Developing Children’s Socio-Emotional Competencies Through Drama Pedagogy Training: An Experimental Study on Theory of Mind and Collaborative Behavior

Drama pedagogy training (DPT) is a drama-based-pedagogy focused on socio-emotional-learning (SEL) development, over academic or artistic. This study aims to see if DPT promotes theory of mind (ToM) and collaborative behavior in 126 French children aged 9-10 years old, randomly assigned to an experimental group (DPT), either a control group for 6 weeks. Post-tests showed large effects of training on ToM, F(1, 124) = 24.36, p < .001, η² =.16, and collaborative behavior, F(1, 124) = 29.8, p < .001, η² = .19. T-test showed significant differences on ToM (t = -4.94, p < .001) and collaborative behavior (t = -5.46, p < .001), higher for DPT. Effects of type of school and grade are discussed. Results confirm the hypotheses.

Nevertheless, a review on pretend play and its effects on child development (Lillard et al., 2013), found that correlations between pretend play and social skills, and pretend play and ToM were inconsistent, although most of the studies reviewed showed "important failures to replicate" (p. 16), which could be an explanation to the lack of positive results. In a similar way, the meta-analysis of Lee et al. (2015) studied the effect of drama based pedagogies (DBP) on children from preschool to adolescence, finding that DBP had a significant impact on educational achievement outcomes but that the impact on social skills was not significant, concluding that the belief that DBP might promote prosocial behaviors was unfounded.
On the contrary, it was supported that DBP had a significant positive effect on other children's social skills, like attitudes toward achievement and the so-called 21st century skills (which includes collaborative behavior).
Moreover, several drama-based trainings were considered to be out of the scope and thus weren't examined in the meta-analysis, we missed, for example, some well-known drama-based methodologies, such as the Speech Bubbles project or the Drama for Learning and Creativity (D4LC), which we believe could have influenced some of the negative results regarding DBP and its impact on social competencies. In the same line, the meta-analysis of Lee et al. (2015) explains that they had to exclude several studies from their analysis due to a lack of reporting information, demanding that researchers must better report their methods, as it is fundamental for replicability. They found that most of the studies analyzed presented important bias as they were mostly quasi-experimental studies that rarely presented matching control groups, being difficult to assure that DBP was or was not the cause of social outcomes enhancement, concluding that there is a need of better reporting the trainings and strategies used in order to be able to have consistent material to analyze and thus have consistent results.
To our knowledge, and based on these meta-analyses, except for the recent work from Goldstein and Lerner (2018) and a study on fantasy play (e.g. Thibodeau, Gilpin, Brown, & Meyer, 2016), studies that have been done in order to examine the impact of DPT methods and elements on socio-emotional competencies relies on studies a) without an active control group well matching the experimental one, b) where children volunteer to participate in the experimental group and so there's no randomised assignment, and c) where the researchers are involved in the training. In this line, several interventions and trainings are excluded from scientific analyses and thus, it remains difficult to provide consistent evidence of the impact of DPT for the development of socio-emotional competencies. The study of Goldstein and Lerner (2018) conducted on 97 pre-school children, aimed to provide evidence on this and thus controlled the variables shown before, seeking the impact of a pretend play group (PPG) on ToM, altruism, distress towards others and helping behavior, emotion matching and social behavior. The results showed that children in PPG group had lower levels of neutral social behaviors and higher levels of positive social behaviors than block play or reading groups. They also concluded that as children had lower levels of emotional control than the average, the pretend play training did not help with ToM and compassion, but it did help them with emotional regulation and social behaviors.
The importance of developing socio-emotional competencies through drama can be found in the origins of the emotional intelligence concept and in the fundamental role that socio-emotional competencies play in nowadays' education. Today, there are still several approaches when defining socio-emotional competencies, although they are all coming from the concept of emotional intelligence (EI; Salovey & Mayer, 1990;Mayer & Salovey, 1997). According to Mikolajczak, Quoidbach, Kotsou, and Nelis (2014), the definition of EI still presents some issues within the scientific community, as there is still no consensus among the researchers to identify what means to be emotionally intelligent. Consequently, they proposed to gather these capacities under Celume,Goldstein,Besançon,& Zenasni 709 the concept of emotional competencies instead of talking about an intelligence (emotional intelligence). In this line, they defined emotional competencies as "the practical capacity of identifying, understanding, expressing, regulating and using personal and other's emotions" i (Mikolajczak et al., 2014, p. 3) according to a particular context.
Thus, regarding children's development and its implication in educational settings, emotional competencies would be the skills that children should have in order to identify, understand, express, regulate and use their own emotions (intrapersonal level) and those of their classmates, teachers, friends and people who they interact with (interpersonal level), within the school setting. According to Mikolajczak et al. (2014), the interpersonal variant of the "understanding" level of emotional competencies, can be described as the practical capacity of understanding the causes and consequences of others' perceived emotions. In this line, this capacity would be directly related to the ability of understanding and reacting, in a cognitive, social or affective level, the emotion expressed by another, which is known in academic and popular literature as empathy. According to Davis (1980Davis ( , 1983 empathy is a complex construct that may be divided into (a) emotional empathy, and (b) cognitive empathy. The emotional branch or side, is related to the emotional process and response of an individual facing another's emotional reaction (e.g. Eisenberg & Miller, 1987), while the cognitive branch answers to the ability of understanding others' mental states (e.g. Baron-Cohen & Wheelwright, 2004). The former, is considered by some authors as emotional empathy or just empathy (de Vignemont & Singer, 2006), while the latter, considered as a social cognitive ability to adscribe mental states, such as thoughts, beliefs and emotions to another person is known in literature as ToM (Premack & Woodruff, 1978).
In this line, the development of socio-cognitive skills, such as ToM, and socio-emotional competencies, is intrinsically related to social functioning behaviors and competencies, such as teamwork/collaboration. The latter is considered to be part of the main 21st century skills to develop in education (e.g. Trilling & Fadel, 2009) altogether with creativity and other socio-emotional competencies. In fact, already in 1996, Garaigordobil, Maganto, and Etxeberría (1996) presented evidence on how the experiences lived by children at school had an important impact on children's prosocial behavior. Social competencies, are detrimental for children development and school is an important modulator of them. According to Burrus and Brenneman (2016), most teachers agree that teamwork/collaboration is a fundamental skill to be learned in the classroom; nevertheless, they are not always sure how to define it, and thus, it gets difficult to find useful tools in order to develop it.
In summary, DPT in schools is an active pedagogy focused on the social and emotional world of children.
Studies have presented evidence of the efficacy of DPT related tools and elements in the development of socio-emotional competencies such as ToM and collaborative behavior in elementary school children, although the presented literature regarding the different studies analyzed, concerning drama-based trainings and their impact on socio-cognitive and socio-emotional competencies, do not always present complete reporting, didn't provide a substantial randomised control, and the only one that did, was focused on preschool children (Goldstein & Lerner, 2018). Teachers agree with the fact of the importance to develop socio-emotional competencies and a collaborative behavior in children, although they are not wide aware of the kind of training that could help them reach this objective. Thus, the main objective of the current study is to conduct a randomised pre-post test study with matching control group in order to present evidence of the efficacy of DPT in promoting socio-emotional competencies, such as ToM skills and a collaborative behavior in elementary school children.
Child Theory-of-Mind Collaboration Drama-Pedagogy 710

Design of Study
We conducted a randomised study including pretest -intervention -posttest design with a matching active control group.

Participants
French educational system provides mandatory education for all children (foreign and French) from age 3 to age 16. Elementary school comprises 5 levels going from 1st grade (CP) to 5th grade (CM2). Within the

Materials Independent Variables
Trainings -DPT was created by selecting and adapting some drama pedagogy training activities proposed in the spanish program "Programa Juego" (Garaigordobil, 2003) considering that all the activities on the program were already validated. Some other drama pedagogy training activities were taken from the books of García-Huidobro (2004) and Boal (1989). The training consisted of 6 sessions, adjusted to French school time in order to avoid holidays that could interfere with the sessions. At the beginning of each session there was a transition/warming up activity where children expressed how they felt, leaving the standard activities of school and getting prepared for the training. Then, there was the main drama pedagogy training activity(ies), and finally a feedback time in where children expressed their opinions and feelings about the session. Each of the six sessions lasted between 60 to 70 minutes. The differences in the time of each session was due to respect their feedback time. Sessions 2 and 6 had a collective activity, same for both groups. In order to have a clearer idea of the activities within DPT, we will explain a game played in session three: Here, children were asked to create and play a scene with their faces and bodies while the other half of the class had to guess what was going on in the scene, giving details and explanations of their guessing (why they think this or that was going on in the scene). The idea is for children to be able to express something without using words.
CSG was created by selecting popular collective games, taken from Boulo and Olivier (2005)  of a game played in session three for this group, we will explain the collective game called messengers and combatants: Here, children are divided into two groups, one half has to deliver a message (scarf) and combatants have to avoid the message to be delivered by stealing the message. The idea is for children to work together into a strategy to arrive at the other side of the room and get the message delivered.  Plumb, 2001). Accordingly to the original version, in this task, children are presented a series of 28 pictures of a pair of eyes from different people expressing an emotion or a mental state. We projected this images on the board. Children were asked to look at the picture in each item and to circle the emotion or mental state that better described the look of the person, choosing it among four words displayed. Three of these words were distractor words while only one was the correct word that matched the emotion or mental state of the person in the picture. The task was scored by adding up the number of items correctly answered. For each image they had around 15 seconds, so between 7 to 10 minutes in total. Reliability score for our sample was acceptable considering our design (α = .64) although a higher reliability score is preferable.
Collaborative behaviour -In order to measure collaborative behavior, we used the prisoner's dilemma (PD) task from of Garaigordobil (1995). For PD, each child has to be sit in two rows back to back with another. They are given two different cards, a red one and a blue one to answer the dilemma. Each couple of children are partners in a robbery and have to decide if betray his/her partner to save him/herself or not. The objective of the game is to stay in prison the less time that is possible. The police gives them a treat. They have one minute to decide, and then pick the decision by raising the corresponding card. They can't look or talk to their classmates or they are out of the game and lose. Decisions are marked down. The results are said out loud. They cannot comment, and that they are making their decision again, and they can change their decision if they like. Say out loud the results. Explain again: "now you have 30 second to decide if you keep the same answer or if you change your answer. 30 seconds starts now… time is over raise you answer" Write down their answers and say it out loud.

Pre-Test Collaborative Behavior
After the trainings were started, in week 2 (T1), children were measured in collaborative behavior through the prisoner's dilemma game. Results presented a significant difference between the randomised groups (t = 2.79, p < .006); with the CSG group (M = 3.12, SD = 0.13) presenting significant higher mean scores than the DPT group (M = 2.62, SD = 0.13), after one week of started the trainings. Nevertheless, median and mode scores present no differences between the groups (Mdn = 3).
When searching for possible previous Collaborative Behavior differences by class, results showed no significant differences.
When analysing the effect of school on the collaborative behavior variable, measured through prisoner's dilemma, results showed significant effects, F(2, 123) = 7.6, p < .001, presenting a medium to large effect size

Post-Test Correlational Analysis
Pearson's correlation analyses showed significant weak correlations between ToM and collaborative behavior (measured through PD; r = .35; p < .001).

Post-Test ToM
As plotted in Figure 1, an ANOVA 2x2 model was run to see the effect of the type of training (DPT, CSG) on the variable ToM regarding the two times of measurement (T0, T3). Results showed significant large effects, F(1, 124) = 28.79, p < .001, η 2 = .19, with more important increases on DPT. Child Theory-of-Mind Collaboration Drama-Pedagogy 714

Post-Test Collaborative Behavior
As Figure 2 plots, a two way repeated measures ANOVA analysis was conducted in order to compare the effect of the training (CSG, DPT) on the variable collaborative behavior measured through prisoner's dilemma, considering the two times of measurement (pre-test and post-test). Results showed significant differences as well as large effect sizes, F(1, 124) = 74.83, p < .001, η 2 = .38, higher for DPT.

Interaction Effects
Type of training by class -As shown in Figure 3, ANOVA analysis presented a significant effect of the type of training on class, with a small to medium effect size on variable ToM, F(1, 122) = 6.32, p < .013, η 2 = .05, with higher means for DPT in both classes.
Celume, Goldstein, Besançon, & Zenasni 715 No significant effect and a small effect size on collaborative behavior-PD (η 2 = .03) was found by class. Figure 4 plots the significant effect and large effect size of the type of training on the type of school on ToM post-test scores, F(2, 120) = 6.83, p < .002, η 2 = .102, with higher mean scores for DPT on the three schools. Further analyses showed no effect of training by type of school on collaborative behavior PD post test scores, nevertheless small effect size was found, presenting scores of η 2 = .030 with higher means for DPT for three schools.

DPT training on ToM by school -
Considering the significant effects of the type of training on ToM scores by school type, further ANOVA analyses were made. Looking for interaction between the times of measurement and the type of school in variable ToM, results showed a significant interaction effect, F(2, 58) = 5.28, p < .008, with large effect size (η 2 = .15). Highest differences in pre-post scores were found for public (M = 18.54, SD = 3.9) and REP (M = 16.11, SD = 3.9) schools.

DPT training on ToM and collaborative behavior by class -Analysis showed no significant effect of DPT
training on ToM by class, reporting a small effect size (η 2 = 0.02). No significant effect of DPT training on collaborative behavior by class was found, no effect sizes were reported.

Discussion and Conclusion
Studies have shown that socio-emotional learning competencies can be taught and learnt at school (Durlak et al., 2011;Ebert, Hoffmann, Ivcevic, Phan, & Brackett, 2015) and at the same time, some researchers insist on the importance to train children as well as teachers on socio-emotional learning competencies, so the benefits can be positive in a personal way but also in a collective way (Fernández-Berrocal, Cabello, & Gutiérrez-Cobo, 2017). More specifically, some authors sustain that ToM and prosocial behavior are two main domains in the field of socio-emotional learning that are important to be developed among children and youngsters (Beauchamp & Anderson, 2010;Rose-Krasnor, 1997). As a recent meta-analysis showed, dramatic play seems to have a positive impact on both of these competencies, even though no rigorous studies have been conducted yet in order to corroborate these effects (Lillard et al., 2013). In the present study, we carried out a randomised protocol in which specific measures for both ToM and collaborative behavior were applied.
ToM can be defined as the ability to read and understand another's mental states through body, face and vocal expression as well as previous knowledge of the other (Goldstein, 2009). In our study, this ability was measured through the RMET-G in which children had to infer mental states through the observation of eyes zones pictures. 1991). A more homogeneous French socio-cultural background was present in the public school, with a more foreign socio-cultural background one in the priority school, thus, suggesting that public school children have a more direct access to French vocabulary because they speak the same language at home in contrast to priority school children whose parents (mother or/and father) do not speak French as a first language. We also support this idea based on the fact that private school, where children's socio-cultural background was very heterogeneous, did not exhibit significant differences with neither of the two other schools.
It was also found a significant difference between both grades' scores being higher for 5th grade than 4th grade. Nevertheless, this was kind of expected, and can be simply explained by the fact that as children grow up, they acquire more language skills, thus having 5th grade children a larger vocabulary than 4th graders.
Post-test analyses presented encouraging results regarding our hypotheses.
First, correlation analyses showed that there was a significant but weak correlation between ToM and collaborative behavior. This results suggest that it seems to be an important relationship between ToM and collaborative behavior competencies in children that occurs inside a DPT setting. Nevertheless, this relationship is not strong enough to ensure a clear correspondence of ToM with collaborative behavior, thus suggesting more investigation and precisions in this area that should be considered for further research addressing the links between the social competencies that might be enhanced in DPT.
Further on, when comparing the type of training, in order to see the impact of DPT on ToM, both ANOVA analyses and t-tests showed effects and significant differences of the type of training on ToM. These significant differences, were statistically higher for DPT group, suggesting that DPT enhanced ToM in both 4th and 5th grade children from all three schools. These results are consistent with previous studies that established a relationship between drama-related games and the development of ToM skills (e.g. Goldstein & Winner, 2012) even though, their results were achieved on a high-school sample. Type of training also showed an effect between pre-test and post-test, confirming post-test analyses. DPT group showed higher differences in scores than CSG, the latter presented a difference of pre-test and post-test scores that was almost invariable. These results are inside the scope of what we predicted, as children who participated in the DPT were trained in emotion and mental states identification through the different DPT methods and activities played. Sessions 1, 3, 4 and 5 were dedicated to emotional identification, expression and communication through play and mini-pretend-plays in where children had to choose an emotion or emotional state and to represent it in an invented collective scene. In this way, children were "obliged" to identify the different characteristics of emotions and the mental states they were to represent, and so learning was acquired through playing, using their minds and bodies to recreate, permitting embodiment, which helped with the understanding of the emotions and mental states. This idea, which defends the theory that embodiment is a direct contributor of the development of the cognitive process of empathy, in which ToM is a fundamental part, is supported by the work of several authors (e.g. Gieser, 2008;Sofia, 2016). Moreover, for some authors, mind cannot be separated from body, as the body is evidence of the mind (Bateson, 2000;Butterworth, 1994). In this means, body and mind cannot be divided because the body is the support of mind expressions.
Regarding the class level, the highest difference in ToM post-test is presented in 5th grade. While DPT group in 4th grade is slightly higher than CSG in 4th grade, scores get clearly higher for DPT in 5th grade. These results suggest that DPT would work better for 5th graders. In both cases, these results only reflect our specific sample and results should be taken carefully as they do not represent the whole French population. Besides, the fact Child Theory-of-Mind Collaboration Drama-Pedagogy 718 that DPT worked better in 5th grade for developing ToM can be only due to the fact that understanding mental functioning and ascribing mental states to others is in part, a cognitive ability, thus, it should naturally evolve along with children's cognitive development. In this means, 10-year-old children, that were already turning 11 by the time of the post-tests, have these cognitive abilities more developed. In fact, the theory of stages (Piaget, 1929) describes an evolutionary process of the intellectual capacities which establishes a transformation from concrete operational stage to formal operational stage at the age of 11. Moreover, regarding ToM, Perner and Wimmer (1985), confirm a natural evolution in the attribution of second-order beliefs in their work with children aged 5 to 10-year-olds.
Regarding the difference between pre-test and post-test repeated measures ANOVA found that DPT on ToM task had differences considering the type of school. As shown in results, PU school showed almost no increase between pre and post ToM tasks in comparison with REP and PR schools which post scores were between 2 and 3 points of difference.
On the other hand, collaborative behavior may benefit children's social relationships (Schellenberg, Corrigall, Dys, & Malti, 2015) as is encompassed in prosocial behavior which appears to be one of the key elements that benefits social interaction (Eisenberg, Spinrad, & Knafo-Noam, 2015). Moreover, in 2006, Barry & Wentzel (2006 established that collaborative behaviors develop a sense of security among friends and peers. In our study, we analysed collaborative behavior from two perspectives; an objective one through task performance and a more subjective one, through the observation of children's behaviors.
The state of the art showed that there were already differences among children from different schools, being always PR school children who had the more higher scores in collaborative behavior. After questioning teachers and directors from the three schools involved, there's still no clear answer that could explain this higher collaborative behavior in comparison to the other schools. No specific program was held before our intervention, and all three schools offered the same kind of extracurricular activities.
Girls showed to be more collaborative than boys. Observations in both pre-test and post-test showed that they tended to be more encouraging than boys in their groups, but also dynamics observed among groups constituted mostly of girls presented less anti-collaborative behaviors and showed a more federated way of work. This is consistent with Garaigordobil (2009) findings who established that girls are more collaborative than boys.
Even though group randomisation was made completely blinded and through an online free software, t-test results showed a significant difference between intervention groups. This means that there were already differences in collaborative behavior scores between the experimental (DPT) and the active control group (CSG).
CSG started the interventions presenting an advantage over the DPT group. When teachers were asked about differences they could perceive between both groups of their class after they were constituted, some of them agreed that DPT group was composed by "more difficult students than the other group" and that they felt the difference of working in the class with the other half (CSG) while the DPT session was being carried out. They reported they were able to "work better without that part of the class". Nevertheless, this wouldn't be sufficient to explain these differences, so we rely on the fact that the evaluation of collaborative behavior was carried on one week after the groups were constituted. With this timing in mind, we might infer that after a week of intervention, children already had created some particular relationships, group identification or bonding links Celume,Goldstein,Besançon,& Zenasni 719 with their classmates in their corresponding groups (DPT and CSG), which would had interfered on the second week, when the evaluation of collaborative behavior was carried on.
Finally, when testing our hypothesis, for collaborative behavior we found significant differences between the two groups, being higher for DPT. Both t-test analysis and ANOVA confirmed the hypothesis. This means that children in the DPT group increased their collaborative behavior during the intervention showing important differences when compared to their CSG classmates. PD task measure how the children respond and solve a situation in which they have to decide whether to collaborate to their partner or no. Children are offered the possibility to escape from prison, even though this means that their partner will stay in prison for some years. At the beginning, DPT group presented less collaboration, showing answers that in great majority chose not to collaborate and so escape from prison. After DPT was conducted, children appeared to develop more consciousness on the need to collaborate with others, and they decided to collaborate in the PD task, staying a few years in prison in order to not let their partner alone in prison for even more years.  Lee et al. (2015) who explained that even though drama-based pedagogies have been supported as being an effective way to foster prosocial behaviors studies analysed in their meta-analyses were not consistent.

Limitations and Future Perspectives
One of the main concerns we find in this study is related to the way instruments were used. The fact that the pre-and post-tests were identical could imply that there were learning effects on children for the post-tests and that increased scores were in part due to this fact. Nevertheless, we consider that if this was the case, this learning effect would have affected both groups equally as both groups answered the exact same tasks at the same time, so it shouldn't have a direct impact on the significant differences between the groups. Another limitation we appreciate is that the RMET-G task only measures one dimension of ToM, and moreover, its reliability can be discussed. Unfortunately there are not many tasks for measuring ToM in the range of age we worked with, as most reliable tasks are constructed for younger children and only few are translated and validated in France. It could also be argued that as DPT is a methodology focused on the work with the internal world of participants it might have a direct impact in socio-cognitive tasks such as the RMET-G. We propose, for future studies, an intervention considering different forms of the same task in order to discard the issue of learning effect, and the construction of a different scale that can measure better what was intended in the beginning of the study, or even the use of additional tests for measuring ToM in order to address the issue of reliability and measured dimensions of ToM. This also opens a possibility for further research considering the different approaches to emotional vocabulary and content that might be involved depending on the different socio-cultural backgrounds. On the other hand, the significant differences found in the pre-test scores for collaborative behavior should also be considered carefully. We propose a future intervention that considers Child Theory-of-Mind Collaboration Drama-Pedagogy 720 carrying on all variable testing tasks before any session had already started in order to avoid the bias of bonding in the different groups.
Another limitation we observed is related to the analyses chosen. Considering that differences among children were significant among the type of schools, it could have been interesting to run further analyses in order to control results regarding this issue. A future study should consider a multi-level analysis.
Finally, no follow-up study was conducted in order to measure both of the developed competencies. In this line, we propose to conduct a new study, probably using a longer intervention of DPT in order to see the possible follow-up impact on ToM and collaborative behavior scores, and why not, on other SEL competencies, as previous studies suggest.
In summary, DPT showed to be an effective method to develop ToM and collaborative behavior in 9 to 10-year-old children engaged in three different French school contexts, although further work is needed in order to see the long term impact of this pedagogy.

Funding
The authors have no funding to report.