The Games They Play: Observations of Children with Autism Spectrum Disorder on the School Playground

Abstract

The playground may be an important context to examine the social functioning of children with autism spectrum disorder (ASD). Previous literature on playground peer engagement has used quantitative methods, but there is limited research using qualitative observations to understand the nuances of playground behavior. Using a mixed-methods approach, 55 elementary school-aged children with ASD who are primarily included in general education settings were observed on the school playground using the Playground Observation of Peer Engagement. Quantitative and qualitative data were examined using a mixed methods approach. The results showed that children with ASD: engage in solitary and peripheral activities; demonstrate appropriate initiations and responses to peers; display self-stimulatory, motoric behaviors most frequently during solitary activities; and often have neutral affect on the playground. These findings suggest that intervention and supports for children with ASD may be important to deliver at recess to address peer engagement.

Autism spectrum disorder (ASD) is a pervasive developmental disorder characterized by impairments in social interactions, language and communication, and the presence of restricted and repetitive behavior (American Psychiatric Association, 2013). Social impairments may be in the areas of social-emotional reciprocity, nonverbal and verbal language, and forming and maintaining relationships (American Psychiatric Association, 2013). Children with ASD may be unaware of their social environment and experience greater loneliness (Bauminger and Kasari, 2000; Bauminger et al. 2003), have flat affect (American Psychiatric Association, 2013; Begeer et al., 2008; Loveland et al., 1994; Samson et al., 2014; Strid et al., 2013), and experience a number of challenges with perspective taking and sharing emotions with others (Bauminger et al., 2010; Tager-Flusberg, 1999).

Social interactions for children with ASD are particularly significant in school settings. Social communication deficits (initiating, responding, adjusting to different interaction styles) may affect children with ASD in the classroom (Kangas et al., 2011) as well as with peers on the playground (Calder, Hill, and Pellicano, 2012; Bauminger and Kasari, 2000). Children with ASD often are less connected in peer social networks and experience less acceptance, companionship, and friendship reciprocity (Chamberlain et al., 2007; Bauminger et al., 2010; Tager-Flusberg, 1999). Emotional competence (i.e. expression, perception, responding, and understanding) also varies for children with ASD based on severity, context, age, and cognitive ability (Begeer et al., 2008). These social outcomes are important as they may affect activity selection and peer engagement on the playground.

Research using playground observations have found that children with ASD often are on the periphery of activities (Kasari et al., 2011), have decreased interactions, are less physically active, and gravitate to adult figures over their peers (Lang et al., 2011; Memari et al., 2015). Bauminger et al. (2003) demonstrated that in comparison to typically developing children, children with ASD spend only half the time in a social interaction during recess or snack time. In addition, Locke and colleagues (2016) reported that children with ASD spent approximately 30% of recess in solitary activities and 40% of recess in joint engagement with peers. Other studies have found that children with ASD often exhibit more stress during interactions with peers and decreased social motivation during free play (Macintosh and Dissanayake, 2006), which may negatively impact their social communication and relationships with peers (Corbett et al., 2014; Schupp et al., 2013). While these findings quantify the amount of time children with ASD spend with peers on the playground, more evidence is needed to determine what activities and behaviors children with ASD actually engage in on the playground.

The playground setting is important, as it is an opportune time for children to socialize, exercise, gain play skills, and form friendships (Couper et al., 2013). Recess is a critical period for children with ASD to practice social skills with their peers and often is where children with ASD display self-stimulatory behaviors (Lang et al., 2011; Couper et al., 2013; Samson et al., 2014). Commonly observed self-stimulatory behaviors include rubbing, rocking, vocalizations, flapping, smelling, squeezing, and chewing (Mays et al., 2011). Self-stimulatory behaviors may interfere with social communication and peer engagement, as these behaviors may be difficult to manage during times of stress (Mays et al., 2011). It is vital to consider how the playground can positively or negatively impact a child with ASD as the child may be excluded from play with peers (Couper et al., 2013). Negative treatment by classmates may further exacerbate social communication challenges and self-stimulatory behavior (Ochs et al., 2001). Understanding the playground activities and behaviors of children with ASD may point to the ways in which school personnel and providers can facilitate a more inclusive and supportive playground context that may complement strategies used in the classroom.

Many studies have documented the playground behavior of children with ASD (Dean et al., 2017; Kasari et al., 2015; Kasari et al., 2012; Kretzmann et al., 2015; Locke et al., 2016). Locke and colleagues (2016) quantitatively measured peer engagement using playground observations to evaluate how the percentage of intervals spent in solitary and joint engaged activities differed between children with ASD and their peers. They found that children with ASD spend more time in solitary activities and less time joint engaged compared to typically developing children. While quantitative research measures the amount of peer engagement of children with ASD during recess, there is a need to use qualitative measures in order to understand the nuances of playground engagement. Few studies have qualitatively described the activities of children with ASD on the playground during recess. Recently, Dean and colleagues (2017) used mixed methods to explore gender differences in the social behaviors and activities of children with ASD on the playground. They found that girls with ASD spent most of their time in joint engaged activities with peers but also frequently moved to different groups, whereas boys with ASD primarily engaged in solitary activities such as wandering but also preferred organized or structured games. This study builds off Dean and colleagues’ (2017) work by exploring the activities children engage in and the ways in which they engage. In this study, we used qualitative data to better understand the nuanced details of children’s experiences on the playground and evaluate commonalities in the data. Given the gaps in the literature, this mixed methods study explored the activities of elementary-aged children with ASD during recess. Using both complementary and expansion techniques (Palinkas et al., 2011), we examined activities children with ASD participated in, their social communication, self-stimulatory behaviors, and affect. This research explores these themes in the context of the playground in order to better understand what influences the recess experience of children with ASD.

To this end, this study had four aims. Our first aim was to describe the playground activities that children with ASD participated in during recess. We predicted that children with ASD would more frequently engage in solitary activities such as playing alone or with an object (e.g., ball, jungle gym) over joint engaged activities as the literature documents that children with ASD spend more time solitary and unengaged (Kasari et al., 2011; Lang et al., 2011; Memari et al., 2015; Locke et al., 2016). Our second aim was to examine the social communication of children with ASD with peers during recess. We expected that children with ASD who had higher peer engagement would have more social communication interactions such as conversations with peers or conflicts over objects or activities, which is consistent with the literature on theory of mind and conflict resolution (Bauminger et al., 2010; Larson et al., 2012; Tager-Flusberg, 1999). Our third aim was to examine how self-stimulatory behavior (e.g., arm flapping, inappropriate body orientation, and self-orientated behaviors, etc.) was associated with the activities of children with ASD on the playground. We hypothesized that children with ASD would display more self-stimulatory behaviors while in solitary activities as compared to joint engaged activities. Our last aim was to examine how affect was associated with peer engagement states. We hypothesized affect would change with engagement and type of activity because of previous findings that children with ASD frequently demonstrate decreased joint attention and coordinated positive affect (Landa, Holman, & Garrett-Mayer, 2007). For example, we anticipate that neutral or sad affect would occur more frequently when children with ASD were unengaged at recess, whereas positive affect would occur more frequently when children with ASD were jointly engaged.

Methods

Participants

Fifty-five school-aged children with ASD participated in this study. All participants were included in kindergarten through fifth grade from 16 public elementary schools in the eastern United States. Participants were an average age of 8.68 (SD = 1.76) years. We focused on elementary school-aged children as this group typically has a recess or playground period during the school day. Eighty-four percent of students were male, and the majority was White (55%) with some students identifying as African American (25%), Multi-racial (9%), Asian (5%), Hispanic/Latino (4%), or Other (2%). Inclusion criteria included: 1) a referral from school administrators; 2) an ASD diagnosis confirmed by a school provider or clinician in the community; and 3) inclusion in a general education classroom for 80% or more of the school day. All students had an individualized education program (IEP); however, these records were not shared with the research team and ASD severity and intelligence quotient (IQ) scores could not be obtained. See Table 1.

Table 1.

Participant baseline characteristics

	Participants (n = 55)
Sex
Male	46
Female	9
Race/Ethnicity
African American	14
Hispanic	2
White	30
Asian	3
Multiethnic	5
Other	1
Grade
Kindergarten	4
1st	6
2nd	16
3rd	6
4th	6
5th	17
Age (M=8.65; SD=1.73)

Measures

Playground Observation of Peer Engagement (POPE; Kasari et al., 2005)

The POPE is a timed-interval behavior coding system designed to evaluate peer engagement and social communication of children with ASD during the lunch period or on the playground (Kasari et al., 2011). Independent observers documented each child’s engagement and social communication with peers on the playground. Observers watched children with ASD for 40 consecutive seconds and coded for 20 seconds. The duration of the observation interval ranged from 10-35 minutes across the recess and lunch period with an average duration of 15.89 minutes. Participants were observed once during the course of the study during the middle of the school year. Observations were not videotaped but were recorded in vivo via the POPE. Engagement states were measured through the percentage of intervals children spent in solitary play (when the child was not engaged with peers or solely engaged with an adult) and joint engagement (turn-taking, reciprocal conversations, and group activities) with peers. Meanwhile, social communication included initiations and responses. Initiations were coded in two ways: successful or unsuccessful initiations. Successful initiations were coded when the child with ASD used nonverbal or verbal language to at least one peer who responded either nonverbally or verbally. Unsuccessful initiations were coded when the child with ASD communicated nonverbally or verbally to at least one peer, but the peer did not respond. Furthermore, responses to a peer’s initiation were coded in two ways: appropriate and missed. Appropriate responses were pragmatically suitable (e.g. a peer said, “Thank you!” and the child said, “You’re welcome”) while in missed responses, the child did not respond to the peer’s initiation. Conversations were coded when there were at least four exchanges between the child and a peer or peers.

Observers were trained and considered reliable with percent agreement ≥ 80. Reliability for the study was collected on 20% of total play periods with an average 87% agreement (range 80-97%). In addition to the quantitative ratings, each interval included qualitative data that detailed additional information regarding each child’s social communication, activities, affect, engagement states, self-stimulatory behavior, and miscellaneous notes regarding the child’s play and social functioning. Observers wrote comments as brief words or sentences describing the child’s behavior while on the playground. This study closely examined the qualitative comments to explore trends in activities, affect, engagement, and self-stimulatory behaviors while using quantitative data to expand or complement these findings.

Procedure

School district officials were initially contacted to obtain a list of schools that had children with ASD included for 80% or more of the day in a general education Kindergarten through fifth grade classroom. Subsequently, principals were contacted and provided with study recruitment materials. The institutional review board approved this study, and all school districts agreed to participate. Parents and guardians gave written consent and children assented after being informed about the study. Observers measured each child’s peer engagement during the child’s recess and lunch period using the POPE, which were used for analysis. Given the duration of recess was variable and merged with lunch periods, observations included both the recess and lunch periods.

Data Analysis

This study used a mixed-methods approach. Based on Palinkas and colleagues (2011), data were examined through “collecting, analyzing and merging both quantitative and qualitative data” (p. 44). First, the structure of the research was “QUAL + quan” suggesting that quantitative and qualitative data were collected simultaneously, although qualitative data were weighed more heavily in this study given its exploratory nature. Second, the data’s function was largely complementary, where qualitative data were used to elaborate on the quantitative findings. The quantitative data offered a better understanding of the qualitative data and aided in exploring trends among participants. The function also included expansion of the results as qualitative findings were used to explain quantitative data (e.g., POPE comments expanded upon the engagement states identified in the quantitative data). Third, the process was to connect the data as the qualitative dataset built upon the quantitative dataset (Palinkas et al., 2011). Data were analyzed in this way to better understand the nuances of playground activity among elementary aged children with ASD.

The first author (SG) reviewed all qualitative data to generate the initial codebook with a systematic approach. Operational definitions were provided for each code along with examples. Subcodes for activity were created to categorize solitary and joint engaged activities; subcodes for self-stimulatory behavior were created to parse motor, vocal, and sensory behaviors; and subcodes for affect were created to reflect positive, neutral, and negative affects. Coders were trained in NVivo QSR 10, a qualitative data management program, and data reliability was collected on 20% of qualitative comments from an independent transcriber. A blind independent rater coded 20% of overlapping data. Reliability between raters was excellent (96.2%). Disagreements were discussed and a consensus was reached with 100% agreement between coders. The codebook was revised to include discrepancy in coding and to reflect additional subcodes. See Table 2 for an inventory of codes with its corresponding definitions and examples.

Table 2.

Qualitative Data Codes

Code	Definition	Example
Activity	Activities refers to popular activities children with autism participated in. Activities were categorized by solitary or joint engaged play.	Soccer
Social Communication	Social communication refers to the social interactions and pragmatics of children when on the playground. Social interaction may include social rules and conflict resolution. Pragmatics includes verbal and nonverbal communication such as discourse, body language, and eye contact.	“Discussing fart noises”
Self-stimulatory Behavior	Self-stimulatory behavior refers to any noticeable differences in play. Self-stimulatory behaviors are not characteristic of typically developing children; These behaviors are often characteristics of an autism diagnosis, but can also be unique to the child.	Staring
Affect	Affect refers to the emotional state of the child	Neutral affect

Descriptive analyses were conducted using SPSS Version 24. The quantitative data focused on the percent of time in solitary and joint engagement, successful and unsuccessful initiations, appropriate and missed responses, and total initiations and responses. Frequency was calculated as the number of participants who demonstrated a sub-code, whereas duration of time was calculated as the average number of minutes each sub-code occurred in the field notes. After separate analyses of the qualitative and quantitative data were completed, the data were connected in which the qualitative data were the primary source of data and the quantitative data were used as a secondary source. This process was complementary and focused largely on the observational notes and used quantitative data to make sense of those notes. This was particularly useful in interpreting the quantitative data and bridging any gaps the quantitative data presented. Data were triangulated by referring to each individual file and evaluating what each child did and when. Specifically, the interplay of activities, social communication, self-stimulatory behaviors, and affect that occurred during a given time interval were analyzed. Thus, the quantitative and qualitative data were analyzed separately and later connected to better understand not only the percentage of solitary activities or joint engagement on the playground but also what their activities, social communication, self-stimulatory behaviors, and affect looked like during solitary and joint engagement.

Results

Activities

Activities were categorized into solitary (i.e. solo activities or engaging only with an adult) or joint engaged (i.e. activities involving joint engagement between two or more peers) activities. Engagement states (i.e. solitary and joint engaged) were analyzed using descriptive measures. Quantitative data indicated that children with ASD spent 25% (SD=29) of the time in solitary engagement and 30% (SD=33) of the time in joint engagement. Solitary activities were activities performed alone or with no peer interaction as well as activities solely involving an adult. Activities were labeled: 1) peripheral; 2) object-focused play; 3) motor-oriented; 4) adult engagement; 5) academic-focused; and 6) singing. Peripheral activities were observed among 56% of participants for an average duration of 4.44 minutes. Peripheral activities consisted of sitting or standing on the periphery of the playground (36% participants, for an average duration of 6.42 minutes), observing or looking around (35%, 2.76 minutes), eating alone (29%, 6.92 minutes), and laying down or napping (5%, 1.67 minutes). Object-focused play was observed among 40% of participants and an average of 4.78 minutes involving playing on a structure (25%, 10.00 minutes), playing with a toy such as Legos, blocks, counting bears, or hula hoops (13%, 2.00 minutes), and playing with natural items such as leaves or rocks (9%, 2.33 minutes). Motor-oriented activities occurred in 38% of participants and on average 1.96 minutes consisting of wandering (29%, 1.56 minutes), running alone (13%, 2.29 minutes), skipping (4%, 1.00 minutes), and dancing (2%, 3.00 minutes). Adult engagement, such as talking to an adult, was observed among 20% of participants and for average duration of 2.22 minutes. Next, academic-focused tasks were observed among 9% of participants and an average of 7.83 minutes with activities such as reading (4%, 2.50 minutes), playing on a laptop (2%, 15.00 minutes), and completing homework (2%, 6.00 minutes). Lastly, children also were observed singing (2%, 1.00 minutes).

With regard to playground structures, children with ASD gravitated toward equipment with tires, seesaws, spiral ladders, swings, monkey bars, swinging poles, bridges, and slides, where 29% of participants played on the equipment for an average duration of 8.77 minutes. However, children with ASD also had interest in playing with natural items such as gathering sticks, playing in sand, pulling grass, picking at tree bark, and digging in the dirt (9% of participants for an average duration of 2.33 minutes). The majority of the qualitative notes indicated that children with ASD (65%) spent a portion of each recess interval in some sort of solitary activity. It is important to note that in a few instances (4% of the participants; 2 observations), solitary activities qualitatively became joint engaged activities. In one instance, a child with ASD was on a seesaw across from a peer but only initially engaged with their aide. Eventually, the child with ASD had a conversation with their peer on the seesaw. In the other example, a child with ASD was dancing alone in their seat and later joined his peers in singing the song, “Hey Soul Sister”.

Children with ASD also engaged in the following joint engaged activities: talking to peers, playground games, ball games, miscellaneous activities with a peer or group of peers, and pretend play/fantasy games. Talking to peers was the most frequently observed joint engaged activity (45% of participants, for an average duration of 2.17 minutes) and comprised conversation with one peer or multiple peers on the playground. Meanwhile, playground games were general activities that involved rules. Playground games occurred among 24% of total participants for an average duration of 4.67 minutes. These playground games included tag or chase (13%, 4.50 minutes), board games (4%, 8.50 minutes), eeny meeny miny mo (2%, 1.00 minutes), hide and seek (2%, 1.00 minutes), duck duck goose (2%, 7.00 minutes), rock paper scissors (2%, 5.00 minutes), hand games with physical contact such as Patty Cake (2%, 4.00 minutes), hopscotch (2%, 5.00 minutes), and four square (2%, 5.00 minutes). Ball games were observed among 20% of participants for an average of 8.30 minutes consisting of traditional sports or games played with a ball such as basketball (5%, 7.33 minutes), kickball (5%, 10.50 minutes), soccer (4%, 14.00 minutes), ball tag (4%, 4.00 minutes), and volleyball (2%, 4.00 minutes). Meanwhile, miscellaneous activities involved an array of play activities that a small number of participants performed. Though few children with ASD participated in these activities, they were notable in that they involved peer interactions for the majority of recess. Miscellaneous activities occurred among 16% of total participants for an average of 3.62 minutes including playing with toys such as figurines, hula hoops, cars, spinning tops, gears, and loom bands (9%, 4.25 minutes) with a peer or group of peers, interacting with a peer on a structure including seesaws and slides (5%, 9.67 minutes), dancing to music (4%, 2.20 minutes), wrestling with a peer (2%, 1.00 minutes), and singing (2%, 1.00 minutes). Lastly, pretend-play/fantasy games were observed on the playground among 5% of participants for an average of 6.67 minutes. Pretend-play involved acting like robots (2%, 2.00 minutes), pretending to be lions (2%, 13.00 minutes), and playing a Pokémon game where children acted like the characters (2%, 5.00 minutes). These pretend activities were infrequent on the playground and each activity was specific to a different child’s playground interest.

In regard to duration, ball games occurred for the longest time period (8.30 minutes) followed by academic-focused tasks (7.83 minutes). The activities shortest in duration included singing (1.00 minutes) and motor-oriented activities (1.96 minutes). Solitary activities occurred for an average of 3.71 minutes. Meanwhile, joint engaged activities occurred for an average of 5.08 minutes. Thus, children with ASD remained in the same activity for a longer duration when they were playing with peers compared to playing alone. See Table 3.

Table 3.

Activities

Solitary	Ave. Duration (Min.)	% of Participants	Joint Engaged	Ave. Duration (Min.)	% of Participants
PERIPHERAL	4.44	56	TALK TO PEERS	2.17	45
Sit or stand on periphery	6.42	36
			PLAYGROUND	4.67	24
Observe or look around	2.76	35	GAMES
			Tag or chase	4.50	13
Eat alone	6.92	29	Board games	8.50	4
Lay down or nap	1.67	5	Eeny Meeny Miny Mo	1.00	2
			Hide and Seek	1.00	2
OBJECT-FOCUSED PLAY	4.78	40	Duck Duck Goose	7.00	2
			Rock Paper Scissors	5.00	2
Play on structure	10.00	25	Hand games	4.00	2
Play with toy	2.00	13	Hopscotch	5.00	2
Play with nature	2.33	9	Four Square	5.00	2
MOTOR-ORIENTED	1.96	38	BALL GAMES	8.30	20
Wander	1.56	29	Basketball	7.33	5
Run alone	2.29	13	Kickball	10.50	5
Skip	1.00	4	Soccer	14.00	4
Dance	3.00	2	Ball tag	4.00	4
			Volleyball	4.00	2
ADULT ENGAGEMENT	2.22	20
			MISCELLANEOUS	3.62	16
			Play with toy	4.25	9
ACADEMIC-FOCUSED	7.83	9	Structure	9.67	5
			Dance	2.20	5
Read	2.50	4	Wrestle	1.00	2
Play on laptop	15.00	2	Sing	1.00	2
Homework	6.00	2
			PRETEND PLAY	6.67	5
SINGING	1.00	2	Act like robots	2.00	2
			Pretend to be a lion	13.00	2
			Pokémon game	5.00	2

Social Communication

Social communication referred to the initiations and responses of children with ASD. Particularly, social communication included social interactions (e.g. social rules and conflict resolution), pragmatics, and social cognition of children with ASD when on the playground. Children with ASD had a total of 7.47 initiations (SD=8.441) and 6.93 responses (SD=7.645). On average, 5.05 (SD=6.288) of their initiations were appropriate to peers while 2.25 (SD=4.120) were missed or inappropriate. The mean of appropriate responses was 5.56 (SD=5.999) and the mean of missed opportunities was 1.35 (SD=2.444). The success rate of initiations and responses for children with ASD was 80% (SD=26) and 72% (SD=28), respectively.

With respect to grade, kindergarten through 2^nd grade (K-2) students had a total of 8.31 initiations and 7.19 responses while 3^rd grade through 5^th (3-5) grade students had 6.72 initiations and 6.69 responses; 5.15 initiations were appropriate and 2.77 were missed on average for the K-2 cohort while 4.97 initiations were appropriate and 1.79 were missed for the grades 3-5 cohort. The mean of appropriate responses and missed opportunities was 5.85 and 1.31, respectively for the K-2 cohort, whereas in the grades 3-5 cohort, the mean of appropriate responses and missed opportunities was 5.31 and 1.38, respectively. Of interest, however, were the nuances of social communication on the playground such as the content of children’s conversations with peers and how children with ASD communicated to peers.

The most frequently noted social communication comments involved conversations with peers, which occurred among 45% of children with ASD and an average of 2.17 minutes. Conversations involved something they were mutually watching, an item a peer was holding, or even topics as broad as Halloween or fart noises. There were no notable differences in conversations among children in different grade categories (i.e. participants in kindergarten to second grade compared to participants in third to fifth grade) as the qualitative data did not include detailed notes of all conversations. However, in 80% of instances where social communication was noted, children with ASD were involved in a joint engaged activity such as playing a ball game or sharing toys such as a spinning top or figurine. Infrequently, comments were related to conflict among peers (9% of participants; average duration of each conflict was 4.0 minutes). Four children with ASD were observed yelling at a peer and in one instance, peers yelled at the child with ASD. When peers yelled at a child with ASD, it was due to not sharing a play item and policing them for breaking the rules during the same play activity. When children with ASD were yelling at a peer, they were either attempting to gain the attention of a peer and were not successful or refused to listen to a peer’s request. Children with ASD were observed to demonstrate obedience to peer requests or followed their peer’s lead in three instances. The qualitative comments indicated peers requested the child with ASD to move to a different spot during kickball or to cease screaming.

Though children with ASD were seen following peer actions and requests in several situations, one child with ASD was observed as the leader among his classmates. Peers joined him in joint engaged activities and when he wanted to move to a new task, his friends followed. Lastly, one child with ASD was observed introducing peers to each other while another child with ASD was observed helping his friend after she fell. Most social communication (i.e. initiations and responses) occurred in games such as ball games or pretend games and when talking to peers. There were no distinguishing characteristics of these two children compared to the rest of the sample.

Self-stimulatory Behavior

Self-stimulatory behaviors were relatively infrequent. We categorized self-stimulatory behaviors into motor, sensory, or vocal. Motoric self-stimulatory behaviors included flapping (18% of participants, average duration of 1.60 minutes), finger flexing (11%, 1.14 minutes), rocking side to side (5%, 1.00 minutes), fidgeting (5%, 1.50 minutes), flailing (5%, 1.67 minutes), hand motions (4%, 1.00 minutes), running in circles (4%, 1.00 minutes), and running on toes (2%, 1.00 minutes). Second, sensory-related self-stimulatory behaviors included staring (18%, 1.42 minutes), rubbing own body parts (5%, 1.00 minutes), putting fingers in mouth to chew or lick (5%, 1.00 minutes), rubbing structures (4%, 1.50 minutes), self-hugging or inappropriate hugging (4%, 1.00 minutes), covering ears or ear plugging (4%, 1.00 minutes), skin picking (2%, 1.00 minutes), and putting items in nose (2%, 1.00 minutes). Third, vocal self-stimulatory behaviors included self-talk (9%, 1.50 minutes), scripting (4%, 1.00 minutes), self-humming (2%, 1.00 minutes), and various noises (2%, 1.75 minutes). Among all the participants, only 53% of children with ASD displayed self-stimulatory motoric behaviors for an average of 1.29 minutes. Meanwhile, 36% displayed self-stimulatory sensory behaviors for an average 1.24 minutes and 13% displayed self-stimulatory vocal behaviors for an average of 1.45 minutes. Twenty-seven percent of participants demonstrated two or more types of self-stimulatory behavior.

In total, 13% of the fieldnotes documented children with ASD engaging in self-stimulatory behaviors. On average, children demonstrated these behaviors for 1.21 minutes. Self-stimulatory behaviors most frequently occurred when children were engaged in solitary activities. In addition, a slightly longer duration of self-stimulatory behaviors occurred during solitary activities (1.32 minutes) compared to joint engaged activities (1.17 minutes). Particularly, staring was the most frequent solitary self-stimulatory behavior which comprised 15% of the total number of observed self-stimulatory behaviors. Next, self-talk was observed when playing alone on a structure or walking around the periphery 13% of the instances, and flapping was observed 11% of the instances while at lunch playtime, walking on the periphery, or on a structure. Less than 4% of the time, all other self-stimulatory behaviors mentioned with the exception of self-humming and various noises were observed while solitary (eating, watching a movie, or in the yard). Most often, self-stimulatory behavior during a solitary activity was observed while the child with ASD was eating lunch.

When kids were in joint engaged activities, the most frequent self-stimulatory behavior was flapping which comprised 8% of the total self-stimulatory behaviors. Flailing and finger flexing occurred in 3% of the observed self-stimulatory behaviors. Less than 2% of the time, staring, self-humming, various noises, inappropriate hugging or self-hugging, or finger licking or chewing were observed. Most often, self-stimulatory behavior during a joint engaged activity was observed during a ball game but also seen while playing tag, duck duck goose, sharing figurines, or during pretend play. See Table 4. Mostly, finger flexing and flapping were exhibited in excitement as indicated on the POPE during an activity in which the child was overstimulated. This behavior was paired with positive affect and observed almost equally in both solitary and joint engaged activities.

Table 4.

Self-stimulatory Behaviors

Self-stimulatory Behaviors	Average Duration in Minutes	% of Participants
MOTOR	1.29	53
Flapping	1.60	18
Finger flexing	1.14	11
Rocking	1.00	5
Fidgeting	1.50	5
Flailing	1.67	5
Hand motions	1.00	4
Running in circles	1.00	4
Running on toes	1.00	2
SENSORY	1.24	36
Staring	1.42	18
Rubbing own parts	1.00	5
Putting fingers in mouth	1.00	5
Rubbing structures	1.50	4
Self-hugging or inappropriate hugging	1.00	4
Covering ears or ear plugging	1.00	4
Skin picking	1.00	2
Putting items in nose	1.00	2
VOCAL	1.45	13
Self-talk	1.50	9
Scripting	1.00	4
Self-humming	1.00	2
Various noises	1.75	2

Affect

Affect referred to the emotional state of the child. Positive affect was defined as a state of happiness, excitement, or contentment while negative affect was coded when the child with ASD was angry, sad, disappointed, or upset. Neutral affect was documented when the child displayed a flat or emotionless facial expression. The authors calculated average duration based on time intervals and accounted for each type of affect if it occurred for the majority of that interval. Children with ASD exhibited a range of affect: 46% (SD=39) of children with ASD demonstrated neutral affect, 44% (SD=37) demonstrated positive affect, and only 9% (SD=20) demonstrated negative affect. The results suggest most of the children with ASD (33%) had neutral affect during a solitary activity. Less frequently (23%), children with ASD displayed positive affect in solitary activities as well as joint engaged activities. During joint engaged activities, a small number (15%) of children with ASD had neutral affect and only 3% had negative affect. Similarly, only 3% of children with ASD displayed negative affect in a solitary activity. Children with ASD demonstrated positive affect for an average of 5.87 minutes, neutral affect for 5.24 minutes, and negative affect for 1.05 minutes. Though more children displayed neutral affect, they maintained positive affect for a slightly longer duration than neutral affect. When children demonstrated negative affect, the duration was relatively short.

Discussion

This study explored the activities, social communication skills, self-stimulatory behaviors, and affect of elementary school children with ASD during recess. Our results indicated that children with ASD participated in a variety of age appropriate activities both alone and with peers. As expected, children with ASD engaged in both positive (appropriate initiations and responses, reciprocal conversations) and negative (conflict) social communication interactions. Children with ASD also most frequently displayed motor and sensory-related self-stimulatory behavior, and neutral affect while being observed on the playground. By furthering knowledge of playground behaviors, these data have implications for strategies to better support children with ASD on the playground.

The quantitative data suggest that children with ASD spent 25% of time in solitary activities, where they were on the periphery of activities or alone. However, we also found that children with ASD spent 30% of time in joint engaged activities, which included reciprocal conversations and games with peers. Meanwhile for the qualitative data, children with ASD were found to participate in more solitary activities (65%) compared to joint engaged activities (35%). This discrepancy can be attributed to how the POPE defined engagement states. In the quantitative data, engagement states are coded based on the majority of an interval. However, engagement states change every 3 seconds (Kasari et al., 2005). As a result, children can be engaged in solitary activities but be coded as joint engaged should the activity occur for the majority of the observed interval (Kasari et al., 2005). The qualitative observations accounted for both solitary and joint engaged activities as they occurred in each interval, regardless of their domain over the interval period. Similarly, the majority rule could explain why there was a low rate of joint engagement (30%), yet a high success rate for social communication (80% and 72%). Joint engagement was not be coded if it did not occur for the majority of the interval, but communication could have occurred during that observation interval. There were no identifiable themes for why engagement state changed. However, it is documented in the literature that engagement states tend to vary and fluctuate quite frequently, as often as every three seconds (Kasari et al., 2015; Kasari et al., 2011; Kasari et al., 2012; Locke et al, 2016). Identifying the mechanism of change in engagement may further inform intervention strategies.

Results from both the quantitative and qualitative data describe how children with ASD spend their time on the playground, indicating an opportunity to increase their involvement with their peers. The results from this study also may be important for informing intervention type (e.g. applied behavioral analysis (ABA), speech-language therapy, etc.) as well as intervention target (e.g., activity, social communication, self-stimulatory behavior, and affect) as school personnel and providers may facilitate opportunities for children with ASD with their peers on the playground. There is some literature that suggests social deficits can be addressed using adult-facilitated and peer-mediated interventions (Kretzmann et al., 2015; Locke et al., 2015; Kasari et al., 2016). Kasari and colleagues (2012) found that children with ASD who received peer-mediated instruction were more included in their peer social networks and had higher levels of peer engagement as compared to children receiving adult-facilitated intervention. However, Kasari and colleagues (2016) found that children randomized to a didactic skills-based curriculum improved peer engagement, particularly for children with higher behavior problems and lower teacher-child closeness. These studies suggest that intervention is not “one size fits all” and should carefully consider individual needs. Interventions such as ABA or speech therapy, two interventions that are commonly used in schools to address social communication and atypical behaviors, also may be important to consider when targeting social impairments in school.

This mixed-methods approach can assist implementation of intervention strategies and highlight various aspects of treatment development by providing a deeper understanding of the process and findings (Aarons et al., 2012). The results from this study have important implications to guide intervention as it shows broad activities where children with ASD most frequently engage, which can help school personnel and providers build off children’s strengths and common interests. The findings can be useful for schools to facilitate opportunities for children with ASD who want and need more structured social opportunities with peers. This may include providing more play options that are suitable for children with ASD and fit their unique desires and needs.

It is important to highlight the positive social communication skills observed on the playground such as being a leader or helping others. Contrary to studies that indicate children with ASD have significant social impairments (Bauminger and Kasari, 2000; Bauminger et al., 2003; Bauminger et al., 2010; Kangas et al., 2011; Tager-Flusberg, 1999), we observed several instances where children with ASD engaged in prosocial behaviors with peers. This engagement is important as it shows children with ASD can be peer role models, spearhead activities, and promote group participation. Social skill interventions can begin with acknowledging prosocial behaviors and building off children’s strengths, talents, and abilities.

Self-stimulatory behaviors mostly occurred when children were in solitary engagement. The most common self-stimulatory behaviors were motor or sensory-related with the most frequent behaviors being flapping and staring. As suggested by Samson et al. (2014), self-stimulatory behaviors often appear when children with ASD are emotionally dysregulated. Since flapping and other motor behaviors were demonstrated during states of excitement regardless of activity, the data suggest that these behaviors were influenced by their affect. Though flapping is noted as a self-stimulatory behavior, it is important to consider how flapping during excitement can be a typical and expected central nervous system response (e.g., people often jump and wave their arms in excitement during a surprise birthday party or when hearing good news like a wedding engagement, etc.). The literature has yet to inform community adaptation and response to motor atypicalities, which would be important to explore. Furthermore, these data indicate instances on the playground where self-stimulatory behaviors may occur. Particularly, self-stimulatory behaviors were most observed during lunch. This would be an opportune time for school personnel and providers to provide additional supports for children who want or need more structure in the cafeteria (e.g., helping children find a peer at lunch). Additionally, Macintosh and Dissanayake (2006) identify the playground as a stressful period of the day for children with ASD. Although our data did not indicate if self-stimulatory behavior discouraged peers from interacting with children with ASD, self-stimulatory behavior may point to activities where children with ASD are stressed and understanding what situations on the playground may be most stressful to children with ASD may help school personnel and providers identify and respond to these situations quicker.

The data also indicated that children with ASD mostly displayed neutral affect. This finding is consistent with previous literature where children with ASD often show a flat, emotionless expression (American Psychiatric Association, 2013; Begeer et al., 2008; Loveland et al., 1994; Samson et al., 2014; Strid et al., 2013). However, children with ASD almost as frequently demonstrated positive affect. In this study, children with ASD showed positive affect when engaged with peers, which highlights the importance of peer engagement to some children with ASD. It is vital to be mindful of affect, so school personnel can target and facilitate activities and intervention strategies as well as create opportunities to maximize positive affect as this may aid social motivation and attention (Chevallier et al., 2012). Interestingly, although infrequent, children with ASD displayed negative affect, which was predominantly a result of peer conflicts. Awareness of affective states of children with ASD on the playground is important for socio-emotional wellbeing and to foster a safe space for everyone.

Our findings highlight both the challenges and successes of children with ASD on the playground. We observed that some children with ASD could be flexible in following their peers’ ideas as well as direct the play themselves. We also observed that some children with ASD may need more support on the playground, particularly around how to navigate appropriate conflict resolution. These findings highlight areas where some children with ASD may need support on the playground but also inform the ways in which to support children with ASD in inclusive and general education classrooms. For example, what is observed on the playground may manifest in the classroom environment. With US trends emphasizing more academic and less recess time (Murray and Ramstetter, 2013), it also would be useful to consider how shorter recess time may impact a child with ASD. As the literature shows, there are implications of keeping recess in curriculum as it can help develop social skills and peer relationships (Couper et al., 2013). This study shows that children with ASD have broad interests of play activities and have opportunities to interact with peers through these different activities. The playground encourages interactions with classmates, which may not be as accessible in the classroom environment. Overall, knowledge of activity, social communication, self-stimulatory behaviors, and affect on the playground can provide rich opportunities for peer interaction and can inform how children with ASD may be observed in other environments. These characteristics can help educators, parents, clinicians, and caregivers know how to best support children with ASD.

Limitations and Future Directions

Although this is one of few studies that examine playground engagement of children with ASD using both quantitative and qualitative methods, several limitations are noted. First, this study lacks a comprehensive understanding of the social behavior of children with ASD as observations were based solely on the conjoined lunch and recess periods. Importantly, future studies should tease apart the lunch versus playground contexts given the potential for differences in each setting (structured/confined vs. unstructured/free). Second, this study comprised predominantly male children with ASD, and it was not possible to conduct sex analyses. Since females were underrepresented in this study, future research should focus on recruiting more female participants and noting differences among sexes (Dean et al., 2014). Third, these data were cross-sectional. Future research should triangulate data from several observations in order to produce more robust results as different patterns of playground behavior may unfold over time. In order to determine consistency of playground activity, longitudinal studies are warranted to see more nuanced changes in engagement, social communication, self-stimulatory behavior, and affect. Fourth, the time frames of each construct were not specifically documented within each observation interval. Future studies ought to analyze duration and how these variables change over the course of the observation. Fifth, affect was not recorded for each participant during every observation interval; therefore, it was not possible to assess how affect changes across the total observation period. Sixth, this study did not have a typically developing comparison group and it may be important to explore how each of these constructs differs between children with ASD and typically developing children. Lastly, due to school district policies, we were unable to access student’s records of ASD severity, IQ, or language skills. It would be useful to conduct testing in these areas in order to describe each child’s cognitive or language abilities. Importantly, it would be useful to recruit a more diverse sample as this study focused on fully included verbal children with ASD on the playground. Future studies may include participants that are nonverbal, more ethnically diverse, or cognitively variable. Increased variation among participants can improve generalization of findings to broader contexts and across the autism spectrum.

Conclusion

Previous literature has examined playground engagement using quantitative methods, but few have incorporated qualitative observations to supplement, support, and unpack what we know about peer engagement of children with ASD. This study provided a mixed-methods analysis of children with ASD on the playground where qualitative data provided a more nuanced understanding of peer engagement. These data highlight a deeper understanding of the activities with which children with ASD engage, how they communicate, and their self-stimulatory behavior and affect on the playground. Lastly, these findings point to areas where schools can intervene or provide additional support.

Table 5.

Affect

Affect	Average Duration in Minutes	% of Participants
Positive	5.87	44
Neutral	5.24	46
Negative	1.05	9