1. Introduction
In a dynamic and uncertain world, SRL is essential in preparing young students to become active, lifelong learners who can competently navigate change [
1]. Research highlights the importance of SRL not only for school success, but also for the continuous development of competences and skills [
2]. SRL is generally defined as an active, cyclical process [
3]. During this process students set goals, monitor, and control their cognition, motivation, and behavior based on their objectives and the environmental context [
4] (p. 453). However, despite the presence of SRL practices in many schools, teachers rarely foster metacognitive knowledge or engage in strategy instruction by providing insights on why and how students could use strategies for learning [
5,
6]. To address this, several interventions were conducted to support teachers in fostering SRL among students. These studies reported varying effect sizes, ranging from none [
7] to small [
8] to medium [
9]. The results show that while it may be challenging to demonstrate positive student outcomes, an intervention can potentially foster SRL.
Since SRL should preferably be promoted within specific subject areas [
10], this intervention intends to promote SRL within reading tasks among 5th and 6th graders, recognizing the relevance of reading comprehension for overall school achievement. While reading can become a partially automized process, as with practice word recognition and decoding demand less conscious effort, comprehending more complex texts requires certain aspects of SRL, such as planning or monitoring [
11]. Previous interventions promoting SRL among 5th and 6th graders have understated cooperative learning among students, even though research has highlighted its beneficial impact on academic achievement and other dimensions of learning, including meta-cognitive, cognitive, and social aspects [
12].
Additionally, it is crucial to include both teachers and parents as important actors in developing students’ SRL [
13,
14]. However, research in this area remains limited, as Otto’s [
15] study is the only SRL intervention in primary school involving parents and teachers. To address these gaps, a ten-week school-based training program based on Stöger and Ziegler [
16] was adapted to include cooperative learning and parental involvement in order to foster SRL within reading tasks among 5th and 6th graders. Furthermore, prior intervention studies fostering SRL within reading tasks [
8,
17] gave only minimal consideration to treatment integrity. Nevertheless, it is essential to account for treatment integrity since it can affect the success of an intervention [
18]. In the present study, treatment integrity was taken into consideration.
3. The Current Study
Building on the SRL training of Stöger and Ziegler [
16] that showed positive effects on preference for SRL and reading comprehension, the current study added the following two components to the training: cooperative learning and parental involvement. While teacher-led interventions tend to have lesser effects than researcher-led interventions, the former are valuable for supporting knowledge transfer in authentic learning environments [
10]. Hence, in this study, teachers were trained to administer the training.
The aim of this study was to examine whether the adapted training led to positive effects on students’ outcome variables (SRL, reading comprehension). Effects were analyzed twice, in the post-test (analyzing effects right after the main training) and in the follow-up (analyzing effects of the whole training, including booster sessions in the long-term). The following questions were addressed:
- (1)
Did the intervention significantly increase the reported SRL activities of the experimental group in comparison with the control group in the medium (post-test) and long term (post-test) (while controlling for gender, cognitive ability, first language, parental educational level, and participation in parental training)?
Hypothesis 1: The intervention significantly increased the reported SRL activities of the experimental group compared to the control group in the medium term. In the long term, the effects of the intervention on the reported SRL activities were maintained, if not increased.
- (2)
Did the intervention significantly increase reading comprehension in the experimental group in comparison with the control group in the medium and long term (while controlling for gender, cognitive ability, first language, parental educational level, and participation in parental training)?
Hypothesis 2: The intervention significantly increased the reading comprehension of the experimental group compared to the control group in the medium term. Furthermore, booster sessions sustained the effect of the intervention on long-term reading comprehension.
5. Intervention Program
5.1. Teacher Training
Before implementing the training in their classes, the intervention group teachers attended three 3.5 h training sessions (outside of regular school hours). These interactive trainings (e.g., group discussions, self-reflection periods) were designed and delivered by SRL experts. The first training addressed SRL models, theories, and learning strategies. The second presented the structure of the two introductory weeks on reading comprehension strategies and the SRL cycle. In the third session, the concept for the eight-week training was explained and the teachers were instructed on cooperative learning. Teachers received a manual with lesson plans and required materials (e.g., learning diaries) and were invited to contact researchers with any queries about the training. Two optional online meetings were organized to discuss any implementation issues.
The control group teachers received written information about the data collection process before the pre-test and at regular intervals throughout the project. They did not receive any further instruction about SRL, but they were invited to attend the same training free of charge after the intervention.
5.2. Parental Training
For some classes, a parental training session was offered to parents on supporting their child’s SRL at home. The content of the training was based on social-cognitive learning theory [
19] and conceptual change theory [
61]. Parents were encouraged to reflect upon their attitudes toward school and failure at school (failure mindset). Furthermore, strategies on how to facilitate SRL at home, such as organizing the learning space, were discussed. Parents were also presented with methods to provide motivational support and encourage autonomy amongst their children, without placing an emphasis on active parental involvement [
38,
62]. Custom-designed videos were used to give examples of constructive and counterproductive parental behaviors in the context of learning. Parents received handouts covering the topics discussed. Those parents who had been initially excluded from attending the parent training were invited to attend the training after the intervention.
5.3. Students’ Training
The students’ SRL training was based on Stöger and Ziegler’s [
16] training, which was enhanced with cooperative learning and parental involvement. The ten weeks of training were delivered by the regular teachers over the course of five months (see
Figure 2). The training was structured as follows: two introductory weeks were immediately followed by five weeks of training. After a break of a few weeks, three further weeks of training were given, using the same content as before, over a longer period of time (see booster sessions,
Figure 2).
During the first introductory week in class, teachers addressed the importance of text comprehension strategies and introduced three text reduction strategies. These included the following: 1. underlining and making lists of the main ideas, 2. drawing mind maps that contain the main ideas, and 3. summarizing the main ideas in their own words [
8,
16]. Students learned and practiced how to use these strategies effectively when reading. At the end of this week, students were given an overview with advice on text reduction strategies and encouraged to refer to this summary as required throughout the program. During the second introductory week, students applied their knowledge about reading comprehension strategies to become familiar with the SRL cycle (see
Figure 1). Teachers explained the eight SRL steps using “Sail the Mouse”. This fictional character was used as a first-person narrator to help students better understand the SRL steps, as storytelling supports students’ understanding [
63]. The lessons contained examples from everyday situations, such as preparing for a sports competition, which include setting measurable and achievable goals. Finally, teachers gave students an overview of what the training program consisted of in the following weeks.
During the training weeks, the students applied the reading comprehension strategies and SRL steps four days a week (no training on Wednesdays). Over this period, students were expected to work through the eight steps of the SRL cycle repeatedly and actively. The learning process was supported by a learning diary, helping students to reflect on their strategy use and supporting their SRL development [
64].
Since the reading texts covered natural science topics (e.g., animals, human body), the training took place in German and science classes. Every week, the students were expected to read three texts at school and apply the SRL cycle when reading. The texts developed by Stöger and Ziegler [
16] contained ten main ideas and students could set a goal for each, defining how many main ideas they would find. Each one was approximately 420 words long and of similar difficulty. To accommodate students in different grades, a second simplified version of each text was generated by the study’s authors according to specific guidelines. These guidelines included the use of shorter words and favored active over passive sentences. This gave students a choice best suited to their reading level.
Two types of cooperative learning were implemented: (1) discussing experiences regarding strategy use, and (2) questioning about the text. Research indicates that questioning enhances learning as it encourages learners to engage deeply with the content [
42]. Moreover, dialogic peer feedback about strategy use requires students to verbalize their thoughts and actions, promoting metacognition [
36]. To support this self-directed group work, students used flashcards to provide a clearly defined procedure of their cooperative tasks and to allocate specific roles to each student (e.g., questioning, providing feedback). In addition, teachers gave students instruction and support on how to elaborate their thoughts and provide peer feedback. As (peer) feedback should be provided frequently to further support the development of SRL [
33], cooperative learning opportunities were scheduled three times a week.
5.4. Treatment Integrity
Throughout the intervention, teachers provided weekly self-reports by completing online questionnaires on how they were implementing the training and additional information about the students’ motivation, the number of cooperative learning sessions, and their perception of the implementation quality. The treatment integrity analysis revealed that most of the teachers implemented the training program as intended, as they all completed the two introductory weeks, and their students read at least 17 of the 24 texts during the training weeks. One class had to be excluded since the implementation of their training diverged significantly (they did not use the standardized learning diary).
In the follow-up test, the control group class teachers were asked whether they fostered SRL in class during the school year and if so, how, and to what extent. Approximately 60% indicated that they fostered SRL in class during the school year through a range of activities, such as lessons in which students could plan their daily or weekly activities, choose difficulty levels, and reflect on their learning. Some teachers also gave lessons in which students worked on self-chosen content or individual projects. Topics such as fostering independence, help-seeking, and responsibility were also mentioned. However, none of these teachers used systematic SRL training. Additionally, 42.9% of the control group teachers said that they did not foster SRL in class during the school year.
7. Statistical Analyses
Due to the hierarchical structure of the data (students nested in classes), multilevel analyses were conducted using the software ‘Mplus 8.10’ [
70]. To examine the development of the reported SRL activities and the reading comprehension, respectively, two analyses were conducted—one predicting outcome at post-test, and one predicting outcome at follow-up—to investigate for medium- and long-term effects. SRL was modeled as a latent variable to explicitly model measurement errors. In line with the eight-step SRL model, each step was conceptualized as a first-order factor and calculated as a second-order factor. All other variables were used as manifest variables. Relevant control variables (gender, first language, cognitive abilities, and parental education level) were included in all analyses. Parental participation in the parent training was also used as a control variable, as not all parents of the SRL training and parental involvement condition attended the training. Subsequently, all students who received the SRL training formed the experimental group.
The effect of the intervention on the development of SRL (Hypothesis 1) was analyzed with a multilevel structural equation model (ML-SEM), while the effect on reading comprehension (Hypothesis 2) was analyzed with a (manifest) multilevel analysis. To evaluate the model fit of the ML-SEM models, several fit indices commonly applied to latent-variable models were used, namely, the Satorra–Bentler scaled chi-square test [
71], 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). Hu and Bentler [
72] suggest that values of CFI/TLI ≥ 0.95 and RMSEA/SRMR ≤ 0.05 indicate a good model fit.
Preliminary analyses testing longitudinal measurement invariance were conducted for SRL. However, longitudinal multilevel confirmatory factor analyses with the second-order SRL model with twenty-four items and eight factors were too complex and failed to converge. Therefore, the measurement invariance was tested with single-level models. Furthermore, the longitudinal measurement invariance was checked following the recommendations of Liu et al. [
73] for ordinal data. This testing includes the assessment of invariance over time through a series of models; it starts with a baseline model to ensure that factor loadings are consistent over time. This first step is followed by more restrictive models: the loading invariance model (same factor loadings across time), the threshold invariance model (consistent thresholds for response categories), and the unique factor invariance model (equal unique factor variances over time) [
73]. In addition, since SRL was conceptualized as a second-order construct, measurement invariance was tested by a series of increasingly restricted models comparing the construct over time (supplements: (a) configural model, (b) first-order factor loadings, (c) second-order factor loadings, (d) thresholds of measured variables, (e) intercepts of first-order factors, (f) disturbances of first-order factors, and finally, (g) disturbances of measured variables) [
74]. When comparing the models [
73,
74], a CFI and TLI decline of 0.01 or less were used as a reference to indicate that the measurement invariance hypothesis should not be rejected [
75].
Due to the complexity of the longitudinal ML-SEM analyses, factor scores were extracted from a first-order measurement model. The factor scores of the eight first-order factors were then used as indicators of the general SRL latent factor, simplifying the measurement model in the subsequent analyses. Factor scores may not explicitly control for measurement error, but they offer a reliable approach to this issue because they assign greater weight to items characterized by lower measurement errors and therefore provide partial control for measurement errors [
76]. The extracted factor scores were used to calculate the longitudinal ML-SEM using the robust maximum likelihood estimator (MLR) [
70]. Individual variables (e.g., gender, first language) were group-mean centered, and level two variables (e.g., condition) were grand-mean centered [
77]. Since the hypotheses of this study were directional, the
p-values were interpreted one-tailed with a significance level of
p < 0.05 [
78].
Furthermore, when longitudinal measurement invariance across time was tested, the preliminary models revealed a negative covariance matrix due to a correlation higher than one between latent variables. Consequently, modification indices were analyzed, and two items were excluded. The excluded items were one item of strategy implementation (“First I think about the best way to approach tasks and then proceed accordingly.”) and one item of ‘outcome evaluation’ (“I think about how my grades have changed from one exam to the next.”). The reasons for their exclusion were twofold: the modification indices suggested an improved model fit upon their removal, and these items might have assessed different aspects of the respective SRL step in comparison with the other items. These adapted models achieved a satisfactory model fit and showed a CFI and TLI decline of 0.01 or less when compared, indicating that the measurement invariance hypothesis should not be rejected [
75] (
Appendix A).
The intra-class correlation coefficients (ICC) varied from a low to medium range (ICC of the eight factor scores of the SRL pre-test: 0.03–0.07; post-test: 0.04–0.07; follow-up test: 0.01–0.07; reading comprehension: pre-test: 0.23; post-test: 0.18; follow-up test: 0.16), whereas [
79] suggest that ICC values from 0.05 to 0.10 represent a small to medium effect, indicating a group effect. Given the class-level administration of the intervention program, multilevel analyses were conducted to control for this aspect.
9. Discussion
The main goal of this study was to examine the effects of an existing SRL intervention program [
16] enhanced with cooperative learning and parental involvement. Contrary to expectations, students in the intervention group did not report more SRL activities in the post-test compared to control group students. Similar results were observed in the follow-up test, as no increase in reported SRL activities was found. This suggests that the intervention, including the booster sessions, did not increase students’ reported SRL activities in the medium or long term. Additionally, while improvements in reading comprehension were observed in both groups, there were no significant differences in the medium or long term between the groups. To gain a better understanding of these outcomes, treatment integrity was examined.
This study’s outcomes are surprising, particularly because Stoeger et al. [
8] found a significant effect on the SRL preference in the training group. Unlike Stoeger and Ziegler’s research, this study measured the reported SRL activities. This difference highlights a critical insight: an improvement in preference for SRL does not necessarily lead to a change in behavior. However, it is important to distinguish whether an intervention changes only the students’ attitudes towards SRL or leads to a change in their learning behavior. Similar results were found in studies focusing on teachers, as noted by Steinbach and Stoeger [
80] and Spruce and Bol [
81]. Both found no significant correlation between teachers’ attitudes towards SRL and their actual SRL teaching practices. Nevertheless, it should be noted that the current study did not assess the actual SRL competences of the students, but rather their self-reported SRL activities, which in turn tend to show low correlations with actual SRL competences [
82].
Moreover, the absence of significant effects in intervention studies is not uncommon in educational research [
83], possibly because transmitting knowledge is complex and influenced by various factors. There are several potential explanations for the lack of intervention effects: one might be the initially high levels of student agreement to engage in SRL activities. This suggests a ceiling effect, meaning that it might be difficult to achieve more SRL activities. Students’ responses regarding the willingness to engage in SRL activities might be influenced by social desirability or overconfidence, which is particularly common among underachieving students [
84]. However, if self-reports are influenced by social desirability or overconfidence, it might be challenging to measure actual progress. Moreover, it is crucial to note that approximately 60% of the teachers in the control group reported promoting SRL in their class, although they did not use systematic training. This suggests a high level of interest and initiative to promote SRL, despite a lack of a structured approach and raises the question of whether this high level of interest could potentially reduce differences in outcomes. Nevertheless, this lack of an approach in fostering students’ SRL underscores the importance of teacher-led interventions that provide teachers with the necessary knowledge to implement SRL effectively and sustainably. Consistent with previous findings that researcher-led interventions generally yield more effects than those led by teachers, this study shows that achieving an impact with teacher-led interventions continues to be challenging; further approaches to enhance the effectiveness of teacher-led interventions should therefore be considered, for example, by cultivating a closer cooperation between researchers and teachers [
10].
The parental training did not have an influence on the students’ SRL. This finding is not surprising given that the parental training consisted of only one 1.5 h session addressing parental (emotional) learning support rather than focusing on specific SRL steps. It is possible, however, that the parental training affected the students’ learning motivation more than SRL activities. Unfortunately, students’ learning motivation was not measured in this study.
The lack of intervention effects regarding reading comprehension may be due to the intervention’s primary focus on SRL. Additionally, the experimental group’s higher proportion of non-native German speakers may have introduced additional challenges, such as limited vocabulary. While the aim of this training was to improve students’ ability to apply text reduction strategies, it did not necessarily lead to an improvement in the results of the reading comprehension test. The test (FLVT) assessed surface-level reading skills (e.g., basic summary of the text) as well as in-depth comprehension (e.g., identifying complex connections within the text). This training focused on fostering text reduction strategies (e.g., drawing a mind map to illustrate connections within the text), thereby promoting some (but not all) abilities assessed by the FLVT. Moreover, it should be noted that reading comprehension also relies on a wider range of skills such as word recognition, vocabulary, or background knowledge, which were only fostered implicitly (and not explicitly) through this training [
85].
Despite the lack of intervention effects, this study presents interesting results about treatment integrity by demonstrating that several implementation variables can significantly affect the outcomes of the intervention (e.g., [
54]). For instance, adherence, which was measured by the amount of cooperative learning sessions about strategy use, had a positive effect on the SRL post-test outcome. This suggests that frequent cooperative learning sessions with discussions about SRL can lead to more SRL activities. This result is in line with several studies that underline the beneficial effects of cooperative learning on diverse aspects of learning (e.g., [
86]). In addition, the number of texts read during the training showed a tendency toward a positive effect on the reported SRL activities in the post-test, highlighting the importance of regular practice. Furthermore, child responsiveness plays a vital role in the success of such a training as higher perceived motivation of students also positively influenced the SRL outcome variable in the post-test. These results indicate that the training can be effective when it is implemented as intended.
To summarize, these results suggest that the implementation of cooperative learning sessions and a motivating learning environment are crucial elements in fostering SRL. Considering the complexity of SRL, which involves planning, monitoring, and reflecting on learning processes, repeated practice plays an important role in developing students into competent self-regulated learners.
Although significant medium-term effects of treatment integrity variables on reported SRL activities were found, these effects were not observed in the long-term data. A potential reason for this could be that there were only three booster session weeks after the main intervention, with no further sessions for nearly two months until the follow-up testing, potentially minimizing the effects of treatment integrity. Moreover, there were no effects of treatment integrity variables on reading comprehension (post- and follow-up). This could be explained by the assessment, as general reading comprehension was measured, and not only the specific competences fostered through the intervention (e.g., finding main ideas).
Limitations and Future Directions
This study offers interesting insights into fostering SRL and treatment integrity; nevertheless, it is important to acknowledge its limitations. For a more objective assessment of SRL beyond self-reports, future studies should consider evaluating students’ learning journals [
87]. However, as journal entries reflect a subjective evaluation of one’s behavior, it would be beneficial to assess changes in students’ SRL using methods such as the think-aloud method [
88] or a microanalytic assessment method [
89]. Furthermore, SRL was only measured in terms of quantity, indicating that a greater number of SRL activities by students would result in a higher score. This should be viewed critically, as learning can be successfully regulated through few SRL strategies if they are suitable for the task and the learner is proficient when using them [
90]. Another limitation of this study is that students’ motivation was only measured through teacher reports, even though motivation is equally essential for SRL as metacognitive and cognitive aspects [
20]. Furthermore, motivation may have been an important variable when analyzing the effects of the parental training.
Additionally, the teachers were informed of the study topic before randomized allocation to the experimental and control groups to reach informed consent, which might have potentially increased the control group teachers’ sensitivity to SRL.
While this study represents an advancement upon previous intervention studies [
8,
17,
41] by assessing and analyzing the effects of multiple components of treatment integrity, it nonetheless relies on self-reported teacher data and single-item assessment. Direct observations are considered as a benchmark in treatment integrity measurement [
91], but they require significant time and financial resources [
57,
92]. Furthermore, they only capture observable, frequent practices [
52], and can distract participants [
56]. Additionally, the assessment of competence of delivery should be viewed critically, given that it was based on self-reported measures of implementation success, which may not measure actual competence. Nevertheless, it offers an insightful and non-intrusive way to observe the teachers’ perceived competence. An essential goal for future research, therefore, is the development of a reliable instrument to comprehensively assess the four components of treatment integrity (adherence, competence of delivery, treatment differentiation, and child responsiveness).
When designing interventions to foster SRL within reading tasks among primary school students, a combination of text reduction strategies and metacognitive components (visually supported by a SRL cycle, as in the present study) proves to be a favorable approach. Furthermore, two core domains of academic learning are targeted: reading, and self-regulation. Based on the current findings about the benefits of cooperative learning when fostering SRL, the next steps in reading instruction could be to implement group sessions over a certain period that give students the opportunity to share insights about their learning processes and advise each other on effectively using text reduction strategies. Teachers should prepare students thoroughly to this task and scaffold this process with questions that stimulate active and meaningful conversation. To enhance reading comprehension further, it would be valuable to incorporate additional strategies beyond text reduction, such as activating prior knowledge [
93] or making inferences.
In conclusion, despite the absence of intervention effects, this study offers valuable insights into the importance of treatment integrity. It highlights the importance of adherence and child responsiveness when fostering SRL. The creation of a motivating learning environment and the integration of consistent, systematic practice appear to be key components in the successful promotion of SRL. Moreover, providing students with frequent, carefully introduced opportunities for cooperative learning to discuss and reflect about their SRL activities is a promising element when fostering SRL.