Effects of Stability Balls on Children’s On-Task Behavior, Academic Achievement, and Discipline Referrals: A Randomized Controlled Trial

Alicia Fedewa; Matthew A. C. Davis; Soyeon Ahn

doi:10.5014/ajot.2015.014829

Abstract

OBJECTIVE. We used a randomized controlled design to investigate whether using stability balls during the school day was associated with higher levels of on-task behavior and academic achievement and fewer discipline referrals.

METHOD. Over 9 mo, students in 2 second-grade classrooms in a southeastern rural elementary school used stability balls as chairs while students in 2 control classrooms used chairs as usual. We collected measures of on-task behavior, standardized measures of literacy and mathematics achievement, and discipline referrals.

RESULTS. We found similar levels of on-task behavior and achievement in treatment and control classrooms and a downward trend in disruptive behaviors in treatment classrooms.

CONCLUSION. This study did not find use of stability balls for entire general education classrooms to be a practical use of resources for schools. More research with rigorous controlled designs is needed to support the use of stability balls for the general education population.

According to an ecological systems perspective, the predictors for academic risk or benefit are associated with a range of interactional effects between student and environment (Bronfenbrenner, 1979; Gutkin, 2012). The contextual factors in settings such as schools can influence the educational, social, and behavioral performance of students across the general population. Classrooms with high instructional quality and positive classroom climate have been found to result in greater behavioral and psychological engagement among students (Dotterer & Lowe, 2011). In addition, teachers must be aware of contextual factors related to the physical classroom setting that improve student learning and behavior; an environment that promotes learning includes control over physical aspects (e.g., noise, lighting, classroom arrangement, temperature) that may influence student attention, on-task behavior, and, in the long term, academic achievement.

One intervention in the classroom environment used to influence student attention, on-task behavior, and academic achievement is the use of stability balls in lieu of student chairs (Carriere, 1998; Fedewa & Erwin, 2011; Schilling, Washington, Billingsley, & Deitz, 2003). Recent studies conducted on the effects of stability balls in the school-age population have focused primarily on students in special education, including students with attention deficit hyperactivity disorder (ADHD; Bill, 2008; Fedewa & Erwin, 2011; Schilling et al., 2003) and those with autism spectrum disorder (ASD; e.g., Bagatell, Mirigliani, Patterson, Reyes, & Test, 2010). Use of a stability ball in the classroom has shown positive effects on participation (Bagatell et al., 2010), attention (Bill, 2008), and in-seat behavior (Schilling et al., 2003).

In 2011, Fedewa and Erwin conducted the largest, most systematic investigation of stability ball use, measuring changes in attention and hyperactivity of 76 students in four general education classrooms as well as in-seat and on-task behavior of 8 students identified with high-risk symptomatology of ADHD. At the end of a 2-wk period of stability ball use by all participants, observers recorded a decrease in hyperactivity and an improvement in attention. Using momentary time sampling (MTS; Rapp, Colby-Dirksen, Michalski, Carroll, & Lindenberg, 2008), a 12-wk intervention with the 8 students with high ADHD probability resulted in remarkable increases in both in-seat and on-task behavior. The two studies (Fedewa & Erwin, 2011; Schilling et al., 2003) using designs with objective measures supported the previous studies, which reported anecdotal conclusions of promising results for the use of stability balls.

There is reason to hypothesize that the positive effects of stability ball use in special education populations might also extend to the general education population. As mentioned, Fedewa and Erwin (2011) found improvements in hyperactivity and attention scores among the general education population. Social validity was also measured, with general education teachers largely in favor of the use of stability balls in the classroom (Fedewa & Erwin, 2011). Stability ball use has been cited as effective because of its perceived use in addressing sensory needs (Bagatell et al., 2010; Schilling et al., 2003), increasing student enjoyment when in seat (Fedewa & Erwin, 2011), and providing needed arousal (Schilling et al., 2003), all of which would benefit children in the general education population. In addition, stability balls are often associated with physical activity and movement, both of which have been linked to student attention (Voelcker-Rehage & Niemann, 2013), cognitive ability (Donnelly & Lambourne, 2011; Fedewa & Ahn, 2011), and academic achievement (Fedewa & Ahn, 2011).

The results from prior stability ball studies hold promise for enhancing children’s attention and behavior in the classroom. However, small sample sizes, short timelines for intervention implementation, inability to provide all students in a classroom with a stability ball, and previous primary focus on special education student populations are methodological limitations that limit external validity in generalizing results to a broader population of students. Only a few studies used objective measures to understand the effects of stability balls, and no researchers used a randomized controlled trial (RCT) design to more accurately assess the effects of stability ball use as a school-based intervention.

The purpose of this study was to address this gap in the literature by using a more rigorous, RCT design to determine whether random assignment of stability balls in classrooms would result in findings similar to those documented in prior research on children’s on-task behavior, achievement, and negative behaviors (e.g., as measured by discipline referrals). The following research questions guided the current study:

Does the use of stability balls during the school day result in higher levels of on-task behavior for children in the treatment classrooms?
Does the use of stability balls during the school day result in higher levels of mathematics and literacy achievement for children in the treatment classrooms?
Do discipline referrals decrease in the treatment classrooms where children are using stability balls in lieu of chairs?

Method

Research Design

The present study used an RCT design (Del Boca & Darkes, 2007) wherein 2 of 4 second-grade classrooms were randomly assigned to the treatment group by the lead investigator. Students in the treatment group received the stability balls to use throughout the school day for the remainder of the school year. The university institutional review board approved all procedures for the study. We obtained parental consent to collect behavioral and achievement data and child verbal assent before the start of the intervention. All students in the treatment classrooms received the stability balls regardless of whether their parent consented; the lack of parental consent waived the right of the researchers to collect any behavioral, achievement, or disciplinary data on their child. One set of stability balls was purchased by the school and the other set was purchased by the authors’ institution.

Participants

Participants included second-grade students in a rural elementary school in the southeastern United States. The lead researcher contacted the principal to assess interest in a study in which stability balls were used in lieu of chairs. The principal and teachers agreed to have 4 of the second-grade classrooms participate in the study. Parental consent was sought from all students in the second-grade classrooms (N = 81). Sixty-seven of the 81 students returned parental consent forms (83% return rate). Of the 67 students who returned consent, 32 were girls (48%). A total of 36 children were in the 2 treatment classrooms (54%). None of the students in the participating classrooms had any physical or cognitive limitations that would prevent them from participating in the study. There was no participant attrition during the study.

Instruments

On-Task Behavior.

Two boys and two girls from each classroom (n = 16) were randomly selected for in-depth behavioral observations. The random assignment at the student level was completed using a random number generator (RANDBETWEEN function) in Excel (Microsoft Corporation, Redmond, WA). These children were matched by gender. In the event that these children would not be present on an observation day, the names of 4 backup children, also matched by gender, were provided to the observers (the primary 4 children were present for 96% of the observations; thus, the backup children were infrequently observed). All of the children had parental consent.

The behavioral observations were conducted using MTS (see Fedewa & Erwin, 2011; Rapp et al., 2008), in which every 30 s, the observer coded the students’ behavior on the basis of several behavioral classifications of on-task behavior: group work with peers, independent seatwork, or interaction with the teacher (e.g., listening to instructions, talking with the teacher to clarify an assignment or ask questions). Observers carried a stopwatch to mark the 30-s time intervals and recorded their observations at the end of each interval on designated worksheets assigned for each child participant. As in prior studies (Fedewa & Erwin, 2011), we selected MTS at 30-sec intervals because it has been shown to reduce the number of false positives for duration events. Observations using this methodology have been shown to be valid and reliable across observers (Rapp et al., 2008).

Achievement.

Students’ performance in math and literacy was measured using the Measures of Academic Progress (Northwest Evaluation Association, Portland, OR) in the fall and spring of the academic year (pre- and postintervention). Using a Rasch unit scale, we tracked student progress using a converted score that shows how much a student knows on the basis of the state’s core academic standards.

Discipline Referrals.

Discipline referrals were collected by each classroom teacher using “clip downs,” a schoolwide disciplinary system. Clip downs represented incidents of serious classroom disruption or misbehavior; students would move their plastic or wooden clips “down” on a behavior chart for each act of negative behavior that resulted in a disciplinary consequence (e.g., loss of privilege, visit to the principal’s office). We had the teachers tally the total number of clip downs per month for students who had parental consent.

Procedures

Training.

One doctoral student (the second author) and four undergraduate students from the lead researcher’s (Alicia Fedewa’s) university served as observers. During observer training, these observers were told that they would be assessing on-task and off-task behavior in 4 classrooms and that in some classrooms, the children would use stability balls instead of chairs. Thus, because of the nature of the study (i.e., use of stability balls in lieu of chairs), the observers were not fully blind to the purpose behind the study, although they were not explicitly told the study’s hypotheses or research questions.

The trained doctoral student observed with the four undergraduate students in each of the four classrooms to conduct reliability analyses among observers at the beginning of the academic school year. Interrater reliability data were collected during 10 observational periods both before and throughout the study. Average reliability among the raters was 96% and ranged from 63% to 100% (mode = 98%). To calculate weekly on-task and in-seat or on-ball scores, we computed mean (M) scores across raters.

Data Collection.

Observers visited the classrooms 1 day per week during literacy and mathematics instruction and observed each of the 4 students for 30 min. The same schedule was used for the washout and intervention phases. The 2-wk washout phase was designed to be a period of acclimation for both the observers and the students using the stability ball; observers collected data in the classrooms, but we did not use these data for analysis. Intervention data collection followed the washout period and lasted 9 mo. Observers were trained to be as discreet as possible; they were initially provided with a description of the children they were observing so they could identify the children without disturbing the students and teacher in the classroom.

Because there were 4 children to observe in each classroom, observers recorded the behaviors of all 4 children within each of the 30-s time intervals during the 30-min observation and marked the behavior in one of six columns on the worksheet (i.e., on- or off-task behavior for work with peers, independent work, and interaction with the teacher). Observers were purposely assigned to different days of the week and different periods of observation so that an equal number of observations occurred for both mathematics and literacy. These rotations were consistent, and the same observer collected data in the same classroom throughout the academic year. With 30-min observations for the intervention phase occurring at least 2×/mo for 9 mo, observers were able to capture a total of 20 observations per student for the academic year. On-task behavior was calculated as the percentage of total observations (in minutes) during which the 16 children were on task during the academic year.

Data Analysis

On-task behavior, achievement, and discipline referrals served as dependent variables, and the use of stability balls by the treatment group was the predictor or independent variable. We first gathered descriptive statistics for classwide achievement and disciplinary and on-task behavioral data. Then we performed two sets of mixed-design analyses of variance (ANOVAs) to compare on-task behavior and achievement between control and treatment classrooms over the 8 mo of the intervention period.

First, we used a 20 (Time: number of observations over 8 mo) × 2 (Group: control and treatment) mixed-design ANOVA to examine the effect of intervention on students’ on-task behavior across the 8 mo of intervention (i.e., percentage of observations when student was on task in working with peers, doing independent work, or interacting with the teacher). Second, we compared treatment and control groups using two sets of a 2 (Time: fall and spring) × 2 (Group: control and treatment) mixed-design ANOVA on literacy and mathematics achievement scores. Last, we graphically compared change in discipline referrals between treatment and control classrooms across 8 mo; discipline referrals were aggregated and thus identical for all children in the same classroom, so we could not perform inferential statistical analyses. All assumptions were checked before comparing differences across and within groups through visual inspection of the graphed data and residuals (Cohen, 1988).

Results

Descriptive Statistics

Two classrooms were randomly assigned to the treatment of using stability balls in lieu of chairs (n = 16 in Treatment Classroom 1, n = 20 in Treatment Classroom 2; n = 16 boys in each classroom), and two classrooms were randomly assigned to the control group using chairs as seats (n = 13 in Control Classroom 1, n = 18 in Control Classroom 2; n = 15 boys in each classroom). The treatment classrooms had a mean of 32 discipline referrals (standard deviation [SD] = 2.63) during the intervention period. The mean achievement score in literacy (n = 35) was 178.66 (SD = 2.52) at baseline (fall) and 194.00 (SD = 11.63) at posttest (spring). The mean achievement score in mathematics was 180.34 (SD = 2.52) at baseline and 194.94 (SD = 9.42) at posttest.

The control classrooms had mean discipline referrals of 22.38 (SD = 3.74) over the intervention period. The mean achievement score in literacy was 173.13 (SD = 2.98) at baseline and 187.13 (SD = 14.62) at posttest. The mean achievement score in mathematics was 178.19 (SD = 2.52) at baseline and 189.10 (SD = 13.54) at posttest.

Mean percentages of time on task for each of the three types of behavior (working with peers, doing independent work, and interacting with the teacher) are presented in Table 1. Students in the control classrooms were on task—on average—more often than children in the intervention classrooms throughout the duration of the intervention (87% in the control classrooms vs. 77% in the treatment classrooms).

Table 1.

Mean Percentage of Time Students Were On Task, by Group (N = 16)

Type of Behavior	Mean % of Time On Task
Type of Behavior	Control Group	Intervention Group
On task	87	77
Working with peers	15	13
Doing independent work	39	29
Interacting with the teacher	33	35
Off task	13	23

View Large

Table 1.

Mean Percentage of Time Students Were On Task, by Group (N = 16)

Type of Behavior	Mean % of Time On Task
Type of Behavior	Control Group	Intervention Group
On task	87	77
Working with peers	15	13
Doing independent work	39	29
Interacting with the teacher	33	35
Off task	13	23

Effect on Discipline Referrals

Figure 1 shows the change in discipline referrals across 8 mo. The data are graphically represented separately by teacher, and intervention and control groups can be compared. Discipline referrals decreased and tapered off to about 20 referrals midintervention for the treatment classrooms. However, the control classrooms showed no change in discipline referrals and continued to follow a staggered pattern averaging between 20 and 30 discipline referrals over the 8-mo period. As is apparent from the graphs, the treatment classrooms began the intervention with higher levels of discipline referrals than the control classrooms.

Figure 1.

Discipline referrals, by teacher and intervention group.

Full Size | Slide (.ppt)

Effect on On-Task Behavior

Percentage of Time Working With Peers.

The 20 × 2 mixed-design ANOVA indicated a significant interaction effect between Time and Group, F(19, 95) = 4.23, p < .01,

η_{p}^{2}

= .46, and a significant main effect of Time on percentage of observations working with peers, F(19, 95) = 5.03, p < .01,

η_{p}^{2}

= .50. “Time” represents weeks; thus, every time point is 1 wk of the intervention (e.g., Time 1 would be Wk 1 of the intervention). Because a higher-order effect (two-way interaction) supersedes a lower-order effect (main effect of Time), a follow-up analysis was performed to examine the effect of the significant interaction. A post hoc analysis using Sidak adjustment showed that children in the control group worked with peers significantly more at Time 4 (M_diff = 29.43, standard error [SE] = 12.92), Time 5 (M_diff = 26.55, SE = 1.90), Time 12 (M_diff = 34.92, SE = 2.48), and Time 15 (M_diff = 62.46, SE = 1.46), compared with children in the treatment group. Overall, children in the control group exhibited more time on task over the course of the intervention in the area of working with peers than did children in the treatment group.

Percentage of Time Doing Independent Work.

A 20 × 2 mixed-design ANOVA indicated a significant interaction effect between Time and Group, F(19, 95) = 4.91, p = .02,

η_{p}^{2}

= .50, and a significant main effect of Time on percentage of doing independent work, F(19, 95) = 3.56, p < .01,

η_{p}^{2}

= .42. A post hoc analysis using Sidak adjustment showed that children in the control group did significantly more independent work at Time 7 (M_diff = 25.83, SE = 9.24), Time 9 (M_diff = 70.49, SE = 14.15), Time 14 (M_diff = 62.13, SE = 15.44), and Time 16 (M_diff = 66.89, SE = 19.35) compared with children in the treatment group. One exception was found at Time 3; children in the treatment group had a significantly higher percentage of doing independent work (M_diff = −28.83, SE = 4.32) than children in the control group. Overall, however, children in the control group exhibited more time on task over the course of the intervention in the area of doing independent work than children in the treatment group.

Percentage of Time Interacting With the Teacher.

A 20 × 2 mixed-design ANOVA indicated a significant interaction effect between Time and Group, F(19, 95) = 2.94, p < .01,

η_{p}^{2}

= .37, and a significant main effect of Time on the percentage of interaction with the classroom teacher, F(19, 95) = 3.52, p < .01,

η_{p}^{2}

= .41 (Table 2). A follow-up analysis on the significant interaction effect using Sidak adjustment shows that children in the control group had significantly more interaction with their teacher at Time 3 (M_diff = 55.26, SE = 9.01), Time 6 (M_diff = 13.95, SE = 4.48), Time 17 (M_diff = 49.48, SE = 15.59), and Time 18 (M_diff = 43.94, SE = 4.37) compared with children in the treatment group. However, children in the treatment group showed a significantly higher percentage of interacting with the teacher compared with children in the control group at Time 9 (M_diff = −67.05, SE = 21.44) and Time 14 (M_diff = −26.23, SE = 7.95). Overall, children in the treatment group exhibited more time on task over the course of the intervention in the area of interacting with their teacher than did children in the control group.

Table 2.

Results of the Mixed-Design Analyses of Variance

Source	SS	MS	F	df	p	η²
Percentage of time working with peers
Time	22,495.48	1,183.97	5.03	19	.000	.50
Time × Group	18,905.59	995.03	4.23	19	.000	.46
Error (Time)	22,374.15	235.52	—	95	—	—
Group	169.90	169.90	0.32	1	.596	.06
Error (Group)	2,652.02	530.40	—	5	—	—
Percentage of time doing independent work
Time	28,416.03	1,495.58	3.52	19	.00	.41
Time × Group	23,719.55	1,248.40	2.94	19	.00	.37
Error (Time)	40,351.22	424.75	—	95	—	—
Group	1,934.28	1,934.28	3.37	1	.13	.40
Error (Group)	2,867.27	573.45	—	5	—	—
Percentage of time interacting with the teacher
Time	23,127.05	1,217.21	3.56	19	.000	.42
Time × Group	31,881.31	1,677.96	4.91	19	.000	.50
Error (Time)	32,469.35	341.78	—	95	—	—
Group	1,059.70	1,059.70	2.13	1	.204	.30
Error (Group)	2,482.34	496.47	—	5	—	—
Literacy
Time	7,077.19	7,077.19	136.47	1	<.01	.68
Time × Group	14.82	14.82	0.29	1	.59	.00
Error (Time)	3,318.94	51.86	—	64	—	—
Group	1,263.67	1,263.67	3.45	1	.07	.05
Error (Group)	23,475.91	366.81	—	64	—	—
Mathematics
Time	5,346.21	5,346.21	140.72	1	<.01	.69
Time × Group	112.33	112.33	2.96	1	.09	.04
Error (Time)	2,431.55	37.99	—	64	—	—
Group	525.45	525.45	2.39	1	.13	.04
Error (Group)	14,059.76	219.68	—	64	—	—

Note. — = not applicable; df = degrees of freedom; MS = mean square; SS = sum of squares.

View Large

Table 2.

Results of the Mixed-Design Analyses of Variance

Source	SS	MS	F	df	p	η²
Percentage of time working with peers
Time	22,495.48	1,183.97	5.03	19	.000	.50
Time × Group	18,905.59	995.03	4.23	19	.000	.46
Error (Time)	22,374.15	235.52	—	95	—	—
Group	169.90	169.90	0.32	1	.596	.06
Error (Group)	2,652.02	530.40	—	5	—	—
Percentage of time doing independent work
Time	28,416.03	1,495.58	3.52	19	.00	.41
Time × Group	23,719.55	1,248.40	2.94	19	.00	.37
Error (Time)	40,351.22	424.75	—	95	—	—
Group	1,934.28	1,934.28	3.37	1	.13	.40
Error (Group)	2,867.27	573.45	—	5	—	—
Percentage of time interacting with the teacher
Time	23,127.05	1,217.21	3.56	19	.000	.42
Time × Group	31,881.31	1,677.96	4.91	19	.000	.50
Error (Time)	32,469.35	341.78	—	95	—	—
Group	1,059.70	1,059.70	2.13	1	.204	.30
Error (Group)	2,482.34	496.47	—	5	—	—
Literacy
Time	7,077.19	7,077.19	136.47	1	<.01	.68
Time × Group	14.82	14.82	0.29	1	.59	.00
Error (Time)	3,318.94	51.86	—	64	—	—
Group	1,263.67	1,263.67	3.45	1	.07	.05
Error (Group)	23,475.91	366.81	—	64	—	—
Mathematics
Time	5,346.21	5,346.21	140.72	1	<.01	.69
Time × Group	112.33	112.33	2.96	1	.09	.04
Error (Time)	2,431.55	37.99	—	64	—	—
Group	525.45	525.45	2.39	1	.13	.04
Error (Group)	14,059.76	219.68	—	64	—	—

Note. — = not applicable; df = degrees of freedom; MS = mean square; SS = sum of squares.

Effect on Academic Performance

Literacy.

A 2 × 2 mixed-design ANOVA indicated a significant main effect of Time on students’ performance in literacy, F(1, 64) = 136.47, p < .01,

η_{p}^{2}

= .68. However, neither the main effect of the intervention, F(1, 64) = 3.45, p = .07,

η_{p}^{2}

= .05, nor an interaction effect between Time and Group, F(1, 64) = 0.29, p = .60,

η_{p}^{2}

= .004, was found on students’ performance in literacy. In other words, students showed a significant increase in literacy achievement from fall (M = 176.06, SD = 15.83, n = 66) to spring (M = 190.77, SD = 13.47, n = 66), regardless of whether they were in the intervention classrooms.

Mathematics.

A 2 × 2 mixed-design ANOVA indicated a significant main effect of Time on students’ performance in mathematics, F(1, 64) = 140.72, p < .01,

η_{p}^{2}

= .69. However, neither the main effect of the intervention, F(1, 64) = 2.96, p = .09,

η_{p}^{2}

= .04, nor an interaction effect between Time and Group, F(1, 64) = 0.29, p = .60,

η_{p}^{2}

= .004, was found on students’ performance in mathematics. In other words, students showed a significant increase in mathematics achievement from fall (M = 179.33, SD = 11.13, n = 66) to spring (M = 190.20, SD = 11.82, n = 66), regardless of whether they were in the intervention classrooms.

Discussion

This study evaluated the effect of using stability balls on 67 second-grade students’ academic achievement and discipline referral levels and on 16 students’ on-task behavior. The findings associated with contextual aspects of on-task behavior and overall discipline referral data in this study indicate that the use of stability balls in lieu of chairs at the elementary level could potentially be considered a promising intervention. Generally, however, findings suggest that the implementation of stability ball use did not result in significantly greater on-task behavior or academic achievement, as we had hypothesized. Therefore, this study demonstrated that for students in the treatment classrooms, stability ball use resulted in mixed findings.

On-Task Behavior

Overall, the children sitting in chairs had higher levels of on-task behavior than those sitting on stability balls. Previous research examining use of stability balls in the classroom has provided potential explanations for this finding. An important consideration is the developmental level of the participant being studied; most previous studies exploring the effect of stability ball use on educational variables focused on elementary-grade students as young as preschool (Schilling & Schwartz, 2004) and as old as fifth grade (Fedewa & Erwin, 2011). Significant improvements in positive behaviors (e.g., task engagement or in-seat behavior) are well documented for elementary-age children with ADHD (e.g., Schilling et al., 2003; Wu et al., 2012). However, this study found that nondisabled, typically developing students showed positive therapeutic trends but not significant improvements resulting from stability ball use (Fedewa & Erwin, 2011; Wu et al., 2012). Thus, students who do not have significant learning or behavioral concerns may not experience similarly positive outcomes from the use of stability balls as students who are not typically developing.

Prior studies have found that students with a diagnosis of or at risk for ADHD using stability balls showed significant improvements in levels of attention (Fedewa & Erwin, 2011), reaction time and latency measures (Wu et al., 2012), and in-seat behavior and legible handwriting (Schilling et al., 2003). Furthermore, children with ASD using stability balls have shown immediate and substantial improvements in levels of engagement and in-seat behavior (Bagatell et al., 2010; Schilling & Schwartz, 2004). Fedewa and Erwin (2011) found that students who had the most significant attention difficulties appeared to benefit most from the use of therapy balls. This line of reasoning suggests that elementary students with academic, social, or behavioral difficulties (e.g., ADHD or ASD) may benefit more from the use of stability balls in lieu of chairs compared with the general education population. For special education populations, stability balls potentially prevent disengagement by addressing sensory or physical activity needs through acceptable and enjoyable movement. Older, typically developing students (i.e., young adults) using therapy balls have also shown improvements in self-reported comfort levels and instructor-rated positive classroom behaviors (Al-Eisa, Buragadda, & Rao Melam, 2013), suggesting that further investigation of stability ball effectiveness for older students would be useful.

A design consideration regarding the on-task behavior observations is that only 4 children in each classroom were observed. In our previous research (Fedewa & Erwin, 2011), average scores of all 76 students with ADHD dropped; in the present study, 4 students may have been too limited a sample, limiting generalizability. Further, although the present study is the longest known intervention examining the effectiveness of stability balls in improving education-related variables, a longer investigation may be needed to capture long-term behavioral or academic change. An additional methodological consideration is that the students were not sitting on their stability balls for the entire time observers collected on-task and off-task observations. Previous studies observed stability ball use for short durations (e.g., Schilling & Schwartz, 2004, conducted several 5- to 10-min data collection sessions), but the 30-min observations conducted in this study often included unanticipated times when students sat on the floor as a class for various activities. Perhaps the movement from the stability ball to the floor affected what observers were able to capture in terms of on-task and off-task behavior.

The finding of lower levels of on-task behavior in the treatment classrooms in the present study may indicate that the stability balls were a distraction to the children. Fedewa and Erwin (2011) found that both observers and teachers reported that students using stability balls were “more active” (p. 397). Although evidence exists of the importance of physical activity for educational factors (e.g., Fedewa & Ahn, 2011; Tomporowski, Davis, Miller, & Naglieri, 2008), and although moving on stability balls might be more natural than moving in chairs, moving more during learning might make being off-task easier. Evidence suggesting that moving (e.g., walking or “shuffling” in one’s seat) can improve learning outcomes requires additional research (Carson, Shih, & Langer, 2001). More research is needed to confirm the reasons general education students seated on stability balls would be more off task than students seated in chairs.

Finally, our evidence suggests that children in the treatment group exhibited more time in on-task behavior specific to interacting with the teacher (e.g., listening to the teacher) compared with the control group. This was the only on-task behavior on which the treatment group scored higher than the control group. This finding could be associated with differences in the structure of classrooms where stability balls are used; for example, students with stability balls may be able to move in more natural ways that allow easier orientation toward the teacher. Use of stability balls might thus be beneficial for classroom situations requiring teacher–student interaction. A final possibility is that the teachers of the treatment classrooms may simply have had a different style of teaching and interaction with their students before the integration of the stability balls and that the observations merely captured this previously existing difference between groups.

Academic Achievement

Stability ball use had no effect on participants’ achievement levels in math and literacy. The present study is the first stability ball study we know of to explore achievement outcomes in standardized tests of math and literacy in elementary school students. A study of stability ball use by female physical therapy undergraduate students found higher scores for the treatment group in measures of participation, comprehension, cooperation, and knowledge (Al-Eisa et al., 2013), so the academic outcomes of stability ball use might improve with age; further exploration of the impact of stability ball use on academic achievement is needed. In addition, a longer-term intervention may be warranted to explore whether student achievement outcomes are influenced by the use of stability balls in the classroom for a longer period.

Discipline Referrals

We found an overall decrease in discipline referrals for the treatment classrooms. This reduction may be associated with the extra physical activity required in using stability balls. Physical activity research has found benefits for executive control task performance, attention, and overall positive behaviors (Fedewa, Erwin, & Ahn, 2012; Mahar, 2011; Tomporowski et al., 2008; Voelcker-Rehage & Niemann, 2013). Stability balls have proved effective for purposes related to strength training, conditioning, and physical fitness (Jakubek, 2007). In addition, stability ball use may help curb the more serious behavioral problems that result in discipline referrals.

Limitations

Although this study is the first to implement a randomized controlled design to examine the effectiveness of stability balls, a number of methodological limitations influence interpretation of the results. As previously mentioned, the sample size of 4 children per classroom to assess on-task behavior was small, limiting the external validity of this finding. The length of time we investigated the effectiveness of the stability ball intervention was longer than that in previous studies, but for the general education population, perhaps more data and a longer time frame are needed to properly assess time-sensitive variables such as achievement and behavior. Moreover, as mentioned by Fedewa and Erwin (2011), it is unlikely that stability balls would be provided to every regular education student in a classroom because of limited school resources, further limiting potential generalizability to most general education classrooms. Last, unlike several prior stability ball studies, this study used no measure of student or teacher acceptability or social validity. Because little research on stability ball use has used general education students as participants, it would have been beneficial to include qualitative data to assess student and teacher perceptions of the advantages and disadvantages of stability ball use in classrooms.

Future Directions

Future researchers should consider using a larger sample size for collection of on-task behavior data. It would also be beneficial to collect data across several different grades within and beyond elementary school to explore developmental aspects related to stability ball use. Developmental aspects associated with age, gender, and disability status might further the understanding of when stability balls used in lieu of chairs in the classroom are most beneficial. Based on the findings from the present study, research assessing the effectiveness of stability ball use in the general education population should be no shorter than 1 yr in duration and would preferably be longer to determine long-term effectiveness, including follow-up measures. An additional consideration is the effect of individual teaching styles and classroom structure within and across grades. Studies should also include teacher and student impressions of stability ball use to further establish evidence of teacher-reported acceptance and student improvement (Fedewa & Erwin, 2011).

Implications for Occupational Therapy Practice

The findings of this study have the following implications for occupational therapy practice:

The use of stability balls in the classroom setting in lieu of chairs may be effective in reducing student discipline referrals. Stability balls potentially provide needed physical activity, which can increase positive classroom behaviors, fitness, and mental health.
Additional research is needed to confirm our finding that children who sat in chairs had higher levels of on-task behavior, but at this time the evidence indicates that stability balls are most beneficial for students with attention or behavioral difficulties or those with ASD. At the second-grade level, the use of stability balls for entire general education classrooms has not yet been shown to be a practical use of school resources.
Despite the academic achievement benefits associated with physical activity such as stability ball use, the current evidence does not promote the use of stability balls to improve students’ achievement levels in math and literacy. Although stability balls did not adversely affect children’s achievement outcomes, the lack of significant difference in children’s achievement scores provides no support for occupational therapists to promote using stability balls as a means to improve academic outcomes for a general education population.

References

Al-Eisa, E., Buragadda, S., & Rao Melam, G. (2013). Effect of therapy ball seating on learning and sitting discomforts among Saudi female students. BioMed Research International, 2013, 1–4. http://dx.doi.org/10.1155/2013/153165 [Article]

Bagatell, N., Mirigliani, G., Patterson, C., Reyes, Y., & Test, L. (2010). Effectiveness of therapy ball chairs on classroom participation in children with autism spectrum disorders. American Journal of Occupational Therapy, 64, 895–903. http://dx.doi.org/10.5014/ajot.2010.09149 [Article] [PubMed]

Bill, V. N. (2008). Effects of stability balls on behavior and achievement in the special education classroom. Unpublished master’s thesis, Southwest Minnesota State University, Marshall.

Bronfenbrenner, U. (1979). The ecology of human development. Cambridge, MA: Harvard University Press.

Carriere, B. (1998). The Swiss ball: Theory, basic exercise, and clinical application. Berlin: Springer-Verlag.

Carson, S., Shih, M., & Langer, E. (2001). Sit still and pay attention? Journal of Adult Development, 8(3), 183–188. [Article]

Cohen, J. (1988). Statistical power analysis for the behavioral sciences. Hillsdale, NJ: Erlbaum.

Del Boca, F. K., & Darkes, J. (2007). Enhancing the validity and utility of randomized clinical trials in addictions treatment research: I. Treatment implementation and research design. Addiction and Its Sciences, 102, 1047–1056. http://dx.doi.org/10.1111/j.1360-0443.2007.01862.x [Article]

Donnelly, J. E., & Lambourne, K. (2011). Classroom-based physical activity, cognition, and academic achievement. Preventive Medicine, 52(Suppl. 1), S36–S42. http://dx.doi.org/10.1016/j.ypmed.2011.01.021 [Article] [PubMed]

Dotterer, A. M., & Lowe, K. (2011). Classroom context, school engagement, and academic achievement in early adolescence. Journal of Youth and Adolescence, 40, 1649–1660. http://dx.doi.org/10.1007/s10964-011-9647-5 [Article] [PubMed]

Fedewa, A. L., & Ahn, S. (2011). The effects of physical activity and physical fitness on children’s cognitive outcomes: A meta-analysis. Research Quarterly for Exercise and Sport, 82, 521–535. [Article] [PubMed]

Fedewa, A. L., & Erwin, H. E. (2011). Stability balls and students with attention and hyperactivity concerns: Implications for on-task and in-seat behavior. American Journal of Occupational Therapy, 65, 393–399. http://dx.doi.org/10.5014/ajot.2011.000554 [Article] [PubMed]

Fedewa, A. L., Erwin, H., & Ahn, S. (2012). Student academic performance outcomes of a classroom physical activity intervention: A pilot study. International Electronic Journal of Elementary Education, 4, 473–487.

Gutkin, T. B. (2012). Ecological psychology: Replacing the medical model paradigm for school-based psychological and psychoeducational services. Journal of Educational and Psychological Consultation, 22, 1–20. http://dx.doi.org/10.1080/10474412.2011.649652 [Article]

Jakubek, M. D. (2007). Stability balls: Reviewing the literature regarding their use and effectiveness. Strength and Conditioning Journal, 29(5), 58–63. http://dx.doi.org/10.1519/1533-4295(2007)29[58:SBRTLR]2.0.CO;2

Mahar, M. T. (2011). Impact of short bouts of physical activity on attention-to-task in elementary school children. Preventive Medicine, 52(Suppl. 1), S60–S64. http://dx.doi.org/10.1016/j.ypmed.2011.01.026 [Article] [PubMed]

Rapp, J. T., Colby-Dirksen, A. M., Michalski, D. N., Carroll, R. A., & Lindenberg, A. M. (2008). Detecting changes in simulated events using partial-interval recording and momentary time sampling. Behavioral Interventions, 23, 237–269. http://dx.doi.org/10.1002/bin.269 [Article]

Schilling, D. L., & Schwartz, I. S. (2004). Alternative seating for young children with autism spectrum disorder: Effect on classroom behavior. Journal of Autism and Developmental Disorders, 34, 423–432. [Article] [PubMed]

Schilling, D. L., Washington, K., Billingsley, F. F., & Deitz, J. (2003). Classroom seating for children with attention deficit hyperactivity disorder: Therapy balls versus chairs. American Journal of Occupational Therapy, 57, 534–541. http://dx.doi.org/10.5014/ajot.57.5.534 [Article] [PubMed]

Tomporowski, P. D., Davis, C. L., Miller, P. H., & Naglieri, J. A. (2008). Exercise and children’s intelligence, cognition, and academic achievement. Educational Psychology Review, 20, 111–131. http://dx.doi.org/10.1007/s10648-007-9057-0 [Article] [PubMed]

Voelcker-Rehage, C., & Niemann, C. (2013). Structural and functional brain changes related to different types of physical activity across the life span. Neuroscience and Biobehavioral Reviews, 37, 2269–2295. http://dx.doi.org/10.1016/j.neubiorev.2013.01.028 [Article]

Wu, W., Wang, C., Chen, C., Lai, C., Yang, P., & Guo, L. (2012). Influence of therapy ball seats on attentional ability in children with attention deficit/hyperactivity disorder. Journal of Physical Therapy Science, 24, 1177–1182. [Article]

Table 1.

Mean Percentage of Time Students Were On Task, by Group (N = 16)

Type of Behavior	Mean % of Time On Task
Type of Behavior	Control Group	Intervention Group
On task	87	77
Working with peers	15	13
Doing independent work	39	29
Interacting with the teacher	33	35
Off task	13	23

View Large

Table 1.

Mean Percentage of Time Students Were On Task, by Group (N = 16)

Type of Behavior	Mean % of Time On Task
Type of Behavior	Control Group	Intervention Group
On task	87	77
Working with peers	15	13
Doing independent work	39	29
Interacting with the teacher	33	35
Off task	13	23

Table 2.

Results of the Mixed-Design Analyses of Variance

Source	SS	MS	F	df	p	η²
Percentage of time working with peers
Time	22,495.48	1,183.97	5.03	19	.000	.50
Time × Group	18,905.59	995.03	4.23	19	.000	.46
Error (Time)	22,374.15	235.52	—	95	—	—
Group	169.90	169.90	0.32	1	.596	.06
Error (Group)	2,652.02	530.40	—	5	—	—
Percentage of time doing independent work
Time	28,416.03	1,495.58	3.52	19	.00	.41
Time × Group	23,719.55	1,248.40	2.94	19	.00	.37
Error (Time)	40,351.22	424.75	—	95	—	—
Group	1,934.28	1,934.28	3.37	1	.13	.40
Error (Group)	2,867.27	573.45	—	5	—	—
Percentage of time interacting with the teacher
Time	23,127.05	1,217.21	3.56	19	.000	.42
Time × Group	31,881.31	1,677.96	4.91	19	.000	.50
Error (Time)	32,469.35	341.78	—	95	—	—
Group	1,059.70	1,059.70	2.13	1	.204	.30
Error (Group)	2,482.34	496.47	—	5	—	—
Literacy
Time	7,077.19	7,077.19	136.47	1	<.01	.68
Time × Group	14.82	14.82	0.29	1	.59	.00
Error (Time)	3,318.94	51.86	—	64	—	—
Group	1,263.67	1,263.67	3.45	1	.07	.05
Error (Group)	23,475.91	366.81	—	64	—	—
Mathematics
Time	5,346.21	5,346.21	140.72	1	<.01	.69
Time × Group	112.33	112.33	2.96	1	.09	.04
Error (Time)	2,431.55	37.99	—	64	—	—
Group	525.45	525.45	2.39	1	.13	.04
Error (Group)	14,059.76	219.68	—	64	—	—

Note. — = not applicable; df = degrees of freedom; MS = mean square; SS = sum of squares.

View Large

Table 2.

Results of the Mixed-Design Analyses of Variance

Source	SS	MS	F	df	p	η²
Percentage of time working with peers
Time	22,495.48	1,183.97	5.03	19	.000	.50
Time × Group	18,905.59	995.03	4.23	19	.000	.46
Error (Time)	22,374.15	235.52	—	95	—	—
Group	169.90	169.90	0.32	1	.596	.06
Error (Group)	2,652.02	530.40	—	5	—	—
Percentage of time doing independent work
Time	28,416.03	1,495.58	3.52	19	.00	.41
Time × Group	23,719.55	1,248.40	2.94	19	.00	.37
Error (Time)	40,351.22	424.75	—	95	—	—
Group	1,934.28	1,934.28	3.37	1	.13	.40
Error (Group)	2,867.27	573.45	—	5	—	—
Percentage of time interacting with the teacher
Time	23,127.05	1,217.21	3.56	19	.000	.42
Time × Group	31,881.31	1,677.96	4.91	19	.000	.50
Error (Time)	32,469.35	341.78	—	95	—	—
Group	1,059.70	1,059.70	2.13	1	.204	.30
Error (Group)	2,482.34	496.47	—	5	—	—
Literacy
Time	7,077.19	7,077.19	136.47	1	<.01	.68
Time × Group	14.82	14.82	0.29	1	.59	.00
Error (Time)	3,318.94	51.86	—	64	—	—
Group	1,263.67	1,263.67	3.45	1	.07	.05
Error (Group)	23,475.91	366.81	—	64	—	—
Mathematics
Time	5,346.21	5,346.21	140.72	1	<.01	.69
Time × Group	112.33	112.33	2.96	1	.09	.04
Error (Time)	2,431.55	37.99	—	64	—	—
Group	525.45	525.45	2.39	1	.13	.04
Error (Group)	14,059.76	219.68	—	64	—	—

Note. — = not applicable; df = degrees of freedom; MS = mean square; SS = sum of squares.