their environment, and control their learning processes, being able to implement self-monitoring (i.e., self-instructions, focusing attention) and self-observation (i.e., self-registration and self-experimentation) strategies. Finally, the third phase self-evaluation occurs when students must assess their performance and adjust their strategies to achieve their learning objectives, i.e., self-judgment (self-evaluation of their learning process and causal attributions, and adaptation or defense) occurs in the final phase. These three phases repeat cyclically and iteratively throughout the learning process (Zimmerman, 2000). This description highlights that SRL is a multidimensional, complex, and recursive phenomenon that involves a set of cognitive, social, metacognitive, and behavioral processes whose interlocking is articulated in a cyclical model (Sáez-Delgado et al., 2022, 2023).

The ability of students to regulate their learning process is related to their commitment and achievement of personal learning objectives (Moreno-Marcos et al., 2020). However, one of the highly documented problems in education is that students often exhibit suboptimal levels of SRL at different levels of their academic trajectory (Sáez-Delgado et al., 2023), which in higher education makes them more likely to drop out of their programs of study (López-Angulo et al., 2023). Therefore, to describe, analyze and intervene in the promotion of this process, it is necessary to have a relevant and valid instrument.

Regarding the above, self-report measures constitute one of the primary methods for assessing SRL due to their relative ease of administration and scoring. However, psychometric properties are a fundamental need when assessing the quality of such measurement protocols (Roth et al., 2016).

A systematic literature review characterized various measurement instruments on SRL designed for students at different educational levels (León-Ron et al., 2020). The findings showed 31 instruments available, of which 57.5% correspond to instruments developed, adapted and validated in Higher Education. About the countries where these psychometric studies have been conducted, 24 countries were identified; however, it is in Europe where more psychometric studies on SRL were developed (40%); on the contrary, Central and South America and Oceania represented the lowest percentage (10 and 3%, respectively). This gap has been documented in other research, which indicates that Latin America is in the preliminary stage of developing of instruments to measure SRL and that further studies are required (López-Angulo et al., 2020).

An analysis of the instruments developed to measure SRL in university students allowed the authors of the present study to make a classification. Firstly, some instruments measure specific skills and strategies for learning, for example, a second language. Secondly, instruments focused on specific contexts, such as professions or online teaching modalities. Other instruments have been developed based on theoretical models other than Zimmerman’s. Finally, there are instruments developed based on Zimmerman’s theoretical model. In the following sections, we will analyze each category in detail.

Among the instruments that aim to know how students use specific SRL skills and strategies to learn a second language, an item example is: For me, successful learning of English is more important than my grades or opinion of my teachers, family, or Friends (e.g., Salehi and Jafari, 2015; Teng and Zhang, 2016). Other instruments along the same lines are those focused on assessing the use of SRL in specific skills, such as writing. An item example is: When writing, I use some literary devices to make the composition more interesting (e.g., Teng and Zhang, 2016). For reading comprehension, an item example is: From that first glance, I question myself to the point of clarifying what I know about the subject (e.g., Solano-Pizarro et al., 2004). In these cases, the content of the items covers aspects of language learning and writing. Thus, the items not only assess the process of self-regulation but also reflect the context in which they are being assessed.

Other instruments focus on specific contexts, either linked to a specific profession, career, or course modality (e.g., Iyama and Maeda, 2018). An item example is: Using my own words, I want to explain important points I have learned in my clinical nursing practice. A significant limitation is that the scale is titled “Development of the Self-Regulated Learning Scale in Clinical Nursing Practice for nursing students,” and it is composed of only two subscales, one for motivation and one for learning strategies. Another instrument of this group is that of Barnard et al. (2010), which evaluates learning in online courses. An item example is: I set standards for my assignments in online courses (Barnard et al., 2010).

There are also instruments developed with theoretical models other than Zimmerman (2000), such as the Motivated Strategies for Learning Questionnaire (MSLQ) (Pintrich et al., 1993), which is based on Pintrich (2000) model and comprises two dimensions: Motivation and Learning Strategies.

Of the questionnaires that were developed based on Zimmerman’s theoretical model, the Academic Self-Regulated Learning Scale (A-SRL-S) (Magno, 2010) is composed of seven factors: memory strategy, goal setting, self-evaluation, help seeking, structuring the environment, responsibility for learning, and planning and organization. Similarly, the Learning and Study Strategies Inventory (LASSI) (Weinstein and Palmer, 2002) covers three dimensions: ability, disposition, and self-regulation. Although both instruments explicitly measure SRL, they do not exhaustively evaluate the aspects contained in Zimmerman’s model. Also, there is the Inventory of Self-Regulation of Learning Processes (IPAA) (Rosário et al., 2007), has a significant limitation: the items that make up the scale are not univocal, some items evaluate two actions in one sentence, which can cause confusion. Therefore, the items must be specific and capable of capturing the abstract construct they intend to measure to guarantee the validity of the scales (Löhr, 2022; Diamantopoulos et al., 2012; Lilienfeld and Strother, 2020; Tavakol and Wetzel, 2020). Conversely, it may have significant consequences in the measurement error of latent variables (Rhemtulla et al., 2020).

A scale that overcomes these limitations and is based on Zimmerman’s theoretical model is the Instrument to Measure Self-Regulation of Learning Phases (Sáez-Delgado et al., 2022), which was designed to measure the three phases of the SRL process (disposition, performance and evaluation) in Chilean secondary students. This scale has several advantages: (1) It considers the different dimensions (behavioral, motivational and cognitive) that are part of the study and learning processes of students (Roth et al., 2016); (2) It is based on Zimmerman (2000) theoretical model of SRL phases; (3) It showed good adjustment indexes in the original study; and (4) It allows identifying those variables with inadequate levels. Therefore, it is presented as an alternative and valuable scale to measurer SRL in university students.

Despite the relevance of the SRL, to our knowledge, there are no validations of the Instrument to Measure Self-Regulation of Learning Phases (Sáez-Delgado et al., 2022) in university students. Therefore, the general objective of this research was to validate and estimate the psychometric properties of an instrument to measure the phases of self-regulation of learning in Chilean university students. Specifically, we sought answers to the following questions:

An instrumental design was used to carry out the study, which according to the classification of Ato et al. (2013), it considers the psychometric characteristics analysis of the Instrument to Measure Self-Regulation of Learning Phases (Sáez-Delgado et al., 2022) in a sample of Chileans university students.

Participants comprised 348 first-year university students from STEMS careers university in southern Chile participated. Of the total 136 were men (39.1%), 209 women (60.1%), and 3 preferred not to say (0.9%). Their ages ranged from 18 to 21 years (M = 18.85; SD = 0.773). Non-probability convenience sampling was used, considering the cohort of first-year students.

Scale to measure self-regulation of learning phases (Sáez-Delgado et al., 2022): is composed of three subscales that account for the process of self-regulation of learning in consonance with the cyclical model proposed by Zimmerman (2000). It is a self-report instrument with a 7-point Likert scale, ranging from 1 meaning never to 7 meaning always. The validation with high school students showed adequate psychometric properties, with a structure composed of three factors. The first scale, before studying (Learning Disposition Phase), is composed of 5 items that measure the frequency with which students use self-regulation strategies to prepare for their study, e.g., “Before starting to study, I set short-term academic goals (daily, weekly),” and reliability was adequate (α > 0.79; Ω > 0.82). The second scale, while studying (Learning Performance Phase) is composed of 6 items and measures the frequency with which students use strategies to control their study based on a previously established planning that includes the goals to be achieved, e.g., “While studying, I check if I am learning,” and the reliability was adequate (α > 0.87; Ω > 0.91). The third scale, after studying (Learning Self-evaluation Phase), is composed of 5 items and measures the frequency with which students reflect on the results obtained in some school task or test, e.g., “When I finish studying, I self-evaluate whether I made progress in relation to my previous knowledge,” and reliability was adequate (α > 0. 85; Ω > 0.87). The items’ average represents the score for each scale. Frequency values for using self-regulation strategies indicate: (a) Learners at optimal SRL levels (6–7 points), (b) Learners at suboptimal SRL levels (3–5), (c) Learners at insufficient SRL levels (1–2).

Meaning in life questionnaire (Steger et al., 2006): consists of 10 items designed to assess the meaning of life. This instrument was translated into Spanish through translation and back-translation, thus ensuring its linguistic equivalence. The questionnaire is composed of two subscales: “Presence of Meaning,” which includes items 1, 4, 5, 6, and 9; and “Search for Meaning,” which comprises items 2, 3, 7, 8, and 10. The answers in the questionnaire are given through a Likert-type scale, ranging from 1 (totally false) to 7 (totally true).

Additionally, sociodemographic data was collected, including sex, age, career, and other data relevant to the study.

Data collection was carried out as part of a larger research project. The project coordinators established contact with the academic authorities to whom the purpose of the study was explained. After obtaining their authorization, dates for administering the questionnaires were agreed upon. The surveyors entered the classrooms, offered students the opportunity to participate after explaining the objectives of the research. Those students who agreed to participate signed an informed consent form before completing the questionnaire. The Ethics Committee CEBB1394-2023 approved this research.

First, descriptive analyses of the items that make up the instrument were performed to evaluate their distribution using descriptive statistics, together with skewness and kurtosis coefficients. To examine the factorial structure, a Confirmatory Factor Analysis (CFA) was implemented to evaluate its underlying structure and compare its settings (Bollen, 1989). The Robust Maximum Likelihood Estimation (MLR) method was used, which does not require normality of the data due to its robustness in identifying significant effects (Lloret-Segura et al., 2014). The analysis was carried out with MPlus 8.4 software, evaluating the model fit using several indices: a non-significant Chi-square value (p ≥ 0.05) (Tabachnik and Fidell, 2007), an Root mean square error of approximation (RMSEA) and Standardized root mean square residual (SRMR) below 0.07, and a 90% confidence interval (CI) between 0.000–0.050 (Hu and Bentler, 1999; Steiger, 2007), and Comparative fit index (CFI) and Tucker-Lewis Index (TLI) values above 0.94, in addition to requiring that the factor loadings of the items be significant and above the threshold of 0.40 (Hair et al., 2014). Correlations between dimensions and items were also performed to deepen the analysis. In addition, as part of the construct validation, discriminant validity analyses were performed, implying that two theoretically unrelated variables reflected on low or non-existent correlations. In this research, which explores validity based on discriminant measures, correlations of magnitude <0.50 indicate that the instruments assess different constructs. Validity was estimated with the Meaning in Life Questionnaire (Steger et al., 2006). Reliability was calculated using McDonald (1978) Omega coefficient (Ω), a measure considered more accurate than the alpha coefficient since it considers factor loadings rather than simply the number of items or response options (Hayes and Coutts, 2020), expecting to obtain values above 0.70. The JASP software version 0.8.3.1 was used for this calculation.

Exploratory CFAs were performed to evaluate the factorial structure of the instrument form measuring the self-regulation of learning phases since these analyses allow the modeling of different factorial solutions (Bollen, 1989; Schmitt, 2011). Table 2 presents the fit results obtained for the different models tested.

First, a CFA was performed considering the three subscales suggested by the creators of the instrument (Sáez-Delgado et al., 2022); as can be seen, only subscale 2 had adequate fit indices, so it was decided to test a solution of three first-order related factors. However, it did not fit as expected.

Given this scenario, it was decided to respecify the model, considering the trajectories indicated by the Mplus through the analysis of the modification indexes. The relationships were only added considering their theoretical sense; since this instrument accounts for a theoretical model of three related and sequential phases, items sharing similar contents, although belonging to different factors, tended to show linked residuals. In this case, the residuals of items 10 and 4 (MI = 69.347), 13 and 12 (MI = 45.193), 16 and 3 (MI = 27.149), and 14 and 10 (MI = 22.738) were correlated. These pairs of items allude in their wording to the conditions of the place chosen to study, self-evaluation of the fulfillment of study goals, elaboration of a schedule to organize study time, and conditions of the place chosen to study, respectively. Thus, a better fit was obtained in the model. As shown in Table 2, the last solution is the one that presented the best fit and is consistent with the theoretical approach of (Zimmerman, 2000), a structure of one second-order factor and three first-order factors is confirmed.

Table 3 reports the factor loadings obtained for the three factors proposed for the analysis, all of which presented significant loadings in the second order factor. All items had loadings above 0.40; this reflects adequate levels of association in all items of the instrument with respect to the dimensions, thus confirming three dimensions: (1) Learning Disposition, (2) Learning Performance and (3) Learning Self-evaluation (see Table 3).

As for the correlations of the instrument items, Table 4 shows that all items correlated significantly with the dimension to which they correspond, reflecting adequate levels of association in all instrument items concerning their dimensions.

Considering this confirmed structure, it is different from what the authors proposed in the original version of the instrument (Sáez-Delgado et al., 2022), presenting a structure composed of three interrelated dimensions, which in turn forms a global construct that accounts for the process of self-regulation of learning, its cyclicality, and functioning by phases (Figure 1).

This confirms a second-order factorial model composed of 16 items, where factor 1 (learning disposition) measures aspects related to the self-regulation strategies used by students to prepare their study, for example, setting academic goals, preparing materials, making a schedule, defining a suitable place and prioritizing a to-do list; factor 2 (learning performance) refers to the self-regulatory learning strategies that students use to monitor their study based on the previously established planning that includes self-monitoring while studying, i.e., checking if the time is sufficient, if the learning strategies they are using are effective, if the materials are sufficient, if the place is adequate, if they are meeting or not meeting the study goals, and checking if they are learning; factor 3 (learning self-evaluation) captures the self-regulation strategies of learning that students use to verify and self-evaluate the effectiveness of the study planning executed, i.e., self-evaluate if they made progress with respect to previous knowledge, if they met the goals they set, if the place was adequate, if the learning strategies were effective, and if they met the schedule. These factors make up a general factor, which based on the theoretical meaning of the items, will be called Phases of Self-Regulation of Learning, see Annex 1.

With the Meaning in Life scale (Steger et al., 2006), validity based on divergent measures was explored. The results in the table indicated the existence of significant correlations between the factors of the instruments, with a range from r = 0.303 to r = 0.359; despite this, it has a low magnitude (Table 5). These findings suggest that both instruments are assessing different constructs.

As shown in Table 6, each dimension of the scale has adequate indices, being higher than 0.70 and lower than 0.93. These coefficients suggest that the dimensions and the instrument are reliable for measuring the phases of self-regulation of learning in Chilean university students.

This study has important theoretical and practical implications. From the theoretical edge, it contributes to the advancement in psychometric research on measuring SRL, overcoming limitations of previous studies that have shown low rigor in the content validation of their instruments (López-Angulo et al., 2020). Additionally, it contributes to a theoretical gap regarding the development of validated questionnaires in Latin America (Hernández et al., 2017; León-Ron et al., 2020; López-Angulo et al., 2020).

From the practical perspective, a valid and reliable instrument to measure SRL in university students is of utmost relevance, demonstrating practical applications beyond the research. Specifically: (a) It allows an accurate assessment of the student’s self-regulatory skills, facilitating an in-depth understanding of how they manage their learning process; (b) It facilitates the design of educational interventions, especially when a group of students with difficulties in their phases of the self-regulatory process is identified while applying the instrument, allowing to visualize in which of them to emphasize for their promotion and development. Based on this, specific strategies can be developed to improve these skills (Sáez-Delgado et al., 2018); (c) Progress Monitoring, having a valid and reliable instrument allows monitoring the progress of the student body over time, thus observing whether their self-regulatory skills improve, remain stable or require attention (Mella-Norambuena et al., 2021); (d) Identification of Individual Needs, this is especially relevant, as each student is unique in how they self-regulate, which guides the adaptation of teaching strategies accordingly for the faculty (Contreras-Saavedra et al., 2024); (e) Improvement of teaching practice, making available an SRL phasing instrument available allows teachers to assess and understand their students’ self-regulatory skills, helping them to adapt their pedagogical approach and offer specific support when needed (De La Fuente et al., 2020; López-Angulo et al., 2022).

In summary, this study provides an adequately validated instrument to measure SRL becoming a valuable tool for research and educational practice. It thus contributes to the comprehensive development of students and their academic success.

The obtained results provide valuable information on the validity and reliability of the SRLI-U. However, it is important to consider some limitations and areas for future research. A relevant element that must necessarily be mentioned is the type of instrument that was validated in this study, which is a self-report. Considering that it was applied to university students, it is likely to present, to some extent, the so-called social desirability bias in the process of collecting the responses obtained (Durmaz et al., 2020; Fisher and Katz, 2000). Specifically, this implies questioning the scores obtained by the scale that could obtain a high average in the different phases of the SRL process. Now, while there are other alternatives for measuring SRL in college students, for example, observational instruments, measurements through the learning analytics approach, and more recently with measurement challenges based on artificial intelligence (Molenaar et al., 2023), possible difficulties or limitations are identified with these instruments. Therefore, the recommendation is that future research that seeks to measure SRL in university students should consider integrating more than one measure method to achieve a more accurate understanding of this phenomenon.