The Effects of Goal Difficulty on Exercise Achievement Among Individuals with Autism Spectrum Disorder

Author: Margaret Dannevik Pavone, PhD, BCBA, LBA
Affiliation: Lindenwood University School of Education

Abstract

In this study, four participants with autism spectrum disorder volunteered for a group exercise program in the community. The purpose of the study was to see whether goal difficulty affected performance for participants on four exercise tasks. Each individual was given a different type of goal for each of four tasks. The assignment of goal type to exercise task was randomized so that each individual was exposed to four goal types and was given the chance to perform all four exercises. For all four skills and four participants, difficult goals produced the largest improvement in performance. This is aligned with findings in the general population from recent studies of goal setting in organizational contexts (Roose & Williams, 2018). Findings from this study suggest that individuals with ASD may perform at their best when given challenging goals regardless of initial capability. Implications for practice, policy, and future research are discussed.

Introduction

The purpose of this study was to examine the effect of goal difficulty on the achievement of fitness-related goals among adults with Autism Spectrum Disorder. To accomplish this, a carefully constructed single subject design experiment was conducted. To control for extraneous variables and individual differences, each subject learned four novel exercises under each of four different goal conditions. The order of exercises and assignment of exercise to goal condition was counterbalanced throughout the subjects. The specific dependent variables under study included four exercise movements that were unfamiliar to the participants. Specific behaviors and dimensions are outlined in Table 1. Rest was operationally defined as stopping the exercise task for 5 seconds or engaging in any other behavior other than the exercise task for 5 seconds. 

Table 1

The independent variable in this study was the goal difficulty assigned to each exercise. Goal difficulties were chosen based on recent research (Roose, 2014; Roose & Williams, 2018), as well as anecdotal information shared by participants. Specifically, participants and their guardians reported that in previous attempts to achieve fitness related objectives, participants were often given arbitrary goals or told simply to “do your best.” As these were commonly reported practices in the applied setting, “do your best” and “arbitrary” goals were included as additional independent variables in this study. Independent variables are listed in Table 2.

Table 2

Method

To control for individual differences in performance baselines across each of the different exercise tasks and participants, IV/DV combinations were counterbalanced carefully. Specifically, each participant was exposed to all four exercises and all four goal varieties, but no participant had the same combination of IV and DV in any condition as another participant. The assignment of IV and DV combinations for each participant is outlined in Table 3. This counterbalancing not only controlled for extraneous variables but allowed the researcher to use a flexible single subject design to separate the effects of goal difficulty from other influential factors (Connell & Thompson, 1986; Holcombe, Wolery, & Gast, 1994).

Table 3

To determine the effect of goal difficulty on performance, an embedded alternating treatment design was used within a multiple baseline across participants design. In this design, participants were exposed to all four independent/dependent variable dyads during each session. The order of exercises was varied so that a participant that started the session with pushups on Monday would start Tuesday’s session with walking, Wednesday’s with wall sits, Thursday’s with pushups, etc. The multiple baseline design across participants was implemented concurrently, so that session one occurred on the same calendar day for all participants. This controlled for history effects while the short time frame of the experiment helps to control for maturity effects (Kennedy, 2005).

Participants, Setting, and Materials

Participants included four young adults with a diagnosis of Autism Spectrum Disorder. The mean age of participants was 23.25 years with a range of 19 years to 28 years old. All participants communicated using vocal, verbal English and could participate in typical age-appropriate conversations. None of the participants reported having experience with any of the exercises that would be taught during the experiment. In addition, none of the participants reported engaging in regular physical activity in their typical routines.

Participants were all part of a community-based fitness program conducted at a local park in the Midwest. Participants and their families were provided information about the study, its risks and benefits, and details about participation. Four participants and guardians volunteered to take part and gave informed consent. Participants also had to provide a statement from their physician stating they were medically cleared to exercise within the parameters of the fitness program.

Participants were paired with a coach who worked 1:1 with them for 9 weekly exercise sessions, each lasting 45 minutes. Coaches received basic training regarding safe performance of each exercise and were instructed to interrupt any unsafe behaviors and provide verbal redirection. In addition, coaches provided non specific verbal praise for effort and for continuing to work during session on a variable interval schedule. Participants were free to refuse an exercise and to stop participating in a session at any time. Water breaks were provided non contingently and snacks in the form of fresh fruits were available during all sessions.      

Each session was divided into 4 components. During baseline, each component included 10 minutes of instruction, modeling, rehearsal, and feedback-as is typical of a behavioral skills training procedure (Miles & Wilder, 2009). Following BST, a baseline measure was taken and recorded by the coach. Participants were not made aware of their achievement during baseline.

After 3 sessions of baseline, the BST portion of each component was changed to a goal-setting component. During goal-setting, the participant was given a visual representation of their performance in the previous session and a verbal goal stated by the coach. The participant-coach dyad then participated in the exercise and recorded data on the coach’s data spreadsheet. After the performance, comparative feedback was given. An example of stated goals and comparative feedback for each condition is included in Table 4.

Table 4

Data Collection and Interobserver Agreement

Data was collected on the dependent variables by each participant’s coach during session. Coaches used the stopwatch feature on their smartphones to collect duration data. Coaches used a smartphone application to enter performance data on a Google Spreadsheet so that all data collection occurred in real time. Coaches entered data 4 times during each session, each immediately following their participant’s performance on one of the exercise tasks.

Interobserver agreement (IOA) data was taken by the author during three of the nine sessions for each dyad. Exact count per session method was used to ensure maximum sensitivity (Kennedy, 2005). Mean IOA was 85%. Lower IOA scores were most common during the wall sit exercises. This is likely due to the more sensitive dimension (seconds rather than minutes) and the conservative method of IOA. Treatment integrity (TI) data was collected by the author during three of the nine sessions for each dyad. Treatment integrity feedback was delivered to the coaches following data collection. TI scores for all dyads and all sessions was 100%.

Results

Aggregate results of all participants’ performances in each goal condition are represented in Figure 1. For all four participants, regardless of the exercise with which it was paired and regardless of initial performance level, the largest improvements in performance were seen under the difficult goal condition. Each participant had a more significant improvement from baseline performance when difficult goals were set. Easy goals, arbitrary goals, and do your best goals affected performance differently for each participant.

The results for each individual participant are displayed in figures 2 through 5. For participant one, performance improved in all four conditions with the largest gains occuring in the difficult goal condition. For participant two, the largest performance improvement occurred in the difficult goal condition. Performance in the easy goal and arbitrary goal conditions improved to a lesser degree, while performance in the do your best goal condition actually declined over time. Participant three improved in all four conditions, with the largest gains occurring in the difficult goal condition. Participant four had the most variable performance but still demonstrated improvements under all four goal conditions. Participant four also experienced the largest gains in performance under the difficult goal condition.

Results according to each dependent variable are displayed in figures 6 through 9. As evidenced below, participants varied considerably in initial performance ability. For walking, wall sits, and pushups all four goal conditions produced improvements in performance. For squats however, only arbitrary goals, difficult goals, and easy goals resulted in improvements. Performance on the squat task actually decreased in the do your best goal condition.

Discussion

In this study, the largest improvements in exercise performance were attained in the difficult goals (200%) condition. This difference in improvement was seen consistently, regardless of individual performance during baseline and independent of the type of exercise attempted.Some future directions to explore include the following:

• Does goal acceptance mediate the results of performance under difficult goal conditions? Is there a difference in performance if verbal goals are received positively by the performer? (Erez & Zidon, 1984)
• When do difficult goals become too hard and end up impeding performance? There is some evidence that impossible goals are negatively related to performance among certain populations and tasks. Future researchers should use parametric analysis to determine ideal levels of difficulty when setting goals for individuals with Autism Spectrum Disorder as well as neurotypical populations.
• Does goal difficulty affect quality of performance in addition to quantity? The measures used in this study were primarily of quantity of exercise, not quality. Some evidence suggests that goal difficulty may differentially affect task performance quality when compared to task performance quantity (Bavelas & Lee, 1978). In tasks where quality of performance is more important than repetition or duration, this may be a more important question to answer in applied settings. 

In addition to suggesting future areas of inquiry, the results indicate a need for clinicians, policy-makers, and educators to re-evaluate the way that the autism support community sets goals for individuals on the Autism spectrum. Because performance improved most for exercise tasks paired with difficult goals, this should be considered when creating individual support plans or individualized education plans. Instead of choosing norm-based, arbitrary, or “do your best” type goals, higher levels of achievement might be attained using difficult goals. The results of this study support the need to evaluate performance ceilings that may impede achievement without being reflective of ability or potential. Performance ceilings related to procedures, goal setting procedures specifically, limit the achievement of individual performers (Binder, 2010). This research suggests that arbitrary goals, easy goals, or vaguely defined goals-like do your best goals- may introduce artificial limit on individual performance that inhibit maximum achievement.

For individual clinicians or support persons that work one on one with individuals on the ASD spectrum, this knowledge could create a paradigm shift in the way goals are set on a day to day basis. Specifically, clinicians or support persons may want to encourage clients and loved ones to set higher goals in areas such as fitness related achievement. By setting lower, more easily-achievable goals, an individual with ASD may perform at a level lower than their potential best.

In some cases, difficult goals have been reported to cause frustration, low rates of reinforcement, and even refusal to attempt a performance (Cooper, Heron, & Heward, 2007; Fellner & Sulzer-Azaroff, 1984). However, this research suggests difficult goals can also produce higher levels of performance overall. Clients experiencing frustration may need to work on approximations of the final goal to increase the rate of reinforcement produced by their efforts (Fellner & Sulzer-Asaroff, 1984), while clinicians continue to encourage high rates of final performance.

References

Bavelas, J. B., & Lee, E. S. (1978). Effects of goal level on performance: a trade-off of quantity and quality. Canadian Journal of Psychology/Revue canadienne de psychologie, 32(4), 219.

Binder, C. (2010). Building fluent performance: Measuring response rate and multiplying response opportunities. The Behavior Analyst Today, 11(4), 214.

Connell, P. J., & Thompson, C. K. (1986). Flexibility of single-subject experimental designs. Part III: Using flexibility to design or modify experiments. Journal of Speech and Hearing Disorders, 51(3), 214-225.

Erez, M., & Zidon, I. (1984). Effect of goal acceptance on the relationship of goal difficulty to performance. Journal of applied psychology, 69(1), 69.

Fellner, D. J., & Sulzer-Azaroff, B. (1984). A behavioral analysis of goal setting. Journal of Organizational Behavior Management, 6(1), 33-51.

Holcombe, A., Wolery, M., & Gast, D. L. (1994). Comparative single-subject research: Description of designs and discussion of problems. Topics in Early Childhood Special Education, 14(1), 119-145.

Kennedy, C. H. (2005). Single-case designs for educational research. Pearson/A & B. Miles, N. I., & Wilder, D. A. (2009). The effects of behavioral skills training on caregiver implementation of guided compliance. Journal of Applied Behavior Analysis, 42(2), 405-410.

Roose, K. M. (2014). An Examination of the Effects of Performance Improvement Goals and Feedback on Performance and Persistence on an Analog Work Task (Doctoral dissertation).

Roose, K. M., & Williams, W. L. (2018). An evaluation of the effects of very difficult goals. Journal of Organizational Behavior Management, 38(1), 18-48.

Sitkin, S. B., See, K. E., Miller, C. C., Lawless, M. W., & Carton, A. M. (2011). The paradox of stretch goals: Organizations in pursuit of the seemingly impossible. Academy of Management Review, 36(3), 544-566.

Dr. Pavone is an assistant professor and the program chair for the behavior analysis graduate program at Lindenwood University in Saint Charles, MO. Research interests include the application of the science of human behavior to socially significant issues including education, organizational behavior management, prevention of obesity and chronic illness, and dissemination of behavior analysis to underserved populations.