Abstract
Peer tutoring as an effective inclusion strategy is decades old and widely used in integrated classrooms, particularly for socialization (Ernst & Byra, 1999). Surprisingly, very little research has been done on academic outcomes of peer tutoring, especially in physical education classes. The current research hints at positive results, but is inconclusive. Thus far, peer tutoring has proven to be a highly effective method for increasing activity levels in children with severe and moderate disabilities (SMD) (Klavina & Block, 2008), deaf students (Liebermann, Dunn, van der Mars & McCubbin, 2000), and visually impaired students (Wiskochil, Lieberman, Houston-Wilson & Peterson, 2007), but only slightly effective in improving the motor performance of autistic students (Ward & Ayvazo, 2006) in inclusive elementary physical education classes.

Introduction
Peer tutoring is a type of collaborative learning strategy in which students support each other’s learning rather than relying solely on an adult teacher or aid for instruction. Peer tutoring is often used in classrooms as a recommended inclusion strategy for students with disabilities. Research suggests that peer tutoring increases learner achievement and has been an effective system for promoting student’s social and cognitive learning since 1932 (Ernst & Byra, 1999). Peer tutoring in general physical education (GPE) classes typically involves accountability to one another through interaction involving instructions, demonstration, and physical assistance, and feedback such as praise, encouragement and error correction. Opponents of peer tutoring in GPE classes worry that peer tutoring decreases practice time and physical engagement, increases chance of giving and receiving incorrect feedback, and increases opportunity for conflict or small-talk (Ernst & Byra, 1999). Proponents of peer tutoring in GPE classes agree that “by observing the performance of the doer, comparing the performance against the criteria, drawing conclusions about the accuracy of the performance, and giving appropriate feedback, the observer better understands the process of learning a task” (Ernst & Byra, 1999, p. 26). The following is a review of recent literature regarding the effectiveness of peer tutoring in increasing activity levels and improving skill performance of students in inclusive elementary physical education classes.

Research Method Evaluation: Data Collection and Analysis
The strongest aspect of Ward and Ayvazo’s (2006) study of two autistic kindergartners, in which Ben and Peter were paired with typically developing peers, Holly and Sarah, to assess the effectiveness of classwide peer tutoring (CWPT) “during a 26-lesson motor skill unit focusing on catching and striking,” (p. 235) is the data collection (Ward & Ayvazzo, 2006). The procedures for data collection were well planned, organized, and executed. First, the observers were trained and evaluated, and “were required to meet a criterion level of 95% accuracy or better” (Ward & Ayvazo, 2006, p. 236). They knew exactly what to look for - level of engagement and correct motor performance - and how to record the data accurately. The amount of data gathered was substantial in that students were observed over a 13 -week, 26-lesson plan unit. Also, each lesson was designed with the same format so as to prevent confusion and random results. Lastly, the entire class was given instruction in and practiced CWPT just before the study began.
Two separate studies that are equally strong proponents of peer tutoring have similar strengths. Research by Wiskochil et al. (2007) on the effects of trained peer tutors on the academic learning time – physical education (ALT-PE) scores of visually impaired children in a GPE classroom found a mean increase in scores of 20.8% (Wiskochil et al., 2007). Likewise, research by Lieberman et al. (2000) on peer tutors’ effects on activity levels of deaf students in inclusive physical education classes revealed that both the deaf students and tutors increased their moderate to vigorous physical activity (MVPA) from 19% to 41.5% (Lieberman et al., 2000). In both studies, data collection procedures and analysis were as strong and reliable as both Ward and Ayvazo’s (2006) research. Liebermann et al. (2000) collected data using a widely accepted System for Observing Fitness Instruction Time (SOFIT) method and recorded 32 observations over a 5-month period. Interobserver agreement (IOA) ranged from 89.75% to 91.25% but probably could have been higher had the classes been videotaped like they were in Wiskochil et al.’s (2007) study (Liebermann et al., 2000). Interobserver reliability in Wiskochil et al.’s study was 95.8% although only random checks were made to arrive at this data (Wiskochil et al., 2007). Wiskochil et al. (2007) used the widely accepted ALT-PE coding sheet which allowed them to record whether students were motor engaged (ME) and not motor engaged (NE) (Wiskochil et al., 2007). Additionally, a sighted classmate “who was determined to be the closest in motor skills to the participant” served as “a comparison for the amount of time that sighted students spent engaged in physical education” (Wiskochil et al., 2007, p. 343). Data analysis could have been affected in Wiskochil et al.’s (2007) study because each observation was in different activities ranging from soccer, volleyball, dance, basketball, kickball, ultimate Frisbee, to wrestling and archery (Wiskochil et al., 2007). However, researchers evaluated the differences between “open-skilled” and “closed-skilled” activities and found that “the increase of ALT-PE scores during open-skill activities was 16.6% and…during closed-skill activities was 29.4%” so “peer tutors were effective in helping to increase the ALT-PE scores during both types of activities” (Wiskochil et al., 2007, p. 347). Lieberman et al.’s (2000) research did not account for the types of activities the students participated in, but did distinguish between activity levels coded “1 for lying down, 2 for sitting, 3 for standing, 4 for walking, and 5 for very active” (p. 24) and coded 4 and 5 were combined to give a total MVPA score (Lieberman et al., 2000).
Klavina and Block’s (2008) study also implemented very specific data collection procedures. Data was collected for 46 thirty-minute GPE classes in which 20 minutes was of peer-mediated instructional support with trained peer tutors and 10 minutes was of voluntary, untrained peer support. Data was separated into categories depending on persons involved, interactions, and behavior subcategories. Computerized Evaluation Protocol of interactions in Physical Education (CEPI-PE) was used to code the recorded sessions after two pilot studies validated it as an effective tool to analyze and describe instructional, physical, and social interactions. Due to the careful method of the study, observers were able to track differences of interactions with teachers, trained peer tutors, and classmates quite effectively.

Research Method Evaluation: Participant Selection and Setting
While data collection procedures are strong across the board, the setting and participant selection seems skewed and may have resulted in some biased conclusions. Ward and Ayvazo (2006) selected a K-8 charter school “specializing in inclusion placements for children with autism” (p. 235). That is problematic in that the teachers and the students already have a vast amount of experience with inclusion and are, therefore, more likely to be successful than a randomly selected school. Also, the “students were in classrooms that used peer tutoring, [but] it had not been used previously used in their physical education lessons in formal ways” (Ward & Ayvazo, 2006, p. 235). Studying students who are accustomed to peer tutoring and are already comfortable with inclusive classrooms may have also influenced the data. The study observed two of six autistic students, Ben and Peter, within a classroom of sixteen children. This classroom is unique in that nearly half the students are autistic, which is not a representative sample of most inclusive classrooms across the country. Clearly, this school is better equipped to handle the challenges of inclusion. The study states that it was a kindergarten class, but both Ben and Peter are reported as being 8 years old. This is problematic in that the comparison of 8-year-old autistic boys to typically developing kindergarten girls seems odd, at best. Lastly, the two typically developing peers were “purposely selected” by the investigator’s request for students who “would be good peer tutors for Ben and Peter” (Ward & Ayvazo, 2006, p. 235). Randomly selecting peer tutors or selecting average peer tutors would have provided more accurate results. Unfortunately, because of these choices, the conclusions of the study are inherently biased. The results are useful to specific schools that cater to autistic children, but more studies need to be conducted if they are to make a more general statement that can be applied to schools across the nation.
Participant selection is also a weakness that may be skewing results in the studies by Liebermann et al. (2000) and Wiskochil et al. (2007). In both studies, participants were “purposive, meaning criteria were used to select children who were representative of…students and classmates...in inclusive elementary schools throughout the country” (Lieberman et al., 2000, p. 22). In the study of deaf students, eight students, half boys and half girls, four from Grade 4 and four from Grade 5/6, were selected to work with peer tutors were matched based on gender and selected if he/she was already a classmate who demonstrated good behavior, wiliness to participate, successful completion of training, and had no previous close friendship with the deaf student (Lieberman et al., 2000). Here again, the problem with the selection of participants is that they all attend the same school, which is more likely to be better equipped to meet the needs of the deaf students. Additionally, both Grade 4 classes have two deaf students and the Grade 5/6 class has nine deaf students, although only four participated. In fact, “many of the hearing children knew a significant amount of sign language” (p. 34) before the study (Lieberman et al., 2000). However, “over 4,412 schools in the United States serve only one deaf child” (Lieberman et al., 2000, p. 22). Clearly this not a carefully selected sample of the schools across the U.S. and may have favorably altered the results.
The largest weakness with Wiskochil et al.’s (2007) participant selection is similar to Klavina and Block’s (2008) study in that they simply didn’t observe enough participants. Wiskochil et al. (2007) selected four participants: two with low vision, Sally Grade 3 and Roland, Grade 8, and two who were blind, Billy, Grade 6, and Betty, Grade 11 (Wiskochil et al., 2007). Obvious limitations arise in the data of such a small and varied sampling of students.
Klavina and Block (2008), while limited by the small sample, were careful about student selection, ensuring a fair sampling of the SMD population that had specific physical education goals of increased participation in GPE class written into in their Individual Education Plan (IEP). All three students with SMD had an adaptive physical education (APE) teacher with whom they were accustomed to working with. Peer tutors were selected by first asking for volunteers, then selecting the students “thought to be most appropriate based on their opinion on students’ skills and abilities to follow the requirements of the peer tutoring program” (p.136) from among the thirteen who returned permission slips (Klavina & Block, 2008). Questions do arise as to the applicability of the same results on randomly selected peer tutors. The nine peers were trained using the five TIP-TAP steps for peer tutoring, which includes instructions, demonstration, physical assistance, feedback, and error correction. Similar outcomes may also depend on the effectiveness of this training. The other classmates were not specifically trained but were given opportunities to voluntarily offer peer support after the peer mediated instruction time was over. “During teacher directed (i.e., baseline) conditions, the APE teacher provided assistance to students with SMD applying the same instructional strategies they used with their students before the study. No additional instructions were given to teachers” (Klavina & Block, 2008, p. 137). They concluded, “during peer-mediated and voluntary peer support conditions, the instructional and physical interaction behaviors between students with SMD and their peers increased, while social interaction remained low. The activity engagement time data increased for all target students throughout intervention sessions” (Klavina & Block, 2008, p. 132). Conclusions cannot yet be generalized to all GPE classes because of admitted limitations such as the number of participants, the presence of an observer, the requirements of the specific sport unit, recording difficulties, and the maintenance of a long-term peer tutoring program. However, the data presents strong evidence for the effectiveness of peer tutoring for increasing physical interaction of students with SMD.

Clinical Outcomes & Conclusion
Ward and Ayvazo (2006) concluded that “CWPT intervention was more effective than whole group instruction in meeting the goal of including [autistic] students into physical education…[and that] Ben and Peter performed more correct catches during CWPT than during regular whole group instruction” (Ward & Ayvazo, 2006, p. 241). However, “the results for the typically developing peers were mixed” (Ward, & Ayvazo, 2006, p. 233). The positive social outcomes of inclusion for autistic children have been extensively researched, but academic outcomes need more investigation. There is “initial evidence” (p. 233) to support the practice in physical education classes to improve motor performance, but that evidence is mixed (Ward & Ayvazo, 2006). On the other hand, peer tutoring showed no effectiveness in increasing social interaction of children with SMD, but did increase activity levels substantially. “The mean scores of activity engagement data…increased (by) 11-23.7%” (p. 154) which is significant given the limited physical abilities of the three target students (Klavina & Block, 2008). Similarly, both deaf and visually impaired students increased their ALT-PE or MVPA levels by at least 19% and as high as 41.5% with the use of trained peer tutors (Lieberman, et al, 2000; Wiskochil, et al, 2007). This is fantastic news for both physical education teachers and for special needs students who most often display lower physical activity levels than their typically developing peers. Overall, the research demonstrates strong data collection techniques and analysis. However, participant selection is problematic, at best. Lastly, the inclusion of intensive training for the peer tutors, the students’ awareness of the observer, and the lack of any long-term studies remain unresolved. Variances in the success of peer tutoring on activity levels in inclusive physical education classrooms are inevitable. Only with broader implementation and anecdotal records from teachers across the country will we truly know the effectiveness. One clarification that seems to be surfacing in the data is that peer tutoring benefits both the tutee and the tutor, which resolves any fears that tutoring may cause activity decline in typically developing students. Thus far, any physical education teacher confronted with low activity from both disabled and typically developing students in inclusive classes may consider implementing peer tutoring or CWPT as an effective method to increase activity levels of the entire class.

References

Ernst, M., & Byra, M. (1998). Pairing learners in the reciprocal style of teaching: Influence of student skill, knowledge, and socialization. The Physical Educator, 55, 24-37.

Klavina, A., Block, M. (2008). The effect of peer tutoring on interaction behaviors in inclusive physical education. Adapted Physical Activity Quarterly, 25, 132-158.

Lieberman, L., Dunn, J., van der Mars, H., & McCubbin, J. (2000). Peer tutors’ effects on activity levels of deaf students in inclusive elementary physical education. Adapted Physical Activity Quarterly, 17, 20-39.

Ward, P., Ayvazo, S. (2006). Classwide peer tutoring in physical education: Assessing its effects with kindergartners with autism. Adapted Physical Activity Quarterly, 23, 233-244.

Wiskochil, B., Lieberman, L., Houston-Wilson, C., & Petersen, S. (2007) The effects of trained peer tutors on the physical education of children who are visually impaired. Journal of Visual Impairment & Blindness, 339-350.