The importance of self-regulation and the need to support the development of self-regulated learners is widely acknowledged [
1,
2]. An important aspect of self-regulated learning is the knowledge and application of learning strategies (LSs), i.e., the activities carried out during learning [
3,
4]. It is valuable for students to know the advantages and disadvantages of different LSs in specific learning situations, and thus, to possess good metacognitive knowledge of LSs [
5]. Besides their knowledge of LSs, student’s self-regulatory choices and learning are influenced by motivational factors [
1,
6]. An influential theoretical framework for studying school-related motivation is the situated expectancy–value theory that differentiates two types of motivational beliefs: expectancies and values [
7,
8]. While a plethora of studies have shown the importance of subject-specific expectancies and values in learning and learning outcomes in various subject fields [
7], learning-related strategy motivation [
9] is far less studied [
10,
11]. However, as strategy motivation is critical for learning complex topics that presumes the use of deep LSs [
10,
12], more studies focusing on different school levels and subject domains are needed.
The aims of the present study were (1) to examine differences between middle and high school students’ perceived effectiveness of LSs and their motivation to use effective LSs and (2) to explore the relations among these constructs and math and history grades in middle and high school. We focused on middle and high school where successful learning presumes good knowledge of LSs and high motivation to use deep LSs. Math and history grades were used as indicators of achievement in subject domains that possibly differ in the specific LSs required for success.
1.1. Learning Strategies and Their Effectiveness
Learning strategies (LSs) may be defined as a form of procedural knowledge or as goal-directed procedures that are intentionally and effortfully applied to support the regulation, execution, or evaluation of a particular learning problem or task [
3]. With time, strategy usage becomes automated and is referred to as a skill [
3]. Nonetheless, the use and differentiation of these two terms is not consistent among all researchers [
4,
13]. When put into practice, strategies and skills are the actual processes that students employ during learning, which directly influence both the process and outcomes of learning, being better predictors of learning outcomes than cognitive abilities [
3,
4].
LSs have broadly been categorized as supportive of deep or surface learning. Surface strategies focus on understanding or solving a problem, while deep strategies aim at integrating and transforming [
3,
13]. When using surface LSs, a student perceives new information and tends to mechanically repeat or memorize this information without attempting to integrate it with their prior knowledge [
13,
14]. With the use of deep LSs, a student not only perceives but also elaborates on new information, differentiates between more and less important information, and actively tries to integrate the new knowledge with their prior knowledge, resulting in the construction of new knowledge that is not simply a restatement of the new material [
3]. It is well known that the use of deep LSs—compared to surface LSs—tends to support the comprehension of new material in a way that it can later be recalled and flexibly used when solving novel tasks [
3]. The learning of complex topics is especially enhanced by the use of deep LSs rather than surface LSs [
15,
16].
However, these two types of strategies rarely form a clear-cut opposition due to differences in learner characteristics, tasks, and contexts [
13]. The strategies each individual student uses differ across contexts and tasks [
17], and each learning task may be successfully solved with the use of several different strategies [
3]. Furthermore, each LS can be applied more or less effectively, which is related to deep or surface learning, respectively [
18]. While some strategies are subject-specific (e.g., calculation strategies in math), others (e.g., self-testing) are more domain-general, i.e., useful across a variety of subject domains [
13]. Still, the context matters, and the effectiveness of the same strategy may differ between domains—a strategy may be a highly adaptive or an optimal strategy in one domain but relatively less effective in another domain [
3]. Also, the learner’s prior subject-specific knowledge influences the efficiency of using specific LSs [
3]. Importantly though, being a successful self-regulated learner requires metacognitive knowledge of learning strategies and skills to apply them adaptively. Such knowledge and skills develop as students mature and learn.
Primary school children use easily applicable surface LSs (rehearsal and rereading) that do not require well-structured prior knowledge and a high working memory capacity [
19]. The use of such strategies may be sufficient for learning basic skills and factual knowledge but does not support learning and understanding complex material taught in subject lessons in higher grades. Although middle school students’ psychological processes—that have matured and developed during learning—allow the use of more complex and effective LSs that support deep learning [
20,
21], students tend to value and use surface LSs even at the end of middle school and later [
16,
22,
23]. Still, cross-sectional studies have indicated that grade 9 students value and report using deep LSs more than younger students [
22], and high school students tend to value deep LSs more than middle school students [
24].
Students may discover the advantages of deep LSs independently when solving complex learning tasks and obtain information about learning from different sources like the internet and their parents. This may be more likely to happen during high school studies when students are required to comprehend and memorize large amounts of material and carry out homework projects, where successful completion depends on the context-specific application of different LSs. Also, high school students, compared with primary and middle school students, tend to have better prior knowledge, reasoning skills, and working memory capacity. Still, to support the development of metacognitive knowledge of LSs and a systematic adaptive use of LSs, explicit teaching and discussions about LSs are needed [
25,
26,
27]. While teaching domain-general LSs, it is also critical to practice strategies for solving different tasks and explain why certain LSs are more adaptive in specific contexts [
26,
28]. However, authentic classroom observations have shown that teachers rarely explicitly talk about learning processes and the advantages of specific LSs [
29,
30]. Accordingly, in their study among grade 8 and 9 students, Olop et al. [
31] found that only half of the participants were able to give examples of teachers’ recommendations for preparing for complex exams. Low emphasis on the explicit teaching of LSs may be one of the reasons for the lower knowledge and rare application of deep LSs in middle school students.
1.2. Learning Strategies Addressed in This Study
We focused on three widely used deep strategies (distributing learning, retrieval, and integrating new knowledge with prior knowledge) and three surface strategies (massing, rereading, and highlighting). Nonetheless, as explained earlier, in certain situations, surface strategies may be more effective, and deep strategies may be less effective in terms of learning outcomes. In the following paragraphs, we describe the findings of empirical studies on the efficiency of these specific LSs and students’ knowledge and use of these strategies.
Learning complex topics does not happen in a single shot, but rather, for deep learning, learners must repeatedly return to previously studied information. As such, distributing learning over time (also known as spacing) promotes better long-term retention than massing [
32,
33,
34,
35]. The advantages of distributing over massing have been shown both in experimental studies and educational settings, but its efficiency depends on the interval between study sessions [
35,
36,
37]. In contrast, studies have indicated that students tend to believe that massing leads to better learning and test results than distributing [
23]. Wiseheart et al. [
35] argued that this may be true as massing supports short-term retention, which is often assessed in traditional education. Additionally, massing creates an illusion of learning, while, in contrast, the beneficial effect of distributing learning becomes visible after a longer period [
38]. Differently from these studies, Granström and Kikas [
24] found that both middle and high school students tended to perceive distributing learning for a test coming up in a few weeks to be more effective than studying intensively before the test.
Retrieval, i.e., testing what has been learned, for example, by answering questions or generating explanations, has been demonstrated to be an effective LS (also referred to as self-testing, testing effect, and practice testing [
32,
39,
40,
41]). Retrieval has been shown to consolidate learned material, improve recall, and enhance meaningful learning at all ages and for learners of different ability levels [
41,
42]. Studies have also indicated that students value self-testing quite highly [
24].
Rereading refers to reading a text or its parts repeatedly—something that may be carried out quite mechanically [
33]. Although people can learn more from a second reading compared to the first reading, these gains tend to be small [
43], and rereading is a much less efficient LS compared to those that require the learner to be more actively involved, for instance, retrieval [
44,
45]. Rereading has been shown to be a commonly reported study strategy [
44]. Some studies that have manipulated retrieval practice versus rereading have found that students underestimate the benefits of retrieval practice and perceive rereading to be a more effective strategy [
45], while others have found that students value retrieval more than rereading [
24].
The essence of deep LSs is to actively recall what is known and to integrate new knowledge with prior knowledge. As such, creating verbal associations is a highly effective deep LS [
46]. When new information is integrated with existing knowledge, it can be more easily applied in new contexts [
5]. However, the efficiency of this LS also depends on what the student already knows [
3], and thus, for each student, this strategy may be more effective in some domains (where the student possesses correct and well-structured prior knowledge) but less effective or even non-effective in other domains (where the student has low knowledge or misconceptions).
Underlining or highlighting refers to marking specific parts of the text. This LS has the potential to benefit learning in two ways: (1) selecting what is important in the text elicits elaborative thinking (generative function), and (2) underlining or highlighting important parts makes it easier to identify them later (storage function [
18]). However, students’ ability to use this strategy in an effective way varies a lot, and they often highlight/underline without really engaging in the selection of important information and thus use this strategy as a surface strategy [
18]. Empirical studies have shown both the advantages and disadvantages of highlighting [
33,
47]. Students who highlight may later process the highlighted parts only, which makes reading more fluent and thus creates the illusion of learning [
47]. Marking the text is a widely valued and practiced LS [
18,
24,
47]. Still, Granström and Kikas [
24] also found that high school students valued underlining less than more active LSs like creating associations.
1.3. Learning-Related Motivational Beliefs and Their Relations to Learning
Motivational factors influence students’ self-regulatory choices and learning [
1,
6]. Intensive research has been carried out in the situated expectancy–value theoretical (SEVT) framework that describes two types of motivational beliefs: expectancies and values [
7,
8]. Students’ expectancies—their beliefs about how well they will perform in future tasks in a specific field—are closely related to self-efficacy [
48]. Attainment value is the perceived personal importance of achievement and is related to one’s identity [
7]. Intrinsic value refers to interest or inherent enjoyment gained from a given task. Utility value is the perceived relevance or usefulness of a given task or subject area regarding students’ current or future plans. Cost has been conceptualized as the time and energy that one must give up in order to engage in a given task, but it also involves the negative emotional or psychological consequences of this engagement [
49] and has also been conceptualized as a separate construct from values [
50,
51]. Value components are interrelated and, in turn, related to expectancy or self-efficacy [
7].
Motivational beliefs may support learning and outcomes via two pathways: the quantity and the quality of learning [
52]. Learning quantity can be measured via persistence, frequency, intensity of study, etc. Learning quality refers to the use of more adaptive and deep LSs in specific learning situations. Since deep LSs are effortful and time-consuming, students who value certain tasks or domains (e.g., have higher interest in and see their usefulness in their lives) are more likely to employ these strategies. Previous research has shown that students with higher subject-related expectancies (academic self-efficacy) and subject-related task values tend to show higher task-persistent learning behavior and greater value and use of deep LSs [
10,
53,
54,
55]. Recently, researchers have started to examine the motivational beliefs related to learning, LSs, and their application. Namely, Karabenick et al. [
9] differentiated between subject- or outcome-related and learning- or strategy-related motivation. Outcome-related motivational beliefs refer to motivation to study a specific topic, while strategy-related motivation (further strategy motivation) in an educational context refers to the motivation to use learning strategies. They emphasized that students decide to use a given strategy when they see its value and feel confident to use it in the specific learning situation. Regarding the cost component, since the use of deep LSs requires increased cognitive effort—including attention, working memory capacity, and higher reasoning skills [
19,
56,
57]—the cost of using these strategies may also be perceived to be higher than that of applying surface LSs. However, students may still use deep LSs if they perceive the effectiveness of these strategies to be high, i.e., if they perceive the utility value of deep LSs to be high. Karabenick et al. [
9] examined ninth-grade students’ LS-related utility and cost and the use of LSs in math lessons and found that, for most students, the strategies most often used were also the ones considered to be the most useful. However, reported LS use was not consistently related to perceived cost—for some students, a higher cost was linked to less frequent use of the LS, for some, cost and strategy use were relatively unrelated, and for others, the relations were even positive.
As to changes in students’ motivation during school years, in most countries, subject motivation has been shown to clearly decline throughout middle school [
7,
58]. However, there is individual variability in such changes in motivation: person-oriented studies have described different trajectories of motivation, showing that the decline is more apparent for some students than for others [
59,
60]. So far, there are no studies on strategy motivation differences between middle and high school. Thus, our study aims to contribute to filling this gap.
1.4. Network Approach to Examining Interrelations between Motivational Beliefs, LSs, and Achievement
Psychometric network models allow for the interrelations between a group of variables to be examined so that the relationship between any two variables is conditioned on all other variables in the system [
61]. As such, this approach allows for a deeper understanding of the relationships between variables as a holistic and complex system. A network comprises nodes (variables) and edges between the nodes—namely, the associations between any two variables, when accounting for all other variables in the network, in contrast to bivariate correlations that account only for the relationship between any two variables. Although rather new in educational psychology research, this approach has recently been utilized in the SEVT framework to gain a better understanding of the interrelations between the components of motivational beliefs and achievement.
In one such study, Tang et al. [
62] examined networks of multiple components of subject-specific motivational beliefs (expectancies, intrinsic value, attainment value, utility value, and cost) and academic achievement among Finnish and German sixth- to ninth-grade students in math and languages. By comparing the networks of different grade levels, subject domains, and countries, they showed differences as well as similarities in the interrelations of motivational beliefs and achievement across these networks. More specifically, concerning the similarities, across all of the networks, achievement had the strongest connection with expectancies and only weak associations with interest, utility, and attainment values. In another study, Lee et al. [
63] used networks to examine the interrelations between math achievement and motivational beliefs concerning math among ninth-grade students. They found that expectancy was strongly connected to interest value, while it was weakly linked to utility value. They also found only expectancy to be connected to achievement.
While the network approach has been used to examine the interrelatedness of motivational constructs and achievement as a holistic system, to our knowledge, such an approach to study the interrelations between strategy motivation, perceptions of multiple specific LSs, and achievement as a holistic system has not yet been used.
1.5. The Cultural–Educational Background of This Study
The current study was carried out in Estonia, a small country that was part of the Soviet Union from 1940 to 1991 and has since undergone rapid social changes. In Estonia, compulsory formal education starts at the age of seven and lasts for nine grades. During the first three years (in some schools, up to six years), the core subjects (i.e., math, literature, language, and science) are taught by the same class teacher. Later, different subjects are taught by different subject teachers. After grade 9, students can decide whether they want to continue their academic studies in high school (gymnasium) or in vocational school or whether they want to opt out of further academic studies.
The aims and requirements of education are specified in the national curricula [
64,
65]. Besides the demands regarding academic skills, the curricula describe general (key) competencies and emphasize the importance of supporting the development of these competencies in all subjects. Learning to learn is one of the key competences included in educational policy documents as a main aim of learning to be developed in school [
66,
67,
68]. Knowledge of learning—including knowledge of adaptive LSs—, skills to adaptively use LSs, and adaptive motivational beliefs form vital dimensions of learning to learn and thus should, according to curricular demands, be supported in all subjects in school.
1.6. Aims
The aim of the current study was to examine the similarities and differences between middle and high school students’ perceived effectiveness of specific LSs and strategy motivation and the interrelations among these constructs and math and history grades. Middle and high school students’ learning differs both due to curricular demands (e.g., more complex, substantial, and integrated topics, but also more voluntary subject choices in high school) and learning procedures (e.g., longer periods devoted to learning specific subjects and longer lessons). Moreover, while middle school studies are obligatory, high school studies are voluntary, and thus, high school students may be more engaged in learning. The following research questions were used:
First, how do middle and high school students differ in their perceptions of the effectiveness of deep (distributing, retrieval, and integrating) and surface LSs (massing, rereading, and highlighting)?
Second, how do middle and high school students differ in their motivational beliefs regarding LSs?
Third, how are students’ perceived effectiveness of six LSs, their motivational beliefs, and their math and history grades interconnected? Are there differences in these associations between middle and high school students?