Socioscienti�c �ecisionMaking in the Science �lassroom� The Effect of EmbeddedMetacognitive Instructions on Students ’ Learning Outcomes

e purpose of the present study was to examine the effects of cooperative training strategies to enhance students’ socioscienti�c decision making as well as their metacognitive skills in the science classroom. Socioscienti�c decision making refers to both �describing socioscienti�c issues� as well as �developing and evaluating solutions� to socioscienti�c issues. We investigated two cooperative training strategies which differedwith respect to embeddedmetacognitive instructions that were developed on the basis of the IMPROVE method. Participants were 360 senior high school students who studied either in a cooperative learning setting (COOP), a cooperative learning settingwith embeddedmetacognitive questions (COOP+META), or a nontreatment control group. Results indicate that students in the two training conditions outperformed students in the control group on both processes of socioscienti�c decision making. However, students in the COOP+META condition did not outperform students in the COOP condition. With respect to students’ learning outcomes on the regulation facet of metacognition, results indicate that all conditions improved over time. Students in the COOP+META condition exhibited highest mean scores at posttest measures, but again, results were not signi�cant. Implications for integrating metacognitive instructions into science classrooms are discussed.


Introduction
Over the past decades curriculum authorities as well as science educators and researchers worldwide have called for changes in the way science is taught at schools (e.g., [1][2][3][4]).Modern science education should not only foster the acquisition of scienti�c content knowledge but engage students in scienti�c inquiry, in lifelong learning and in discussions about modern science problems, their technological applications as well as their personal and societal implications [1][2][3][4][5].In a similar vein, the implementation of socioscienti�c issues into the science classroom has been proposed for more than two decades (e.g., [6][7][8][9][10]).Socioscienti�c issues represent modern science problems, such as global climate change or the loss of worldwide biodiversity, that are tightly linked to social, political, and economical concerns (e.g., [11]).ey are complex, real-world scenarios at the interplay between science and society and thus, can no longer be solved by relying on scienti�c knowledge only [8,10,11].Consequently, they fundamentally challenge the aims and scope of traditional science instruction.
A growing body of research within the area of science education highlights the notion that the implementation of socioscienti�c issues into science classrooms can enhance students' learning outcomes with respect to conceptual scien-ti�c knowledge as well as reasoning and argumentation skills (e.g., [8,[12][13][14][15]). Ottander and Ekborg found that interest in socioscienti�c issues correlates with self-reported learning outcomes in science education [16].In addition, they have the potential to prepare students' becoming literate citizens (e.g., [8,10]).However, working with complex socioscien-ti�c issues also poses high cognitive demands on students, because students need to engage in various information search and evaluation processes as well as argumentation, reasoning, and problem solving processes.is also involves the ability to take perspectives and to integrate multiple perspectives into the development of solution strategies.us, implementation of learning settings that enable students to engage in peer interactions and motivate them to argue, to reason, and to negotiate how to solve these problems and hereby participate in discourse on modern science problems is crucial (e.g., [7,11,17,18]).
Moreover, learning about complex issues needs to be carefully structured, as prior research also showed that students can easily be distracted when working on socioscien-ti�c issues where the outcome is uncertain [16,19].Embedded metacognitive guidance or self-regulated scaffolds have widely been regarded as one means to meet these ends.Among the most prominent approaches that use cooperativemetacognitive settings are Palinscar and Browns' reciprocal teaching method to enhance reading literacy, King and Kitcheners' re�ective �udgment model, or Kings' �uided Peer Questioning [20][21][22].Based on the seminal work of Polya, Mevarech and Kramarski developed the IMPROVE method to activate students' metacognitive skills during mathematical problem solving to enhance students' mathematical achievement [23,24].Within science education, Mevarech and colleagues also used this method to enhance students' scienti�c inquiry skills [25].Azevedo and colleagues could show that facilitation of self-regulated learning can improve student achievement on complex science topics [26].However, still only few studies exist that analyse the effects of such metacognitive or self-regulated learning settings on students' socioscienti�c decision making and reasoning (e.g., [27,28]).e present study aims to contribute to this research need.It analyses the effects of embedded cooperative-metacognitive trainings on senior high school students' reasoning and decision making about socioscienti�c issues.

Effects of Cooperative-Metacognitive Learning Settings on
Student Achievement.Cooperative learning has been on the international agenda for more than half a century by now both in educational research and in educational practice (e.g., [29][30][31][32]).Research on the effects of cooperative learning is traditionally rooted either in social or cognitive psychology.While social psychologists take on a motivational or social cohesion perspective on cooperative learning, cognitive psychologists oen refer to mental information processes that are stimulated by cooperative learning (e.g., [32]).From a constructivist point of view, new knowledge can only be attained if it is connected to and integrated into prior knowledge (e.g., [33]).While learners interact with each other, they provide explanations, engage in discussions, develop arguments about complex problems, and re�ect upon the topic and tasks at hand.ese peer-to-peer interactions can lead to deeper processing of information, facilitation of higher-order thinking skills, and construction of profound knowledge.us, they are likely to enhance individual achievement (e.g., [34,35]).
Numerous research studies could actually show that cooperative learning has bene�cial effects not only on student achievement, but also on student interest as well as social skills [30-32, 36, 37].As a consequence, cooperative learning has oen been euphorically advocated as the optimal learning strategy [34,38].However, empirical research also highlights the notion that cooperative learning is not per se more ben-e�cial than other learning settings [34,38].Merely putting learners into small groups will not lead to interactive group work and meaningful learning (for an overview see [34]).Referring back to the works of Johnson and Johnson as well as Slavin, cooperative learning settings need to account for positive interdependence and individual accountability, promote face-to-face interaction, and foster interpersonal and social skills to be successful (e.g., [29,32]).Moreover, cooperative groups need to be able to monitor and re�ect upon their learning processes [29,32].Especially this last aspect has been identi�ed as one crucial factor for successful collaboration (e.g., [20,22,23,35]) Typically, these studies provide support measures in the form of metacognitive guidance or selfregulated learning trainings to support students' elaboration and learning processes [20,22,23,35,39].
Metacognitive guidance has been extensively used and analysed in the area of mathematics education (e.g., [40]).Mevarech and Kramarski developed the IMPROVE method to enhance students' mathematical reasoning [23].Central to IMPROVE are metacognitive questions that can be differentiated into comprehension, connection, strategic and re�ection questions [23].Comprehension questions address the main idea of the problem or the task to be solved.Connection questions support students in analysing similarities and differences between the current task and tasks that were solved in the past.Strategic questions ask students to re�ect on the speci�c strategy that might be appropriate to solve the task.Finally, re�ection questions ask students to either monitor their learning or problem solving process during or at the end of the process.Mevarech, and colleagues showed in a series of studies that students who studied under the IMPROVE method outperformed students who studied under traditional, more individual instruction or under cooperative instruction that was not additionally structured by metacognitive guidance (e.g., [23,24]).In addition, they could show that a metacognitive instruction using IMPROVE did not only have immediate but also delayed effects [41].Furthermore, Mevarech and Fridkin showed that an intervention using IMPROVE did not only foster students' mathematical knowledge and reasoning but also their metacognitive skills [42].
Within the area of science education, fewer studies explicitly implement cooperative-metacognitive trainings or selfregulated learning in science classrooms.Zion and colleagues transferred the use of IMPROVE to an intervention study on scienti�c inquiry in microbiology [25].ey showed that students who studied under IMPROVE in a network technology environment outperformed those groups that had no metacognitive guidance.Moreover, Azevedo and colleagues showed that students who studied about complex scienti�c issues in self-regulated learning settings with embedded scaffolding outperformed students who studied in self-regulated learning settings without any additional scaffolding [26,43].

Education Research International 3
Metacognition and self-regulation are oen treated as two separate concepts in the literature [44, page 223].However, this is not due to the fact that they are different concepts but that they originally stem from two research �elds� developmental psychology and educational psychology [44].ere are still ongoing discussions about de�ning the relations between metacognition and self-regulation, but a common ground seems to be that metacognition can be seen as a part of self-regulation in that self-regulation can be described as the dynamic interaction of cognitive, metacognitive and motivational aspects of learning [44][45][46].Metacognition is typically de�ned by two components� knowledge of cognition and regulation of cognition (e.g., [44,47]).e former is described as knowledge about one's own cognitive functions and is oen differentiated into declarative, procedural, and conditional knowledge (e.g., [44,47]).e latter is typically regarded as control of one's own cognitive activities and typically refers to processes such as planning, monitoring and evaluation [44,47,48].As the present study aims to enhance students' learning outcomes by using the IMPROVE method, the theoretical basis of the present study refers more to the concept of metacognition than to the concept of self-regulation.

1.�. �nhancing Students� Socioscienti�c �easoning and �eci� sion
Making.Socioscienti�c issues represent controversial issues of modern science that involve social, political, economical, and ethical considerations [8,10,49,50].Examples for socioscienti�c issues are loss of worldwide biodiversity, but also bioethical dilemmas or biotechnology issues such as genetic engineering.ey oen represent issues of �rst frontier science or "science in the making" [49, page 294].ey have their basis in science, but can no longer be solved by relying on scienti�c evidence only [8,11].Instead, they are factually and ethically complex and do not have a clear-cut solution (e.g., [8,10,15,51,52]).Moreover, multiple solutions exist that all have their drawbacks [8,10,15,52,53].New solution strategies have to be developed by integrating multiple, oen competing, perspectives.In addition, socioscienti�c issues and solution strategies are subject to ongoing inquiry and are oen based on uncertain, fragile and con�icting evidence [8,10,50,53].
�orking with socioscienti�c issues in the science classroom poses high processing demands on students because they are engaged in various information search and evaluation processes as well as argumentation and reasoning processes (e.g., [12,15,52,54]).As socioscienti�c issues cannot be solved on the basis of "simple cause and effect reasoning" [10, page 375], students �rst need to understand and describe a socioscienti�c issue in its complexity.Second, they need to be able to generate solutions that account for multiple perspectives on the issue, and third they have to be able to critically evaluate developed or existing solutions (e.g., [55]).
ere is empirical evidence that students can be promoted with respect to socioscienti�c decision making and reasoning.Several studies focused on the quality of argumentation and reasoning processes while dealing with socioscienti�c issues (e.g., [7, 12-15, 27, 54, 56]).Results showed that students can be trained in developing pro and contra arguments, in using trade-offs to compare possible solutions and in weighing arguments or decision criteria to reach an informed decision [7, 12-15, 27, 54, 56].
Few studies exist that analyzed the effect of embedded metacognitive or self-regulated trainings on students' socio-scienti�c decision making and reasoning.�resch and colleagues showed in a pre-post-follow up control-group design that a web-based training program with additional metacognitive prompts to support task analysis enhances students' socioscienti�c decision making with respect to "evaluating solutions" [28].Labuhn and colleagues showed again in a pre-post-follow up control-group design that self-regulated learning elements can be successfully integrated into science classrooms.In addition, they showed that students who studied in a self-regulated learning condition outperformed students who studied under traditional instruction on a knowledge test about decision-making processes [57].Eggert and colleagues used the IMPROVE method in an intervention study among seventh graders to enhance socioscienti�c decision making ("evaluating solutions") with respect to the issue of river assessment and renaturation [27].Results showed positive effects in both training groups.Students in the IMPROVE condition performed better at posttest measures, but the effect was not statistically signi�cant.However, results from this study are promising that metacognitive guidance can have a positive impact on students' socioscien-ti�c reasoning and decision making.

Objectives of the Current Study.
On the basis of existing research, we aimed to investigate the effect of two cooperative training strategies on students' socioscienti�c reasoning and decision making.As described above, working on socioscien-ti�c issues is a complex endeavor.�e assume that cooperative learning settings will provide learners with multiple opportunities to engage in peer-to-peer interactions that are needed to reason and argue about complex socioscienti�c problems.is may than lead to deeper information processing as well as elaboration processes and eventually to better individual performance.Referring to Kirschner and colleagues [34] who postulate that cooperative groups are most successful in terms of effective learning when task complexity is high, we assume that cooperative learning settings are especially adequate for working on socioscienti�c issues.As socioscienti�c issues are not only complex, and solutions need to be developed by integrating multiple perspectives, individuals might bene�t from the advantage to distribute information processing and thus, to reduce cognitive load (cf.[34]).
In more detail, we hypothesize that students who study in cooperative learning settings will produce better learning outcomes with respect to socioscienti�c reasoning and decision making than students who study under more traditional, individual instruction.
Similar to existing research that highlights the importance of metacognitive guidance to support group processing, we also assume that individual student achievement will be enhanced through an additional metacognitive training that explicitly supports students in formulating and answering questions.Referring to Mevarech and Kramarski's work on mathematical problem solving [19,23], we assume that students who work on metacognitive questions will gain a deeper understanding of the problems they work on.In more detail, we assume that students who learn in a cooperativemetacognitive setting will produce better learning outcomes with respect to socioscienti�c reasoning and decision making than students who study in a cooperative learning setting.All participants were from grades 11-13 (last three years of senior high school in Germany).Students studied in three different conditions: cooperative learning (COOP), cooperative learning with embedded metacognitive instruction (COOP+META), and a nontreatment control group with traditional instruction.Due to restrictions concerning school and classroom settings, participants could not be randomly assigned to the different conditions, but assignment took place at the class level.In total, 112 students from 7 classes were assigned to the COOP condition, 129 students from 8 classes to the COOP+META condition, and 119 students from 8 classes to the control group.21 teachers (12 females) participated in the study (mean age = 43 years; age range from 29 to 63 years; mean teaching experience = 13.2 years).

Training Conditions and Learning Material.
Both training conditions (COOP and COOP+META) were identical in terms of lesson structure and time as well as context and tasks.ey only differed with respect to the presence or absence of metacognitive instruction.While students in the COOP+META condition spent time on the metacognitive guidance, students in the COOP condition had time to elaborate on the socioscienti�c issue of palm oil production in Indonesia (see below).Both, the COOP as well as the COOP+META condition used the same set of cooperative learning methods such as the �igsaw and the �shbowl method.In addition, think-pair-share processes were included in all of the lessons [29].
e COOP+META condition was developed using the IMPROVE method [23,24].On the basis of IMPROVE we integrated comprehension, connection, strategic, and re�ection questions into the learning material.ese questions were given to students prior to and during learning activities as well as aer having �nished learning activities.Appendix A shows an example for the implementation of these metacognitive questions into one of students' group work.
e socioscienti�c issue addressed in both training conditions was the issue of palm oil production in Indonesia.ere is an increasing demand on palm oil worldwide as an ingredient for cosmetics and food, but especially with respect to its potential as a biofuel.Palm oil is typically produced on monocultures within the Indonesian rainforest.Due to the increasing demand, more and more plantations emerge.Many people on Sumatra, one of the main islands in Indonesia, work on these plantations to earn their living.As a consequence, the Indonesian rainforest decreases.In addition, indigenous people who traditionally live in and subsist on the rainforest are negatively affected.e described problem represents a typical socioscienti�c issue.It is factually and ethically complex and needs to be addressed by incorporating ecological, economical, and social aspects and perspectives [51].Various social groups play a role within this problem situation such as workers on the plantations (representing nonsustainable users), indigenous people (representing sustainable users) who live in the rainforest, but also external stakeholders such as governments and the consumer in general.With respect to the problem, students need to understand the situation in its complexity.ey need to understand the interdependence between the nonsustainable users and the decrease of the rainforest as a natural resource.e indigenous people in turn suffer from the overuse of the rainforest.us, both social groups are interrelated.Oen, such problem situations are described as socioecological dilemmas [58].
Students in the nontreatment control group received traditional, individual instruction.ey studied according to their regular school curriculum, which did not include the speci�c socioscienti�c issue of palm oil production in Indonesia.�owever, working on socioscienti�c issues is mandatory according to the national educational standards and all training conditions were obliged to teach to these standards [4].
Teachers in both training groups received an introductory training on the learning material.All teachers were familiar with cooperative learning and implemented it regularly in their classrooms.Teachers in the COOP+META condition received a one day introductory training on the IMPROVE method.e training was designed in the biology education research group and administered to the teachers by the researchers themselves.Teachers were �rst introduced to the theoretical construct behind the IMPROVE method and then worked on exemplary student tasks that included the four different metacognitive questions.Teachers in the control group received no speci�c training.

�ocioscienti�c �ecision Ma�ing.
Students' learning outcomes were measured using two 45 min paper-and-pencil tests on socioscienti�c decision making prior to and aer the intervention.e pre-as well as the posttest consisted of three socioscienti�c issues (SSI) that were identical in structure but used different contexts in order to keep students motivated at the posttest.In addition, different contexts were used to counteract increases in students' learning outcomes at the posttest that are only due to training effects on the questionnaire (Appendix B shows two example socioscien-ti�c issues from the pretest questionnaire).Table 1 shows the distribution of the different contexts in the pre-and posttest.
All test items to these socioscienti�c issues were presented in an open-ended format.With respect to the �rst two socioscienti�c issues, students had to describe the problem as well as to develop sustainable solutions to the problem.With respect to the third socioscienti�c issue, students were asked to evaluate presented solutions in terms of their sustainability and to suggest improvements to these solutions.
Students' responses to the open-ended questions were coded independently by two of the researchers.e �nal scoring guide consisted of 10 items (Table 2).Interrater-reliability was found to be sufficient (Cohen's Kappa: ≥ .88).In case of disagreement, discussions took place until agreement on the score could be reached.e socioscienti�c decision making questionnaire (preand posttest version) includes two scales.Scale 1 consists of four items that represent the description of SSIs (items 1-2 for issue no. 1, items 3-4 for issue no.2; Table 2).Scale 2 consists of six items that represent the development and evaluation of sustainable solutions to SSIs (item 5 for issue no. 1, item 6 for issue no.2, items 7-10 for issue no.3; Table 2).Both scales for the pre-and posttest were analysed in terms of reliability in previous studies.Reliability indices were found to be acceptable (  . ..In addition, item difficulties were checked to allow comparisons between pre-and posttest scores.With respect to the present study reliability indices for the pretest were   .(scale 1) and   .(scale 2) and for the posttest   .(scale 1) and   .(scale 2).

Metacognition.
To assess general metacognition a questionnaire developed by A. Kaiser and R. Kaiser [59] was used.e original questionnaire consists of 19 items.Seven of these items refer to the regulation of cognition (planning, monitoring, and debugging), the facet of general metacognition that is relevant for the instructional approach addressed in the present study.Exemplary items were "I check my knowledge in detail that can be helpful to work on the assigned task" or "If I realise that I'm stuck, I will check whether another strategy will be more successful".Each item was scored on a four-point Likert type scale ranging from "I completely agree" to "I completely disagree".Cronbach's Alpha was found to be   .for the pretest and   .for the posttest.

3.�. �ocio�cienti�c �eci�ion Ma�ing.
With respect to the socioscienti�c decision making scales, data were analysed as follows.Concerning "Describing Socioscienti�c Issues", we conducted a one-way ANOVA to examine group differences between the control group and the two experimental groups at the pretest.Results indicated no signi�cant differences prior to the beginning of the intervention with respect to scale 1 ( (2, < .,   .).is legitimated us to conduct a  × 2 (treatment × measurement points) repeatedmeasures ANOVA with "Describing Socioscienti�c Issues" (scale 1) as the dependent variable.Table 3 presents the mean scores and standard deviations with respect to scale 1 by time and treatment.e repeated measures ANOVA indicated a signi�cant main effect for time ( (,  2.2, < .,eta 2  .)and a signi�cant main effect for treatment ( (2,  ., < .,eta 2  .).e interaction effect between treatment and time was also signi�cant (� (2,  .9, < .,eta 2  .8.Post hoc Tukey tests revealed that students in both experimental groups (COOP and COOP+META) performed signi�cantly better than the control group (both Ps <.001).However, the two experimental groups did not differ signi�cantly, thus, indicating that the COOP+META group did not bene�t from the embedded metacognitive instruction.
With respect to "Developing and Evaluating Solutions" we also conducted a one-way ANOVA to check group differences on the pretest scores.Results indicated that there was a signi�cant difference between groups ( (2,  .,  < .).Post hoc Tukey tests revealed that the control group differed signi�cantly from the COOP group on pretest scores at the 5% level of signi�cance.Consequently, we used a multiple regression analysis with prior knowledge (pretest score) and treatment conditions as independent variables and the posttest score of scale 2 as dependent variable.Concerning treatment conditions, two contrast variables were coded.Contrast one examined the difference between the control group and both experimental groups (Control versus COOP and COOP+META).Contrast two examined the difference between the two experimental groups (COOP versus COOP+META).Predictor variables were entered blockwise into the regression analysis.Table 4 shows the mean and standard deviations on "Developing and Evaluating Solutions" (scale 2) by time and treatment.
Results from regression analyses showed that prior knowledge as well as both contrasts predict students' learning outcomes at posttest measures.regression models.e �nal statistical model accounted for 27% of the variance with prior knowledge accounting for 19%, the �rst contrast variable accounting for 7% percent of the variance and the second contrast variable accounting for 1%.Interestingly, the second contrast variable, which represented the difference between the COOP and the COOP+META condition, exhibited a negative relationship with posttest performance.

Metacognition.
With respect to the regulation facet of general metacognition, a one-way ANOVA was conducted to examine possible differences between groups at the pretest.Results indicated no signi�cant differences prior to the beginning of the intervention ( (2,357) < 1.00,   .05).us, data were analysed using a 3 × 2 (treatment × measurement points) repeated-measures ANOVA.Table 6 presents the mean scores and standard deviations by time and treatment.e repeated-measures ANOVA indicated a signi�cant main effect for time ( (1,352) = 58.76, < .001,eta 2 = .14),but no signi�cant main effect for the treatment ( (2,352) = 2.062,  = .13).e interaction effect between treatment and time was signi�cant ( (2,352) = 4.090,  < .02,eta 2 = .023).Post hoc Tukey tests revealed no signi�cant differences between the groups.

Discussion
e major purpose of the present study was to examine the effects of two cooperative training settings (COOP and COOP+META) on students' socioscienti�c decision making and metacognition.Socioscienti�c decision making refers to the description of socioscienti�c issues as well as to the development and evaluation of solutions to socioscienti�c issues.Findings show that both training groups outperformed the nontreatment control group on both scales.is is in line with a large body of research that identi�ed bene�cial effects of structured cooperative learning settings on students' learning outcomes (for an overview, e.g., [30]).It also re�ects �ndings from studies that used the IMPROVE method in mathematics education [23,24,35,41].
However, with respect to the COOP+META treatment condition, �ndings did not meet our expectations.Students who studied in the COOP+META condition did not bene�t from the embedded metacognitive training, as there were no differences on "Describing Socioscienti�c Issues" (scale 1) between the COOP and the COOP+META condition.With respect to "Developing and Evaluating Solutions" (scale 2), �ndings even exhibited a negative relationship between the corresponding contrast variable and students' performance on the scale at the posttest.is, at �rst side astonishing, negative impact raises several questions.Why did students not bene�t from the additional embedded metacognitive training and in more detail, which factors can be identi�ed that lead to the decline in posttest measures when compared to the COOP condition?What can we deduce with respect to future research?
Empirical research on the effects of cooperative learning settings on student achievement suggests that students bene�t most from collaboration if task complexity is high because individuals are more willing to distribute information processing among group members to reduce cognitive load [34].ese bene�cial effects were found mostly in highly structured cooperative groups (e.g., [34,60]).On the basis of these �ndings, we assumed that the COOP+META condition would outperform the COOP condition.However, this was not the case.One possible explanation might be that cooperation between group members in the COOP+META condition was overly structured so that natural cooperation was disturbed or even disrupted.Students were not able to cooperate naturally but were forced into a script that they felt was arti�cial or too detailed (cf.[60]).While students worked on complex socioscienti�c issues, overly structuring their group processes may even have hindered them from employing higher-order thinking skills and being creative (cf.[36]).
When confronted with socioscienti�c issues students have to perform a variety of information search and evaluation processes as well as reasoning and argumentation processes.Students may likely have experienced cognitive overload during group work as they had to solve a complex socioscienti�c issue, collaborate with their peers, and understand and work with the metacognitive instructions (cf.[60]).us, they may have concentrated more on solving the socioscienti�c issues or on working with the metacognitive instruction.As posttest measures on "Developing and Evaluating Solutions" were lower compared to the COOP condition, metacognitive guidance may even have hindered students from dealing with the socioscienti�c issue.Consequently, for future research, we need to carefully reconsider the design of the metacognitive guidance to ensure an adequate balance between group autonomy and provision of additional support measures.Another important aspect addresses the issue of successful implementation of metacognitive trainings into the science classroom.Referring to existing research three fundamental principles have to be acknowledged: ensuring connectivity, being explicit about the function of metacognitive guidance, and extensive and prolonged metacognitive training [61, page 9], [62, page 85].Possible explanations to the unexpected results refer to a combination of these three principles.Although the metacognitive questions were integrated into teaching materials at various stages in the teaching unit and contextualised with respect to the issue taught, students did probably not make use of these questions to their full extent.Although teachers in the COOP+META condition explained the metacognitive questions and their function, students may not have acknowledged their usefulness.us, they did not invest the extra effort that is needed for successful metacognitive instruction [61, page 9].is is in line with �ndings from Hogan, who argues that "simple immersion" of metacognitive guidance in the task is not sufficient to build students' metacognitive knowledge [56, page 1101].Instead, an intervention that explicitly focuses on the use of metacognitive guidance and its functions seems to be more successful [56].
Moreover, written comments on lesson plans from teachers who taught in the COOP+META condition revealed that the use of re�ection questions especially at the end of lessons oen fell short.is has two possible reasons.On the one hand, lesson plans were quite packed with respect to learning goals on socioscienti�c decision making, the socioscienti�c issues taught as well as the extra metacognitive guidance.Although teachers aimed faithfully to implement the lessons according to our instructions, it is only reasonable that they considered the curricular requirements with respect to socioscienti�c decision making �rst.
On the other hand, teachers in the COOP+META condition had no prior experience with metacognitive instruction, which might have led to difficulties during the intervention.Although they were trained in using the metacognitive questions, a shortcoming of the present study is that teachers were not additionally supported during the intervention in their classrooms.
With respect to posttest measures for metacognition, results indicate that all groups, including the control group, improved on the metacognition scale at posttest measures.Although the COOP+META condition had the highest mean score, differences between groups were not statistically signi�cant.us, one has to be cautious about interpreting these results.However, they might still contribute to the discussion described above that students either concentrated more on solving the socioscienti�c issues or on working with the metacognitive questions.Given the highest mean scores on the posttest measures, one might argue that students who studied under the COOP+META condition focused more on working with the metacognitive guidance.
Another explanation to this �nding is that the intervention itself aims to enhance students' critical thinking and re�ection on socioscienti�c issues and possible solutions strategies.While developing and evaluating solution strategies, students need to engage in critical thinking to be able to identify nonsustainable solutions, to incorporate multiple perspectives and to monitor and regulate their own problemsolving processes, especially during group work.Developing solutions to a socioscienti�c issue can be described as a special problem solving process, which can be divided into three main aspects.First, students need to understand and describe the problem situation, second they need to develop possible solutions on the basis of relevant information, and third, they need to evaluate possible solutions in order to reach an informed decision [8,52,55].Especially with respect to the second and third aspect students need to monitor their information search as well as their decision making process as a whole.e described phases were also taught and discussed with students in both training groups.In line with existing research, this may likely have enhanced performance on the metacognition scale at posttest measures (e.g., for an overview see [45]).
Implications for future research are diverse, but one major aspect refers to the improvement of teacher support during the intervention.Apparently, the introductory training to the metacognitive guidance was not sufficient to enable teachers to implement both socioscienti�c decision making and metacognitive instruction into their science classrooms as we had wished.In future research, we need to be even more aware of the "teachers' dilemma" [56, page 1104] to teach according to curricular requirements as well as to focus on the metacognitive instruction and especially become aware of the potential bene�ts of such instruction for students' learning processes.In terms of biology education, it is highly important that the teachers themselves conduct the training and not the researchers, although the latter might lead to better results with respect to student achievement [63].
With respect to the methodological limitations of the present study, a mixed-method approach should be applied in future research that converges both product and process data (e.g., [43]).e analysis of process data would have given deeper insights into possible difficulties with respect to the instruction of metacognitive guidance on the teachers' side as well as the actual learning outcomes on the students' side.
In addition, research needs to be done to shed more light on the relationship between the two concepts of socio-scienti�c decision making and metacognition.As described above, dealing with socioscienti�c issues can be described as a problem solving process.Consequently, students are already engaged in re�ection and monitoring processes.us, it would be extremely important to analyse in depth which processes with respect to the regulation of cognition are being promoted by socioscienti�c decision making in the science classroom.erefore, process data are absolutely vital (cf.[64]).

Appendices A. Metacognitive Guidance
Note. e following questions were given to students in the COOP+META condition during one of their group works.e overall task was to develop a solution to the problem of palm oil production in Indonesia and its side effects with respect to the rainforest and the indigenous people who live in the forest.Each group took up the perspective of one of the groups that are part of the problem.A panel discussion, which succeeded this group work, aimed to integrate all the different perspectives and solutions developed.
Here are some questions that can help you before you actually start your task.
(i) What are the goals of our task?(ii) Can we describe the current situation of the person we are dealing with?Take some notes.
(iii) Which aspects are essential to develop a good solution from the perspective of …? Take some notes.
(iv) How should we proceed to develop a solution and in which way can we apply the strategies from previous lessons?Quote some.
Here are some questions that can help you while working on your task.
(i) Are we still on task or are we running off the track?
(ii) Are we incorporating all essential aspects?
Here are some questions that you should consider just before completing your task.
(i) From the perspective of …, did we consider all important aspects for our solution?
(ii) If our solution to the problem was implemented, how would the situation improve from our perspective?Take some notes.
(iii) Anticipate the consequences that our solution would have for the other social groups!Take some notes.
�. ��a�p�es �or �ocioscienti�c �ssues �ro� the Pretest Questionnaire In Sulawesi, one of the Indonesian islands, indigenous people, who live in and subsist on the rainforest, collect Rattan.ey use Rattan to make ropes for �shing or for farming.ey also use Rattan for building their houses.[…].When collecting Rattan, they take care that some of the shoots will not be harvested, so that the Rattan plants are able to resprout.
Other Sulawesian people also collect Rattan.ey do not only need it for their own supply but they collect and sell it to agents who in turn sell it to the furniture industry.ese people also depend on collecting Rattan to assure their livelihood.e money that they get depends on the amount as well as the weight and diameter of the harvested Rattan stems.ey harvest Rattan in large groups so that they can collect a large amount of Rattan per day.Oen, they collect all of the Rattan shoots in one area.e harvest is being collected and then transported out of the rainforest.
Rattan is a palm that climbs through and over other vegetation.Depending on the Rattan species, it takes about 5-25 years to harvest Rattan for the �rst time.It grows 0.2-1.5 meters per year.For resprouting, it is important that enough Rattan shoots remain in the forest.Otherwise, Rattan species will likely decline within the area.
Although it is forbidden to collect Rattan within the national park on Sulawesi, illegal harvesting still takes place as park rangers oen cannot control the whole area.As a consequence, Rattan species are also under threat in the national park.
Tasks are mentioned below.
(i) Describe the problem situation and explain the interrelations of central aspects.(ii) Develop a possible solution to this problem that acknowledges these interrelations.Shrimp farms provide jobs for many people but also have negative side effects on mangrove areas as well as on the people who live in these areas and subsist on the mangroves.

B.2. Abbreviated Version of
Solution A: Shrimp production in Europe.Shrimp production will be moved to shrimp farms in Europe.Existing shrimp farms in Southeast Asia will be closed and no new aquacultures will be built.us, mangrove areas in Southeast Asia and the people living in mangrove areas won't be affected any longer.
In Europe, indoor shrimp aquacultures will be built that simulate conditions of mangrove areas, in particular marine water conditions.In such aquacultures, shrimps can be raised up to their requested size and then be sold to the food industry.
Solution B: Installing sustainable shrimp production in Southeast Asia.Shrimp farming in Southeast Asia will be shied towards sustainable production.e overall aim is to receive a certi�ed label for shrimp production in these shrimp farms.To receive such a label, shrimp farms have to meet a variety of requirements: At least half of the farming area needs to be covered with mangroves.Existing mangroves must not be cut down, otherwise new mangroves need to be planted.To assure sustainable production permanent controls need to be put through.Due to these new-less intensive-working conditions, less workers will be needed on shrimp aquacultures.
Tasks are mentioned below.
(i) Evaluate both solutions with respect to their sustainable development.Consider positive and negative outcomes in case these solutions would be considered for implementation.
(ii) Develop suggestions for improvement for both solutions.Explain!

T 1 :
Contexts for the different SSIs used in the pre-and posttest questionnaire on socioscienti�c decision making.
Table 5shows the unstandardized beta values and their standard errors as well as standardized beta values with respect to the different T 5: Multiple regression predicting posttest performance on "developing and evaluating solutions" (scale 2) by prior knowledge and treatment condition.P < .05,* * * P < .001. * Rattan is a very popular material for the production of chairs, armchairs, or outdoor furniture.In the 1980s furniture that was made out of Rattan became popular in Europe and North America and has been popular ever since.90% of the Rattan that is used in the furniture industry stems from Indonesian Rattan.[…].
[65]Abbreviated Version of Issue No. 1: Uncontrolled Collection of Rattan in the Indonesian RainforestNote.edescription of this socioscienti�c issue is based on research and �ndings from �och and colleagues[65].B.1.1.Rattan from Indonesia.
Issue No. 3: Shrimp Aquaculture in Mangrove Areas in Indonesia B.2.1.Shrimp Aquaculture in Southeast Asia B.2.2.Introductory text to the problem.Due to the high demand for shrimps in Germany and worldwide, shrimp farming in mangrove areas of Southeast Asia is steadily increasing.