Teachers’ Knowledge of Students’ Learning Strategies: Recommendations and Evaluations

The objective of this research was to identify which learning strategies pre and in-service teachers ( N = 340) recommend to students in two hypothetical learning situations, in order to examine how teachers evaluate students’ learning strategies, and to study relations between recommendations and evaluations. Among the recommended and evaluated learning strategies, four tapped deep learning, and two tapped surface learning. Results showed that teachers mainly recommended either deep-learning strategies or a combination of deep and surface-learning strategies. Teachers gave better evaluations of deep-learning strategies compared to surface-learning strategies. Differences in evaluations of surface-learning strategies and interleaving were found between teachers who recommended only deep-learning strategies and those who recommended both types of strategies.


Introduction
The strategies students use to learn directly affect their learning process and learning outcomes (Dinsmore & Hattan, 2020).The importance of good learning skills in academic success and psychological well-being became especially salient during the COVID-19 pandemic, when students were forced to plan and execute their learning independently (Holzer et al., 2021;Salmela-Aro et al., 2021).A plethora of studies have indicated that students tend to use strategies supporting surface learning instead of more complex strategies that facilitate deep learning (Dirkx et al., 2019;Dunlosky et al., 2013).Because of limited knowledge and low cognitive skills, young children tend to use only simple learning strategies like rehearsal (Schleepen & Jonkman, 2012).When first using a complex strategy, students may experience utilization deficiency (Clerc et al., 2014), and assume that poor outcomes are related to low strategy efficacy rather than their low skills.Thus, students need teacher support in learning to use various learning strategies, particularly complex ones.Teachers may suggest specific learning strategies, demonstrate their efficacy for specific learning tasks, discuss potential challenges when using them, and support practicing their use (Dunlosky et al., 2013;Schnellert & Butler, 2020).Throughout this process, teachers' knowledge of learning strategies is of critical importance (Schnellert & Butler, 2020).Thus, examining teachers' knowledge of learning strategies has practical value when planning training courses for both pre and in-service teachers.
Previous research has indicated that teachers' knowledge of learning strategies tends to be modest (see Dignath & Büttner, 2018).Past research has assessed this knowledge primarily via teacher evaluations of the effectiveness of certain strategies (Glogger-Frey et al., 2018;Michalsky, 2021), although a few studies have also examined teachers' recommendations of learning strategies (Hattie, 2003;McCabe, 2018).
Throughout these studies, little attention has been given to recommendation and evaluation consistency.The aim of this study was to examine teachers' knowledge of learning strategies using open-ended questions and evaluations.Teachers were asked to recommend learning strategies for two learning tasks, and to evaluate the effectiveness of different learning strategies.

Learning strategies
Learning strategies refer to the processes learners use to enhance their learning (Hattie & Donoghue, 2016).Learning strategies involve different levels of strategic processing, deep and surface learning (Frey et al., 2017;Weinstein et al., 2019).Surface learning strategies mainly support basic knowledge, while those that support deep learning promote better understanding of the material, and help connect new ideas with previous knowledge (Chi & Wylie, 2014;Hattie & Donoghue, 2016).
Deep learning is enhanced when creating associations between new ideas and existing knowledge (Bjork et al., 2013;Diamond 2013).Associations can be created via verbal, visual, or integrated summaries (Dunlosky et al., 2013;Eitel & Scheiter, 2015;Fiorella & Mayer, 2015), or using explanatory examples related to previously learned material or real life (Baldassari & Kelley, 2012;Caplan & Madan, 2016).Another deeplearning strategy is self-testing, when a learner activates previous knowledge in order to answer questions or solve tasks, then restructures this knowledge into a coherent response so that new knowledge can be integrated with previous knowledge (Fiorella & Mayer, 2015).Distributing learning -dividing study material into smaller parts that can be learned over a longer period of time -allows more time for the study material to become embedded, and studies have proven its efficacy in students of different ages (Agarwal et al., 2021;Carpenter et al., 2012;Dunlosky et al., 2013).Another way to diversify learning is through interleaving, during which a learner alternates practicing one skill with practicing a different skill (Chen et al., 2021).The value of interleaving has been recognized and studied in recent decades (Chen et al., 2021;Dunlosky et al., 2013).
In contrast, massing -studying large quantities of material within a short period of time -leads to surface learning.However, many students frequently use massing and value it highly as it creates an illusion of learning (McCabe, 2011).Rereading, when a student reads the text several times, is also related to surface learning (Dunlosky et al., 2013).Rehearsal -repeating learning material -may be carried out along with creating associations, but it may also be done quite mechanically.Simple rereading and massing are often associated with rehearsal, and are used when there is not enough time left to implement complex strategies (Rawson & Kintsch 2005;Weinstein et al., 2019).
Employed mechanically, rehearsal is related to surface learning.Strategies like underlining text, reading text out loud, and copying text are passive strategies that presume little mental effort (Chi & Wylie, 2014;Fiorella & Mayer, 2015), and are also related to surface learning.For instance, underlining may take one's focus away from the text as a whole, and does not support the forming of associations between different parts of the text (Dunlosky & Rawson, 2015).
It should be noted that there is no single most effective learning strategy or group of most effective strategies, since the usefulness of each strategy depends on the student's current knowledge and abilities, but also on the learning task, and context of learning (Dinsmore & Alexander, 2012).Strategies that support surface learning may be made more effective when used together with other strategies that support deep learning (Frey et al., 2017;Roediger & Karpicke, 2006;Weinstein et al., 2019).However, as surface level strategies create an illusion of learning, because their usage is easier and the learning process appears to proceed smoothly (Roediger & Pyc, 2012), students need explicit teaching about learning processes (Schnellert & Butler, 2020;Dunlosky et al., 2013).

Teachers' knowledge of learning strategies
Studies across different countries have indicated that pre-service and in-service teachers have fragmentary, and sometimes inaccurate, knowledge of learning strategies.Dignath and Büttner (2018) found that teachers had difficulty defining learning strategies.Glogger- Frey and colleagues (2018) found that teachers often rely on misconceptions, and generally have a poor understanding of which learning strategies support deep learning.Some studies have also revealed poor knowledge of learning strategies among university instructors.Morehead and colleagues (2016) found that more than 60% of instructors recommended distributing and self-testing, 33% recommended making diagrams, and 25% recommended using flashcards.However, many instructors also recommended learning strategies that support surface learning, like rereading (40%), outlining/highlighting (40%), and recopying notes (18%).Furthermore, McCabe (2018) asked academic support centers at US colleges and universities to list the top three learning strategies that are recommended to students, and to evaluate the effectiveness of different strategies.The most frequently recommended learning strategies were those that support deep learning: self-testing, answering questions, using flashcards (39%), distributing (23%), and using schemes (including concept maps, 10%).Participants also perceived discussing, self-testing, answering questions, and distributing as highly effective, while perceiving rereading, highlighting/ underlining, and massing as much less effective.Still, some inconsistencies were also found, as participants tended to give low evaluations to some strategies that have been shown to support deep learning (e.g., multimodal coding, imagery, flashcards, and interleaving).

The Current Study
This study was carried out in Estonia, a small country in northern Europe which has undergone rapid social changes since the collapse of the Soviet Union in 1991.Estonian students have shown strong academic knowledge and skills in international comparisons, consistently ranking at the top of PISA (Programme for International Student Assessment) results (see OECD, 2019a).However, Estonian students have reported receiving less support from their teachers than students in other OECD countries (OECD, 2019b), and Estonian students' well-being has also deteriorated over time (OECD, 2019c).Although the aims of school education have widened, and teaching practices have changed (Uibu & Kikas, 2008), some teachers (specifically older ones) tend to value and use teacher-directed didactic approaches, while others tend to use constructivist or combined didactic-constructivist approaches (see Tang et al., 2017; see also Kikas et al., 2023).A preference for teacher-directed approaches -in addition to outdated beliefs that may have developed during a teachers' own school years -can inhibit the application of constructivist teaching approaches that better support students' holistic development (Kikas et al., 2023;Lawson et al., 2019).Solid knowledge of learning strategies is associated with higher psychological well-being in students (Tavakolizadeh et al., 2012), which is crucial for coping with different challenges in school.Thus, solid knowledge related to learning and learning strategies in teachers is also important to be able to support students.
The aim of this study was to identify which learning strategies teachers recommend to students, to examine how teachers evaluate the effectiveness of different learning strategies, and to study relations between recommendations and evaluations.It is important for pre-and in-service teachers to know the advantages of different learning strategies, to be able to distinguish between those that support surface learning and those that support deep learning, and to be able to recommend the most effective learning strategy for students according to a given learning situation.Solid knowledge of different learning strategies and their meaningful use is an important part of self-regulated learning (Dinsmore & Hattan, 2020;Fiorella & Mayer, 2015).We used two learning scenarios: studying for a test that would take place the next day (next-day test or Scenario A); and studying for a test that would take place three weeks later (major test or Scenario B).Our research questions and hypotheses were as follows.
First, which strategies do teachers recommend to students?Do recommendations depend on the test date?We presumed (H1a) that teachers would primarily suggest strategies that support deep learning (McCabe, 2018;Morehead et al., 2016).We also expected (H1b) that teachers would recommend more surface-learning strategies for Scenario A due to the limited time frame of this scenario (Roediger & Pyc, 2012).
Based on prior research among higher education teachers (McCabe, 2018;Morehead et al., 2016), we presumed (H1c) that strategies related to self-testing would be recommended most often.
Second, do teachers combine strategies and recommend both deep and surfacelearning strategies in Scenario A? We expected (H2) that some teachers would recommend only surface-learning strategies (McCabe, 2018).
Third, do teachers' evaluations of learning strategies load onto the six theoretically expected latent variables (creating associations, self-testing, distributing, interleaving, rehearsal, and passive strategies)?We expected (H3) that a six-factor model of teachers' evaluations would have an acceptable model fit.
Fourth, are teachers' recommendations and evaluations of learning strategies in consonance?We assumed (H4) that teachers who recommend a specific strategy would evaluate its effectiveness higher than teachers who do not recommend it.
Fifth, do evaluations differ among groups of teachers who recommend only deeplearning strategies, only surface-learning strategies, or both types of strategies?
We hypothesized (H5) that teachers who recommend only deep-learning strategies would evaluate surface-learning strategies lower than teachers who recommend only surface-learning strategies or both deep and surface-learning strategies.

Participants
This study involved 340 Estonian pre-service and in-service teachers (89.3% female).

Procedure
The survey was conducted from autumn 2019 until autumn 2020.Data were gathered via a web-based questionnaire using the SurveyMonkey online survey program.
Teachers were invited to participate through university lectures or via school principals.The questionnaire took an average of 15-30 minutes to complete.Teachers were first informed of the study objectives, and all teachers had the option to discontinue the questionnaire at any time.

Measures
The entire questionnaire consisted of three parts (Surma et al., 2019).In the current study, we analyzed two parts: recommendations and evaluations.The third part included different learning scenarios in which two students used either more or less effective strategies, and respondents were asked to evaluate the effectiveness of strategies and justify their answers.These findings are reported in another paper (Granström et al., 2022).

Recommendations and coding
Teachers were given descriptions of two learning scenarios, and asked to recommend strategies students could use for each scenario.The scenarios were as follows: First, teachers' answers were categorized into specific categories (Appendix 1) and coded by two coders.Interrater reliabilities between the two coders were calculated separately for each scenario and were very good (Scenario A: Kappa = .905(p < .001),95% CI (.870, .940);Scenario B: Kappa = .835(p < .001),95% CI (.798, .872);Cohen, 1988).Based on the aforementioned differentiation of learning strategies (see also Agarwal et al., 2021;Fiorella & Mayer, 2015), these specific categories were grouped into deep-learning strategies (creating associations, self-testing, distributing, interleaving) and surface-learning strategies (rehearsal, passive strategies) (see Table 1).

Evaluations
Teachers were asked to evaluate the effectiveness of 21 different learning strategies on a 5-point Likert-type scale (1 -the strategy is ineffective, to 5 -the strategy is very effective).The 21 strategies are reported in Appendix 2.

Data analysis strategy
To answer the first two research questions, descriptive statistics and z-tests were used.To answer the third research question, confirmatory factor analysis (CFA) was used.Model fit was examined using five indices: chi-square (χ2); the comparative fit index (CFI); the Tucker-Lewis index (TLI); the root mean square error of approximation (RMSEA); and the standardized root mean square residual (SRMR).Cohen, 1988).A repeated measures ANOVA was used to examine differences between evaluations in groups of teachers who recommended only deep strategies, and both deep and surface strategies.Eta squared was used as a measure of effect size (small effect η 2 ≥ .01;medium effect η 2 ≥ .06;large effect η 2 ≥ .14;Cohen, 1988).SPSS Statistics 27 and the Mplus statistical package (Version 8.6;Muthén & Muthén, 1998-2017) were used for statistical analysis.

Teachers' recommendations
The number of answers in each category are provided in Table 1 for both Scenario A and Scenario B. Teachers were asked to provide up to three strategies for Scenario A and one strategy for Scenario B. If the respondent mentioned different strategies within the category, the response was counted once.In several cases, respondents gave more strategies than asked for.Thus, the total number of answers is larger than the expected number of answers.Percentages are taken from the total number of answers.
Teachers tended to recommend deep-learning strategies for both scenarios.The most popular strategy for Scenario A was creating associations (45.7%), and for Scenario B distributing (30.2%).Interleaving was not mentioned.Deep-learning strategies were suggested more for Scenario B than for Scenario A. Z-tests indicated that, except for rehearsal, there were statistically significant differences between percentages of recommendations given for Scenarios A and B (Table 1).

Recommendations of different strategies in Scenario A
Teachers were asked to recommend three strategies in Scenario A. The number and percentage of given recommendations and their combinations are shown in Table 2.As some teachers recommended fewer than three strategies and some more than three, the total number of answers is different from the expected number of answers.A score of 1 was given if teachers mentioned one or more strategies in a given category.The majority of recommendations included both deep and surface-learning strategies (52.2%), followed by strategies that support deep learning (45.4%), and then surfacelearning strategies (2.4%).Teachers often combined creating associations with selftesting (15.4%), but also cited creating associations alone (12.8%).

Teachers' evaluations
Means and standard deviations of all evaluations are provided in Appendix 2. We carried out confirmatory factor analysis with six latent variables that corresponded to theoretically described variables: creating associations, self-testing, distributing, interleaving, rehearsal, and passive strategies.These were the same variables used when categorizing recommendations (Table 1).Items could only load onto one factor.The first unstandardized loading was fixed to one, while the rest of the loadings were freely estimated.All six latent constructs were allowed to correlate.The model fit was not good [χ 2 (237) = 728.138,p < .001;CFI = .83,TLI = .81,RMSEA = .08,SRMR = .07].Two items ("creates their own learning strategies" and "gives lecture on learned material to family members or peers") were removed due to low factor loadings (.40).The item, "asks who-what-why-how-questions about the learnt material" loaded onto the expected factor (self-testing; .51),but modification indices suggested that it could also load onto three other factors (creating associations, distributing, and rehearsal).Thus, we excluded these three items from further analyses.Creating associations was correlated with self-testing (r = .79;p < .001),distributing (r = .76;p < .001),interleaving (r = .21;p = .002),rehearsal (r = .26;p < .001),and passive strategies (r = p < .001).Self-testing was also related to distributing (r = .83;p < .001),interleaving (r = .17;p = .018),rehearsal (r = .62;p < .001),and passive strategies (r = .75;p < .001).Interleaving was not related to passive strategies or rehearsal.Rehearsal was highly related to passive strategies (r = .82;p < .001).

Relations between teachers' recommendations and evaluations
To examine if teachers who recommended a strategy from one of five categories (Table 1) rated its effectiveness higher than teachers who did not recommend it, we carried out t-tests (see Table 3).Significant differences were found only for distributing and rehearsal.Second, we examined differences in evaluations in groups of teachers who recommended only deep strategies, and those who recommended both deep and surface strategies.Since just three teachers recommended surface strategies only, we did not include them in subsequent analyses.We carried out a 2 (group: recommended deep strategies only, and recommended both deep and surface strategies) × 6 (strategy: creating associations, self-testing, interleaving, distributing, rehearsal, and passive strategies) factorial univariate analysis of variance (ANOVA) with repeated measures for the last factor.The ANOVA revealed a significant main effect of strategy, F (5, 1665) = 638.05,p < .001,η2 = .66,and group, F (1, 333) = 9.05, p < .05,η2 = .03.There was a significant interaction between strategy and group, F (5, 1665) = 20.72,p < .001,η2 = .06.A series of one-way ANOVAs on each of the six dependent variables were conducted as follow-up tests to examine between-group differences (see Table 4).Teachers who only recommended deep strategies evaluated rehearsal and passive strategies lower, and interleaving higher, than teachers who recommended both deep and surface strategies.

Discussion
This study analyzed learning strategies that teachers recommended to students, how teachers evaluate the effectiveness of different learning strategies, and the relations between recommendations and evaluations.We used two learning scenarios: learning prior to a next-day test, and learning prior to a major test.Results showed that teachers recommended more strategies that support deep learning in both learning situations.
Surface-level strategies were generally recommended together with deep-learning strategies.Teachers also believed deep-learning strategies to be more beneficial than surface-level strategies.Consistency between recommendations and evaluations was found for some learning strategies, but not all.

Teachers' recommendations
As expected (H1a), teachers mostly recommended strategies supporting deep learning, and the percentage of these answers was higher (H1b) in Scenario B (learning prior to a major test) than in Scenario A (learning prior to a next-day test).Although our sample included only pre-service and in-service teachers, our findings are consistent with similar studies involving university instructors (McCabe, 2018;Morehead et al., 2016).Comparisons between recommendations for short and long-term learning tasks -an area that has not been studied previously -indicated that teachers adapt their recommendations task specifically.For example, teachers might argue that a next-day test checks factual knowledge and specific problems, while a major test taps conceptual knowledge and includes complex problems from different subtopics.Thus, surface learning may be sufficient for a next-day test, while active integration of new and existing knowledge is needed for long-term knowledge acquisition and flexible use later on (Chi & Wylie, 2014;Hattie & Donoghue, 2016).About 8% of teacher recommendations for the next-day test (compared to 1.7% for the major test) included various passive strategies like underlining and copying text, which tend to be effective only in the short term (Chi & Wylie, 2014;Fiorella & Mayer, 2015).
In addition, distributing is an effective learning strategy when a learner has enough time (Carpenter et al., 2012;Dunlosky et al., 2013).Distributing was mentioned in about one-third of recommendations for the major test, and in fewer than 12% of recommendations for the next-day test.Distributing is also needed in university studies, where knowledge may be checked only at the end of the term (Morehead et al., 2016).
Teachers did not point out interleaving, which may indicate insufficient knowledge of this strategy.Interleaving is a complex strategy that can only be used in specific scenarios, and when topics are carefully selected (Chen et al., 2021).Still, it is a powerful and effective strategy when properly used, thus more attention should be paid to raising teachers' awareness of this strategy.
While self-testing was the most common suggestion on the university level (McCabe, 2018), which was not anticipated (H1c), it was not the most popular strategy overall.Less than one-fifth of suggestions for the next-day test involved self-testing, a surprising finding considering Estonian textbooks usually include test questions at the end of each chapter, and using this strategy is considered both easy and effective.
This may be because teachers treat self-testing as more of an assessment than a learning strategy (Smith & Karpicke, 2014).
Although it was recommended for both learning scenarios, creating associations was mentioned more for the next-day test than for the major test.The reason for this may be that teachers consider students' prior knowledge to be rather good, and believe that it is easy to relate new material to previous knowledge.Teachers may think that students are able independently to create associations with previous study material, everyday life, schematics, and so on.
Rehearsal was recommended for both learning scenarios by 15% of teachers.
Rehearsal is an easy-to-use strategy that has a positive effect on short-term learning (Chi & Wylie, 2014).However, when combined with deep-learning strategies, it can also be an effective long-term learning strategy (Frey et al., 2017).As evidenced by responses to Scenario A (see next section), teachers tended to combine rehearsal with other strategies.

Recommendations of different strategies in Scenario A
Teachers were asked to recommend three different learning strategies to use before a next-day test (Scenario A), which allowed us to examine how teachers combine different strategies.As expected (H2), just a few teachers (2.4%) recommended only surface-learning strategies.Thus, even for short-term learning, the vast majority of teachers appear to consider further aims of learning that can be achieved via deep learning.This finding is similar to those carried out with pre and in-service teachers (Glogger-Frey et al., 2018).
Over 50% of participants combined learning strategies that support both surface and deep learning.As the task was to prepare for a next-day test, using easier and less time-consuming surface strategies may be adequate, but teachers would ideally also consider how to support students' long-term learning, by recommending strategies for deep learning.The majority of answers combined rehearsal with deep-learning and surface-learning strategies.Although rehearsal can lead to good learning outcomes in the short term (Chi & Wylie, 2014), mechanical repetition does not promote long-term learning.However, when rehearsal is spread out over time, this strategy can also be useful for long-term learning.In addition, if rehearsal is combined with an active strategy, this combination can support deep learning (Weinstein et al., 2019).
For example, Roediger and Karpicke (2006) found that rereading while practicing retrieval is more effective than simply rereading alone.Our study also revealed that teachers who recommended strategies that promote both deep learning and surface learning mentioned rehearsal most often in their combinations.In 12 of the 16 combinations, rehearsal was mentioned as one of the strategies.
About 45% of respondents recommended only deep-learning strategies even for short-term learning.The most frequently suggested combination was creating associations and self-testing.Both are strategies that support deep learning: when creating associations, the learner has to associate new information with preliminary knowledge (e.g., when summarizing); and self-testing enhances knowledge acquisition (Fiorella & Mayer, 2015).

Teachers' evaluations and their relation to recommendations
In addition to providing recommendations, teachers were also asked to evaluate the effectiveness of different learning strategies.As expected (H3), teachers' evaluations could be organized into six different categories that aligned with theoretical assumptions: creating associations, self-testing, distributing, interleaving, rehearsal, and passive strategies (Chi & Wylie, 2014;Dunlosky et al., 2013;Fiorella & Mayer, 2015).
With the exception of massing and rereading, teachers tended to evaluate all strategies quite highly.Teachers valued surface-learning strategies the least, that is rehearsal and passive strategies.
To examine the relationship between teachers' recommendations and evaluations, we first analyzed (H4) whether teachers who recommended a specific category evaluated its effectiveness higher than teachers who did not recommend it.It is important to emphasize that teachers first had to give recommendations, then evaluate the efficacy of specific strategies.The hypothesis was confirmed only for distributing and rehearsal.By contrast, teachers evaluated creating associations and self-testing (both deep-learning strategies) very highly, regardless of whether they recommended these strategies or not.However, it is worth mentioning that only a few teachers (21) did not recommend creating associations.Although not many teachers (83) recommended passive strategies, they were still evaluated quite highly.Analyzing both recommendations and evaluations enabled us to determine inconsistencies between teachers' theoretical knowledge and practical knowledge.This has been discussed in earlier studies (see Glogger-Frey et al., 2018).Inconsistencies may also be related to the age of students and the subjects teachers taught, neither of which were examined in this study.
Additionally, we examined differences in evaluations among groups of teachers who recommended only deep-learning strategies, or both deep and surface-learning strategies.As expected (H5), teachers who recommended only deep-learning strategies evaluated surface-learning strategies (rehearsal and passive strategies) lower than teachers who recommended both types of strategies.Consonance in answers was more visible here than in the earlier comparison with specific strategies.In real-world classroom lessons, teachers combine practices and strategies -this should be taken into account during analyses.We may speculate that teachers who recommended only deep strategies can distinguish deep and surface strategies better than teachers who recommended both.These teachers may be aware of the fact that rehearsal and passive strategies support surface, short-term learning, and thus do not recommend these.
An unexpected difference between groups was also found.Teachers who recommended only deep-learning strategies evaluated interleaving higher than teachers who recommended both deep and surface-learning strategies.There are two ways to diversify learning: distributing practice and interleaving (Agarwal et al., 2021).As previously discussed, no teachers recommended interleaving, but many recommended distributing.It is possible that teachers who only recommend strategies that support deep learning are generally more familiar with these strategies and their effectiveness.Thus, a higher rating of interleaving could be associated with better knowledge (note: the effect size was low).The finding that teachers did not recommend interleaving, but rated its effectiveness higher than average, may indicate confusion and uncertainty about this strategy.Only recent empirical studies have examined its value (Agarwal et al., 2021), and researchers still debate its psychological mechanisms (Chen et al., 2021).

Limitations, conclusions, and future directions
The current study had some limitations that must be addressed.First, the sample was quite small, which limits the ability to generalize our findings.In addition, because the study design involved questionnaires, our results do not reflect which strategies teachers actually employ in the classroom with their students.And because it was never asked directly, we cannot firmly conclude which combinations of strategies teachers actually use.
There are many different approaches to learning and teaching (Chi, 2009).However, teachers can only support the multifaceted development of their students by taking into account both student-centered and subject-specific factors, while keeping in mind the aims of education (Kikas et al., 2023;Tang et al., 2017).Teacherdirected didactic approaches reduce the role of the learner.By contrast, constructivist approaches emphasize students' mental activity during the learning process, while stressing the importance of teacher activities in creating a supportive learning environment, choosing appropriate learning tasks, and uniquely supporting each student (Chi & Wiley, 2014;Kikas et al., 2023).In this process, teachers' knowledge of learning, learning strategies, which learning strategies support deep learning, and which support surface learning is vitally important.The appropriate use of learning strategies creates an environment in which students' learning can be more effective, and in which newly acquired knowledge can be used flexibly later.It is important that teachers are aware of different learning strategies, and when they should be used in the context of different learning tasks.Based on our results, it can be concluded that teachers prefer deep-learning strategies and also evaluate these strategies higher.Since we assessed teachers' knowledge using two different methods, we were able to acquire a broad picture of their knowledge.As a practical conclusion, we can recommend that teacher training should emphasize different strategies and the ways in which they can be combined, in order to achieve various learning objectives.
In the future, classroom observations should be carried out, in order to assess which strategies teachers actually recommend and use with students, if any.It would also be important to explore further if and how teachers implement interleaving in the classroom as a strategy to support deep learning (and if not, why not).
Scenario A: Imagine an average, basic school student.The student has to comprehend three pages of text from their textbook for the next day.The student knows that their knowledge will be tested.The student has thoroughly read the text once.What would you recommend doing next?Describe up to three effective learning strategies.Describe these in the most detail you can, and explain why they would be suitable.Scenario B: Imagine an average, basic school student.The student has to prepare for a major test, which is coming up in three weeks.The test covers one chapter of their textbook (several dozen pages), and the student must be ready to answer questions and solve tasks related to this chapter.The student has studied the chapter and solved tasks related to this chapter, and now they have three weeks until the test.What would you recommend doing next?Describe one effective strategy in the most detail you can, and explain why it would be suitable.
CFI and TLI values above .95,RMSEA values below .06,and SRMR values below .08 indicate excellent model fit (Hu & Bentler, 1999), while CFI and TLI values above .90,and RMSEA and SRMR values below .10 are indications of acceptable model fit (Kline, 2015).To answer the fourth research question, t-tests were used.Effect sizes were estimated by Cohen's d (small effect d ≥ 0.20; medium effect d ≥ 0.50; large effect d ≥ 0.80; The model with 21 items still had poor model fit [χ 2 (174) = 538.78,p < .001;CFI = .86,TLI = .83,RMSEA = .08,SRMR = .06].Step by step, we included covariances between the residuals of the same factor suggested by modification indices.The final model included three covariances: "creates a summary, concept map, or diagram/ drawing on the topic" and "uses concept maps, summaries, and drawings"; "copies the text" and "reads the text out loud"; and "underlines or highlights the more important parts of the text" and "repeats the parts he/she has underlined or highlighted".The final model with six factors had mostly acceptable model fit [χ 2 (171) = 410.74,p < .001;

Table 1 :
Number and percentage of teachers' recommendations in two scenarios

Table 2 :
Number and percentage of different category recommendations

STRATEGIES AND THEIR COMBINATIONS n % Strategies that support deep learning 153 45.4
Note. n = number of answers, % = percentage of answers.

Table 3 :
Differences between recommendations and evaluations of learning strategies Note. ** p < .01.No teachers recommended interleaving.

Table 4 :
Teachers' evaluations of learning strategies in different recommendation groups