Development and Validation of the Marburg Self-Regulation Questionnaire for Teachers (MSR-T)

: This study presents the development and validation of a new measurement tool, the Marburg Self-regulation Questionnaire for Teachers (MSR-T). The questionnaire measures the self-regulatory processes of planning, monitoring and reﬂection in teacher-relevant contexts and aims to help assess and understand teachers’ self-regulatory skills in a broader context. In order to provide a reliable and valid instrument, an integrative framework model was developed and items were generated. In two online studies ( N = 255; N = 356) involving German-speaking (trainee) teachers and student teachers, the psychometric properties were examined and the items were revised after conducting conﬁrmatory factor analysis. Further, discriminant and convergent validity was examined by administering our scale along with indicators of professional restraint, self-efﬁcacy and other constructs that are related to self-regulation. The MSR-T encompasses 27 items and showed acceptable to good internal consistencies. Moreover, validity analyses indicated good convergent and discriminant validity, making it an economical yet speciﬁc tool that can be used to assess self-regulatory processes, including planning, monitoring and reﬂection, within a speciﬁc context. The context-unspeciﬁc wording of the questionnaire allows for transferability to other professional contexts, opening up avenues for the examination of self-regulation research in different ﬁelds.


Introduction
Self-regulation is an everyday process through which we can achieve our personal goals by influencing and repeatedly adjusting our thoughts, feelings and actions [1]. Several studies have focused on the influence of self-regulatory skills on health (mental health [2]; weight under stress [3]; chronic diseases [4], academic [5,6]) or professional success [7]. For example, the study by Shalev and Sulkowski [2] shows that low self-regulation skills can play a role in a variety of impulsive and obsessive-compulsive disorders and Lohrig and colleagues [4] showed in a pre-post cohort study that several weeks of self-management training, which included elements of self-regulation promotion, helped patients with chronic diseases to integrate more health-promoting behaviours into their daily lives and to improve their overall health status. Duckworth et al. [6] showed in a four-year study of undergraduates that self-control (self-report and peer-report) predicted grades (GPA) in the next semester. This demonstrates the importance of self-regulation and related skills for academic careers.
The importance of self-regulation in maintaining mental health is illustrated by Kanfer's self-management training [8] for the treatment of people with mental illness, which focuses on planning helpful behaviours and positively influencing thoughts and feelings. Studies such as Frayne's [9] show that this training also has positive effects on work performance in a non-clinical sample and that self-regulation is relevant in everyday work. As there is also evidence that children's self-regulatory competence (indirectly) predicts later career success [10,11], and as described above, the implementation of each daily action underlies this process [12], early promotion of this competence appears to be central. This is reflected in the number of programs and studies dealing with students and self-regulation, not least Pintrich's [13] or Schunk and Zimmermann's [14] model of self-regulated learning. These models help teachers to instruct students in these complex regulatory competences in a targeted way. However, there is comparatively little research on self-regulation among teachers who are not only mediators but also role models for students. This is also surprising given that there is already research on the link between teaching quality and teachers' mental health, both of which can be supported by good self-regulation skills [15].
Particularly in view of the increased health risks in the teaching profession [16,17], it seems necessary to identify relevant and trainable competencies such as self-regulation in a differentiated way, in order to be able to identify risks and promote competencies in prospective and practicing teachers. So far, studies have assessed self-regulation rather globally, not paying attention to potentially different underlying competences in different domains. With the current study, we therefore present the development and validation of the Marburg Self-Regulation Questionnaire for Teachers (MSR-T) which aims to measure teaching-related self-regulatory skills in a meaningful and differentiated way.
The self-regulation questionnaires that are widely used in German-speaking countries often mix several constructs or focus more on the outcomes of successful self-regulation than on the strategies used. It is therefore not surprising that there seems to be a separate self-regulation questionnaire for different contexts (e.g., Hannover Self-Regulation Inventory, HSRI [18] for clinical psychology). Other questionnaires, such as the AVEM (Allgemeine Verhaltens-und Erlebensmuster [19]), which is described as a "multidimensional personality diagnostic instrument", focus more on stable personality traits and less on the use of modifiable strategies, even though the items represent different self-regulation processes and other authors also refer to it as a self-regulation questionnaire (e.g., AVEM Short Inventory [20]). Overall, the AVEM provides more information about health-promoting or health-threatening behaviours and experiences in the work context and does not relate these to process areas of self-regulation. The Locomotion Assessment Questionnaire [21], which has been translated into German [22], is based on independently functioning dimensions of self-regulation, the executive dimension "locomotion" ("just doing it") and the evaluative dimension "assessment" ("doing the right thing"). It partially correlates more highly with achievement orientation from the FPI-R (Freiburg Personality Inventory [23]) than with other SR constructs. Items here are, e.g: "I like to judge other people's plans". Or 'I often compare myself to others'. The FIT-L (Fit für den Lehrberuf [24]) is a questionnaire developed specifically for teachers, consisting of the three scales Resilience and Coping Behaviour, Socio-emotional Engagement and Basic Skills, which in turn are divided into a total of eleven subscales. These include, but are not exclusively focused on, self-regulation strategies. Examples of items are "I really come alive under stress". The English-language Motivation and Engagement Scale (MES) by Liem and Martin [24], which has not yet been translated into German, is a strategy-focused questionnaire that is closely aligned with self-regulation-relevant constructs and allows for concrete training recommendations. In addition, the assessed items can be divided into adaptive and maladaptive strategies, so that a differentiated strength/weakness profile can be created, on the basis of which training recommendations can be made. This makes the MES a suitable questionnaire for successful training practice.
In summary, the German-language questionnaires available to date [18][19][20]22,24,25] tend to measure actual states or personality traits, making it difficult to derive recommendations as to which strategies should be specifically promoted. In addition, the questionnaires are not well suited for investigating how exactly self-regulation competence is built up. It is conceivable that, analogous to the self-concept of ability, there are different levels of competence in different occupationally relevant areas, instead of one global self-regulation competence. This, in turn, would have implications for the training to be developed, which in this case would have to be individualised for area-specific weaknesses and cannot be implemented in a one-size-fits-all intervention. In order to develop a self-regulation training for teachers or to measure the success of the training, a suitable questionnaire is currently needed that is more oriented towards common self-regulation models and focuses more on behavior or the use of strategies while simultaneously taking the specific affordances of teaching contexts at schools into account. This questionnaire could also be used to explore the construct of self-regulation in more detail. For example, it has not yet been extensively investigated whether self-regulation is a general or context-specific competence (analogous to the ability of self-concept). However, this knowledge is also necessary for the design of training, in order to avoid unnecessary duplication of training content without reducing effectiveness. The present study therefore aims to develop and validate a questionnaire on teacher self-regulation based on an integrative theoretical model of self-regulation, drawing on common models of self-regulation, in order to provide a more nuanced assessment of the malleable constructs of self-regulation in two online samples of teachers. We are investigating the following research questions: Does our newly developed questionnaire show adequate internal consistencies (1); and is it a valid measuring instrument (2)? The final questionnaire will also be used to investigate whether self-regulation is a context-specific competence (3).

General Materials and Methods
The MSR-T was developed and validated in two studies. An overview of the proceedings of the two studies can be found in the Supplementary Material (https://data. uni-marburg.de/handle/dataumr/240). The first study served as an initial check of the items and item reduction. The second study validated the shortened questionnaire and subjected it to further item reduction. Both studies were supported by "ProPraxis", a project funded by the German Federal Ministry of Education and Research as part of the "Quality Offensive Teacher Education" (QLB) in order to develop teacher education in Germany. The studies were pre-registered with AsPredicted (https://aspredicted.org/KWK_VIQ, accessed on 11 November 2020).

Study 1: Scale Development and Procedure
For the first study, prior to the actual item construction, the authors mapped an overview of the theoretical facets of self-regulation based on theoretical models [13,14,26] and already-established questionnaires on self-regulation and related constructs (e.g., selfcontrol). Based on this, an integrative theoretical framework model of self-regulation was developed (ITR; Figure 1, for detailed information see Appendix A). The three processes of planning, monitoring and reflection were considered in the item construction. In addition, adaptive and maladaptive facets were distinguished for all three process areas. In a first step, 79 items were constructed by the authors in consultation with a panel of experts, each of which could be related to the four contexts of teaching, self-motivation, self-care and communication, and which served as the initial pool in Study 1. All items referred to the four contexts of teaching, self-motivation, self-care and communication, as these are four professionally relevant areas in the teaching profession [27].
Before the interview began, participants were shown a video or, if they wished, an explanatory text with a definition of self-regulation and an explanation of the above processes and contexts (see Appendix B).

Sample Recruitment
The survey was conducted via the online platform Soscisurvey (https://www.soscisurvey.de, 11 November 2020). Data were collected between 8 December 2020 and 15 March 2021. The inclusion criterion was being a teacher, trainee teacher or student teacher. Recruitment took place via social media, letters to schools and study seminars in various federal states, and the university's internal student mailing list. As an incentive to participate, participants who reached the study through our social media call were offered the chance to take part in a draw for €20 gift vouchers.

Sample Recruitment
The survey was conducted via the online platform Soscisurvey (https://www.soscisurvey.de, 11 November 2020). Data were collected between 8 December 2020 and 15 March 2021. The inclusion criterion was being a teacher, trainee teacher or student teacher. Recruitment took place via social media, letters to schools and study seminars in various federal states, and the university's internal student mailing list. As an incentive to participate, participants who reached the study through our social media call were offered the chance to take part in a draw for €20 gift vouchers.

Instruments
Demographic variables. At the beginning, the variables age, sex (gender), work experience (in years), employment (student, trainee teacher (Teachers who have graduated but have not yet completed their training are referred to in this study as 'trainee teachers'. Teacher training in Germany begins with a course of study that leads to a so-called "Referendariat" or internship. During this internship, teachers gain further practical experience at a particular school and acquire additional relevant knowledge in a training format. Trainee-teachers receive support but are already responsible for their own classes), teacher) and type of school were recorded.
Self-regulation. Self-regulation was measured with the newly constructed Marburg Self-Regulation Questionnaire for Teachers (MSR-T). The first version of this questionnaire comprised 79 items, each of which could be assigned to one of the three process areas of planning (e.g., "I think about what I want to achieve."), self-observation (e.g., "I adapt my strategies to the respective situation.") and reflection (e.g.," I check how well I am progressing."). All items were asked in parallel for four contexts: teaching, motivating, selfcare and communication, as these are essential contexts of teachers' work. Participants were asked to indicate how often (Likert scale from 1 = "hardly ever"-5 = "almost always") they engage in the thoughts and behaviours described in the items in each of the four contexts, in order to consciously adapt their behaviour towards a specific goal. For each of the four contexts, sum scores were calculated for all three processes, resulting in a total of 12 subscale scores. In addition, a total score was calculated for all processes and contexts. A higher total score indicates higher self-regulation competence. The internal consistencies

Instruments
Demographic variables. At the beginning, the variables age, sex (gender), work experience (in years), employment (student, trainee teacher (Teachers who have graduated but have not yet completed their training are referred to in this study as 'trainee teachers'. Teacher training in Germany begins with a course of study that leads to a so-called "Referendariat" or internship. During this internship, teachers gain further practical experience at a particular school and acquire additional relevant knowledge in a training format. Trainee-teachers receive support but are already responsible for their own classes), teacher) and type of school were recorded.
Self-regulation. Self-regulation was measured with the newly constructed Marburg Self-Regulation Questionnaire for Teachers (MSR-T). The first version of this questionnaire comprised 79 items, each of which could be assigned to one of the three process areas of planning (e.g., "I think about what I want to achieve."), self-observation (e.g., "I adapt my strategies to the respective situation.") and reflection (e.g.," I check how well I am progressing."). All items were asked in parallel for four contexts: teaching, motivating, self-care and communication, as these are essential contexts of teachers' work. Participants were asked to indicate how often (Likert scale from 1 = "hardly ever"-5 = "almost always") they engage in the thoughts and behaviours described in the items in each of the four contexts, in order to consciously adapt their behaviour towards a specific goal. For each of the four contexts, sum scores were calculated for all three processes, resulting in a total of 12 subscale scores. In addition, a total score was calculated for all processes and contexts. A higher total score indicates higher self-regulation competence. The internal consistencies of the MSR-T scales before (79 items) and after item reduction (55 items, see text below) are shown in Table 1.
Job success and health. Even though the first study aimed at a first review and reduction of the items, at least first correlations to performance satisfaction and health should be checked. Since the unabridged MSR-T had a large scope, only 5 single-items were collected for review to save time: (1) "I am satisfied with my work-related performance"; (2) "I am often unmotivated to approach my work"; (3) "I achieve what is important to me in meaningful conversations"; (4) "I feel physically and mentally healthy overall"; (5) "I am not able to recover properly from work". All items were assessed on a five-point Likert scale ranging from 1 ("strongly disagree") to 5 ("strongly agree").

Data Preparation and Statistical Analysis
Due to our theory-based approach to item selection and the low number of complete data sets in relation to the number of items, we decided to deviate from the pre-registration, to not conduct an exploratory factor analysis and to use the sample from Study 1 only for an initial item selection and validation, due to the small sample size. In order to use the data from Study 1 for the investigation of the factor structure, we decided to calculate a CFA with the pooled data from Study 1 and 2.
Item analyses. First, the items were combined into context-specific scales according to the theoretically assumed structure and in the next step reduced on the basis of item difficulty, discriminatory power and internal consistency in all four contexts.
Linear regressions. For a first check of the practical implications of the questionnaire and the context-specific scales, several multiple regression analyses were carried out. All analyses included work experience and gender as control variables. Since only 204 people provided information on work experience and gender, the analyses were calculated again without these two control variables in order to exclude the possibility that people who did not provide information might be people without work experience, which could change the results. Since small differences were only found in two out of five regression analyses, the analyses are reported as standard with the inclusion of the control variables and in the two exceptional cases both results are reported. All analyses can be found in the Supplementary Material (https://data.uni-marburg.de/handle/dataumr/240).
The assumptions were checked using collinearity diagnostics, Shapiro-Wilk test, QQ plots and residual plots. Cook's distance did not reveal any outliers in any case (all values < 1), nor did the variance inflation factor (all values < 10). As the Shapiro-Wilk test challenged the assumption of normality in two cases, the QQ plots were visually checked and robust parameter estimation (5000 bootstrap samples) was performed in JASP in both cases.

Results
Sample. The characteristics of the sample can be found in Table 2. N = 209 women (age M = 30.33 years, SD = 8.23 years) and n = 46 men (age M = 33.74 years, SD = 9.71 years) participated in the study. In total, 5.1% of the participants were student teachers, 62.7% were trainee teachers and 32.2% were teachers. On average, participants scored 4.12 out of 5 points on the abbreviated MSR-T in the context of teaching, 3.44 points in the context of motivating oneself, 3.34 points in self-care and 3.88 points in communicating. Trainee teachers had descriptively lower cumulative scores than teachers, although none of these scale differences were significant (ps > 0.05).

Item analyses.
To identify items that could be removed, item difficulties were first calculated for all items (see Appendix C: Table A1). None of the items showed a value <0.2 or >0.8 in at least three of the four contexts. Subsequently, the items per context were combined into three (planning, self-observation, reflection) and six (planning adaptive and maladaptive, self-observation adaptive and maladaptive, reflection adaptive and maladaptive) scales, whereby only the more general three-process scales were used for the item reduction, since at this point no statements can yet be made about the actual factorial structure and we therefore worked with the simpler three-factorial model. By calculating the item rest correlation and Cronbach's alpha for the three scales per context (see Table 2), it was possible to remove items that had poor discriminatory power in at least three of four contexts and whose removal increased the scale reliability in at least three of four contexts. Seven items were removed (see Figure 1).
In a second reduction, the content of the items was checked to see which items were similar and shortened accordingly. For example, the item ADS2 ("I react flexibly to changing conditions.") was removed due to its content proximity to ADS3 ("I adapt my strategies to the respective situation.") to reduce redundancy. In this process, 17 items were excluded. A recalculation of the internal consistency of the scales (context and process specific) shows overall acceptable to excellent [29] values of α = 0.795-α = 0.923. A total of 55 items remain in the scale, which were further used in Study 2 (see Appendix C: Table A2). For illustration purposes, the process of item reduction is shown in Figure 2. Since we assume on the basis of our theoretical framework model that a further subdivision of the three processes into adaptive and maladaptive strategies is possible, reliability analyses were also calculated with a subdivision into six scales. The corresponding values can also be found in Table 2. Overall, the further subdivision into six instead of three scales shows a slight deterioration of the internal consistency, but the values are overall in the acceptable to excellent range.
Linear regressions. The multiple regression with work-related performance satisfaction as dependent variable and the three scales planning, self-observation and reflection from the context of teaching as predictors, as well as professional experience and gender as control variables is significant (F (5, 197) = 11.80, p < 0.001). The coefficient of determination R 2 for the overall model is 0.231 (corrected R 2 = 0.2101) and according to Cohen [30] corresponds to a medium goodness-of-fit. Work experience and reflection are significant predictors of Since we assume on the basis of our theoretical framework model that a further subdivision of the three processes into adaptive and maladaptive strategies is possible, reliability analyses were also calculated with a subdivision into six scales. The corresponding values can also be found in Table 2. Overall, the further subdivision into six instead of three scales shows a slight deterioration of the internal consistency, but the values are overall in the acceptable to excellent range.
Linear regressions. The multiple regression with work-related performance satisfaction as dependent variable and the three scales planning, self-observation and reflection from the context of teaching as predictors, as well as professional experience and gender as control variables is significant (F (5, 197) = 11.80, p < 0.001). The coefficient of determination R 2 for the overall model is 0.231 (corrected R 2 = 0.2101) and according to Cohen [30] corresponds to a medium goodness-of-fit. Work experience and reflection are significant predictors of performance satisfaction (work experience: β = 0.019; t (197) = 2.23; p = 0.027; reflection: β = 0.542; t (197) = 2.89; p = 0.004). While gender, planning and introspection do not significantly influence performance satisfaction (see Appendix D: Table A3). Preconditions for regression (no multicollinearity, homoscedasticity, normal distribution) were checked and can be found in our Supplementary Material (https://data.uni-marburg.de/handle/dataumr/240). In this case, the Shapiro-Wilk test shows a violation of the normal distribution assumption. With very large samples, even small and trivial deviations from the normal distribution become significant which is why an additional visual check may be necessary. The inspection of the QQ plots did not suggest a violation of the normal distribution assumption, yet a robust parameter estimation (5000 bootstrap samples) was performed in JASP, which led to similar results (see Supplementary Material, https://data.uni-marburg.de/handle/dataumr/240).
The multiple regression with work motivation as a dependent variable and the three scales planning, self-observation and reflection from the context of motivating oneself as predictors, as well as work experience and gender as control variables, is significant (F (5,197) = 12.55, p < 0.001). The coefficient of determination R 2 for the overall model is 0.242 (corrected R 2 = 0.222) and according to Cohen corresponds to a medium goodness-offit. Work experience and self-observation are significant predictors of work motivation (work experience: β = 0.032; t (197) = 2.84; p = 0.005; self-observation: β = 0.891; t (197) = 3.13; p = 0.002). Meanwhile, gender, planning and reflection do not significantly influence work motivation (see Appendix D: Table A3). Preconditions for the regression were checked and can be found in our Supplementary Material (https://data.uni-marburg.de/handle/ dataumr/240). In this case, too, the Shapiro-Wilk test shows a violation of the normal distribution assumption, which, however, was not confirmed by checking the QQ plots. Nevertheless, a robust parameter estimation (5000 bootstrap samples) was also carried out here in JASP, which led to similar results (see Supplementary Material, https: //data.uni-marburg.de/handle/dataumr/240).
The multiple regression with well-being as a dependent variable and the three scales planning, self-observation and reflection from the context of self-care as predictors, as well as professional experience and gender as control variables, is significant (F (5,197) = 12.33, p < 0.001). The coefficient of determination R 2 for the overall model is 0.238 (corrected R 2 = 0.219) and according to Cohen [30] corresponds to a medium goodness-of-fit. Selfobservation alone is a significant predictor of well-being (β = 0.571; t (197) = 2.3; p = 0.022). Work experience, gender, planning and reflection have no significant influence on well-being (see Appendix D: Table A3).
The multiple regression with recovery ability as a dependent variable and the three scales planning, self-observation and reflection from the context of self-care as predictors, as well as work experience and gender as control variables, is significant (F (5,197) = 5.92, p < 0.001). The coefficient of determination R 2 for the overall model is 0.131 (corrected R 2 = 0.109) and according to Cohen [30] corresponds to a low-medium goodness-of-fit. However, none of the predictors used is significant (see Appendix D: Table A3). However, in the more parsimonious regression analysis without gender and work experience, with a similar model fit (F (3,251) = 13.9, p < 0.001) and a coefficient of determination of 0.143 (corrected R 2 = 0.132), planning is shown to be a significant predictor of recovery ability (β = 0.613; t (251) = 2.23; p = 0.026). All other variables have no significant influence on recovery ability (see Appendix D: Table A3).
The multiple regression with assertiveness in conversations as a dependent variable and the three scales planning, self-observation and reflection from the context of communicating as predictors, as well as professional experience and gender as control variables is significant (F (5,197) = 10.68, p < 0.001). The coefficient of determination R 2 for the overall model is 0.213 (corrected R 2 = 0.193) and according to Cohen [30] corresponds to a medium goodness-of-fit. However, none of the predictors used is significant (see Appendix D: Table A3). However, in the more parsimonious regression analysis without gender and work experience, with a similar model fit (F (3,251) = 22.6, p < 0.001) and a coefficient of determination of 0.213 (corrected R 2 = 0.204), reflection is shown to be a significant predictor of assertiveness in conversations (β = 0.379; t (251) = 2.46; p = 0.015). All other variables have no significant influence on assertiveness in conversations (see Appendix D: Table A3).

Study 2
In the second study, a further shortening was made for a final version of the questionnaire by means of factor analysis and the revised questionnaire was validated by means of convergent and discriminant questionnaires.

Participating Teachers
The survey modalities, participation requirements and recruitment were similar to those in Study 1. In addition to the raffle of vouchers as an incentive to participate, feedback on individual self-regulation competence in comparison to other participants was also offered, as well as an exercise booklet with suitable self-regulation strategies.
Data were collected from 11 May to 22 November 2021. The questionnaire battery was the same for all participants and the order of items within a measurement instrument was randomised.
In Study 2, 356 of 765 data sets were included in the analyses. In total, 402 participants were excluded according to pre-registration due to incomplete participation and 7 participants as outliers. There were no exclusions due to suspicious response patterns. Seven participants were excluded because they stated that they were not teachers.

Instruments
Demographic variables. With the help of a self-constructed questionnaire, the variables age, sex (gender), professional experience as a teacher (in years), employment relationship and type of school to be taught were recorded analogously to Study 1. In addition, internship and professional experience of student teachers (in months) was also recorded.
Self-regulation. Self-regulation was measured with the Marburg Self-Regulation Questionnaire for Teachers (MSR-T), which was shortened to 55 items in Study 1. For each of the four contexts, sum scores were again calculated for all three processes. Internal consistencies of the scales of the MSR-T with 27 items (after further shortening by CFA) can be found in Table 3.
Self-efficacy. The teacher self-efficacy scale (LSS) by Schmitz and Schwarzer [31] was used to assess teachers' self-efficacy. This scale comprises ten items and is intended to capture the teachers' subjective certainty that they can successfully manage demanding or difficult tasks, even when problems arise. The areas of professional performance, professional development, social interaction with students, parents and colleagues as well as dealing with professional stress, were covered. The ten statements were rated on a four-point scale from 1 = "not true" to 4 = "true exactly". A higher value on the scale thus corresponds to high self-efficacy. The scale's quality criteria were reported to be satisfactory, with Cronbach's α-values ranging from 0.76 to 0.81 and test-retest reliability at a test-retest interval of one year being 0.66 [31]. In our study, the test has an internal consistency of α = 0.75.
Occupational stress. To examine subjective feelings towards work, the job strain scale (BB) by Enzmann and Kleiber [32] with 15 items was used. This scale was divided by the authors into three content areas: Excessive demands, feelings of control and job satisfaction. Excessive demands were to be measured with six items, the experience of control with three items and job satisfaction with six items. All 15 statements were to be assessed on a five-point scale. The authors recommended the arithmetic mean of the answers as the total value of job strain. In our study, the questionnaire has an internal consistency (Cronbach's-α) of 0.89. Table 3. Statistics on the Scale Reliability of the different Scales from Study 2. Self-and motivation regulation. The present study used the previously unvalidated German translations of the Motivation and Engagement Scale (MES) by Liem and Martin [25] to survey cognitive and behavioural dimensions of motivation and engagement in students and working persons (here teachers). Parallel versions of the scale were used for both groups, differing only slightly in the wording of the items. The 44 items are divided into four main areas (Adaptive Cognitions, Adaptive Behaviour, Maladaptive Cognitions, Maladaptive Behaviour) and eleven subordinate factors. The original English scale has satisfactory quality criteria, as Cronbach's α-values of 0.77 to 0.79 were observed as indicators of internal consistency and fit indices of the confirmatory factor analysis as indicators of a good data fit to the theoretical factor structure [25]. The participants were each asked to rate 44 statements on a seven-point Likert scale (from "not at all true" to "completely true"). The study showed satisfactory internal consistency for the overall scale (teacher version: Cronbach's α = 0.89; student version Cronbach's α =.93).

Context
Work-related coping behaviour. The present study used the short version of Menge and Schaeper's Work-Related Behaviour and Experience Pattern (AVEM) [20] to assess occupational self-regulation by means of 13 items. In contrast to the original AVEM scale, the short version only includes the two areas "occupational commitment" and "resilience". The authors justified the omission of the third domain "emotions" by stating that other research had shown that the two domains "professional commitment" and "resilience" were sufficient to capture the four patterns of work-related behaviour and experience. Moreover, there is a conceptual confound with possible outcome variables such as job satisfaction and emotional exhaustion. For the two areas, the authors selected the characteristics "Subjective meaningfulness of work" and "Professional ambition" for "Professional commitment" and "Distancing ability" and "Resignation tendency (in case of failure)" for "Resilience", which were able to clarify the most variance in each case. The scale thus consists of 13 items distributed across four characteristics, which in turn can be subordinated to the two domains. To answer the 13 statements, the subjects were asked to use a five-point Likert scale. The authors reported satisfactory goodness criteria with a Cronbach's α of at least 0.77 for each of the four characteristics [20]. In our study, the scales also had satisfactory goodness criteria (Cronbach's α of 0.76 to 0.83).
Using the MPlus program (version 8.6 [33]), a confirmatory factor analysis was carried out with the items in order to check the factor structure on the one hand and to further reduce the items and design an economic final version of the questionnaire on the other.
Convergent and discriminant validity was tested by calculating the correlations of the MSR-T sum score with the above instruments. Since the LSS [31], AVEM [20], BB [32] and MES [25] violated the normal distribution assumption, Spearman's rank correlation was used.

Results
Sample. The characteristics of the sample can be found in Table 1. There were N = 279 women (age M = 32.94 years, SD = 10.36 years), 79 men (age M = 35.15 years, SD = 10.56 years) and one person who identified as diverse (age = 30 years) who participated in the study. Overall, 1.7% of the participants were student teachers, 66.7% were trainee teachers and 31.7% were teachers. On average, participants scored 3.96 out of 5 points on the abbreviated MSR-T in the context of teaching, 3.50 points in the context of motivating oneself, 3.45 points in self-care and 3.82 points in communicating. Trainee teachers had lower cumulative scores than teachers who were already working.
Final version of the MSR-T. For further economisation, CFA was used to identify additional items that could be removed from the scale. Items with factor loadings < 0.40 were removed from the respective scales, whereby additional care was taken to ensure that each scale (a) still consisted of at least 4 items and (b) was represented by as many different item contents as possible.
A recalculation of the internal consistency of the scales also shows acceptable to good values [29] of α = 0.715-α= 0.825 (subscales, Table 3) for the final version and excellent values for the context-specific total scores: α =.909-α = 0.927 (Table 4). A total of 27 items remain in the scale (see Appendix E, Table A4), which is also used to examine the factorial structure of self-regulation.
Investigation of the factor structure. Finally, the factor structure of self-regulation was examined. On the one hand, the considerations of the theoretical framework model (see Figure 1), which assume a division into adaptive and maladaptive strategies in the three process areas, were to be examined. On the other hand, it was to be answered whether self-regulation is a general overarching competence or whether self-regulation can be pronounced differently in different areas, which in turn has different implications for training. As already described in point 2.1.3, the data of the participants from Study 1 and 2 were pooled for this purpose. Different models of the structure of self-regulation that can be derived from theory were tested, with model 3 representing the assumptions from the theoretical framework model:

1.
A one-factor model of self-regulation, which postulates self-regulation as a general and cross-domain competence.

2.
A three-factorial model that maps the three processes of planning, monitoring and reflection, but does not take into account the subdivision into different areas.

3.
A domain-and process-specific model with a total of 24 factors that map the adaptive and maladaptive strategies of the three process areas of planning, monitoring and reflection in the four contexts of teaching, motivating oneself, self-care and communication.

4.
A bi-factor model that assumes two bi-factors (adaptive and maladaptive) in addition to model 3.
Neither the single self-regulation factor model (Model 1) nor the model differentiating processes only (Model 2) showed acceptable fit indices (CFI value of 0.90 or higher is often considered an indicator of acceptable fit [34]. TLI value of 0.90 or higher is often considered an indicator of acceptable fit [35]. RMSEA values between 0 and 0.05 indicate an excellent fit, values between 0.05 and 0.08 indicate a good fit, values between 0.08 and 0.10 indicate an acceptable fit and values above 0.10 indicate an inadequate fit [36]. SRMR values from 0 to 0.05 are considered a good fit, while values above 0.10 may indicate an inadequate fit [37]). This indicates that, on the one hand, a further differentiation into the different domains makes sense, since self-regulation does not seem to be a cross-domain competence. The only acceptable model fit is for Model 3 and Model 4 ( Table 5), which is an extension of Model 3 by two bi-factors (adaptive and maladaptive). Both models show that self-regulatory competencies can vary across different domains and that a division into the three processes is useful. Furthermore, they show that the items of the three process areas can additionally be divided into the subscales adaptive and maladaptive strategies. Accordingly, in addition to the overall MSR-T score, the individual subscales of the MSR-T are also meaningful and can be used for an individualised profile of strengths and weaknesses.

Discriminant and convergent validity.
To test validity, the MSR-T was administered with the measures of Occupational Stress (ρ = 0.436, p < 0.001), the Teacher Self-Efficacy Scale (ρ = 0.469, p < 0.001), the Motivation and Engagement Scale (ρ = 0.707, p < 0.001) and the short version of the AVEM (Subjective Meaningfulness subscale ρ = 0.046, n.s., Professional Ambition ρ = 0.212, p < 0.001, Distancing Ability ρ = 0.338, p < 0.001, Resignation Tendency to Failure ρ = 0.416, p < 0.001) correlated. The results can be seen in Table 6. As expected, the correlations are significant and support the validity assumption of the new measurement instrument. As expected, the subscale Subjective Importance of the AVEM, which is defined as "Importance of work in personal life" [20], and thus has no overlap with self-regulation in terms of content, is not significant. The MSR-T has the highest correlation with the MES, which comes closest to the construct of self-regulation.

Discussion
In this study, we developed a new questionnaire to assess self-regulation in the teaching profession and validated it using a sample of working teachers, trainee teachers and student teachers from a German-speaking population. For this purpose, we examined the psychometric properties of a preliminary version with 79 and 55 items in the context of two online studies. The final version of the MSR-T consists of 27 items that can be used contextspecifically and divided into three process areas and adaptive and maladaptive strategies.

Factor Structure
The examination of the factor structure with the data from Study 1 and 2 shows that both the three-way division of the self-regulation process into planning, monitoring and reflection assumed in the theoretical framework model and the division of the three processes into adaptive and maladaptive strategies make sense: Model 1, which assumes a single factor of self-regulation, does not show acceptable fit indices; the same applies to Model 2, which only assumes a differentiability into the three processes of planning, monitoring and reflection, but not into different areas. This indicates that, on the one hand, a further differentiation into the different domains makes sense, since self-regulation does not seem to be a cross-domain competence. Self-regulation seems to be context-specific, which is underlined by the model fit of the 24-factorial model, which assumes the six process areas in the four contexts (research question 3).
For the evaluation, this means that both the total value of the scale and the subscales can be interpreted and used to depict an individual profile of strengths and weaknesses. If we want to promote self-regulation in the future, we must not only improve the acquisition and application of adaptive strategies, but also target the reduction of already ingrained maladaptive strategies.

Reliability and Validity
Across both studies, there are acceptable to excellent internal consistencies for the various context-specific total scores as well as for the subscales (Research Question 1). Validity was tested in Study 2 by means of correlation with convergent measurement instruments and a divergent subscale of a measurement instrument. The correlations are significant and in the expected direction, so that it can be assumed that the questionnaire is an adequate measure of the construct under investigation (Research Question 2).

Limitations and Future Directions
Our studies were designed as a first step in the evaluation of a new measurement tool, so despite several strengths, such as the inclusion of teachers with low and high professional experience, some limitations should be discussed.
In terms of recruitment, there is a clear over-representation of female teachers in the study (80% female). Overall, however, just over a quarter of German teachers are male, so the sample is not fully representative in terms of gender ratio. As other research has shown, men and women have different coping profiles [17,20], and a stronger representation of male teachers would be desirable in order to investigate possible differences in the frequency of use of self-regulation strategies. As more women felt attracted to participate in our study, measures should be taken to recruit more male teachers in followup studies. This could be achieved by tailoring the study application to male teachers. In terms of age, it should also be noted that this is still a relatively young sample, with few teachers with very high levels of professional experience, but also few students with no professional experience.
Since in the second study we offered feedback on self-regulatory competence in comparison to other teachers and a workbook as an incentive to participate, it cannot be ruled out that this led to self-selection of participants, as teachers with a self-perceived low self-regulatory competence and a desire for development may have felt more addressed by this offer than teachers who rated their competence as high.
Due to the cross-sectional design of both studies, we cannot make any assumptions about the stability of self-regulation over time. For this, longitudinal studies are needed, which would ideally also look at the development of self-regulatory competences over the course of a professional career and thus also allow conclusions to be drawn about the critical periods for the acquisition of these competences or examine the influence of self-regulatory competences on professional success more closely. It is conceivable that high levels of self-regulatory competence may predict career success, but also that excessive demands (as part of career success) may influence the use of self-regulatory strategies.
Furthermore, in both studies, only process-related items and hardly any other components (e.g., goals), which are assumed to be the "starting point" for situation-specific self-regulation in the theoretical framework model, were measured due to the already high number of measurement instruments.

Conclusions
This study presents the development and investigation of the reliability, validity and factor structure of the Marbug Self-regulation Questionnaire for Teachers (MSR-T). The MSR-T is a reliable and valid instrument for the contextual assessment of self-regulation. Both the subscales, which measure the processes of planning, monitoring and reflection within a context, and the context-specific total score can be interpreted. With 27 items, it is a very economical yet specific instrument that can be used to answer further questions. These include, in particular, testing the effectiveness of self-regulation training and investigating the development of teachers' self-regulation competencies over the course of their professional careers. The general context-unspecific wording of the individual items makes it possible to transfer the questionnaire to other contexts. It is also conceivable that the questionnaire could be used in other professions that require a high degree of planning and strategy application, and this should be explored in further studies. As self-regulation is a complex process that must always be assessed against the background of self-efficacy expectations, higher-level motivation and long-term goals, the present questionnaire is only suitable for mapping the cyclical process and should be extended in future studies to include existing measurement instruments for the above-mentioned constructs.

Institutional Review Board Statement:
The studies were conducted in accordance with the Declaration of Helsinki, and approved by the Institutional Ethics Committee of Philipps-University Marburg.
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
Publicly available datasets were analysed in this study. These data can be found here: [https://data.uni-marburg.de/handle/dataumr/240].

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.

Description of the Integrative theoretical framework model of self-regulation
The coexistence of different models of self-regulation shows how complex the mapping of this construct is. For this reason, an integrative theoretical framework model was developed for the construction of the questionnaire, which attempts to integrate common models and make process-specific strategies visible. The model described below does not claim to be exhaustive. Rather, it is intended to provide a framework or taxonomy of the most important processes that frequently play a role in regulation in teachers' everyday lives and that are at the same time individually measurable and changeable.
The core of the model is the cyclical process similar to Pintrich's model [13,38] of planning, introspection and reflection, each of which includes specific adaptive and maladaptive strategies as suggested by the Motivation and Engagement Wheel [25] and Boekaerts [26]. This process is influenced by personal and situational factors that ultimately lead to situation-specific goals. These may be, for example, correcting exams in a given time or to avoid failure in the implementation of a curriculum. Depending on how realistically achievable goals are considered in the current context, adaptive or maladaptive strategies are more likely to be employed. Different strategies are used depending on the context, as goals, self-efficacy expectations and the effectiveness of strategies individually differ in different contexts.
Self-regulation occurs at different levels. On the one hand, it can be at the execution level, where a concrete action is evaluated in detail and modified if necessary. On the other hand, at the strategy level, there may be a change in the chosen behaviour (strategy adaptation), or at the goal level, there may be a change or shift in the originally intended goals.

Translation of the participant information on self-regulation and the four contexts:
What is self-regulation?
In everyday life and at work, we always set ourselves goals that we want to achieve. For example, we might want to teach a certain amount of material by the end of the school year, reduce disruptions in class, or mark exams quickly.
For many of these goals, we already have established routines and automatically do what is necessary. However, we may encounter obstacles or problems along the way.
In order to achieve personal goals, we need to keep an eye on our thoughts, feelings and actions and adjust them as we go. This mundane but important process is called self-regulation.
The following questionnaire asks you to indicate how often you engage in certain behaviours or have certain thoughts that are related to self-regulation.
The questions relate to your professional activities and are divided into different areas:

When teaching
This includes activities such as teaching, explaining and educating, as well as preparing and following up lessons, designing and correcting examinations. Appropriate methods must always be selected and used. These may include, for example, targeted interventions to deal with classroom disruptions, appropriate didactic methods to convey lesson content, or planning and scheduling the correction of examinations.

When communicating with pupils, parents or colleagues and superiors.
Empathy is required to actively and positively shape important conversations. At the same time, such important conversations require the ability to make one's own point of view clear and to assert it in a compatible way. This could be a discussion with parents about their child's problems in class and possible solutions. Or a discussion with colleagues who need to be convinced about an important school project.

Self-care
In order to stay healthy in the face of the demands of everyday life, it is necessary to deal appropriately with stress and negative emotions such as anger, frustration or sadness. It is also important to assess one's own resources, recognise limits and take good care of oneself. This enables you to respond flexibly and proactively to difficult or stressful situations and life events.
Self-care includes situations in which one counteracts stress and strain in a preventive and regenerative way through one's own actions and supportive thoughts. This can mean, for example, making time for things that are good for you, such as hobbies, regular exercise or relaxation. It also means encouraging positive thoughts and being kind to yourself.

Motivating yourself
There will be times when you have little desire or motivation to tackle the work-related activities ahead of you.
This could be because you're tired, a task is taking longer than expected, or you don't like doing an activity even though it's necessary. To maintain or increase your motivation, you can, for example, take precautions against distractions, divide tasks, reward yourself for your work or increase interest by highlighting exciting aspects.
These four areas are interrelated and are important core activities for teachers. The following questionnaire therefore refers to your thoughts and behaviour when teaching, communicating, caring for yourself and motivating yourself.     Note. The item labels are composed of the context (U = teaching, M= motivation, K = communication, S = self-care), the respective processes (PL = planning, MO = monitoring, RE = reflection) and another abbreviation (e.g., ADS3). The ending "recode" indicates that the item has been inverted. An "x" marks all items that were excluded in the course of the analyses.    Note. The item labels are composed of the context (U = teaching, M = motivation, K = communication, S = selfcare), the respective processes (PL = planning, MO = monitoring, RE = reflection) and another abbreviation (e.g., ADS3). The ending "recode" indicates that the item has been inverted. An "x" marks all items that were excluded in the course of the analyses.