An Effort to Strengthen the Objectives of Education for Sustainable Development, Based on the Use of the Cosmos–Evidence–Ideas Model

Abstract

The effective teaching of the basic concepts of ecology is a goal of science teaching and education for sustainable development around which various researchers have been active since it is estimated that it can help people to understand and deal with the sustainability issues of our time. The outcomes of the empirical studies typically provide design recommendations that enhance the effectiveness of any later work that incorporates them. The present study examines whether the Cosmos–Evidence–Ideas (CEI) epistemological model could improve the efficacy of a teaching learning sequence (TLS) for important ecological concepts when utilized as a design tool in an effort to support students’ understanding of ecology. A TLS for key concepts of ecology was designed without considering the CEI model and then modified based on it, resulting in a revised second TLS. The two TLSs were implemented in different groups, evaluated, and compared. It was found that while in both groups the performance of the students showed a statistically significant improvement, in the group where the revised TLS was applied, this improvement was greater, and it seems that students were helped to abandon more of the alternative ideas they originally held. Thus, there are indications that the model might improve the efficiency of a TLS for fundamental ecological concepts.

1. Introduction

In a complex and rapidly changing world, the science of ecology is equipped to deal with current environmental issues and is expected to make a substantial contribution to the understanding of and response to environmental and sustainability problems arising both locally and globally [1]. This claim justifies the growing interest in strengthening ecological literacy, environmental literacy, and ecoliteracy in society. The above terms evolved historically and succeeded each other as the needs of society changed and as environmental and ecological education adapted its objectives to focus on meeting those needs. However, these are terms that show considerable overlap since their scope is common: the environment and the relationship and attitude of humans towards it. The most widely accepted definition of environmental literacy is that it includes awareness and concern about the environment and problems related to it as well as the knowledge, skills, and motivation needed to solve current problems and prevent new ones [2]. Environmental literacy aims to create conscientious citizens who will participate in making informed decisions or taking action on environmental issues [3]. That is why it is encouraged to integrate it into formal education from the first grade [4] or in general biology courses [5]. Ecological literacy is differentiated from environmental literacy because of its focus on the basic ecological knowledge necessary for making informed decisions, which is acquired through scientific research and systems thinking [6]. In an effort to close the gap between these ecological and environmental education study areas, ref. [7] suggested that ecological literacy is a subset of environmental literacy, meaning that environmental literacy is basically a combination of ecological literacy and civic literacy. At the same time, ref. [8] developed an idea of literacy that emphasizes the creation of sustainable human communities and calls for a fundamental reconstruction of the entire educational system, leading to the formation of a new idea, ecoliteracy. The term sustainable development, on which the concept of ecoliteracy is based, is described by the World Commission on Environment and Education [9] as “development that meets the needs of the present without compromising the ability of future generations to meet their own needs”. Capra [10] first used the word ecoliteracy to describe the ability to comprehend the fundamentals of ecosystem organization and apply them to build sustainable human comsmunities and societies. What further differentiates ecoliteracy from ecological literacy is the explicit emphasis on sustainability.

The concept of sustainability was prioritized when, in 2015, the United Nations defined the 17 Sustainable Development Goals (SDGs) to work in a coordinated way to address global challenges such as poverty, inequality, climate change, environmental degradation, peace, and justice [11,12]. The goals, set up to 2030, are interconnected and have been shaped by the pursuit of a future that is better and sustainable for all [11]. Depending on the perspective from which they focus on sustainability, the objectives fall into three broad categories: environmental objectives, which focus on the protection of the planet and the sustainable management of resources [11]; social objectives, which aim at social inclusion, equality, and justice [11]; and economic objectives, which seek to develop the economy in a way that ensures viability [11]. Beyond taking a holistic approach to its goals [13], education for sustainable development requires an understanding of fundamental ecological concepts to raise environmental awareness [13,14,15]. Developing systems thinking skills [15] and incorporating real-world applications into lessons [14] are also recommended as effective practices. Moreover, key issues of education for sustainable development were identified so as to facilitate educators in the implementation of the above general principles [16]. To conclude, it has become clear over the years that societies need to be restructured so that the way they operate does not create problems that need to be solved after the fact but instead takes the perspective of managing available resources with the least possible negative impact on their evolution. A prerequisite for adopting appropriate attitudes and behaviors towards sustainability is an understanding of the fundamental scientific knowledge relating to the structure and functioning of ecosystems since the achievement of environmental objectives is fundamental to achieving social and economic objectives. This should therefore be one of the main objectives of education: to prepare citizens who base their choices and decisions on scientifically based data.

Furthermore, it is argued that science concepts such as those relating to the ecosystem may offer teachers an opportunity to help their students become familiar with understanding the nature of science [17]. For students, comprehending the nature of science is essential because it fosters scientific literacy, gives them the tools to think critically, and helps them make informed decisions [18,19].

However, teaching of the basic concepts of ecology has several difficulties. Initially, students, before starting their formal education, have already formed alternative understandings of these concepts, such as ecosystem structure and function [20,21,22,23], the interactions of organisms [23,24,25,26,27,28], energy flow as well as the food chains and food webs with which it is represented [20,21,23,26,29,30,31,32,33,34,35], and plant growth and nutrition [33,36,37,38,39,40], which altogether make understanding difficult. Also, difficulties in understanding are caused by the common finding that, in recent years, an escalating disconnection of humans, especially children and young people, from nature has been observed [41]. As a result, secondary school students often find ecology education detached from real life and unappealing to them, as they are unfamiliar with the species and dynamics presented. Additionally, while systems thinking is increasingly central to being scientifically and ecologically literate [42], ecosystem functioning is not perceived as an interconnected whole by the students [43,44].

There is intense research activity to address the above challenges. Interesting steps for the design of teaching have been proposed in the literature, such as the identification and formulation of a set of ecological principles that will be the subject of teaching [45]; defining learning objectives clearly and explaining them to the students from the beginning [46]; developing multiple ways of assessing students’ understanding of principles, concepts, information, and ideas and explaining them to them [46]; the selection of learning activities that are structured in such a way as to activate the equal participation of students in work groups [46]; and simplifying complex systems using models and interactive classroom design [47].

All the aforementioned empirical research highlights specific elements in instruction that are considered essential for enhancing learning. The present research argues that the effectiveness of the above proposals could be strengthened if the selected teaching activities were analyzed prior to their implementation on the basis of the proposed model and modified if this was deemed necessary. The proposed model analyzes each educational activity as a collection of interactions, each of which alone and collectively contributes equally to the activity’s effectiveness. If some of these interactions are absent or insufficient, it may have a negative impact on the effectiveness of an educational activity. The advantage of the current proposal is that it provides the chance for prediction and targeted action during planning to close such gaps and boost activity efficiency. Since it may be used in any activity, regardless of the design principles on which it is based, another advantage is that it can include the results of empirical research conducted in the field. To conclude, the present research investigates whether the use of the CEI model as a design tool has the potential to enhance the learning gains of a TLS that is based on the current trends of science teaching.

Theoretical Background

The present proposal is based on the utilization of two epistemological models as tools for designing teaching learning sequences (TLS) for basic ecological concepts. A TLS is an interventional research activity that lasts a few weeks and includes well-validated teaching–learning activities empirically adapted to student reasoning as well as teachers’ guidance with well-documented teaching approaches and expected student reactions [48,49]. The first model is the model of Educational Reconstruction [50], which was chosen because it is well known and frequently used in science instruction as a design tool for TLSs. Figure 1 summarizes its essential elements.

The second model, Cosmos–Evidence–Ideas (CEI), presented in Figure 2, is based on Hacking’s [51] classification of laboratory practice entities. Hacking’s classification refers to the laboratory sciences that “construct” part of the material world in the laboratory in order to study it [51]. The use of the CEI framework in educational settings [52,53] has the advantage of allowing a fruitful analysis of teaching–learning activities in terms of scientific practice; it does not, however, imply that the variety of possible patterns is precisely similar for students and scientists. According to Hacking [51], scientific activity is characterized by an almost autonomous “inner life”, which has an internal dynamic that is hardly disturbed by external factors such as social or cultural relationships. The entities that are part of this inner life interact with each other and are constantly transformed as a result of this interaction. They are grouped into three categories: Ideas, Cosmos, and Evidence. Ideas refers to theoretical concepts, beliefs, questions, fundamental knowledge, theoretical models, and systematic theories, all related to the natural phenomenon being studied. They are linguistic entities whose aim is to represent the physical phenomena to which they refer. The Cosmos entity includes the sample, data generators, devices that interact with the sample, and raw data such as change graphs, photos, etc. All of the above give substance to the phenomenon in the real world, to which the scientist himself belongs. Sometimes, the presentation of the content of the material world can only be done in language, for example, when a teaching activity is described. The Cosmos concept does not contain a conceptual content by itself; this is provided by language when we interact with this entity. For this reason, the terms used in these descriptions are as neutral as possible so that they are understood to correspond to entities that in themselves have no conceptual load. They will acquire specific content (Idea or Evidence) only after appropriate manipulation by students. Evidence comprises data that have undergone some form of processing (e.g., estimation, reduction, analysis, or interpretation). It also includes linguistic entities, just like the entity of Ideas, that aim to represent the laboratory phenomenon.

In educational settings, specific types of activities that promote the connections between these entities are proposed [54], aiming at the improvement of science understanding (

Table 1. CEI model. The relationships in educational settings [54].

Representational	I → E	Linking Ideas with expected Evidence. Predictions of Evidence based on one’s own ideas.
	E → I	Linking Evidence with Ideas. Explaining specific Evidence in terms of some specific Ideas. These Ideas can be scientific or common.
	C→ E	Linking a piece of Cosmos with a piece of Evidence. Descriptions of what is happening in Cosmos in terms of observed or recalled Evidence.
Interventional	E → C	Linking Evidence with a piece of Cosmos. Constructing, intervening, or modifying a specific segment of the material world on the basis of a specific piece of evidence.
	I → C	Linking Ideas with Cosmos. Interventions in the material world. Using scientific ideas, construct a piece of Cosmos with specific characteristics.
Representational	C → I	Linking Cosmos with Ideas. Describing a piece of Cosmos on the basis of one’s own Ideas.

E—Evidence/I—Ideas/C—Cosmos.

). The activities are of two types: representational (R) and interventional (I). The representational activities have as a basic feature the use of language to represent Ideas, Evidence, and the effect of Cosmos on them. The interventional activities guide students to take action and modify a part of the Cosmos concept, based on ideas or evidence.

The model has been used in previous studies as a TLS assessment tool. Kallery et al. [54] analyzed real activities in preschool classes in terms of CEI connections and found that the interventional ones (E → C and I → C) were missing. Psillos et al. [52] developed a TLS in the thematic area of fluids in physics, successive versions of which followed the trends in science teaching through the years. These successive versions of the TLS were analyzed a posteriori according to the CEI model, and it was found that after each modification aimed at improving it, the number of connections promoted by the model increased. In addition, the variety of connections rose along with the overall number of connections. Intervening-type connections, which were missing from the activities’ initial version, were added in each TLS improvement. Therefore, since the effectiveness of activities increases as they adapt to the features of the model, the use of the model as a planning tool seems promising. This finding prompted the present research, which uses the model in advance, in the design phase, in order to achieve improvement in the effectiveness of the activities before they are implemented for the first time. As a result, the requirement for numerous subsequent changes to the same TLS could be diminished. There is not any related research in the literature testing whether using the model as a design tool could increase a TLS’s effectiveness. The study aims to determine whether a TLS for fundamental ecological concepts with all the connections promoted by the model could weaken students’ alternative ideas more and improve learning outcomes more as compared to an equivalent TLS in which only a portion of the connections promoted by the model are recorded. If it is found that the revised TLS has better learning outcomes, the model could be proposed as a tool for describing and modeling the teaching activities of a TLS, with the aim of identifying and improving their weak points.

2. Materials and Methods

2.1. Participants

The research was carried out in a junior high school in northern Greece, where the first author taught in the school year 2021–2022. The study groups comprised the students of two classes of the third grade. Each class was a separate group. Each group comprised 18 students. Thus, the total number of participants was 36. The participants were chosen by chance since they were the only third grade students that the main researcher taught at that period of time.

2.2. The TLSs

Initially, a TLS of five teaching scenarios for basic concepts of ecology (TLS 1) was designed. Design principles were based on the outcomes of previous empirical studies in the field [45,46,47]. The theoretical framework that guided the design was the Educational Reconstruction model [50] (Figure 1). Each teaching scenario focused on the teaching of specific ecological concepts; however, for reasons of coherence and better consolidation, the concepts were not strictly distributed among the scenarios but dispersed between them. The main ecological concepts approached were ecosystem structure and characteristics; energy input to ecosystems and photosynthesis; and energy flow, ecological succession, and biodiversity. Each teaching scenario was designed with the purpose of contributing to the management of specific alternative ideas of the students regarding concepts of ecology, with the aim of moving the latter closer to scientific positions. The duration of each scenario was 45 min. A brief description of the scenarios is presented in Table 2. TLS 1 was retrospectively analyzed based on the CEI model. In other words, the CEI model connections that were present in the original TLS were recorded, but those that were absent were also highlighted. It was then modified by adding selected activities in order to include in each scenario all possible connections between the individual entities promoted by the model. For example, in the first teaching scenario on ecosystem structure and function, the main activity of which was to describe an ecosystem model (Figure 3) in observational terms (e.g., bottle, water, grass/C → E connection), the students participating in the revised TLS 2 were additionally asked to describe the model with theoretical terms (e.g., biotic and abiotic factors, bio community/C → I), construct concept maps using the main concepts mentioned in the description (I → C), and comment on their classmates’ concept maps (E → C). With similar modifications to all scenarios, TLS 2 emerged. After completing this process, TLS 1 and TLS 2 were implemented in the two different groups (comparison and implementation, respectively) that participated in the research, and the learning outcomes, in relation to the teaching goals, were compared to each other to see which students benefited more cognitively. The tested hypothesis was that if the students who attended TLS 2 benefitted more, then the CEI model, when used as a design tool, increased the effectiveness of the designed TLS. In order to answer this question, specific axes were defined on which the evaluation of learning progress and therefore the comparison of the two groups was based. These axes are, briefly, the following:

• Improving overall understanding;
• The enhancement of knowledge regarding particular topics around which the instruction was centered;
• The weakening of their alternative ideas;
• Their ability to incorporate new knowledge into their reasoning.

Table 2. Matching learning scenarios–activities–connections.

Teaching Scenarios	Main Activities	Connection
		TLS 1	TLS 2
1 Structure—ecosystem characteristics	An ecosystem model is presented to the students, and they are asked to perform several activities in relation to it (Figure 4), e.g., description with observational or theoretical terms, construction of concept maps, crossword completion, etc.	C → E I ↔ E	All
2 Plant growth—nutrition	The historical experiments of Helmont and Priestley are presented step by step. Students predict experiments’ progress—draw conclusions. They build chemical molecules using digital simulation.	C → E I ↔ E	All
3 Energy flow—food relationships	They study food chains, webs, and pyramids; make calculations; and come to conclusions. They role-play and simulate a food web.	C ↔ E I ↔ C	All
4 Ecological succession	They study pictures and videos of the successive forms of an ecosystem and focus on the differences. They place in chronological order corresponding images of an ecosystem and recognize the stages of development of another in a case study text.	I ↔ E C → I	All
5 Biodiversity	They focus on the biodiversity that surrounds them. They record how many different organisms contribute ingredients to make a specific food, how many human activities rely on interaction with other organisms, etc. They undertake to present, in a way of their choice, the reasons why we should protect biodiversity.	C ↔ E I → E C → I	All

Italics describe activities added to TLS 2, and “All” means all the kinds of possible connections (C → E, E → C, I → E, E → I, C → I, and I → C). I stand for Ideas, C for Cosmos and E for Evidence. One direction (→) and bidirectional links (↔) are described.

Table 2. Matching learning scenarios–activities–connections.

Teaching Scenarios	Main Activities	Connection
		TLS 1	TLS 2
1 Structure—ecosystem characteristics	An ecosystem model is presented to the students, and they are asked to perform several activities in relation to it (Figure 4), e.g., description with observational or theoretical terms, construction of concept maps, crossword completion, etc.	C → E I ↔ E	All
2 Plant growth—nutrition	The historical experiments of Helmont and Priestley are presented step by step. Students predict experiments’ progress—draw conclusions. They build chemical molecules using digital simulation.	C → E I ↔ E	All
3 Energy flow—food relationships	They study food chains, webs, and pyramids; make calculations; and come to conclusions. They role-play and simulate a food web.	C ↔ E I ↔ C	All
4 Ecological succession	They study pictures and videos of the successive forms of an ecosystem and focus on the differences. They place in chronological order corresponding images of an ecosystem and recognize the stages of development of another in a case study text.	I ↔ E C → I	All
5 Biodiversity	They focus on the biodiversity that surrounds them. They record how many different organisms contribute ingredients to make a specific food, how many human activities rely on interaction with other organisms, etc. They undertake to present, in a way of their choice, the reasons why we should protect biodiversity.	C ↔ E I → E C → I	All

Italics describe activities added to TLS 2, and “All” means all the kinds of possible connections (C → E, E → C, I → E, E → I, C → I, and I → C). I stand for Ideas, C for Cosmos and E for Evidence. One direction (→) and bidirectional links (↔) are described.

2.3. The Evaluation of the TLSs

The implementation of the above (same in terms of conceptual content) TLSs in the two different groups went hand in hand with the evaluation of learning outcomes. The main tool from which data were obtained for the assessment was a questionnaire completed by students before and after instruction. Since a questionnaire containing only the set of concepts and principles included in the research was not available from previous studies, an idea-detection questionnaire was created specifically for the research. The questionnaire comprised of eight multiple-choice questions (MCQs) and two open-ended questions (see Appendix A for examples). Two different types of questions were chosen because each type gives a different kind of information on the development of students’ understanding. Thus, the inclusion of both types was deemed necessary in order to form a more complete picture.

The entire questionnaire was tested prior to the implementation of the TLSs in a group of sixteen students of the same grade irrelevant to the study participants. It was also given to a group of experts in Didactics of Biology (a university professor, a Ph.D. holder, and four Ph.D. candidates) to evaluate its validity. Following feedback from the evaluation of the two groups, it was modified in order to improve any ambiguities, misinterpretations, or difficulties in its use. Then, it was piloted again with a group of 68 students, and Gronbach’s a was calculated for the multiple-choice questions so as to evaluate its reliability. Its value was 0.712, which is considered acceptable according to [55]. Since the target group and the pilot group had similar features (e.g., same region, age, and school type), the questionnaire was then used in that form without any further alterations.

Multiple-choice tests are arguably a key part of educational testing, ranging from student assessment in today’s classrooms and schools to large-scale, high-stakes certification and licensing tests in the professions [56]. Multiple-choice exams are widely used in education because they have numerous benefits over alternative evaluation methods. Multiple-choice exams, for instance, are quite simple to administer, provide more objective scoring, cover more material, and cut down on the amount of time it takes for test takers to respond to the questions. Additionally, because multiple-choice questions are simpler, students typically like them over other types of assessments [57]. Despite the use of multiple-choice tests at every level of education, developing high-quality multiple-choice items remains a challenge for educators and test designers [56]. Research [58] shows that while multiple-choice questions can enhance the reliability and validity of assessments, there often remain significant flaws in their design that can undermine their effectiveness. The first and most common recommendation regarding the selection of alternative responses focused on identifying common misconceptions related to thinking, reasoning, and problem solving [59,60,61]. Other best practices include avoiding common pitfalls, such as the use of “none of the above” and “all of the above” options, as these can complicate the assessment and reduce its discretion. It is also recommended that complex question types or response procedures are avoided. Furthermore, each item in the test should require the activation of a specific cognitive process and have three possible answers available, and efforts should be made to make the test attractive but not too difficult. Overall, when carefully designed, multiple-choice tests can be a useful tool for assessing students’ understanding [58]. In designing the questionnaire for this study, efforts were made to follow as many of the above recommendations as possible.

The eight MC questions, in pairs, shared a common theme: two dealt with energy flow and the ways we represent it, two with plant growth and nutrition, two with population interactions, and two with ecological succession. In all of these questions, widespread alternative ideas were given as possible answers as they emerged from the literature research. Answers were coded as true or false in order to compare the performance of the two groups. Each correct answer was scored one point. The total score for the MC questions was obtained by summarizing the scores on all questions (maximum score 8 points).

At a second level of analysis, MC questions helped to determine the qualitative differences in the cognitive development of the two groups. The recording of performance improvement per question gave an insight into how students’ understanding of specific concepts, the clarification of which was included in the teaching objectives, developed. In addition, multiple-choice questions enabled the capture of students’ main alternative ideas. This analysis was based on the fact that the possible answers to these questions were widespread student misconceptions drawn from the literature. Misconceptions that were selected by three or more students before instruction were thought strong. The abandonment of these ideas by three or more students after teaching was seen as a sign of their weakening.

The two open-ended questions included in the questionnaire were used as a means for the students to spontaneously record their ideas, knowledge, and opinions about the environmental issues they were asked about. The purpose of the open-ended questions was to find out whether students understood why biodiversity loss is a concern and whether they had any recommendations for how to stop it (see Appendix A). Student responses to the open-ended questions were analyzed through bottom-up content analysis so as to draw qualitative conclusions about their cognitive development. Particularly, in order to evaluate students’ ability to integrate the new knowledge, their answers to the open-ended questions were grouped based on their common characteristics. Then, the number of categories and the distribution of responses in the individual categories before and after the teaching were compared for each group. Furthermore, the responses were quantified in order to allow comparisons between the two groups.

For the first question, each answer was analyzed qualitatively (with one sentence as the unit of analysis) and assigned a score using a rating scale presented in Figure 5. This scale emerged from the study of the students’ answers as an attempt to group common elements of the answers given. If a response included more than one phrase, their scores were summed to obtain the response’s aggregate score. In the second open-ended question of the questionnaire, students were asked to suggest three ways in which we could reverse the situation described in the previous question. Their responses were scored one point for each valid proposal on how to stop biodiversity loss.

All replies were scored in the same way from two different researchers: the researcher who taught the TLSs and a university professor in Didactics of Biology. Then, a comparison of the scores was performed, and in case of differences, a discussion followed so as to achieve consensus. This procedure is thought to ensure the validity of the open-ended questions’ analysis. The score for open-ended questions was the sum of the scores of the two questions of that type. The total score of the questionnaire was calculated by adding the scores of both question types.

Concerning the evaluation of the TLSs as a whole, a comparison of the two groups’ initial performance was performed in order to find out whether the two groups included students with different cognitive starting points. Next, each group’s performance was evaluated separately by comparing the scores to the questionnaire pre- and post instruction. To determine whether the students of one of the two groups benefited the most from the teaching, a new variable, i.e., the performance improvement, was calculated, deducting the final score from the initial. The comparisons mentioned were carried out both for the overall score of the questionnaire and for the scores of the two question types: the MC and open-ended.

The quantitative comparisons described above were carried out using independent-samples and paired-samples t-tests, while in the case of non-parametric tests, Mann–Whitney U and Wilcoxon signed-rank tests were used, respectively. The tests of the normal distribution of the samples were carried out using Shapiro–Wilk’s test. The calculation of the size of statistically significant differences was conducted with Cohen’s d and r coefficients. The level of significance for conducting the tests was set at 0.05 (95%). All the above calculations were carried out using IBM SPSS Statistics V.29.

3. Results

3.1. Comparison of TLS 1 and TLS 2 Groups Before the Instruction

First, the comparison of the two groups’ initial performance was carried out so as to find out whether the students had the same cognitive starting point.

To compare the total scores of the students of the two groups, the independent-sample t-test was applied since the distribution of values was normal for both TLS 1 and TLS 2 (

Table 3. Means, standard deviations, and Shapiro–Wilk’s test p-values for TLS 1 group.

TLS 1	Pre			Post			Change
	Mean	SD	Shapiro–Wilk’ s	Mean	SD	Shapiro–Wilk’ s	Mean	SD	Shapiro–Wilk’ s
Total sum	6.65	3.408	0.371	7.29	3.177	0.784	0.65	3.673	0.649
Sum 1–8	3.29	1.572	0.703	4.29	2.054	0.155	1	1.768	0.432
Sum 9	3.35	2.644	0.126	3	2.424	0.024	−0.35	2.999	0.022

and

Table 4. Means, standard deviations, and Shapiro–Wilk’s test p-values for TLS 2 group.

TLS 2	Pre			Post			Change
	Mean	SD	Shapiro–Wilk’ s	Mean	SD	Shapiro–Wilk’ s	Mean	SD	Shapiro–Wilk’ s
Total sum	7.78	3.282	0.437	9.94	2.388	0.464	2.17	2.149	0.576
Sum 1–8	3.61	2.062	0.717	5.72	1.406	0.139	2.11	1.641	0.230
Sum 9	4.17	2.282	0.148	4.22	1.833	0.600	0.06	1.474	0.426

). The average performance of TLS 1 compared to TLS 2 (Table 3 and Table 4) did not differ statistically significantly (difference mean = −1.333, 95% CI [−3.603, 0.936], t (34) = −1.194, p = 0.241).

In order to investigate whether there was a difference in the performance scores in the multiple-choice questions between the two groups before the implementation of the TLS, an independent-samples t-test was applied, as the distributions of the two groups were normal (Table 3 and Table 4). The performance of the TLS 2 group in the multiple-choice questions was better (Table 4) in relation to the TLS 1 group (Table 3), but the difference was not statistically significant (mean difference = −0.333, 95% CI [−1.562, 0.896], t (34) = −0.551, p = 0.585).

Furthermore, it was investigated whether there was a also difference in the mean scores for the open-ended questions between the two groups before the implementation of the TLS. Examples of the analysis of the open-ended questions are presented in Appendix A. The distributions of both groups were normal (Table 3 and Table 4), so a parametric independent-sample t-test was implemented. The performance of the TLS 1 group on the open-ended questions (Table 3) relative to the TLS 2 group (Table 4) had no statistically significant difference (mean difference = −1, 95% CI [−2.688, 0.688], t (34) = −1.204, p = 0.237).

According to the above results, the level of prior knowledge that the students had on the topics examined with the questionnaire did not differ significantly between the two groups.

3.2. Cognitive Development of TLS 1 and TLS 2 Groups

Next, the evolution of each group was investigated separately.

3.2.1. For the TLS 1 Group

The performance of the students in the entire questionnaire followed normal distribution, so a paired-samples t-test was carried out for the comparison. The mean score increased after instruction compared to that before (Table 3), but the performance improvement observed (mean = 0.65) was not statistically significant (95% CI [−2.536, 1.242], t (16) = −0.726, p = 0.478).

Regarding the multiple-choice questions, the majority of students increased their total number of correct answers, while there was also a small number of participants who decreased it (

Table 5. Students’ changes in the correct answers to the MCQs.

	TLS 1		TLS 2
	Frequency	Percent	Frequency	Percent
Stable	2	11.1	2	11.1
Decrease	5	27.8	0	0
Increase	11	61.1	16	88.9
Total	18	100.0	18	100.0

). Overall, performance on the MC questions was better after the TLS was implemented than before (Table 3). Paired-samples t-test was used in order to determine if that increase was statistically significant, as distributions were normal (Table 3). The increase in mean (mean dif. = 1) was statistically significant and moderate (95% CI [−1.773, −0.004], t (17) = −2.120, p = 0.049, d = 0.5).

For open-ended questions, of the 18 students in the TLS 1 group, 5 had better performance after the implementation of the TLS, 5 showed worse performance, and 8 remained stable (

Table 6. Students’ changes in open-ended questions score.

	TLS 1		TLS 2
	Frequency	Percent	Frequency	Percent
Stable	8	44.4	6	33.3
Decrease	5	27.8	5	27.8
Increase	5	27.8	7	38.9
Total	18	100	18	100

). A comparison was made of the mean scores before and after the implementation of the TLS. The difference of the means did not follow a normal distribution (Table 3), so a non-parametric Wilcoxon signed-rank test was chosen for the comparison. Τhere was no statistically significant increase in the mean rank after the implementation of the TLS (mean rank = 6.5) compared to before (mean rank = 3.8) (z = −0.417, p = 0.677).

So, it is clear from the results that TLS 1 students improved statistically significantly but moderately in their performance in the multiple-choice questions of the questionnaire. No corresponding increase was observed for open-ended questions.

3.2.2. For the TLS 2 Group

The performance of TLS 2 students in the total questionnaire questions was compared with paired-samples t-test because the distribution of mean scores was normal (Table 4). It was found that the increase in performance after the application of the TLS compared to before (Table 4) was statistically significant and very large (95% CI, [−3.235, −1.098], t (17) = −4.278, p < 0.001, d = 1.008).

In the TLS 2 group, the large majority of students increased the number of correct answers (Table 5) on multiple-choice questions. Distribution of the difference in means was normal (Table 4), so a paired-samples t-test was performed to see if there was a statistically significant difference between the mean score before the implementation of the TLS and after for the TLS 2 group in terms of multiple-choice questions. The mean score in the MC questions of the TLS 2 group students increased after the teaching scenario compared to its value before (Table 4), and this increase was statistically significant and very large (95% CI [−2.927, −1.295], t (17) = −5.458, p < 0.001, d = 1.286).

For open-ended questions, a small number of students increased their score (Table 6). The mean score of the students after the teaching scenario slightly increased (mean difference = 0.06), but this increase was not found to be statistically significant (95% CI [−0.789, 0.678], t (17) = −0.160, p = 0.875) using paired-samples t-test because the distribution of the difference in means was normal (Table 4).

In summary, TLS 2 students had a statistically significant, large improvement in their overall performance in the questionnaire after teaching. The same was observed for the improvement in MC questions after the implementation of the TLS.

3.3. Comparison of Improvement Between the Groups

Suitable case-by-case statistical tests were applied to compare the performance improvement of the two groups.

Independent-sample t-test was applied to compare the improvement in the total scores of the questionnaires since the condition of the normal distribution was met (Table 3 and Table 4). The mean difference found in favor of TLS 2 (mean = 1.520) was not proven to be statistically significant (95% CI [−3.628, 0.589] t (25.5) = −1.483, p = 0.142).

The same test was performed to compare the improvement in the MC questions group because the condition of the normal distributions was met (Table 3 and Table 4). A difference in means was observed in TLS 1 and TLS 2 groups (mean = 1.222), which was statistically significant and moderate (95% CI [−2.381, −0.063], t (34) = −2.143, p = 0.039, d = 0.714).

The performance improvement in open-ended questions was compared with the non-parametric Mann–Whitney U test because the normal distribution requirement for the TLS1 group was not met (Table 3). The difference of the two groups was not statistically significant (U = 144, z = −0.309, p = 0.782).

Thus, it appears that TLS 2 students had statistically significantly greater improvement in their performance than TLS 1 in terms of the topics covered by the MC questions in the questionnaire.

3.4. Concerning Comprehending Each of the Distinct Ideas

In addition to the quantitative, qualitative differences were also identified in the responses to the MCs in the two groups. The coding as “true or false” of students’ responses was used to track their cognitive development in relation to specific ideas included in the instruction. Figure 6 provides an overview of the concepts whose understanding seemed to be enhanced for students in each group since it depicts the change in the number of correct answers per question. It appears that the students in the TLS 1 group increased the correct answers to the questions concerning the energy flow and the evolution of an ecosystem, while the TLS 2 students chose the correct answer more frequently for all questions after instruction. The largest increase was observed for questions concerning the interactions of organisms and populations as well as the flow of energy.

3.5. Concerning Alternative Ideas

It was also of interest to determine the number of alternative ideas that seem to be weakened in each group after the instruction. This result was inferred directly from the students’ choices in the MC questions. Alternative ideas that, at the beginning, were chosen by three or more students but, after the teaching, were abandoned by at least three after the teaching include the following:

3.5.1. For the TLS 1 Group

• Plants get their food from the soil through their roots [37];
• No energy is lost in transporting the food [21];
• The top of a food chain has the most energy because it accumulates along the chain [29].

3.5.2. For the TLS 2 Group

• A change in an organism’s population will only affect those it is directly linked to through a food chain [23,25,26];
• A change in the population of one organism will affect all other organisms to the same extent [26];
• Populations exist in states of either steady growth or decline depending on their position in a food chain [23];
• Carbon dioxide is a source of energy for plants [62];
• The succession involves distinct stages that ultimately lead to a deterministic climax [62];
• The top of a food chain has the most energy because it accumulates along the chain [29].

3.6. Concerning the Integration of New Knowledge

The information on whether the students managed to integrate the new knowledge into their reasoning was derived from the analysis of the open-ended questions (see Appendix A for examples). In the first open-ended question of the questionnaire—which concerned the reasons why the loss of biodiversity is a problem—there was a remarkable homogeneity in the responses of the students. Almost all of them looked for interpretation in food relationships and made reference to food chains. Some explained the consequences of disturbing the balance, such as the lack of food, the increase in some populations or the decrease in others, and the loss of products or professions. No significant change was observed in student responses before and after instruction in any group (see Appendix A for examples).

The second open-ended question—which looked for ways to tackle the problem of declining biodiversity—saw a wide range of responses across both groups (see Appendix A for examples).

Table 7. Grouping of student responses to the 2nd open-ended question—TLS 1.

Answer Groups	Answer Category	Number of Students–TLS 1
		Pre		Post
Limitation of activities	Limiting the killing of animals	13	34	12	22
	Pollution control	7		3
	Reducing fires	4		2
	Not destroying their homes	8		3
	Changing consumption and food habits	2		1
	Restricting logging	0		1
Taking action	Human care (e.g., feeding them, making them houses, parks, and breeding them)	5	8	6	7
	Informing people about the value of biodiversity	0		1
	Tree planting	1		0
	Strict laws	1		0
	Recycling	1		0

and

Table 8. Grouping of student responses to the 2nd open-ended question—TLS 2.

Answer Groups	Answer Category	Number of Students—TLS 2
		Pre		Post
Limitation of activities	Restriction of hunting—fishing	12	33	10	19
	Pollution control	9		4
	Restriction of logging	3		0
	Reducing fires	2		3
	Reducing product consumption	3		2
	Limiting energy consumption	4		0
Taking action	Characterization of protected areas	0	12	4	9
	Creating places of care (e.g., national parks)	6		2
	Use of renewable energy sources	1		0
	Recycling	0		1
	Information on the value of biodiversity (seminars or media)	3		0
	Tree planting	0		2
	Strict environmental protection laws	2		0

present the results of grouping the responses.

It is clear that the students were unable to integrate the new knowledge to explain why biodiversity loss is a problem and to formulate valid proposals to limit it.

4. Discussion

Summarizing the above results, it seems that the students in the two groups had a similar cognitive starting point for the subjects under consideration. Students from the TLS 2 group improved their performance on the total score of the questionnaire in a statistically significant way, and their improvement on MC questions scores was also very statistically significant. On the contrary, TLS 1 group’s improvement was statistically significant and moderate only for the scores concerned to the MC questions. The comparison of the improvement of the two groups on the MC questions showed that the greater improvement in the performance of TLS 2 students was statistically significant and moderate. In other words, students in this group improved their performance more after teaching compared to students in TLS 1. Moreover, the qualitative differences observed in the two groups’ responses on the MC questions are in line with the inference retrieved from the statistical analysis. The number of alternative ideas that seem to be weakened through instruction in TLS 2 group was greater than those of the TLS 1 group.

Regarding the scores of the students in open-ended questions, no statistically significant difference was found in their performance before and after in any of the two groups. Additionally, no discernible qualitative differences were identified in their responses. This indicates that when it came to debating the environmental issues that were posed to them, both student groups were unable to incorporate the new knowledge. However, the nature of this particular type of question has several limitations that may affect the final result. Open-ended questions may lead to irrelevant or unnecessary information and may be too open-ended for the respondent so that they may not know where to focus their answer. They may also require much more time for the respondents to formulate their answer (which may therefore lead to refusal) or create the feeling that the questionnaire is long and daunting [63]. Furthermore, the results might have been affected by the number of participants in addition to the time requirements of open-ended questions. More diverse comments from a bigger group of students would have been more informative in terms of their learning trajectory. Additionally, group student interviews would be a useful tool to track students’ reasoning pathways and clarify their responses in a way in which the questionnaire was unable. In order to minimize the limitations of the research, these suggestions should be considered for future implementation.

An alternative interpretation of this result, namely the failure of students to improve their performance on open-ended questions, may emerge if the instructional goals and question objectives are compared and viewed through the prism of Bloom’s revised taxonomy as proposed by Krathwohl [64]. The use of open-ended questions is based on the initial assumption that respondents are sufficiently and equally able to articulate their thinking and capture it on paper. However, it is doubtful whether students have had sufficient practice to develop such skills. Biology lessons in particular are often criticized for giving unequal focus on goals from the lower levels of Bloom’s taxonomy hierarchical system, such as recall of facts and concepts that can hardly be memorized, compared to other types of skills [65]. In the present research, the open-ended questions included in the questionnaire traced the achievement of objectives related to the category of creation. This category is the highest hierarchically of the proposed categories, and the achievement of the goals it includes requires the consolidation of the goals of the previous categories. However, teaching was not strongly oriented towards the achievement of objectives in this category. Students were asked to use the scientific concepts taught in order to provide interpretations and propose solutions to environmental problems that were not extensively discussed during instruction. To answer correctly, they would have had to recall many of the scientific concepts on which the teaching was focused, which they would have understood and would have mastered with the ability to apply, analyze, and evaluate based on them. This objective was probably quite ambitious for a TLS of only five teaching hours. Moreover, in order for the students to reach this point, they should have been given appropriate opportunities to practice during the course of the teaching. However, in this case, this task was given to them as an assessment tool without any prior familiarity with it. In a subsequent implementation of the TLS, therefore, it is suggested that its content be enriched with activities that engage students in such cognitive processes. To summarize the results, the comparison of the improvement of the two groups showed that the better performance of the TLS 2 group was statistically significant and moderate only as far as the MC questions are concerned. In conclusion, the results are encouraging, as it seems that students who participated in the revised TLS 2 benefited more from instruction compared to those of the TLS 1 group in terms of the expected learning outcomes. Thus, it seems that the initial hypothesis is confirmed: the CEI model has the potential to increase the effectiveness of a TLS, which is based on the dominant trends of science teaching [49] but also on the proposals of empirical research in the field that bridge the theory with the teaching practice [45,46,47]. This claim is in agreement with the research results of Psillos et al. [52]. These researchers retrospectively analyzed the different versions of a TLS in the subject area of fluids in physics, which emerged after successive implementation of a formative assessment that incorporated the current propositions of the evolving teaching of natural sciences (discovery, constructionism, and inquiry). They found that as the number of connections between model entities increased students’ understanding of the taught concepts improved. Moreover, this research inspired the present study, as it led to the idea of using the CEI model as a tool for planning TLS in advance in order to predict and improve possible deficiencies in the design of teaching activities. The objective is to reduce the number of consecutive cycles of implementation of a TLS required for formative evaluation [49].

In terms of how the model can enhance the learning outcomes of a TLS, we need to consider what exactly its use adds to a teaching scenario. With a critical look at the activities included in the revised TLS, one can observe that each new connection added guides students to reinforce one more science skill. For example, the construction of conceptual maps, which was carried out in the first scenario by students participating in the revised TLS, in order to connect Ideas to Cosmos, prompted them to reflect on the relationships between the individual components of the ecosystem to co-contextualize that it is a set of interactions (systems thinking). It is also important to note that the explicit distinction the model makes between the representational and interventionist nature of science enhances epistemic awareness and promotes deeper understanding by engaging students in interactions with the world of interventions and the world of representations.

Beyond the enhancement of the learning objectives of a TLS, however, there are important reasons why it could be argued that the ICE model could be used as a tool to promote sustainability objectives. While there is a variety of empirical studies [35,66,67,68,69,70,71,72] that attempt to isolate which component of a teaching practice is essential to increasing its effectiveness, the ICE model takes a completely different approach: it treats each teaching activity as a set of interactions as well as each one individually, all of which contribute equally to increasing the effectiveness of any teaching effort. Its use as a planning tool, which is proposed here, also gives it the ability to predict the weaknesses of a teaching intervention and to allow for normative interventions. This approach is in alignment with some of the key competences deemed necessary for sustainable education, such as systems thinking, anticipatory competence, integration competence, and normative competence [73]. While systems thinking is becoming increasingly central to scientific and ecological literacy [42], it is difficult for students to understand, and it is extremely difficult for them to reason about the complex interactions involved in an ecosystem [74]. In a review of research on skills related to promoting sustainability, Redman and Wiek [73] found that scientists, including those involved in sustainable development, continue to break down holistic processes (e.g., problem solving) into their component parts. The integration competence opposes this tendency and urges an emphasis on training for the connections between competences. Integration is closely linked to the development of systems thinking, a key competence for sustainability but also a characteristic of scientific thinking. The skills that the ICE model seeks to reinforce are therefore already identified as necessary to promote sustainability, so its use as a design tool raises expectations for enhancing sustainability goals related to qualitative education (4th STG) and climate action (13th STG), where, as in this research, it is used to teach key concepts of ecology. Finally, it is important to mention the main limitations of the present research. At first, the small number of participants made it difficult to generalize the results. Secondly, the time available for the implementation of the teaching scenarios was limited since in the Greek junior high school, biology is taught only for one hour per week. This resulted in practical difficulties, the overcoming of which requires a constant redesign of teaching. A proposal to overcome this obstacle is the implementation of the TLSs within the framework of an educational learning group, whose operation exceeds the school timetable. However, because participation in learning groups is optional, the number of participants may decrease.

In conclusion, it could be said that the use of the CEI model as a design tool for a TLS for basic concepts of ecology might have the potential to enhance its effectiveness and, as a consequence, to foster the goals of education for sustainable development. However, the findings from related studies should support this conclusion in order for it to be generalizable. It is worth noting that, so far, no other research has been carried out in which the model was used as a design tool. Studies involving activity analysis based on the model [52,54] have used it for their post-implementation analysis. In fact, only the research of Psillos et al. [52] connected the model with the development of students’ understanding. The potential of the model as a planning tool therefore needs to be explored more widely. The results of future research will strengthen or weaken the present conclusions.