Self-Efficacy Beliefs of Teaching ESD: A Theory-Driven Instrument and the Effectiveness of ESD in German Teacher Education

Abstract

Self-efficacy beliefs are important for teachers to conduct Education for Sustainable Development (ESD). Self-efficacy instruments for teaching ESD mainly focus on primary education, and theory-driven instruments are lacking. Thus, we developed an instrument for secondary education reflecting eight subcategories of a science teaching pedagogical content knowledge (PCK) model, transferred to teaching ESD. We conducted an exploratory factor analysis (n₁ = 162) and a confirmatory factor analysis (n₂ = 236) with mainly pre-service biology, politics, and geography teachers for secondary education. We identified seven self-efficacy beliefs of teaching ESD factors. Six of these factors showed an expected low or middle correlation with self-rated content knowledge of SD-relevant issues (divergent validity). Pre-service teachers’ ESD teaching self-efficacy beliefs were rather positive, missing clear differences due to studying biology, politics, and geography. Bachelor students had higher self-efficacy beliefs than Master of Education students regarding four factors. Only 32 of 113 Bachelor students completed any ESD course at that time. They could have overestimated their own skills or underestimated teaching ESD. Current participation in at least one ESD course positively influenced three self-efficacy factors. Since barely any pre-service teacher currently participated in or completed more than three ESD courses at the time of the study, there is room for a stronger focus on ESD in German teacher education.

1. Introduction

Education for Sustainable Development (ESD) equips learners with skills, knowledge, attitudes, and values to participate in Sustainable Development (SD) [1]. ESD aims to create awareness for SD, convey values associated with sustainability, and promote sustainable behavior [2]. SD-focused education might create informed citizens characterized by critically examining SD issues and the associated complexity, uncertainty, and contradictions [2]. Such individuals could become able to bring the necessary transformation towards a sustainable future by promoting political, social, and economic change [3].

Former approaches like the United Nations (UN) Decade of ESD (2005–2014) and the Global Action Programme on ESD (GAP) sought to foster the implementation of ESD internationally in all areas and levels of education [4]. Goal 4 of the Sustainable Development Goals (SDGs) highlights the pivotal role of education in terms of SD [5]. Education is not just a goal itself but also an important driver to achieving all the other SDGs [6]. Based on the 2030 Agenda for Sustainable Development, the current UNESCO program “ESD for 2030” (2020–2030) “focuses on strengthening ESD’s contribution to the achievement of all 17 SDGs, focusing on policies, learning environments, teachers and educators, youth as well as communities” [7] (p. 1). This is even more important since the Sustainable Development Report of 2021 shows a global decline in SD due to the COVID-19 pandemic [8].

Educators can act as change agents by providing sustainability competencies and concepts to their students [4]. Hence, for an adequate implementation of ESD and the promotion of ESD-specific competencies, teachers are of central relevance [7]. Successful implementation of ESD particularly depends on the skills and commitment of teachers toward sustainability [9]. However, according to Evans et al. [10], in most countries, ESD is not an integral part of initial teacher training. Courses in SD in pre-service teacher training can modify students’ beliefs, attitudes, and norms [11]. Referring to Moseley et al. [12,13], Malandrakis et al. [14] conclude that self-efficacy beliefs are one of the most powerful factors in influencing the ability of teachers to initiate ESD. Thus, self-efficacy beliefs in teaching ESD, its identification as learning prerequisites by empirical studies, and its promotion in teacher training are important for implementing ESD.

1.1. Self-Efficacy Beliefs

Self-efficacy beliefs are “beliefs in one’s capabilities to organize and execute the courses of action required to produce given attainments” [15] (p. 3). They are a part of motivational orientations as an aspect of teachers’ professional competence [16]. Furthermore, self-efficacy beliefs are a “key factor of human agency” [15] (p. 3). Teachers’ self-efficacy beliefs are positively related to their job satisfaction and negatively related to emotional exhaustion and depersonalization—two symptoms of burnout (e.g., [17]).

Another important aspect in the teaching context is the predictive effect/the positive relation of teachers’ self-efficacy beliefs on/to their students’ achievement (e.g., in a general context: [18,19,20], in the context of science education: [21]). Self-efficacy beliefs vary regarding different characteristics such as task levels or the generality of the beliefs [15]. Thus, self-efficacy beliefs must be considered in different contexts [22]. Self-efficacy beliefs focus on new or difficult tasks to execute [15,23,24]. One difficult task in the context of teaching is teaching interdisciplinarily, such as required in ESD teaching. Such teaching is challenging because ESD is still not systematically integrated into German teacher education (e.g., [25]).

Further characteristics of self-efficacy beliefs have to be kept in mind when constructing self-efficacy beliefs items. First, they should contain obstacles or demanding situations [15,23]. Second, statements should contain (I) can and ask for present, not possible future skills [15]. Third, self-efficacy beliefs are not a global construct, e.g., referring to teaching as a whole (I can teach ESD.), but a multidimensional construct that needs differentiated consideration [15].

1.2. Requirements for Teaching ESD

One important aspect of teachers’ professional competence is pedagogical content knowledge (PCK). Better teachers’ PCK results in more effective teaching (e.g., [26,27]). Thus, having good PCK is a requirement for teaching ESD as well. Defining PCK has a long tradition, especially in science teaching [28]. The resulting models are developed for science teaching but can be adapted for other contexts such as ESD [29]. Considering the most recent models for PCK in science teaching (the pentagon model from Park and Chen [30], the consensus model from Gess-Newsome [31], and the refined consensus model from Carlson et al. [32]), only the pentagon model focuses on the specific components of PCK. It is the result of a longer empirical developmental process that is built on the model from Magnusson et al. [33]. The pentagon model in the context of science education comprises five categories and corresponding subcategories [30]:

• Orientation to Teaching Science:

  ◦

  Beliefs about Purposes of Learning Science, Decision Making in Teaching, Beliefs about the Nature of Science

• Knowledge of Students’ Understanding in Science:

  ◦

  Misconceptions, Learning Difficulties, Need, Motivation & Interest

• Knowledge of Science Curriculum:

  ◦

  Curriculum Materials, Vertical Curriculum, Horizontal Curriculum

• Knowledge of Assessment of Science Learning:

  ◦

  Dimensions of Science Learning to Assess, Methods of Assessing Science Learning

• Knowledge of Instructional Strategies for Teaching Science:

  ◦

  Topic-specific Strategies (Representations, Activities), Subject-specific Strategies

These categories and their subcategories describe components of teaching that are of importance for teaching every subject or teaching task. Handtke and Bögeholz [34] explain the content of these (sub)categories adapted for teaching science. The model has the potential to be adapted to ESD as well [29].

1.3. Previous Instruments for Self-Efficacy Beliefs of Teaching ESD

In the following, we analyzed the literature on self-efficacy beliefs instruments concerning teaching ESD in order to compile the state of the art and to identify desiderata for further research. We have included instruments published in German and English.

In general, many instruments for measuring self-efficacy efficacy beliefs regarding teaching different subjects and teaching tasks exist. One of the most influential instruments has been the Science Teaching Efficacy Belief Instrument (STEBI; [35,36]) in the versions A and B. The STEBI not only had great influence on science teaching self-efficacy beliefs instruments [34] but also on instruments of self-efficacy beliefs of teaching ESD (see

Table 1. Instruments measuring self-efficacy beliefs of teaching Education for Sustainable Development (ESD) and related contexts (e.g., environmental education). “Author(s), Instrument” presents the author(s) of the instrument, the instrument’s full name, and possible abbreviations. “Based on or modified from STEBI A/B” indicates whether the instrument is derived from the STEBI-A [35] or -B [36]. “Characteristics” indicates the sample size and the number of items, and shows different psychometric features of the instruments and their development. None of the instruments integrates an (explicit) obstacle, as recommended by Bandura [15,23].

Author(s), Instrument	Target Group	Based on or Modified from STEBI-A/B	Characteristics
Sia [37] Environmental Education Efficacy Belief Instrument (EEEBI)	Pre-service elementary teachers	x	Two Samples: n = 21 and n = 19 One factor: Personal EE (Environmental Education) Teaching Efficacy Belief Scale 13 Items Measurement: 5-point Likert scale, positive and negative items Analysis: - α = “yet to be determined” (p. 5)
Taylor [40]; Moseley & Taylor [38] Environmental and General Science Teacher Efficacy Assessment (EGSTEA)	Sample contains in-service high and middle school science teachers, but the underlying instruments focus more on primary education (EEEBI, STEBI)	x	n = 40 One factor: PTE (personal teacher efficacy) for environmental science 13 Items Measurement: 5-point Likert scale, positive and negative items Analysis: ANOVA α = 0.96
Effeney & Davis [43] according to Boon [44]	Pre-service primary teachers		n = 266 One factor: Self-efficacy for EfS (Education for Sustainability) 4 Items Measurement: 5-point Likert scale, positive and negative items Analysis: Principal Component Analysis (KMO = 0.67, Bartlett’s Test of Sphericity significant, 46% explained variance by two factors) α = 0.71
Moseley, Utley, Angle & Mwavita [39] Environmental Education Teaching Efficacy Belief Instrument (EETEBI)	Pre-service elementary teachers	x	n = 255 One factor: Personal environmental education teaching efficacy (PEETE) 13 Items Measurement: 6-point Likert scale, positive and negative items Analysis: Principal component analysis (KMO = 0.88, Bartlett’s Test of Sphericity significant, 50.5% explained variance by two factors) α = 0.92
Malandrakis, Papadopoulou, Gavrilakis & Mogias [14,45]; Mogias, Malandrakis, Papadopoulou & Gavrilakis [46] Teachers Self-Efficacy Scale for Education for Sustainable Development (TSESESD)	Pre-service primary (and early childhood) teachers (Tested with secondary in-service teachers without adaptation of the instrument [14,46])		N (2016) = 305 N (2019) = 924 (804 primary school, 120 early childhood), in addition for validation: 88 in-service primary teachers N (2021) = 267 in-service secondary teachers Four factors: (a) Values and ethics, (b) Systems thinking, (c) Emotions and feelings, (d) Actions 24 Items Measurement: 7-point Likert scale, positive items Analysis: Principal Component Analysis (KMO = 0.96, Bartlett’s Test of Sphericity significant, 72.58% explained variance by four factors) α (2016) = 0.93–0.95, α (2019) = 0.85–0.94, α (2021) = 0.89–0.96

). Three of the presented previous instruments are rather strongly derived from the STEBI-A or STEBI-B [37,38,39,40]. This origin is rather less recommended because the STEBI does not include current demands for teaching and shows weaknesses regarding its construction (e.g., missing obstacle, future tense) [34,41,42].

In summary, we can conclude a focus of previous research about self-efficacy beliefs of teaching ESD on pre-service primary teachers ([14,37,39,43,44,45]; see Table 1). Indeed, the Environmental and General Science Teacher Efficacy Assessment (EGSTEA) was tested with high and middle school science in-service teachers [38], and the Teachers Self-Efficacy Scale for Education for Sustainable Development (TSESESD) was tested with in-service secondary teachers [14,46]. However, the instruments were developed for primary education [14,38,45], and no adaptations to the new target groups are reported [14,38,46]. In addition, some instruments have rather few items (e.g., four items) and, thus, little explanatory power in terms of a differentiated investigation of self-efficacy beliefs of teaching ESD [43]. None of the provided instruments explicitly includes obstacles [14,37,38,39,40,43,45] as demanded by Bandura [15,23].

Another key feature of self-efficacy beliefs instruments is their multidimensional nature [15]. It can be considered by applying a model of teaching ESD (or adapted to ESD) as the base of the instrument development (i.e., theory-driven; cf. for science [34]). Such an approach allows for a more detailed view on the learning prerequisites of prospective teachers regarding different aspects of their self-efficacy beliefs of teaching ESD. Only one of the identified instruments integrated a model for the instrument development [45]. Malandrakis et al. [45] included four domains of ESD competencies for teachers defined by Sleurs [47]: ethics and values, emotions, action, and systems thinking.

Nearly all items of Malandrakis et al. [14] focus on pre-service teachers’ beliefs of being able to develop different abilities of students. Thus, the instrument focuses on the outcome of the teachers’ students. If at all, the instrument implicitly integrates what has to be done by the teacher regarding ESD. The subdomains rather seem to focus on teachers’ self-efficacy beliefs to develop students’ general ESD competencies—an important aspect as well—but not specifically for the teaching of ESD. However, specific instructional strategies, evaluation, or fostering competencies are not integrated into the instrument.

In sum, no instrument for measuring self-efficacy beliefs of ESD seems to exist that represents current demands, focuses on secondary education, includes obstacles in the items, and is theory-driven (regarding teaching) at the same time. The instrument of Malandrakis et al. [14,45] meets some of these requirements. However, the original focus does not seem to be on secondary education, the obstacles are missing, and the theory-driven instrument lacks some central aspects of teaching. In addition, none of the mentioned instruments contains more than 24 items or four factors, and most contain many fewer items and only one factor (see Table 1). Thus, a broad representation of the requirements of teaching ESD does not appear to be given and makes a differentiated diagnosis of learning prerequisites difficult.

1.4. Self-Efficacy Beliefs, Self-Rated Content Knowledge, and Teacher Education

Considering the relations of self-efficacy beliefs with other constructs, self-rated content knowledge plays an important role. For example, self-rated content knowledge in science (subjects) relates to self-efficacy beliefs of teaching interdisciplinary science in primary and secondary education [48,49,50]. Researchers uncovered correlations between both constructs [48,50] and differences in self-efficacy beliefs between groups of different self-rated content knowledge in science (better self-rated content knowledge = higher self-efficacy beliefs) [49].

Focusing on self-efficacy beliefs of teaching ESD, similar results emerge: For example, Malandrakis et al. [14,45] identified substantial correlations between self-efficacy beliefs of teaching ESD (primary education, pre-service) and self-rated content knowledge of ESD concepts and issues (r = 0.39–0.53, p < 0.01 [45]; r = 0.47–0.59, p < 0.01 [14]). Mogias et al. [46] found comparable results with in-service secondary school teachers (r = 0.39–0.58, p < 0.01). Effeney and Davis [43] investigated the correlation between self-efficacy beliefs of teaching ESD (primary education, pre-service) and self-rated content knowledge of sustainability issues. Comparable to Malandrakis et al. [14,45], they identified a significant correlation as well, but much smaller: r = 0.22, p < 0.01 [43]. Hansen and Sillasen [51] also found a positive relationship between pre-service science teachers’ self-efficacy beliefs of teaching ESD and perceived knowledge of sustainability issues, but much stronger (r = 0.69, p < 0.01).

Self-efficacy beliefs can be fostered by direct experiences, indirect experiences, or verbal persuasion [15]. How self-efficacy beliefs develop during teacher education is of interest [52]. For example, more direct experiences with ongoing teacher education should result in higher self-efficacy beliefs (cf. for interdisciplinary science teaching: [49] medium to large effects). In addition, ESD seems to be most integrated into German secondary schools in biology, politics, and geography [53], possibly affecting teacher education. Furthermore, the effect of single courses is of interest. For example, teacher education courses at university resulted in higher self-efficacy beliefs of teaching science (η = 0.8, [54,55]). One ESD course positively influenced pre-service teachers’ beliefs, attitudes, and norms related to SD [11], indicating possibly similar effects for ESD. Evans et al. [56] showed in a quasi-experimental design that ESD courses have a positive influence on (early childhood, primary, middle school) pre-service teachers’ self-efficacy beliefs of teaching ESD (main effect for time, partial η² = 0.42). These results complement the assumption regarding the positive effects of teacher education on self-efficacy beliefs on a more specific level.

1.5. Research Questions and Hypotheses

The analysis of previous instruments revealed that no instrument for self-efficacy beliefs of teaching ESD integrates current requirements, focuses on secondary education, explicitly considers obstacles in the items, and is based on a model regarding teaching ESD at the same time. Among these instruments, one seems to be theory-driven [14,45,46] but does not focus on the act of teaching ESD explicitly. Thus, there was a need for a new instrument developed close to the requirements of teaching ESD (PCK model [30]). It should allow the differentiated diagnosis of prospective teachers’ learning prerequisites regarding the self-efficacy beliefs of teaching ESD. One aim of the instrument development was the investigation of factorial validity. This aim resulted in the following research question:

Research Question 1.

Which self-efficacy beliefs of teaching ESD factors support the subcategories of the PCK model empirically?

In addition to examining the factorial structure of the new instrument, we focused on an often-considered construct in connection with the self-efficacy beliefs: self-rated content knowledge. Different researchers regarding teaching science (primary and secondary education) revealed a relationship between self-efficacy beliefs and self-rated content knowledge [48,49,50]. In the field of teaching ESD in primary and secondary education (pre- and in-service teachers), five studies pointed out positive correlations of different strengths between both constructs [14,43,45,46,51]. Since both constructs are different from a theoretical perspective, we expect rather low or middle positive correlations between both constructs in the context of secondary education. Thus, the following hypothesis examines the divergent validity of the interpretation of the new instruments’ test values [57]:

Hypothesis 1.

The self-efficacy beliefs of teaching ESD are positively correlated with self-rated content knowledge of SD-relevant issues (low or middle effect size).

In addition to the instrument development, the absolute values of the prospective teachers’ self-efficacy beliefs of teaching ESD are of interest. We want to know if their self-efficacy beliefs of teaching ESD are rather positive or negative on the different factors and how these self-efficacy beliefs differ according to their teacher education so far. Thus, we derive the following research question, additionally specified by the following research question and hypotheses:

Research Question 2.

What self-efficacy beliefs of teaching ESD do prospective teachers have?

First, we want to investigate the self-efficacy beliefs of teaching ESD differentiated by the subject studied. We focus on the subjects that seem to contain ESD the most in German secondary schools: biology, politics, and geography [53]. With this research question, we want to gain deeper insight into the development of the self-efficacy beliefs of teaching ESD. Thus, we derive the following research question:

Research Question 2a.

To what extent do self-efficacy beliefs of teaching ESD differ among biology, politics, and geography pre-service teachers?

To further investigate the development of self-efficacy beliefs of teaching ESD, the difference between pre-service teachers in the Bachelor and Master of Education is of interest. With ongoing teacher education, pre-service teachers should make more direct experience in teaching ESD, resulting in higher self-efficacy beliefs of teaching ESD (cf. for teaching science: [49]). This should be especially valid for pre-service teachers in biology, politics, and geography since ESD seems to be integrated the most in these subjects in German secondary schools [53]. Thus, the following hypothesis focuses on the difference between Bachelor and Master of Education:

Hypothesis 2.

Pre-service teachers in the Master of Education have higher self-efficacy beliefs of teaching ESD than those in the Bachelor.

Supplementary to Hypothesis 2, we want to know the influence of participating in at least one ESD course on ESD teaching self-efficacy beliefs. Regarding science teaching, courses in teacher education showed positive effects on the self-efficacy beliefs of science teaching [54,55]. Such effects can be expected for ESD as well. For example, ESD courses have already shown a positive effect on pre-service teachers’ self-efficacy beliefs of teaching ESD for primary, early childhood, and middle school education [56]. Transferring to ESD in secondary education and specifying Hypothesis 2 by focusing on the participation in ESD courses, we derive the following hypothesis:

Hypothesis 3.

Pre-service teachers currently participating in at least one ESD course have higher self-efficacy beliefs of teaching ESD than those not participating in any ESD course.

2. Materials and Methods

2.1. Sample

For instrument development, two samples were used (

Table 2. Description of the two samples of this study. Differences to 100% of the sample are caused by missing values.

	Sample 1 (n = 162)		Sample 2 (n = 236)
Variable	Absolute Value	Relative Value	Absolute Value	Relative Value
Sex
Female	94	58.0%	171	72.5%
Male	64	39.5%	61	25.8%
Federal State
Lower Saxony	110	67.9%	150	63.6%
Others	52	32.1%	86	36.4%
Phase of Teacher Education
Bachelor	78	48.1%	123	52.1%
Master of Education	51	31.5%	112	47.5%
State Examination	9	5.6%	-	-
Finished studies	7	4.3%	-	-
Type of School
Grammar & Comprehensive School	98	60.5%	236	100%
Others	54	33.3%	-	-
Subject
Biology	44	27.2%	131	55.5%
Politics	42	25.9%	21	8.9%
Geography	30	18.5%	60	25.4%
Biology & Politics	3	1.9%	2	0.8%
Biology & Geography	3	1.9%	21	8.9%
Politics & Geography	-	-	1	0.4%
Others	39	24.1%	-	-

). We used the first sample (n = 162) for exploratory purposes (data collection: July to August 2017). The majority of the test persons were in the Bachelor (n = 78) and the Master of Education (n = 51). Some test persons finished their studies recently (n = 7), and a few were trainee teachers for secondary education (n =8). The sample contained 90 pre-service teachers and 8 trainee teachers intending to teach in secondary education, and 54 pre-service teachers preparing to teach in other school types (e.g., vocational school or not specified at the time of the study). Ten participants were not enrolled in their teaching-specific studies at that time. In German secondary schools, ESD seems to be integrated the most in geography, politics, and biology [53]. Thus, pre-service teachers studying biology, politics, or geography were focused on in both samples. Due to the exploratory approach of sample 1, pre-service teachers studying other subjects of interest for ESD (e.g., religion, philosophy, chemistry, history, physics) were included in the sample as well (n = 39). We used sample 1 to investigate Research Questions 1 and 2.

The second sample (n = 236) had a different purpose (October 2018 to May 2019). The sample compilation followed a confirmatory approach for a specific target group (cf. [58]). Thus, only pre-service teachers in the Bachelor and Master of Education studying to teach at a grammar or comprehensive school were considered. Every participant studied at least one of the subjects biology, politics, and geography. Thus, we used sample 2 to investigate Research Questions 1, 2, and 2a as well as Hypotheses 1, 2, and 3.

We asked the test persons in the second sample to indicate how many ESD courses they currently participated in and how many ESD courses they had completed in their studies. Richter-Beuschel and Bögeholz [58] already presented the means of the number of completed ESD courses at the time of the study differentiated by Bachelor/Master and the studied subject, e.g., Bachelor: 0.76 and Master of Education: 1.5. In this paper, we present the frequencies of the completed ESD courses at the time of the study and the frequencies of the ESD courses currently participated in. The following

Table 3. Education for Sustainable Development (ESD) courses participated in and completed differentiated by the studied subject (sample 2). The number of ESD courses (e.g., 1.5) was given by the test persons and taken over like this. par = ESD courses currently participated in, com = completed ESD courses in their studies at that time.

Number of ESD Courses	Biology (n = 115/120)		Politics (n = 21)		Geography (n = 58)		Biology & Geography (n = 20)
	par	com	par	com	par	com	par	com
0	100	66	18	16	21	29	7	7
1	9	23	3	2	20	9	8	2
1.5	0	0	0	0	1	0	0	0
2	4	16	0	2	13	7	4	4
3	2	11	0	0	2	8	0	1
4–6	0	1	0	1	1	5	0	5
7–10	0	3	0	0	0	0	0	1
11–14	0	0	0	0	0	0	1	0

and

Table 4. ESD courses participated in and completed differentiated by the phase of teacher education (sample 2). The number of ESD courses (e.g., 1.5) was given by the test persons and taken over thus. par = ESD courses currently participated in, com = completed ESD courses in their studies at that time.

Number of ESD Courses	Bachelor (n = 111/113)		Master of Education (n = 106/109)
	par	com	par	com
0	76	81	72	39
1	17	10	24	26
1.5	0	0	1	0
2	14	11	7	18
3	2	4	2	17
4–6	1	5	0	7
7–10	0	2	0	2
11–14	1	0	0	0

differentiate these statements by studied subject and pre-service teachers’ phase of teacher education.

Table 3 shows that the majority of pre-service biology and politics teachers did not complete any ESD courses at that time. In addition, 100 of 115 pre-service biology teachers participated in no ESD course at the time of the study. Only a few pre-service politics teachers completed such an ESD courses or participated in one at the time of the study. Most ESD courses (as a percentage) were completed and participated in by pre-service geography teachers and pre-service geography and biology teachers. However, even these pre-service teachers barely completed or participated in more than three ESD courses.

Table 4 shows that 81 of 113 Bachelor students never completed an ESD course at the time of the study. Even 39 of the 109 Master of Education students never completed such an ESD course at that time. In addition, the majority did not attend such an ESD course at the time of the study (76 Bachelor, 72 Master of Education). There are Bachelor and Master of Education students who completed or participated in ESD courses. However, few completed more than three ESD courses in total.

2.2. Instruments

2.2.1. Self-Efficacy Beliefs of Teaching ESD in Secondary Education

Based on our analysis of previous research, we identified the need for a new instrument to measure self-efficacy beliefs of teaching ESD in secondary education. The new instrument was designed by Rosenbaum, Handtke, Richter-Beuschel, Grieger, and Bögeholz [29]. The instrument development described in this chapter is also documented in detail (but unpublished) by Rosenbaum [29]. Figure 1 shows a flowchart concerning the development of the new instrument.

Based on the pentagon model of PCK for teaching science [30], we identified corresponding contents for teaching ESD. Out of the five categories of the pentagon model [30], four categories were used. Comparable to Handtke and Bögeholz [34], the category Orientation to Teaching Science was not applied because this category does not include learn-able competencies. Thus, it does not allow to formulate a demanding situation [34] as demanded by Bandura [15,23]. One important aspect of ESD has to be explained in this context: Gestaltungskompetenz (shaping competence). De Haan [60] (p. 320) explains this construct as follows:

“Gestaltungskompetenz means the specific capacity to act and solve problems. Those who possess this competence can help, through active participation, to modify and shape the future of society, and to guide its social, economic, technological and ecological changes along the lines of sustainable development.”

The twelve subcategories of the four categories of the PCK model for teaching science [30] were summarized and transferred to eight subcategories regarding teaching ESD in

Table 5. Transfer of the subcategories of the original PCK (pedagogical content knowledge) model for teaching science by Park and Chen [30] to teaching Education for Sustainable Development (ESD). Rosenbaum [29] contributed to this model transfer in German.

Subcategory of PCK (Original)	Subcategory of PCK (Transferred to ESD)
Misconception	Learning Difficulties
Learning Difficulties
Need	Need
Motivation & Interest
Curriculum Materials	Curriculum Materials
Vertical Curriculum	Curricular Demands
Horizontal Curriculum
Dimensions of Science Learning to Assess	Dimensions of Gestaltungskompetenz to Assess
Methods of Assessing Science Learning	Methods of Assessing ESD Learning
Subject-specific Strategies	Subject/ESD-specific Strategies and Activities
Activities
Representations	Representations

.

We used research literature (e.g., [61]) as well as several subject-specific and interdisciplinary normative guidelines for students and teachers (e.g., [62,63,64]) to specify these eight subcategories of PCK specific for ESD in secondary education. In a second step, these ESD-specific PCK of secondary education was operationalized for self-efficacy beliefs. For this purpose, we transformed the ESD-specific PCK into actions when teaching ESD in secondary education. These actions were all introduced with “Even in teaching Education for Sustainable Development, I can…”. This approach assured important features of self-efficacy beliefs: the focus on their present capabilities, the integration of a standardized obstacle (teaching ESD), and the multidimensionality [15] by the different subcategories. Specific and different obstacles for each item would result in multiple obstacles. Teaching ESD is already difficult because this teaching task cannot be studied as a specific subject in Germany.

Every subcategory contained at least four items, 50 in total. As previously done by different researchers (e.g., [34,65]), the items were rated on the four-point Likert scale consisting of “Is not right” (1), “Is a little right” (2), “Is rather right” (3), and “Is exactly right” (4). We explained possibly unknown technical terms in the items with footnotes. This should ensure that the instrument measures self-efficacy beliefs of teaching ESD and not knowledge of technical terms. The 50 items developed in this way were used as the starting point with sample 1.

2.2.2. Additional Instruments

In the context of the second study, we used an additional validation instrument of Richter-Beuschel and Bögeholz [59] to consider self-rated content knowledge of SD-relevant issues. They defined 13 topics regarding SD-relevant issues to rate [59]. Important aspects of biodiversity and climate change were focused on (e.g., genetic diversity or sustainable land use) [59]. The rating scale ranged from one to five, comparable with grades in school. One indicated “very good”, two “good”, three “satisfactory”, four “sufficient”, and five “insufficient” (cf. [59]). Thus, the greater values indicate a worse self-rating than smaller values. In addition, we asked the test persons to indicate how many ESD courses they currently participate in (without completed examination) and how many ESD courses they have completed in their studies (with completed examination; see sample description) (cf. [58,59]).

2.2.3. Procedure

We applied the newly developed instrument in two studies. In both studies, it was the last instrument to fill out, besides the instruments of other projects. In the first study, a part of the participants received a payment. In the second study, all participants were paid equally.

2.3. Analysis

One aim of the paper is to identify the factors that constitute the self-efficacy beliefs of teaching ESD. The theory of the pentagon model [30] originally focused on science teaching. Thus, we did not assume a very strong theoretical basis in the context of teaching ESD and applied exploratory factor analysis to generate an initial hypothesis about the factorial structure [66]. The generated model must then be checked with a confirmatory factor analysis using a new sample [67,68]. We used IBM SPSS Statistics 24 (Armonk, NY, USA) and RStudio (1.3.1093; Boston, MA, USA) with the packages lavaan (0.6–3) [69], ggplot2 (3.3.5) [70], and tidyr (1.2.0) [71].

Regarding the exploratory factor analysis, our sample was sufficient, according to Bühner [72]. The prerequisites of the exploratory factor analysis, such as the Kaiser-Meyer-Olkin value (KMO) and the Bartlett’s Test [73], have to be checked. We applied the principal factor analysis to extract latent factors [67]. The maximum-likelihood method would need a normal distribution that is not fulfilled in this sample [74]. We applied the Kaiser-Guttman criterion (i.e., eigenvalues greater than one; [75,76,77]) to identify the number of factors. We considered the scree plot less helpful due to the subjective interpretation [72]. In addition, the scree plot indicated one strong factor, but this would be against the theory of self-efficacy beliefs as a multidimensional construct [15]. We allowed correlations between the factors with the oblique rotation oblimin direct [68] because the pentagon model indicates such relations [30]. We applied a pairwise deletion. We assumed loadings ≥ 0.3 as appropriate [73], but we allowed considerations of items with a loading under 0.3 based on the content of the items [72]. We reran the exploratory factor analysis every time we removed at least one item [73]. The data were handled as metric due to the exploratory approach.

The second and larger sample allowed a confirmatory factor analysis [78] and a more rigorous ordinal data computation. We used a WLSMV estimator for our ordinal, non-normal data [79]. We used pairwise deletion [69] because very few values were missing (232 of 236 persons answered all items). According to Little [78] and Wheaton et al. [80], we used the following values as lower boundary: χ²/df ≤ 5, CFI > 0.90, TLI > 0.90, RMSEA < 0.10.

To investigate Hypothesis 1, we computed one latent factor for self-rated content knowledge of SD-relevant issues in the confirmatory factor analysis. This factor consisted of all 13 self-rated content knowledge items. By specifying this factor and the self-efficacy beliefs of teaching ESD, lavaan automatically calculated the correlations of the self-rated content knowledge of SD-relevant issues with the factors of self-efficacy beliefs of teaching ESD. We used RStudio with the package cocor (1.1–3) [81] to compare correlations.

In the context of Research Question 2, we used an independent t-test. Kubinger et al. [82] recommend using the Welch test (unequal variances) due to its robustness without any normal distribution test for samples from a value of at least 30. To avoid alpha error accumulation, we used the Bonferroni correction [83] for the Welch test. Regarding Hypothesis 2 and 3, we used structural equation modeling in lavaan with one dummy-coded variable each: (i) Bachelor (0) and Master of Education (1), and (ii) currently not participating in at least one ESD course (0), and currently participating in such a course (1).

3. Results

3.1. Research Question 1: Factorial Structure of the Self-Efficacy Beliefs of Teaching ESD

The final exploratory factor analysis contained 35 out of the 50 items and eight factors with an eigenvalue greater than one. The final model resulted after 11 runs of the exploratory factor analysis. These 11 runs were necessary due to the removal of items for several reasons (loadings < 0.3 and content-related decisions).

A KMO value of 0.88 in the final exploratory factor analysis confirmed the adequacy of our sample. The Bartlett’s test of sphericity is significant, χ² (595) = 2618.37, p < 0.01 [73]. The eight factors explained 51.92% of the variance in combination after extraction.

Table 6. Pattern matrix of the exploratory factor analysis (n = 162). Eight factors emerged from the final exploratory factor analysis. Factor loadings greater than 0.25 are displayed. The bold loadings represent the final assignment to the factors. “Item” presents the wordings of all items. All items had the same stem at the beginning: “Even in teaching Education for Sustainable Development, I can…”. ESD = ESD-specific, G = generic. We explained possibly unknown technical terms in the items with footnotes (in this table directly below the corresponding item with superscript numbers). The formatting of the items (bold, underlined, italics) corresponds to that of the original questionnaire. The table includes contributions by Rosenbaum [29], translated into English.

	Item: “Even in teaching Education for Sustainable Development, I can…”	1	2	3	4	5	6	7	8
	Factor 1: Fostering Gestaltungskompetenz and Generic Competence (ESD/G)
1	… foster Gestaltungskompetenz 1 cumulatively 2 and systematically 3 over the course of a single teaching unit. 1 Gestaltungskompetenz describes the capability to actively modify and model the society in which one lives in terms of Sustainable Development—taking into account economic, ecological, cultural, and social aspects. The competence area comprises twelve sub-competencies. 2 Being able to connect content over the course of a single and future teaching units. 3 Fostering competencies in a research-based manner by using competence models and methods.	0.73
2	… foster Gestaltungskompetenz cumulatively 4 and systematically over the course of several school years. 4 Being able to connect content from past and future school years.	0.63
3	… improve students’ generic 5 competencies cumulatively and systematically over the course of several school years (e.g., problem-solving competence, presentation competence, democracy competence, and social competence). 5 Competencies that are not subject-specific and relevant to all subjects.	0.48							−0.25
4	… improve students’ generic competencies cumulatively and systematically over the course of a single teaching unit (e.g., problem-solving competence, presentation competence, democracy competence, and social competence).	0.45				0.26
	Factor 2: Including ESD-specific Instructional Strategies under Consideration of Students’ Motivation (ESD)
5	… integrate cooperative and activating forms of learning to increase self-organized learning (project works, future workshops 6, planning cells 7, Open Space 8). 6 A future workshop is a method in which all participants are experts. The aim is to find solutions to commonly identified problems. For this purpose, three phases are run through: criticism, fantasy, and realization phase (Preußer, 2007b, pp. 3–6). 7 In a planning cell, people representing diverse opinions and interests are brought together at random to develop possible solutions to a given problem (Preußer, 2007b, pp. 3–6). 8 Open Space is a method that enables a discussion process or a creative exchange on a voluntary basis (Preußer, 2007b, pp. 3–6).		0.69
6	… include planning games (e.g., Water for Ganurbia9, World Climate Conference 9, Climate Change in Court 9). 9 Here, the students take on different roles (representatives from politics, economy, science, and NGOs), deepen unsustainable developments in discussions and develop possible solutions (Preußer, 2007a, pp. 9–12).		0.68
7	… conceptualize interdisciplinary projects (e.g., biology and art: nature photography exhibition about forests; natural sciences, geography, and German: debate club on the topic “sustainable resource use”).		0.66
8	… integrate extracurricular partners to include different perspectives in systematic socioscientific decision-making (e.g., experts from politics, the economy, science, and NGOs 10). 10 Non-governmental organizations, such as Greenpeace, Amnesty International.		0.59
9	… create participation opportunities for students (e.g., by running a student company 11). 11 A student company is a company run by students on their own responsibility. In contrast to a simulation, the students participate in the real market situation (KMK & BMZ, 2016, p. 61).		0.51
10	… improve students’ motivation by participating in school and society (e.g., organize a school event with the sale of fair-trade products).		0.47
11	… improve students’ motivation by participating in competitions and campaigns on Sustainable Development (e.g., Transfer-21-Schule 12, Internationale Agenda-21-Schule 12, Umweltschule Europa 12). 12 The “Umweltschule Europa”, “Internationale Agenda-21-Schule”, and “Transfer-21-Schule” awards are conferred to schools for excellent achievements in developing sustainability processes in and outside the school area (BLK, 2005, pp. 13–14).		0.44
	Factor 3: Using ESD-specific Educational Resources for School (ESD)
12	… use journals and materials for teachers when planning lessons (e.g., Unterricht Biologie, Praxis Geographie).			0.76
13	… consider textbooks’ materials and tasks (e.g., Projekt G Gesellschaftslehre, Prisma Naturwissenschaften) when planning lessons.			0.58
14	… integrate current didactical approaches from textbooks of the didactics (e.g., Fachdidaktik Biologie, Geographiedidaktik, Politik-Didaktik).			0.42
	Factor 4: Surveying Dimensions of Gestaltungskompetenz (ESD)
15	… survey students’ ability to show solidarity, to change perspective, and to empathize with humans and the environment.				0.70
16	… survey students’ conceptions about justice and sustainability, which legitimize their social and ethical actions.				0.61				−0.27
17	… survey students’ willingness to motivate themselves and others for topics of Sustainable Development.				0.58
18	… survey students’ private, social, and political participation in realizing Sustainable Development Goals.				0.47
19	… survey students’ critical reflection on their own or foreign value benchmarks in the solution of dilemmatic decision-making situations.				0.39				−0.29
20	… survey students’ interdisciplinary scientific inquiry of solving tasks of Gestaltungskompetenz.				0.30
	Factor 5: Applying Methods of Evaluation (ESD/G)
21	… appropriately select suitable methodical procedures by weighing the advantages and disadvantages of formative and summative evaluation procedures.					0.60
22	… apply methods for summative evaluation 13 (e.g., exam to review performance). 13 Evaluation of effects.					0.58
23	… analyze students’ abilities based on competence models (e.g., model for decision-making 14, sustainability triangle/quad 15). 14 A model that describes three requirements of ESD in the context of socioscientific decision-making: ‘Generating and Reflecting on Factual Information’, ‘Evaluation, Choice, and Reflection’, and ‘Understanding and Reflecting on Values and Norms’ (Bögeholz, 2013, p. 75). 15 Describes the three (social, economic, ecological) or four (plus cultural) dimensions of ESD, which should be considered for a solution to socioscientific issues (KMK & BMZ, 2016, p. 41).					0.52		0.27
24	… use methods for formative evaluation 16 (e.g., climate change concept map, questionnaire to survey everyday concepts on globalization). 16 Supporting the learning process.					0.49
25	… create tasks relating to the three requirement areas 17 using operators appropriate to the situation. 17 Requirement area I: “Reproduction”; Requirement area II: “Reorganization and Transfer”; Requirement area III: “Reflection, Evaluation, and Problem-Solving”.					0.49
	Factor 6: Applying Media (G)
26	… use the (interactive) whiteboard, beamer, and document camera as a medium to present (e.g., to visualize teaching materials and teaching subjects in students’ presentations).						−0.89
27	… incorporate tablets, smartphones, and computers (e.g., apps to execute research 18, take up extracurricular expertises in the form of blogs, simulation games). 18 One example is the app “Get-neutral”, which can be used to scan different everyday products for their carbon footprint. In that way, students can research and assess whether they live sustainably (Rat für Nachhaltige Entwicklung, 2017).						−0.65
28	… use the blackboard and the overhead projector to present charts (e.g., concept maps on climate change, flowchart on CO2 emissions).						−0.45
	Factor 7: Using ESD-specific Materials of the Research (ESD)
29	… consider the guidelines of the Cross-Curricular Framework for Global Development Education when planning lessons.							0.55
30	… integrate materials of subject-specific programs (e. g., BLK 21 19, Transfer-21 20, Eine Welt in der Schule 21) when planning lessons. 19 The aim of the program of the Bund-Länder-Kommission (BLK) was to enshrine ESD in regular school practice as a follow-up to “Agenda 21” (BLK, 2005, p.7). 20 Follow-up program of the BLK 21 program. 21 Bremen-based lessons development project regarding the learning areas “Global Learning” and “Global Development”.							0.54
31	… take into account recent research findings when planning lessons (e. g., from journals such as the Journal of Geography, International Journal of Environmental and Science Education; from PISA studies).							0.29
	Factor 8: Teaching Ethically Relevant Issues of ESD (ESD)
32	… foster students’ handling of dilemmatic situations in the context of socioscientific decision-making competence 22, respectively political judgment competence and capacity for political action 23 cumulatively and systematically over the course of several school years. 22 Socioscientific decision-making is an area of competence to be fostered in science subjects. 23 Political judgment competence and capacity for political action are areas of competence in the social sciences.	0.27							−0.67
33	… consider students’ difficulties with judging dilemmatic situations (e. g., not certified fishing and cultivation of cotton, roses, and coffee beans).		0.27						−0.43
34	… foster students’ handling of dilemmatic situations in the context of socioscientific decision-making competence, respectively political judgment competence and capacity for political action cumulatively and systematically over the course of a single teaching unit.	0.40							−0.40
35	… survey students’ ability to analyze and anticipate unsustainable developments and their global risks.				0.30				−0.37

shows the items of the eight factors after rotation and their loadings in the final exploratory factor analysis. Cronbach’s Alphas of the factors are acceptable; they range from 0.68 to 0.85 in the first sample. One item (item 34) had the same loading on two factors (Table 6). Due to the item’s content, we assigned it to factor 8. Factor 8 contains Teaching Ethically Relevant Issues of ESD. In addition, item 34, focusing on horizontal fostering of handling with dilemmatic situations, is a counterpart to the vertical fostering of handling with dilemmatic situations (item 32). Thus, both items complement each other in factor 8.

After we identified the eight-factor solution of the self-efficacy beliefs of teaching ESD from the results of the exploratory factor analysis, the next task was to test this factor structure with confirmatory factor analysis. The model with eight factors showed an acceptable fit (n = 236): Ratio = 1.65; CFI = 0.89; TLI = 0.88; RMSEA = 0.05 [90% confidence interval: 0.05–0.06]. However, the intercorrelation of factors 1 and 8 was r = 1.14 and, thus, greater than 1. Since this is not possible, at least one of the two factors had to be removed.

Table 7. Confirmatory factor analyses of the self-efficacy beliefs of teaching Education for Sustainable Development (ESD) without factor 1 and without factor 8 (n = 236). An additional decimal place increases the comparability of the fit indices of the confirmatory factor analyses. The table includes unpublished results documented by Lünemann [84] and computed by Handtke.

Factor Removed	Ratio	CFI	TLI	RMSEA
1	1.49	0.916	0.906	0.045 [0.038–0.053]
8	1.51	0.915	0.904	0.046 [0.039–0.054]

shows the fit of a confirmatory factor analysis without factor 1 and a confirmatory factor analysis without factor 8.

Both confirmatory factor analyses show a nearly identical fit. Thus, we made the decision due to content-related reasons (Lünemann [84] documented and contributed to this decision). Factor 1, on the one hand, contained items focusing on generic competencies, and on the other hand, items focusing on Gestaltungskompetenz that are also included in factor 4. Factor 8 includes the ESD central competence of socioscientific decision-making. Thus, factor 1, especially with the generic items, was rather dispensable. Consequently, we chose the model without factor 1. This model has a good fit, as shown in Table 7. Table S1 (Supplementary Material) shows the loadings of the items on the seven factors.

Therefore, the hypothetical model was confirmed, except for the rather dispensable factor 1.

Table 8. Factors of the self-efficacy beliefs of teaching Education for Sustainable Development (ESD) instrument sorted by eigenvalues. The first factor was excluded in the confirmatory factor analysis. Thus, there are no values for sample 2. Names of the factors, corresponding subcategories of the pedagogical content knowledge (PCK) model [30], n = number of items, α = Cronbach’s Alpha, 1 = sample with n = 162 and exploratory factor analysis, 2 = sample with n = 236 and confirmatory factor analysis. ESD = ESD-specific, G = generic. The table includes unpublished results by Lünemann [84].

Self-Efficacy Factor	Subcategory of PCK (Transferred to ESD)	n	α1	α2
Fostering Gestaltungskompetenz and Generic Competence (ESD/G) (excluded after confirmatory factor analysis)	Curricular Demands	4	0.84	-
Including ESD-specific Instructional Strategies under Consideration of Students’ Motivation (ESD)	Subject/ESD-specific Strategies and Activities + Need	7	0.85	0.77
Using ESD-specific Educational Resources for School (ESD)	Curriculum Materials	3	0.77	0.69
Surveying Dimensions of Gestaltungskompetenz (ESD)	Dimensions of Gestaltungs- kompetenz to Assess	6	0.79	0.74
Applying Methods of Evaluation (ESD/G)	Methods of Assessing ESD Learning	5	0.78	0.75
Applying Media (G)	Representations	3	0.73	0.64
Using ESD-specific Materials of the Research (ESD)	Curriculum Materials	3	0.68	0.65
Teaching Ethically Relevant Issues of ESD (ESD)	New (=Learning Difficulties, Curricular Demands, Dimensions of Gestaltungskompetenz to Assess)	4	0.79	0.59

presents the factors identified in the exploratory factor analysis (including the factors confirmed in the confirmatory factor analysis), their corresponding subcategories of the PCK model [30], and the Cronbach’s Alphas for all factors in both samples (exception for confirmatory factor analysis: Fostering Gestaltungskompetenz and Generic Competence).

All factors at least contained three items. Cronbach’s Alphas ranged from 0.68 to 0.85 in the first sample (exploratory factor analysis) and 0.59 to 0.77 in the second sample (confirmatory factor analysis). In the first sample, Cronbach’s Alphas of all factors are above (or almost above 0.7) and, thus, acceptable [57,73]. In the second sample, Cronbach’s Alphas of all factors are worse than in the first sample.

The confirmatory factor analysis revealed seven self-efficacy beliefs of teaching ESD factors (Table 8). All eight subcategories from Park and Chen [30] summarized and transferred to ESD are represented by the seven factors, albeit to varying degrees. Six of these factors can be directly assigned to one or multiple subdimensions of the transferred model from Park and Chen [30]. The seventh factor, Teaching Ethically Relevant Issues of ESD, did not exist in this form in the model from Park and Chen [30]. It emerged as a new factor consisting of three different subcategories.

Some of the factors seemed to be more specific for ESD than others. This is not surprising since, e.g., evaluation methods are used in every subject. The factors Fostering Gestaltungskompetenz and Generic Competence and Applying Methods of Evaluation are partly generic and partly ESD-specific. The factor Applying Media is generic since it only contained a few ESD-specific examples but no ESD-specific competence model (factor 5) or ESD-specific Gestaltungskompetenz (factor 1).

3.2. Hypothesis 1: Correlation of the ESD-Teaching Self-Efficacy Beliefs with Self-Rated Content Knowledge of SD-Relevant Issues

In addition to investigating factorial validity, we analyzed the divergent validity of the instruments’ test value interpretation. We examined the correlation between the self-efficacy beliefs of teaching ESD and self-rated content knowledge of SD-relevant issues as a latent model (Ratio: 1.38; CFI = 0.89; TLI = 0.88; RMSEA = 0.040 [0.035–0.046]). The items of the self-rated content knowledge of SD-relevant issues had loadings on the factor ranging from 0.34 to 0.73 (mean = 0.62).

Table 9. Intercorrelations of the self-efficacy beliefs of teaching Education for Sustainable Development (ESD) factors and correlations with the self-rated content knowledge of SD-relevant issues (srCK) factor (n = 236). The srCK is measured in grades, with 1 being the best and 5 being the worst. Factor 2 = Including ESD-specific Instructional Strategies under Consideration of Students’ Motivation, Factor 3 = Using ESD-specific Educational Resources for School, Factor 4 = Surveying Dimensions of Gestaltungskompetenz, Factor 5 = Applying Methods of Evaluation, Factor 6 = Applying Media, Factor 7 = Using ESD-specific Materials of the Research, Factor 8 = Teaching Ethically Relevant Issues of ESD.

	Factor 2	Factor 3	Factor 4	Factor 5	Factor 6	Factor 7	Factor 8	srCK
Factor 2	1
Factor 3	0.30 **	1
Factor 4	0.59 **	0.31 **	1
Factor 5	0.59 **	0.51 **	0.38 **	1
Factor 6	0.41 **	0.06	0.30 *	0.24	1
Factor 7	0.55 **	0.60 **	0.39 **	0.49 **	0.26	1
Factor 8	0.70 **	0.39 **	0.70 **	0.67 **	0.31 *	0.59 **	1
srCK	−0.26 **	−0.23 **	−0.28 **	−0.24 **	−0.13	−0.25 **	−0.36 **	1

* = p < 0.05, ** = p < 0.01.

shows the resulting correlations. It reveals significant but rather low correlations (r = −0.36 to −0.23, p < 0.01; the negative correlations result from an inverse coding of self-rated content knowledge) for all factors–except factor 6, the only completely generic factor. Using cocor [81], the correlation of Teaching Ethically Relevant Issues of ESD turned out to be the strongest with self-rated content knowledge of SD-relevant issues.

3.3. Research Question 2: Prospective Teachers’ Self-Efficacy Beliefs of Teaching ESD

The absolute values of the mainly pre-service teachers’ self-efficacy beliefs of teaching ESD are shown in Figure 2. The means of the factors range from 2.58 (SD = 0.60) to 3.45 (SD = 0.57) in the first sample and from 2.97 (SD = 0.55) to 3.70 (SD = 0.42) in the second sample. The overall mean in sample 1 is 2.94 (SD = 0.41) and 3.17 (SD = 0.33) in sample 2. Thus, all means are greater than 2.5, which is the theoretical mean of the scale. In the second sample, almost all means are even greater than 3. Therefore, pre-service teachers’ self-efficacy beliefs of teaching ESD seem to be rather positive. This especially seems to be valid for the second sample with exclusively biology, politics, and geography pre-service teachers. Comparing the means between sample 1 and sample 2, a t-test (Welch test) confirmed the significant differences for five of the seven factors in Figure 2 (exceptions due to Bonferroni correction: factors 3 and 5).

Despite the rather positive values on all factors, there are relative differences between some self-efficacy beliefs of teaching ESD factors. In particular, the (mainly) pre-service teachers show the strongest self-efficacy beliefs on the generic factor 6, Applying Media, by far. In sample 1, the Factor 7, Using ESD-specific Materials of the Research, relatively shows the smallest mean. In sample 2, Factor 5, Applying Methods of Evaluation, has the lowest and only mean below 3.

3.4. Research Question 2a: Pre-Service Biology, Politics, and Geography Teachers’ Self-Efficacy Beliefs of Teaching ESD

In addition, we investigated possible differences between ESD teaching self-efficacy beliefs of pre-service teachers in biology, politics, and geography in Figure 3.

The confidence intervals in Figure 3 show hardly any significant differences to expect between the pre-service teachers of the three subjects. Besides possible significant differences, in our sample 2, pre-service geography teachers have the strongest ESD teaching self-efficacy beliefs on all factors. The pre-service teachers of all three subjects had similar high self-efficacy beliefs on the generic factor 6, Applying Media. It is of further note that pre-service politics teachers had rather lower self-efficacy beliefs than biology and geography pre-service teachers concerning factor 3, Using ESD-specific Educational Resources for School, factor 5, Applying Methods of Evaluation, and factor 7, Using ESD-specific Materials of the Research.

3.5. Hypothesis 2: Differences between ESD Teaching Self-Efficacy Beliefs of Bachelor and Master of Education Students

To investigate differences in Bachelor and Master of Education students’ self-efficacy beliefs of teaching ESD, we computed a structural equation model with the phase of teacher education as the predictor. The model showed a good fit (Ratio: 1.43; CFI = 0.92; TLI = 0.91; RMSEA = 0.043 [0.035–0.050]).

Table 10. Structural equation model with the effect of the phase of teacher education on the self-efficacy beliefs of teaching Education for Sustainable Development (ESD) factors (n = 235). Factor 2 = Including ESD-specific Instructional Strategies under Consideration of Students’ Motivation, Factor 3 = Using ESD-specific Educational Resources for School, Factor 4 = Surveying Dimensions of Gestaltungskompetenz, Factor 5 = Applying Methods of Evaluation, Factor 6 = Applying Media, Factor 7 = Using ESD-specific Materials of the Research, Factor 8 = Teaching Ethically Relevant Issues of ESD. β = standardized regression coefficient, CI-95% = 95% confidence interval, SE = standard error. Significant standardized regression coefficients are bold.

Phase of Teacher Education (Predictor)	Factor 2	Factor 3	Factor 4	Factor 5	Factor 6	Factor 7	Factor 8
β	−0.18 *	0.09	−0.18 *	0.01	−0.19 *	−0.21 **	−0.03
CI-95%	−0.32 to −0.04	−0.07 to 0.24	−0.32 to −0.04	−0.14 to 0.16	−0.36 to −0.01	−0.35 to −0.06	−0.19 to 0.13
SE	0.07	0.08	0.07	0.07	0.09	0.08	0.08
p-value	0.014	0.268	0.010	0.891	0.038	0.006	0.696

* = p < 0.05, ** = p < 0.01.

shows the differences between Bachelor and Master of Education students’ ESD teaching self-efficacy beliefs. Meaningful differences occurred regarding four of the seven factors, Including ESD-specific Instructional Strategies under Consideration of Students’ Motivation (Factor 2), Surveying Dimensions of Gestaltungskompetenz (Factor 4), Applying Media (Factor 6), and Using ESD-specific Materials of the Research (Factor 7). The Bachelor students had higher self-efficacy beliefs of teaching ESD regarding these four factors (β = −0.21–−0.18, p < 0.05). For the other three factors, no meaningful difference occurred.

3.6. Hypothesis 3: Influence of Currently Participating in at Least One ESD Course on the ESD Teaching Self-Efficacy Beliefs

To investigate the influence of current participation in at least one ESD course, we computed an additional structural equation model with the participation in at least one ESD course as the predictor. This model had a good fit (Ratio: 1.55; CFI = 0.90; TLI = 0.89; RMSEA = 0.050 [0.043–0.058]).

Table 11. Structural equation model with the effect of the participation in at least one ESD course on the self-efficacy beliefs of teaching Education for Sustainable Development (ESD) factors (n = 217). Factor 2 = Including ESD-specific Instructional Strategies under Consideration of Students’ Motivation, Factor 3 = Using ESD-specific Educational Resources for School, Factor 4 = Surveying Dimensions of Gestaltungskompetenz, Factor 5 = Applying Methods of Evaluation, Factor 6 = Applying Media, Factor 7 = Using ESD-specific Materials of the Research, Factor 8 = Teaching Ethically Relevant Issues of ESD. β = standardized regression coefficient, CI-95% = 95% confidence interval, SE = standard error. Significant standardized regression coefficients are bold.

Participation in ESD Course (Predictor)	Factor 2	Factor 3	Factor 4	Factor 5	Factor 6	Factor 7	Factor 8
β	0.21 **	0.19 *	0.13	0.22 **	0.15	0.03	0.13
CI-95%	0.07–0.35	0.04–0.35	−0.01–0.27	0.08–0.36	−0.04–0.33	−0.13–0.19	−0.04–0.30
SE	0.07	0.08	0.07	0.07	0.09	0.08	0.09
p-value	0.003	0.015	0.077	0.002	0.119	0.710	0.134

* = p < 0.05, ** = p < 0.01.

shows the differences between the pre-service teachers currently participating in at least one ESD course and participating in no ESD course at the time of the study.

Three of the seven factors showed a difference: Including ESD-specific Instructional Strategies under Consideration of Students’ Motivation (Factor 2), Using ESD-specific Educational Resources for School (Factor 3), and Applying Methods of Evaluation (Factor 5). Pre-service teachers currently participating in at least one ESD course had higher self-efficacy beliefs regarding these three factors than the pre-service teachers participating in no ESD course at the time of the study (β = 0.19–0.22, p < 0.05).

4. Discussion

4.1. Theory-Driven Instrument of Self-Efficacy Beliefs of Teaching ESD

One aspect of this paper is the investigation of the factorial and divergent validity of the interpretation of the test values of our newly developed instrument of self-efficacy beliefs of teaching ESD. The instrument is intended to identify the support needs of prospective teachers regarding their ESD teaching self-efficacy beliefs because self-efficacy beliefs are important for the implementation of ESD [12,13,14]. We identified a research gap for such an instrument, especially for secondary education. Previous instruments did not represent current demands, did not include obstacles in the items, and were not developed theory-driven regarding teaching ESD at the same time. One instrument [14,45,46] operationalized a model [47] and, thus, is theory-driven. However, the instrument’s focus is not on different aspects of teaching ESD. This instrument is rather focused on fostering students’ general ESD competencies. In sum, previous instruments did not operationalize a model regarding teaching (e.g., [43,44]), did not integrate obstacles (e.g., [38,40]) or focused primary education (e.g., [37,39]). Our new instrument has these characteristics: (i) it is theory-driven due to the model from Park and Chen [30], (ii) it includes the demanded obstacle(s) due to the phrase “Even in teaching Education for Sustainable Development, I can…” [15,23], and (iii) it focuses secondary education (e.g., due to the used normative guidelines; [62,63]).

We were able to show factorial validity [85] of seven factors of self-efficacy beliefs of teaching ESD with an exploratory and a confirmatory factor analysis. These seven factors integrate (at least parts of) the eight defined subcategories of the transferred PCK model [30]. Thus, they represent a wide range of aspects of teaching ESD and allow a differentiated diagnosis of prospective teachers’ learning prerequisites. In addition, the new factor Teaching Ethically Relevant Issues of ESD extended the aspects in the model from Park and Chen [30]. The confirmatory factor analysis of the seven factors showed a good fit. Therefore, the factors of self-efficacy beliefs of teaching ESD seem to be appropriate, indicating factorial validity. In addition to the factorial validity, we provide the first investigation of divergent validity.

The correlations of the self-efficacy beliefs factors of teaching ESD with the self-rated content knowledge of SD-relevant issues support the divergent validity of our instruments’ test value interpretation. Six self-efficacy factors were significantly correlated with self-rated content knowledge (r = −0.23 to −0.36, p < 0.01; the negative correlations result from an inverse coding of self-rated content knowledge). The factor 6 Applying Media did not correlate significantly with the self-rated content knowledge of SD-relevant issues. We explain this result with the generic character of Applying Media. The generic items of this factor rather relate to every subject and are, despite ESD-specific examples, rather less specific for ESD. This characteristic can explain the missing correlation with (self-rated) content knowledge of SD-relevant issues. Besides this one, another correlation stands out. The correlation of factor 8, Teaching Ethically Relevant Issues of ESD, with self-rated content knowledge of SD-relevant issues was higher than all other correlations. In addition, the ethical aspects of teaching are new compared to the original PCK model [30]. Such a new factor regarding teaching ethically relevant issues emerged in the instrument development of self-efficacy beliefs of interdisciplinary science teaching based on the model from Park and Chen [30] as well [34]. In our study, the factor contains aspects such as dilemmatic situations, which are more related to self-rated content knowledge of SD-relevant issues than other factors. Thus, factor 8, Teaching Ethically Relevant Issues of ESD, probably had the strongest correlation of all self-efficacy beliefs factors. According to the literature (e.g., [14,43]), self-efficacy beliefs of teaching ESD and self-rated content knowledge of SD-relevant issues were related. Self-efficacy beliefs and self-rated content knowledge are different constructs, but the contexts of ESD and SD-relevant issues show certain overlaps in parts since ESD addresses SD issues. Thus, our result of rather lower correlations argues for two related constructs that are different nevertheless, i.e., divergent validity [57]. This argument for divergent validity supplements the arguments for factorial validity. Overall, the results in this paper provide the first arguments for the validity of the interpretation of the test values of the instrument of self-efficacy beliefs of teaching ESD.

Focusing on reliability, the two samples in this paper provide a diverse picture. Cronbach’s Alphas were good in sample 1 with one of eight factors with a value under 0.7 and overall ranging from 0.68 to 0.85 (Table 8). In sample 2, four of seven factors had a value below 0.7 and ranged from 0.59 to 0.77 (Table 8). Thus, we have to focus on why the values of Cronbach’s Alpha decreased in the second but larger sample. The key to explaining these differences lies in the composition of both samples. Considering Table 2, it gets obvious that sample 1 was much more heterogeneous than sample 2. Sample 1, e.g., took pre-service teachers from the Bachelor, Master of Education, State examination, or with finished studies into account. Sample 2 focused on Bachelor and Master of Education. In contrast to sample 1, all pre-service teachers of sample 2 studied to teach at a grammar or comprehensive school (secondary education). Especially regarding the studied subjects, different criteria existed in both samples. While sample 2 exclusively focused on biology, politics, and geography, sample 1 counted additional subjects relevant to ESD (e.g., religion, chemistry, or philosophy). Overall, sample 1 was much more diverse than sample 2. This can be one reason for the lower values of Cronbach’s Alpha in sample 2. Since the homogenous sample tends to cause lower variance and covariance, this can explain the lower Cronbach’s Alphas compared to the more heterogeneous sample 1. Future research should focus on reviewing the reliability of our new instrument with further different samples.

Besides the psychometric characteristics of our new instrument and its advantages compared to previous instruments, we want to underline the purpose of our instrument in the context of ESD. As mentioned initially, self-efficacy beliefs are important for teaching (e.g., [18,21]). Referring to Moseley et al. [12,13], Malandrakis et al. [14] also conclude that self-efficacy is powerful for teachers in the context of ESD. Thus, teacher education should foster these self-efficacy beliefs of pre-service teachers. To do so, an appropriate instrument of self-efficacy beliefs of teaching ESD is needed to diagnose the learning prerequisites of prospective teachers. In this paper, we established a new instrument especially for teaching ESD in secondary education—in contrast to previous instruments, particularly in the context of primary education (e.g., [37,43,44]).

4.2. First Insights into the Effectiveness of ESD in German Universities

For the implementation of ESD, teachers and their skills and commitment to sustainability play a central role [7,9]. Our new self-efficacy instrument allowed the examination of pre-service teachers’ learning prerequisites. Research Question 2 (Absolute means) and 2a (Means differentiated by studied subject) give an impression of (prospective) teachers’ actual self-efficacy beliefs of teaching ESD. The results of Hypothesis 2 (Difference between Bachelor and Master of Education students) and Hypothesis 3 (Influence of current participation in ESD courses) give first insights into the effectiveness of ESD in eight German universities.

The results of Research Question 2 (Absolute means) show that ESD teaching self-efficacy beliefs of sample 2 are higher than in sample 1 concerning the most factors. This is comprehensible since sample 2 focused exclusively on pre-service teachers for grammar and comprehensive schools and on the subjects of biology, politics, and geography, the subjects that integrate ESD the most in German secondary schools [53]. Thus, the pre-service teachers in sample 2 seem to be better prepared for teaching ESD. Both samples have the highest self-efficacy beliefs on factor 6, Applying Media. This is also understandable since it is a generic factor that is improved by every studied subject. In sample 1, factor 7, Using ESD-specific Materials of the Research, shows the lowest mean. This could be due to the more heterogeneous sample 1. Pre-service teachers of biology, politics, and geography probably know research for ESD better due to their subjects. Thus, the mean of factor 7 is rather small in sample 1. In sample 2, factor 5, Applying Methods of Evaluation, relatively shows the lowest mean. Perhaps, applying evaluation methods is less important in teacher education at university than the other topics, since the application is more addressed in the internship by trainee teachers [86]. This could result in little lower self-efficacy beliefs.

The results of Research Question 2a complement these results with the differentiation by the studied subject. The confidence intervals of Figure 3 indicate only a few possibilities for significant differences. This could be due to our sample size and composition (e.g., only 21 pre-service politics teachers). Thus, more research and greater samples are necessary to investigate the subjects’ influence and possible significant differences further. Our sample suggests that pre-service geography teachers have the strongest self-efficacy beliefs of teaching ESD concerning all factors. The rather low values, e.g., on factors 3, 5, and 7, of pre-service politics teachers, could hint at a need for more ESD in teacher education of politics. Compared to the pre-service geography teachers, there seems to be room for improvement for the pre-service biology teachers as well. However, for generalization, these tendencies of our sample need to be investigated in a larger sample with sufficient test persons in every subgroup.

The results of Hypothesis 2 (Bachelor vs. Master of Education) show that regarding four of seven ESD teaching self-efficacy beliefs factors, the Bachelor students had higher self-efficacy beliefs than the Master of Education students. A total of 32 of 113 pre-service teachers in the Bachelor and 70 of 109 in the Master of Education completed at least one ESD course at some point in their studies. Thus, the Master of Education students had more opportunities to learn about teaching ESD or to teach ESD at the time of the study and should have higher self-efficacy beliefs of teaching ESD (cf. for teaching science: [49]). Therefore, our results could argue for a misjudgment of Bachelor students’ skills by themselves.

35 Bachelor students and 34 Master of Education students currently participated in at least one ESD course. Previous research showed that ESD courses at university (primary, early childhood, and middle school) resulted in pre-service teachers’ higher self-efficacy beliefs of teaching ESD [56]. The results of hypothesis 3 support this assumption for secondary education since three factors were positively influenced by the current participation in at least one ESD course: Including ESD-specific Instructional Strategies under Consideration of Students’ Motivation (Factor 2), Using ESD-specific Educational Resources for School (Factor 3), and Applying Methods of Evaluation (Factor 5). This seems plausible because these three factors comprise important facets of teaching. Perhaps, recent ESD courses focus on these aspects of teaching ESD.

These results of Hypothesis 3 (Influence of current participation) could support the assumption that the pre-service teachers in the Bachelor misjudge their abilities, since a similar number of students of the Bachelor and Master of Education currently participated in at least one ESD course. So, there is no strong supposed difference between both groups in this aspect. Overestimating one’s abilities [87,88] or underestimating the difficulty of a task [89,90] are not uncommon when it comes to self-efficacy beliefs. One problem could be the lack of expectations of the skills required to teach ESD (cf. for biology: [91]), addressed in ESD courses. However, Bachelor and Master of Education students currently participated in a comparable number of ESD courses. But, at the time of the study, more than twice as many Master of Education students have completed at least one ESD course in their studies compared to the Bachelor students. This could allow the Master of Education students to better know the requirements of teaching ESD. Thus, the higher self-efficacy beliefs of the Bachelor students rather seem to indicate a misjudgment.

Focusing on the means of the self-efficacy beliefs of teaching ESD, all factors show rather positive values in both samples (greater than the theoretical mean of 2.5). Especially, the second less heterogeneous sample had even higher self-efficacy beliefs (nearly all factor means greater than 3). Due to the studied subjects in sample 2 (at least biology, politics, or geography), this seems to be plausible. Nevertheless, looking at the courses currently participated in, differentiated by the studied subject, misjudgment of one’s own abilities could even be an issue for all pre-service teachers. The majority of pre-service biology and politics teacher did not complete any ESD course, and 100 of 115 pre-service biology teachers participated in no ESD course at the time of the study. Most ESD courses seem to be completed in geography. This matches the result of Research Questions 2a, indicating the strongest self-efficacy beliefs for pre-service geography students in our sample 2. However, even the pre-service teachers in geography barely participated in or completed more than three ESD courses. Thus, overestimation of their own abilities [87,88] or underestimation of the task [89,90] seems possible for the whole sample since all had completed rather few ESD courses. Although self-efficacy beliefs are one of the most powerful factors in affecting a teacher’s ability to implement ESD [12,13,14], “overly confident teachers who are largely ineffective” [88] (p. 118) are not aspired. Wheatley [92] emphasizes that doubts about one’s self-efficacy beliefs do not have to be negative. Thus, rather high self-efficacy beliefs of teaching ESD without the appropriate competencies should be viewed with caution. Looking at the completed ESD courses of our test persons, the self-efficacy beliefs of teaching ESD could go too far beyond their competencies acquired in the few ESD courses.

The number of ESD courses differentiated by the studied subjects gave insight into teacher education regarding ESD. In German secondary schools, ESD seems to be integrated the most in geography, politics, and biology [53]. At the same time, Evans et al. [10] stated in a literature review that in most countries ESD is not an integral part of initial teacher training. Our study also shows potential for more ESD in all three subjects since rarely more than three ESD courses were completed or currently participated in. 146 of 214 pre-service teachers answering this question participated in no ESD course at the time of the study. In addition, the study shows a lack of ESD in biology, where 100 of 115 pre-service teachers participated in no course and 66 of 120 never completed an ESD course at that time. The lack of ESD is even stronger in politics (under the restriction of the small number of pre-service politics teachers). Since the pre-service teachers came from eight universities in five German federal states, these insights could be valid hints for multiple German universities. Due to the pivotal role of education in terms of SD (SDG 4) [5] and as a driver to achieve all other SDGs [6], ESD should be more focused on in teacher education. In addition, more ESD courses seem to be necessary for more realistic self-efficacy beliefs of teaching ESD. Especially the Bachelor and pre-service teachers in biology and politics could need more training in ESD in university.

4.3. Limitations

Despite various advantages, new instrument features, and new insights into teacher education, such as the need for more training in ESD in universities, we want to discuss some limitations of our study. Focusing our samples, especially the second sample contained relatively few pre-service teachers studying politics. In both samples, we could include only a few pre-service teachers with at least two of the subjects of biology, politics, and geography. Overall, the second sample used to confirm the factorial model could be larger (including the mentioned subgroups). This could also allow better investigations of significant differences in the self-efficacy beliefs between pre-service teachers of different subjects. Further, the second sample consisted of 72.5% women. In German teacher education, there are more female students studying biology and rather more female students in geography (e. g., [93]). More than half of our sample 2 are pre-service biology teachers, and about one quarter are pre-service geography teachers. Thus, the rather less pre-service politics teachers could also influence the gender distribution in our sample. Nevertheless, a possible bias in gender distribution should be considered—even if the higher portion of females shows the right tendency.

In addition, the monetary reward for study participation could have had an impact in both samples. We are not able to test the influence in sample 1. In sample 2, all participants were rewarded. Comparing the self-efficacy beliefs of teaching ESD with the self-efficacy beliefs of teaching interdisciplinary science, which were constructed similarly, the latter showed very little effect of the monetary reward on just one factor [50]. Thus, a strong effect of the participation fee can be rather denied. However, by rewarding all participants in sample 2, a positive selection of participants (e.g., due to interest) could have occurred. In contrast, the monetary reward could have motivated participants with low interest in (teaching) ESD.

The possible overestimation of the test person’s own abilities or the underestimation of the task of teaching ESD cannot be empirically tested in this study. Further investigations would be needed to gain deeper insights here.

In addition, the difference between pre-service teachers in the Bachelor and Master of Education have to be treated with caution due to the cross-sectional design of the study. It should also be considered that the survey of participated and completed ESD courses is subjective. It is possible that the pre-service teachers did not remember all ESD courses or did not perceive a course as an ESD-related course. Moreover, we did not make criteria or minimum standards transparent that define which course could count as an ESD course. Thus, the estimations of the study participants to classify one course as an ESD course or not could be rather subjective. Further, we examined the effect of the participation in or completion of ESD courses with dummy-coded variables. This simplification was necessary due to the sample size but should be considered in the interpretation.

One challenge of developing an instrument for self-efficacy beliefs of teaching ESD isthe different perspectives of ESD. ESD is not an independent, interdisciplinary subject such as science. Thus, ESD is important as a teaching task in several subjects (biology, politics, geography, religion, philosophy, chemistry, …). This condition results in the challenge of integrating multiple perspectives—even more than for science—into the instrument. Besides using different normative guidelines, we tried to develop this instrument with researchers of different disciplines. For example, we integrated perspectives of biology, politics, geography, and history. Surely, we could have still integrated more perspectives and other aspects of ESD since ESD is an extremely broad field. We tried to focus concrete on teaching ESD and on corresponding (pre-service) teacher’s competencies, and thereby on PCK in the light of different disciplines. However, one could, of course, integrate further aspects that are important for teaching at school as well or focus more general ESD outcomes of students like Malandrakis et al. [14] did.

4.4. Future Research

Regarding the instrument’s reliability, Cronbach’s Alphas were good in sample 1 and could be improved in sample 2. To confirm the positive results of the heterogeneous sample 1, more studies are needed to review the reliability of our instrument. At this point, the good results of sample 1 lead us to assume sufficient reliability of our subscales. In addition, the confirmatory factor analysis did not confirm all eight factors of the exploratory factor analysis. For future research, it could be reasonable to check the confirmed seven-factor model again in another more diverse sample, e.g., with more test persons in every subject studied. Another research direction could be using the instrument in different educational phases (e.g., trainee teachers and in-service teachers), which could also provide further arguments for validity and reliability.

Moreover, the new instrument allows investigations of learning prerequisites of pre-service teachers in secondary education or pre-post-investigations in the context of teacher education (e.g., certificate at the University of Göttingen [94]). To examine the effects of teacher education, longitudinal research designs could be of great interest as well. Longitudinal studies could reveal developments in ESD-teaching self-efficacy beliefs during teacher education and identify possible influencing factors.

In sum, this theory-driven instrument already sheds light on pre-service teachers’ self-efficacy beliefs of teaching ESD and the influences of teacher education, especially for the rather unnoticed secondary education.