Applied behavior analysis (ABA) is a well-established treatment for autism spectrum disorder (ASD; Eldevik et al. 2009; Reichow 2012; Reichow et al. 2012). ABA has been found to effectively improve performance on measures of intellectual functioning, language, adaptive skills, and social skills for individuals with ASD (Eldevik et al. 2009; Makrygianni and Reed 2010; Peters-Scheffer et al. 2011; Reichow 2012; Reichow and Wolery 2009; Virues-Ortega 2010). Availability of behavioral providers to assist all those in need of services is an area of concern (Wise et al. 2010). Considering the intensive and one-to-one nature of ABA-based services, (Behavior Analyst Certification Board 2014; Eldevik et al. 2009; Reichow et al. 2012) as well as the latest reported prevalence rates of ASD, which estimate that 1 in 59 children in the USA are diagnosed (Autism and Developmental Disabilities Monitoring Network 2018), there is an increasing need for behavior technicians, the direct-care staff that conduct ABA-based treatment.

Typically, ABA-based treatment is supervised by master’s or doctoral-level licensed and/or certified clinicians (e.g., Board Certified Behavior Analyst, marriage and family therapists, speech and language pathologist, etc.), who are responsible for assessing skill deficits, developing treatment plans, tracking progress, and adjusting treatment goals among other responsibilities (Behavior Analyst Certification Board 2014). Supervisors oversee behavior technicians, sometimes referred to as other titles (e.g., therapists, paraprofessionals, tutors, etc.), who are responsible for providing direct services to individuals with ASD. Behavior technicians typically have high school or bachelor’s-level educations and are increasingly being required to obtain certification as they gain experience (e.g., Board Certified Autism Technician, Registered Behavior Technician; Behavior Analyst Certification Board 2014, Behavioral Intervention Certification Council 2015). Behavior technicians are responsible for implementing treatment plans, which are developed by the supervising clinician.

It is often assumed that behavior technician turnover is high; however, estimated rates of turnover for behavior technicians have not been reported. Examining turnover rates in similar professional fields may offer some insight. Rates of turnover for community-based direct-care staff working with individuals with developmental disabilities (DD) have been reported as high as 70.7% to 77% per year (Mitchell and Braddock 1994; Seninger and Traci 2002).

High turnover may have negative implications on organizations, staff, and patients with ASD. Training for behavior technicians is extensive, requiring large commitments of time and resources by both employers and employees. As in the health care industry, loss of productivity and costs of hiring and training new staff places a financial burden on providers (Waldman et al. 2004). Further, high turnover may also negatively impact remaining employees’ morale and workload as experienced staff are replaced by untrained staff (Sulek et al. 2017). Finally, frequent turnover may disrupt the continuity of services, which may potentially have a negative impact on patients’ progress in treatment.

Identifying predictors of turnover may be beneficial to service providers as it gives providers areas to address proactively to reduce turnover. Pertinent information may be gleaned from research on predictors of turnover and burnout in other demanding professions, like child welfare workers (DePanfilis and Zlotnik 2008; Kim and Kao 2014), direct-care staff in services for individuals with intellectual disability (ID) and DD (Gray and Muramatsu 2013; Hatton et al. 2001), and special education teachers (Billingsley 2004; Brunsting et al. 2014); however, working specifically with individuals with ASD poses unique challenges. For instance, teachers working with children with ASD have been found to experience greater levels of burnout in comparison to teachers working with children with ID and children who are deaf (Zarafshan et al. 2013).

The primary purpose of this paper was to conduct a systematic review of predictors of staff retention and turnover for behavior technicians working with individuals with ASD. Due to the scarcity of this research, the current review was expanded to include predictors of burnout and job satisfaction as these factors have been linked to intention to turnover (Kim and Kao 2014). Findings of the current review should be informative for hiring practices, ongoing supervision, and support of behavior technicians working with individuals with ASD.

Methods

The databases PsycINFO and PubMed were searched. Results were limited to peer-reviewed articles with a publication year of 2000 or later. A combination of the following terms was used in the searches: “autism,” “Asperger’s,” “pervasive developmental disorder,” “therapists,” “technicians,” “paraprofessionals,” “tutors,” “staff,” “direct care,” “retention,” “turnover,” “burnout,” and “job satisfaction.”

The titles and abstracts of articles were initially reviewed for relevancy. The methods of relevant articles were reviewed and the article was included if the following criteria were met: (a) was a research study; (b) participants included behavior technicians providing ABA to individuals with ASD; and (c) retention, turnover, burnout, or job satisfaction was a dependent variable. While these procedures may not have produced an exhaustive review, findings should be representative of factors associated with retention, turnover, burnout, and job satisfaction in behavior technicians.

Interobserver agreement (IOA) data were collected for study inclusion. The authors independently followed the procedures discussed previously for study inclusion. Percentage of agreement was calculated by dividing the total number of agreements by the total number of agreements plus disagreements and multiplying by 100 (Kazdin 1982). IOA for study inclusion was 96.9% (i.e., the authors disagreed on one article). The disagreement was discussed and resolved. Procedures were also in place to verify the data collected from the included studies. One author extracted data from the studies (i.e., sample size, setting, methods, dependent variables and measures, independent variables found to be significant, independent variables found to be insignificant). The second author reviewed and verified the extracted data. Any disagreements between the authors were resolved via discussion.

Results

A literature search of PsycINFO identified 31 publications and a literature search of PubMed identified 11 publications. After duplicates were removed, the titles and abstracts of 33 articles were screened. A total of 21 articles were found to be potentially relevant. Methods were reviewed for eligibility and only five articles met the inclusion criteria. Of the 16 articles that were excluded, 10 were excluded because samples did not include behavior technicians providing ABA to individuals with ASD; 4 were excluded because retention, turnover, burnout, or job satisfaction was not a dependent variable; and 2 reviews of literature were excluded. See Fig. 1 for a flow diagram of the search results and Table 1 for summaries of the included studies.

Fig. 1
figure 1

Results of systematic literature searches

Full size image
Table 1 Summary of reviewed studies
Full size table

Participants

Participants across all the reviewed studies included a total of 351 behavior technicians working with individuals with ASD. Samples of participants were predominately female, ranging between 89% and 100% female across studies. Three of the studies provided the geographical region where their sample of participants worked, which included California (Kazemi et al. 2015), Ireland (Gibson et al. 2009), and the United Kingdom (Griffith et al. 2014). The behavior technicians worked in home (Hurt et al. 2013; Kazemi et al. 2015) and school settings (Gibson et al. 2009; Griffith et al. 2014; Kazemi et al. 2015; Kelly and Barnes-Holmes 2013). All studies included behavior technicians practicing ABA; however, a portion of the sample included in Hurt et al. (2013) practiced other therapeutic methods (e.g., Floortime, pivotal response training, Treatment and Education of Autistic and Related Communication Handicapped Children). Finally, portions of the samples in Gibson et al. (2009) and Griffith et al. (2014) were reported to have supervisor-level roles (e.g., lead therapist, supervisor).

Methods

All the reviewed studies administered survey questionnaires or batteries of measures to samples of behavior technicians (Gibson et al. 2009; Griffith et al. 2014; Hurt et al. 2013; Kazemi et al. 2015; Kelly and Barnes-Holmes 2013). One study also administered a computer-based task to participants (Kelly and Barnes-Holmes 2013).

Dependent Variables and Measures

To be included, the reviewed studies must have explored turnover, burnout, and/or job satisfaction as a dependent variable. Burnout was explored most often and was included as a dependent variable in four of the studies (Gibson et al. 2009; Griffith et al. 2014; Hurt et al. 2013; Kelly and Barnes-Holmes 2013). To evaluate burnout across participants, two of the included studies (Gibson et al. 2009; Griffith et al. 2014) used the Maslach Burnout Inventory (MBI; Maslach et al. 1986). The MBI assesses three components of burnout: emotional exhaustion (i.e., feeling emotionally overextended by one’s occupation), depersonalization (feeling impersonal towards those receiving one’s services or care), and personal accomplishment (feeling capable and successful in one’s occupation). The remaining studies used different versions of the MBI to measure burnout. Kelly and Barnes-Holmes (2013) used the MBI—Educators Survey (Maslach et al. 1996) and Hurt et al. (2013) used the MBI—General Survey (Schaufeli et al. 1996). The MBI-GS measures slightly different components of burnout (i.e., exhaustion, cynicism, and personal efficacy).

Intention to turnover was included as a dependent variable in one study (Kazemi et al. 2015). To measure intention to turnover, Kazemi et al. (2015) used two items from Dupre and Day (2007): (1) “I will leave my job if another job becomes available,” and (2) "I will stay at my job for as long as I can.” Finally, job satisfaction was included as a dependent variable in one study (Hurt et al. 2013). Hurt et al. (2013) used the Andrews and Withey Job Satisfaction Scale (Andrews and Withey 1976) to measure job satisfaction.

Descriptive Data

The studies provided some descriptive data on their samples of behavior technicians. Of the behavior technicians sampled by Kazemi et al. (2015), 38% reported being highly or somewhat likely to resign. While low levels of overall burnout were found in behavior technicians working in ABA schools (Gibson et al. 2009), as high as 42% of behavior technicians were found to have low levels of personal accomplishment (Griffith et al. 2014), and as high as 27% were found to have high levels of emotional exhaustion (Gibson et al. 2009). High levels of depersonalization was not common across samples of behavior technicians (Gibson et al. 2009; Griffith et al. 2014).

Predictor Data

The independent variables evaluated in the reviewed studies can be broken down into employee-related variables, organizational variables, and patient- and student-related variables. Each is discussed in turn.

Employee Factors

All of the studies addressed various employee-related factors; see Table 2 for a detailed summary. Several employee-related factors have been identified as significant.

Table 2 Summary of employee variables
Full size table

Personality traits were found to be correlated with burnout and job satisfaction in behavior technicians providing home-based services (Hurt et al. 2013). Four personality traits were found to be correlated with burnout. Extraversion, agreeableness, and conscientiousness were found to be protective traits while neuroticism was correlated with greater levels of burnout. Two personality traits were found to be associated with job satisfaction, with extraversion showing a positive correlation and neuroticism a negative correlation (Hurt et al. 2013).

Implicit attitudes towards children with ASD and typical development (TD) were also found to be significantly correlated with burnout in behavior technicians working in an ABA school (Kelly and Barnes-Holmes 2013). Increased implicit negative attitudes towards children with ASD were found to be positively correlated with burnout, and increased positive attitudes towards children with TD were found to be negatively correlated with burnout. On the other hand, both positive attitudes towards children with ASD and negative attitudes towards children with TD were not found to be correlated with burnout (Kelly and Barnes-Holmes 2013).

Other factors were found to be related to burnout in behavior technicians working in ABA schools. Coping strategies were found to be significant. Particularly, wishful-thinking coping was found to be associated with greater levels of burnout (Griffith et al. 2014). In addition, Gibson et al. (2009) found both commitment to philosophy and perceived therapeutic self-efficacy to be protective against burnout (Gibson et al. 2009).

General demographic factors (i.e., age, education, experience) were not found to be related to burnout (Gibson et al. 2009; Griffith et al. 2014; Hurt et al. 2013) or intention to turnover (Kazemi et al. 2015) in behavior technicians.

Organizational Factors

Organizational factors were investigated by four of the studies (Gibson et al. 2009; Griffith et al. 2014; Hurt et al. 2013; Kazemi et al. 2015). Table 3 provides a summary of the organizational factors explored in the reviewed research.

Table 3 Summary of organizational variables
Full size table

Supervisor support was given the most attention across the reviewed research. Greater satisfaction with supervisor support was found to be associated with lower levels of burnout (Gibson et al. 2009; Hurt et al. 2013), greater levels of job satisfaction (Hurt et al. 2013), and less intention to turnover (Kazemi et al. 2015). Perceived supervisor support was also found to moderate the effects of work demands on burnout (Gibson et al. 2009) and partially mediate the effects of personality traits on job satisfaction (Hurt et al. 2013). Findings on the significance of supervisor support were not entirely consistent. Perceived supervisor support was not found to be associated with burnout in a sample of behavior technicians working in ABA schools (Griffith et al. 2014). Griffith et al. (2014) suggest that their contrasting findings may be due to including a greater portion of behavior technicians with supervisor-level roles as compared to Gibson et al. (2009), and the authors speculate that supervisor support may have less of an impact on those behavior technicians.

Contrasting findings emerged between objective measures versus subjective measures of a couple organizational variables. While report of being under scheduled was correlated with intention to turnover in behavior technicians, the number of hours worked per week was not found to be correlated with intention to turnover (Kazemi et al. 2015). Likewise, level of satisfaction with pay was negatively correlated with intention to turnover; however, wages earned per hour were not found to be correlated with intention to turnover (Kazemi et al. 2015).

Other factors were identified as being related to burnout and intention to turnover in behavior technicians. Level of satisfaction with training was negatively correlated with intention to turnover (Kazemi et al. 2015). High levels of perceived work demands were correlated with greater levels of burnout (Gibson et al. 2009). Finally, greater levels of job satisfaction were found to be associated with less intention to turnover (Kazemi et al. 2015) and lower levels of burnout (Hurt et al. 2013).

Several factors were investigated and not found to be significant predictors. Receipt of benefits and number of hours spent traveling were not found to be correlated with intention to turnover (Kazemi et al. 2015). Furthermore, setting of treatment services (e.g., home, school) was not found to be a predictor of intention to turnover (Kazemi et al. 2015).

Patient and Student Factors

Variables relating to patients and students were explored the least in the reviewed studies. Only two studies evaluated patient- and student-related factors (Gibson et al. 2009; Kazemi et al. 2015). See Table 4 for a summary of patient- and student-related factors. None of the evaluated factors (i.e., age, ASD severity, and time spent working with student) were found to be significant variables.

Table 4 Summary of patient and student variables
Full size table

Discussion

The purpose of the current paper was to review research on predictors of staff retention, turnover, burnout, and job satisfaction for behavior technicians working with individuals with ASD. A total of five articles were identified and reviewed.

None of the studies reported on rates of turnover in their samples of behavior technicians; however, Kazemi et al. (2015) reported that 38% of the behavior technicians in their sample indicated that they were highly or somewhat likely to resign. While this may seem high in comparison to annual voluntary resignation rates reported in the total workforce (i.e., 26%; Bureau of Labor Statistics 2018), it is much lower than annual turnover rates reported for direct-care staff of individuals with DD (i.e., 70.7% to 77%; Mitchell and Braddock 1994; Seninger and Traci 2002). Participants in the included studies were also found to have low levels of burnout as compared to normative data (Gibson et al. 2009). It should be noted that participation in the included studies was voluntary and thus the findings are reflective of staff who are motivated to participate in research. It is unknown if intention to turnover or level of burnout differed between staff who did and did not volunteer to participate (Griffith et al. 2014; Kazemi et al. 2015). Despite this limitation of the reviewed studies, the predictors identified in the studies have important implications. A number of employee and organizational factors were found to be associated with burnout, job satisfaction, and intention to turnover in behavior technicians.

One strategy for improving staff retention is considering whether an organization’s hiring practices are identifying potential employees who are a good fit for the position. While education and experience are often given weight during the hiring process, such demographic variables were not found to be significant with respect to burnout (Gibson et al. 2009; Griffith et al. 2014; Hurt et al. 2013) or intention to turnover (Kazemi et al. 2015) in behavior technicians.

Measurement of character traits of perspective employees may be informative to the hiring process. Preliminary evidence suggests that personality traits (Hurt et al. 2013) and negative implicit attitudes towards children with ASD (Kelly and Barnes-Holmes 2013) are associated with burnout in behavior technicians. Future research should investigate whether such assessments of character traits are associated with turnover. In addition to burnout, behavior technician personality factors and attitudes towards individuals with disabilities have been linked to procedural fidelity (Peters-Scheffer et al. 2013). Staff retention is just one factor of importance to employers. Treatment fidelity and patient outcomes are also highly significant priorities. Tools that may identify job candidates that are less likely to turnover and more likely to have high procedural fidelity are worthy areas of future exploration. Such tools may also be helpful for identifying existing employees who are in need of additional support.

Another strategy for improving retention is evaluating whether ongoing organizational practices play a role in staff retention. Findings of the current review suggest that employers should give special consideration to ongoing support of their staff. For instance, incorporating evidence-based interventions to improve coping strategies (Griffith et al. 2014) and self-efficacy (Gibson et al. 2009) may be beneficial.

Perceived supervisor support received the most attention across the included studies and was found to be correlated with burnout (Gibson et al. 2009; Hurt et al. 2013), job satisfaction (Hurt et al. 2013), and intention to turnover (Kazemi et al. 2015). While perceived supervisor support was not always found to be significant (Griffith et al. 2014), overall, it appears to be meaningful. Perceived supervisor support was even found to moderate the effects of work demands on burnout (Gibson et al. 2009) and partially mediate the effects of personality traits on job satisfaction (Hurt et al. 2013). Perceived supervisor support was also found to increase self-efficacy in behavior technicians (Gibson et al. 2009). Given the benefits of perceived supervisor support, future research should investigate variables that may result in improved satisfaction with supervision in behavior technicians (e.g., frequency of clinic meetings and mentor meetings, management style).

While staff training was only directly explored by one of the reviewed articles, it may be a promising area for further investigation. Kazemi et al. (2015) found greater levels of satisfaction with training to be correlated with less intention to turnover in behavior technicians and found satisfaction with staff training to correlate with reported hours of initial training with 30 h or more corresponding with satisfaction. Furthermore, the significance of commitment to philosophy on burnout and self-efficacy may signify that theoretical-focused training, as opposed to procedural-focused training, may be beneficial (Gibson et al. 2009).

Like satisfaction with supervision and training, organizational variables were often measured via employee perception. In some cases, both subjective and objective measures of organizational variables were explored. For instance, satisfaction with pay was found to be a predictor of intent to turnover; however, actual pay did not correlate with satisfaction of pay and actual pay was not found to be a predictor of intention to turnover (Kazemi et al. 2015). Similarly, report of being under scheduled was correlated with intention to turnover in behavior technicians while number of hours worked per week was not found to be correlated with intention to turnover (Kazemi et al. 2015). Contrasting findings between objective measures versus subjective measures help to illuminate the subjective nature of these drivers of intention to turnover. It appears that an individualized approach may be beneficial. For instance, considering the percentage of an employee’s available hours worked per week may be more valuable than considering their total number of hours worked.

Patient- and student-related factors were explored the least in the reviewed research. Although no factors were identified as significant (Gibson et al. 2009; Kazemi et al. 2015), this area may warrant further investigation. Challenging behaviors, for instance, have been found to impact burnout in special education teachers and support staff of individuals with ID (Hastings and Brown 2002). While behavior technician report of ASD severity was not found to be a predictor of intention to turnover (Kazemi et al. 2015), ASD severity as measured on standardized assessments and child progress may warrant future investigation. Future research should explore whether various patient characteristics (challenging behaviors, symptom severity, rate of progress) contribute to turnover. Such findings may be helpful when considering case assignment, particularly for newly hired behavior technicians.

Some limitations should be considered when interpreting the findings of the current review. First, research evaluating predictors of job retention, turnover, burnout, and job satisfaction in behavior technicians is sparse. For the most part, findings have not been replicated and should be interpreted with caution. In addition, the current review included behavior technicians working in diverse settings. While treatment setting was not found to be related to intention to turnover (Kazemi et al. 2015), school and home settings certainly place differing work demands on behavior technicians. Another limitation is that the behavior technicians in this review were from diverse geographical regions, which may impact the samples differently (e.g., laws concerning work conditions). Finally, no studies investigated actual turnover in behavior technicians. While predictors of burnout, job satisfaction, and intention to turnover were identified in the current review, it is unknown whether these factors are truly related to staff retention. Longitudinal and retrospective analyses of predictors of actual turnover are warranted. Nevertheless, the current review gives meaningful recommendations for how employers may improve their practices.