The choice of a cooperative play-based intervention provided a naturalistic group setting
for elementary school students and enabled the collection of data within this setting through systematic observation. The objective was to observe the use of pro-social skills in real contexts while students were engaged with one another in play. Observation forms (Appendix D) used interval recording, i.e. the observer recorded an occurrence of the behavior if the behavior happened at any time during the interval. Interval recording data are presented as a percentage of the occurrences of the target behavior.
During the 1,3,5,7, & 9 th sessions, the researcher filmed the participants playing a 10 minute game. For interval recording, this 10-minute game was broken into five 2-minute intervals. The researcher reviewed these videos and charted the target behaviors. Target behaviors have been defined in this study and consist of three categories: self-regulating, cooperation, and communication. Each category is measured by specific behaviors. Self-Regulation = accepting consequences, demonstrating patience, adapting, using self-control through considered action. Cooperation = following instructions, helping others, accepting help, problem solving. Communication = listening, making a suggestion, describing thinking/needs/experience, offering encouragement. The researcher viewed each video three times. During a single viewing, behavior occurrence was charted for all students in the group for a single category (i.e. first video viewing = self- regulation behavior tracking, second video viewing = cooperating behavior tracking, and third video viewing = communication behavior tracking). An additional data collection tool used in this study was a behavior rating scale (Appendix G) completed by the students primary classroom teacher. Teachers completed the questionnaire for the first time in the week prior to the first session of the intervention. The teachers completed the questionnaire for the second time two weeks after completion of the intervention. This questionnaire used a 4-point Likert scale and the researcher considered the scale points to be ordinal in the sense that higher numbers represented higher values, but that the intervals between numbers were not necessarily equal. Instead of using terms such as strongly agree or strongly disagree, the questionnaire provided example statements for 1-4 to measure the teachers professional opinion of the students pro-social skill competence. Questions included: How often does the student appear to act impulsively? How well is the student able to manage his/her disappointment or frustration? How well is the student able to follow rules? How well does the student get along with peers? How difficult is it for the student to participate in group activities? Pro-social skills overlap and influence each other and are difficult to represent with a rating scale consisting of so few questions. However, the behavior rating scale questionnaire was specifically created to have few items in order to increase compliance with form completion. Items on the teacher questionnaire were selected based on research of the Skillstreaming Model for Elementary School Children (McGinnis, 2012) and the ProSocial Behavior Questionnaire (Weir & Duveen, 1981).
Data Analysis Data was analyzed along two pathways. The first considered information about pro- social behavior gathered from the observation forms. The second considered information about teacher perception of students pro-social competence gathered from the teacher rating scales. First, we will consider the methods for analysis of observation form data. These methods investigated whether students had demonstrated a change over time in their use of the targeted pro-social skills. The null hypothesis is: the use of pro-social skills does not increase in subsequent cooperative games. The alternative hypothesis is: the use of pro-social skills increases in subsequent cooperative games. A one-way ANOVA was used to determine if there was a significant difference in the occurrence of pro-social behavior between the games. This data reveals whether there has been a change over time in pro-social behavior that would not be attributed to chance at the 95% confidence level. Data was analyzed for each individual student separately, with the exception of three students who did not have sufficient observation data due to absences which resulted in participation in 2 or less games. Data was additionally analyzed for all students as one group, which provided a somewhat larger sample size and more data points to provide a more accurate representation. Finally, a t-test comparing the t-stat values for each game provide a picture of how pro-social behavior differed from game to game. This helps to determine where the increases occur and that it is not just one data set (or one game) that accounts for the increase in pro-social behavior occurrence (Appendix H). The researcher also considered Group 1 and Group 2s average percentage of pro-social behavior occurrence for each game, based on the observation form data. These averages can be reviewed to gain information about which specific behaviors were effected by which games and what was the trend over time. Linear regression was used to analyze the graphs results. This provides a picture of how the behaviors vary from game to game and how the frequency of pro- social behaviors is changing over time. The second focus for data analysis was to investigate how the pro-social behavior of participating students was perceived by their primary classroom teachers pre-intervention and post-intervention. The null hypothesis is: there is no increase in the teachers rating of student pro-social skill competency between repeated measures. The alternative hypothesis is: there is an increase in the teachers rating of student pro-social skill competency between repeated measures. The first analysis of the teacher rating scales was to determine whether the variance of the pre and post test are the same using an F-test two-sample for variances (Appendix H). The next step was to run a two-sample t-test assuming equal variance (Appendix H).
Findings Observation Form Data Results from the Anova data for each individual student indicated that we cannot reject the null hypothesis. In other words, when considering the data for each participating student separately, it cannot be disproven that the use of pro-social skills does not increase significantly in subsequent cooperative games. There is the exception of one student who did show a significant increase.
F(4,55)=1.068916 F critical for F(4,55) = 2.539689 F(4,55)=1.057692
F(4,55)=1.926432
F(4,55)=1.483307
F(4,55)=1.597237
F(4,55)=3.320157
F(3,44)=1.480186 F critical for F(3,44) = 2.816466 F(4,55)=1.450176
F(2,33)=3.198413 F critical for F(2,33) = 3.284918 F(4,55)=1.2241
Figure 1 - Anova Data for Each Student
However, when the Anova data is considered for all the students as a group, we do see a significant increase in the use of pro-social skills in subsequent games and are able to reject the null hypothesis and accept the alternative hypothesis. When considering the whole group, the use of pro-social skills increases in subsequent cooperative games.
ANOVA
Source of Variation SS df MS F P-value F crit Between Groups 102.3505 4 25.58763 8.357729 1.49E-06 2.387878 Within Groups 1711.408 559 3.061553
Total 1813.759 563 Figure 2 - Anova Data for all students grouped together to form one sample
Using the average percentage of behavior occurrence for Group 1 and Group 2 an analysis of the frequency of student pro-social behaviors across games was considered. In this analysis, each behavior was considered separately so that information could be gained regarding which particular behaviors were most utilized during a given game and which behaviors showed the most increase in occurrence across games during the intervention. This data is represented in figures 8 through 13 (Appendix H). Many of the behaviors show a large standard deviation between data points. This indicates that behaviors varied much between games. With a low linear regression value these behaviors are not following a predictable change. However, the majority of the behaviors do show an increase overall in the average percentage of occurrence. The linear regression results are seen in table 2 (Appendix H). From this data, it can be deduced that the four behaviors for the first grade group that can best be represented linearly are: adapting, listening, following instructions, and helping others. By analyzing the figures, it can also be seen that three behaviors (adapting, helping others, and accepting help) all increased tremendously from 20-76%, 16-96% and 8-92% respectively. The behavior that showed the least amount of improvement for the first grade group was problem solving. In the third and fourth grade group, it can easily be seen in the data that the students behavior in the last game differed significantly from previous games. There are several possible mitigating factors, an obvious one being student excitement about the last game and the end of the intervention. Students were also permitted to choose which game would be played in this last session. Because of the issues surrounding this game, this data set was excluded from the analysis. For the group comprised of third and fourth graders, we can deduce from the data that only two behaviors can best be represented linearly: listening and making a suggestion. However, by analyzing the figures we can see that other behaviors increased tremendously. Adapting, helping others, and accepting help increased from 40-70%, 28-75% and 4-75% respectively. The behavior that showed the least improvement in the group of third and fourth graders was Describing thinking, needs, and experience. Considering Group 1 and Group 2 as a whole, listening and following instructions were the two behaviors with the most frequent occurrence. The three behaviors that both groups displayed the least were: problem solving, describing thinking/needs/experience and offering encouragement. Figure 3 and 4 show the bar graph representation of the average percentage of times the students displayed a behavior in each of the three skill categories identified for the present study: self-regulation, communication, and cooperation. For the first grade group, the pro-social behaviors in the area of cooperation show the largest increase. During the first game played together, these students used cooperative skills an average of 29% of the time. During the last game played together, these students used cooperative skills an average 78% of the time. For the third and fourth grade group, the pro-social behaviors in the area of self-regulation show the largest increase. In the first game played together, these students used self-regulatory skills an average of 40% of the time. During the fourth game these students played together, they used self-regulatory skills an average of 72.5% of the time.
Figure 3 - First Grade Group
0 10 20 30 40 50 60 70 80 90 Amoeba Tag Stepping Stones Moonball Road Hazards Stepping Stones 2/18/2014 2/25/2014 3/4/2014 3/11/2014 3/18/2014 Average of Communications Average of Cooperation Average of Self Regulating
Figure 4 - Second Grade Group
Teacher Rating Scale Data Next, findings for the data from the teacher rating scales demonstrated that we cannot reject the null hypothesis because t Stat < t Critical, 1.244083585 < 1.980272249, and p > a, 0.215934425 > 0.05 (Appendix H). In other words, it cannot be disproven that there is no significant increase in the teachers rating of student pro-social skill competency between repeated measures. However, looking at Figure 6, we can see that for most students the mean score was slightly larger for the post test. It is also interesting to look at the pre and post test mean score by question in Figure 7. As a group, students were rated somewhat higher for every question, with Question 3 showing the most increase. Question 3 = How often does the student appear to act impulsively?
0 10 20 30 40 50 60 70 80 Amoeba Tag Stepping Stones Carry the Moon Road Hazards Amoeba Tag 2/18/2014 2/25/2014 3/4/2014 3/11/2014 3/18/2014 Average of Communications Average of Cooperation Average of Self Regulating
Figure 5 - Student Mean Score Pre and Post Test
Figure 6 - Mean Score by Question Pre and Post Teacher Rating Scale
Discussion The purpose of this study was to investigate the effects of a cooperative play-based intervention on the use of social emotional skills by elementary school children who have exhibited behavioral problems. The intervention was specifically designed to pair small group cooperative games with experiential learning, which utilized strategies such as video review, role play, and story-boarding. Three broad categories of pro-social skill were defined and measured: 0 0.5 1 1.5 2 2.5 3 3.5 Before After 0 0.5 1 1.5 2 2.5 3 Question 1 Question 2 Question 3 Question 4 Question 5 Before After self-regulation, cooperation and communication. Specific observable behaviors within each category were tracked and analyzed for trends and changes in frequency of use. Additionally, teacher perception of participating students pro-social skill competence was also measured pre and post intervention and considered in this study. The results of the intervention have been positive. Although statistically significant increases in the frequency of pro-social behaviors across games was not shown when considering individual student data, when data was considered for all students as a group, statistically significant increases were seen. We can conclude that the larger sample provided more data points and therefore a bigger picture of what is happening over time during the intervention. Over time, as students played more games together and participated in social emotional experiential learning sessions, they began (as a group) to more frequently display behaviors associated with pro-social skills. Specifically, both the first grade group and the third/fourth grade group showed encouraging increases in their use of behaviors that demonstrated they were adapting, helping others and accepting help as they played together. An interesting finding was that the first grade group showed the least gains in problem solving, whereas for the third/fourth grade group the least gain was seen in describing thinking/needs/experience. These results suggest that the younger students need more guidance and modeling in problem solving skills. The third/fourth grade students need more guidance in communicating and perhaps identifying their thinking/needs/experience during social interactions. It is useful to look at which games pulled most significantly from these skills. Carry the Moon and Road Hazards appear to be good games for working on describing thinking/needs/experience. Carry the Moon and Stepping Stones appear to be the games that required the most problem solving, however the percentages for this behavior stayed relatively low throughout the intervention, which indicates the need for more research and perhaps different games to encourage this behavior in students. Another interesting finding was that listening and following instructions were consistently high for both groups. Often, students with who exhibit behavior problems at school are said to lack these particular skills. The results from this intervention offer initial support for the possibility that this is not in fact where the skill deficits lie. Future research could explore how behaviors such as problem solving and describing thinking/needs/experience might impact classroom teachers perceptions about how well students are able to listen and follow instructions. Teacher perceptions were used in this study, however the researcher encountered several methodological problems in interpreting the results, foremost of which is the validity of the scale. The rating scale cannot be interpreted to offer assessement of students in distinct areas, but can perhaps offer evidence of the teachers overall perception of the students pro-social skill competence. The work of Spitzberg and Dillard (2002) suggests that pro-social communicative behavior is highly important to teachers perception of social competence in students. The work of Weir and Duveen (1981) provides evidence that different types of pro-social behaviors are positively related to each other. The fact that the teachers ratings between pre and post test did not vary significantly can be evidence that the scale provides an accurate picture of the teachers perception, since we would not expect significant variance in so short a time period. There are many limitations to highlight in this study. Foremost, the present study was not long enough, nor did it have a sufficient sample size to adequately investigate the research question. An additionally significant limitation was the amount of time the researcher/facilitator spent with students during the intervention. Fifteen minute sessions do not provide sufficient time to implement the planned games and instruction with fidelity. However, this was the maximum time permitted for the sessions in the students school day schedule. This is a frequently encountered barrier to social emotional learning interventions in the school building. It is difficult to balance the need for students to be present in the classroom for core curriculum instruction with the need for some students to receive targeted social emotional instruction, which research suggests would better enable them to access the classroom core curriculum. Providing social emotional interventions during recess, as in the present study, is one common solution which prevents students from missing classroom instruction. However, as demonstrated in this study, significant outcomes are unlikely to occur within a few weeks or months of intervention. Therefore, in addition to the need for empirically validated interventions for increasing social competence in students, there is another key research need. We need evidence to guide decisions about how to provide these interventions in a school day. An important future research question would explore whether pulling students from their classroom during the day to provide these interventions does in fact disrupt academic instruction and student learning or whether the benefits of those interventions might perhaps translate into greater academic success despite missed class time. A strength of the present study is its contribution to the evidence that cooperative play can be an effective pedagogical tool for pro-social skill education. Future studies might explore how cooperative play could be integrated throughout students school days, for example on the playground or woven in to interdisciplinary classroom activities. Another strength of the study is the data gained from the observations, which helps to create a picture of how pro-social behaviors are employed by students during the various cooperative games. We can see which behaviors are most responsive to the intervention. By also identifying the behaviors that are least responsive, we can make recommendations for future work in these areas. In conclusion, the experience of this intervention and research taught me a great deal about the importance of helping children practice pro-social behaviors in a natural social setting. I observed shared regulation at work as students played the cooperative games together and I saw how students developed both pro-social skills and relationships through play. I also observed the way play generated an enthusiasm for learning social emotional skills and motivated students to participate. The results of the study reiterate for me the importance of keeping kids with behavior problems in settings where they can play with socially adept peers. Just as the different types of pro-social behaviors appeared to be positively related to each other, individual students seemed to follow the group trend in frequency of pro-social skill usage.