## Abstract

Turnover among direct support professionals (DSPs) in community support settings for individuals with intellectual and developmental disabilities (IDD) has been regarded as a challenge since tracking of this workforce began in the 1980s. This study utilized a group randomized controlled design to test the effects of a competency-based training intervention for DSPs on site-level turnover rates over a one year period. Results suggested that, compared with the control group, sites receiving the training intervention experienced a significant decrease in annual turnover, when multiple factors were controlled. Implications, including the importance of considering quality training as a long term organizational investment and intervention to reduce turnover, are discussed.

## Introduction

Researchers have long noted that the direct support workforce supporting people with intellectual and developmental disabilities (IDD) experiences high rates of turnover (Lakin & Bruininks, 1981; Lakin et al., 1983). More recently, attempts have been made to use competency-based training methods to assist direct support professionals (DSPs) in strengthening their skill sets, with the assumption that more competent DSPs may translate to a more stable direct support workforce for people with IDD. This study, informed by the literature, tested this assumption that competency-based training can help to reduce DSP turnover in residential and day habilitation programs for people with IDD.

### Competency-Based Training for DSPs

Since the 1980s, educators and trainers have made efforts to go beyond traditional knowledge building to develop competency among professionals in the health and human services (Albanese, Mejicano, Anderson, & Gruppen, 2008). Early attempts at competency-oriented training sought to replace memorization of facts with the building of tangible skill sets that practitioners could use in their practice (Smith & Dollace, 1999). In the ensuing years, employers and organizations have adopted competency-based training for workers in many professions, including direct support for people with IDD.

Though wages are typically low, the skill set required of DSPs working with people with IDD is expansive. There have been a number of competency schemas developed to portray the complexity of direct support work. Although there have been a number of conceptualizations of DSP competencies and how to identify or achieve them (e.g.: Hall & Hall, 2002; Harchik et al., 2001; Ricciardi, 2005), there are presently three main schemas of DSP competency that are utilized to guide training practices across service types: the Community Support Skills Standards, the National Alliance of Direct Support Professionals’ competencies, and the tiered competency model from the U.S. Department of Labor. The Community Support Skill Standards (CSSS; Taylor, Bradley & Warren, 1996) were the first major attempt to validate DSP competencies, and organized direct support work into 12 broad areas, with a total of 144 distinct skills needed to adequately conduct direct support work. The National Alliance for Direct Support Professionals (2011) revised and expanded this schema to incorporate the findings of a validation study conducted by Hewitt (1998) and to reflect changes in community supports since the CSSS were initially published. This framework resulted in the addition of new skills and a restructuring of the way in which the competency set was organized. Additionally, cross-sector conceptualizations of direct support that include services for people with IDD among other populations have emerged that identify core competencies for DSPs regardless of the population with whom they work or the setting in which their work is carried out (Hoge & McFaul, 2011; Sedlezky, Taylor, Nord, Hoge et al., 2014; U.S. Department of Labor, 2011).

### Current DSP Workforce

Direct support professionals (DSPs) fill a critical role in the American health and social services systems. Generally defined, the direct support workforce is the segment of the labor force responsible for providing direct personal service and support to people in need of assistance with daily living tasks, personal care, household upkeep, relationship building, and community participation. People who use direct support services are often people with intellectual and/or developmental disabilities (IDD), older adults, people with physical health concerns, or people with severe mental illness. The total population of DSPs in the United States is estimated at 3.38 million individuals, with expected growth to about five million by 2020, when the direct support workforce may become the largest segment of the American labor market (PHI, 2013). The specific population of interest in this study, the direct support workforce supporting people with IDD in community settings, may be found working in individual and small group homes, day training centers, vocational supports, and in family residences.

Since studies of the direct support workforce in IDD settings began in the 1980s (Lakin & Bruininks, 1981; Lakin et al., 1983), general trends of low wages and high turnover have been observed. Though there have been few recent studies, wages reported in surveys in the past several years have shown wages ranging from a low of $8.68 per hour in a nationwide study of residential services providers (Lakin, Polister & Prouty, 2003) to a high of$11.25 per hour in Minnesota’s community supports system (Bogenschutz, Hewitt, Nord, & Hepperlen, 2014). Also recently, a 2011 salary survey of DSPs working in New York State (Maciekowich, Williams, & Canterio, 2012) identified the average base pay for DSPs as $11.17 per hour, with the lowest pay reported at$9.49 and the highest at $15.51 per hour, and PHI (2013) identified the national average wage across direct service jobs at$9.49 per hour. A number of studies have shown that services provided by state-operated agencies in the public sector tend to pay higher wages than do private sector employers (ANCOR, 2010; Braddock & Hemp, 2004; Hewitt, Larson & Lakin, 2000).

With these low wage trends and high skill expectations have come high turnover rates. DSP turnover rates between 45% and 70% annually have been identified within a particular organization providing community supports for people with IDD (Braddock & Mitchell, 1992; Larson, Hewitt & Knoblach, 2005). More recently, a study published by the American Network of Community Options and Resources identified turnover rates between 38-50% among its members (ANCOR, 2010). Turnover within the first six months of hire has also been noted as a problem that complicates workforce stability in many organizations serving people with IDD (Larson, Lakin & Bruininks, 1998). With the cost of replacing a single DSP estimated between $2,413 (Hewitt & Larson, 2007) and$4,872 (ANCOR, 2010), the fiscal costs of DSP turnover in an organization can be dramatic.

In addition to the financial impacts of turnover, there is some evidence to suggest that turnover can impact service quality and consumer outcomes. For instance, among workers in community mental health centers, turnover was found to relate to an organizational inability to adequately implement evidence-based practices (Woltmann, et al., 2008). Hatton and colleagues (2001) also noted that reductions in service quality may result when DSP turnover is high in community services for people with IDD. Additionally, organizations such as The Council on Quality and Leadership integrate staff stability and qualifications into their key factors and success indicators in person-centered supports (CQL, 2014).

### Turnover in General Workforce

The current study specifically concerns the direct support workforce supporting people with IDD. However, to provide context for the IDD workforce it is also useful to look at turnover in other segments of the U.S. labor force that have similar characteristics. The U.S. Bureau of Labor Statistics (2014) tracks annual separation rates across broad sectors of the labor force. In the total U.S. workforce, the annual separation rate was about 38% in 2013, an increase from the recent low of 36.6% in 2011, but still substantially lower than 42.4% in 2008. Health Care and Social Assistance, the Bureau of Labor Statistics’ category that includes direct support workers along with a broad array of other occupations, has had turnover trends that mimic those of the general workforce, though the overall turnover rates in this category tend to be slightly below the national total (U.S. BLS, 2014).

### Turnover Antecedents

The literature on employee turnover and intent to leave suggests a number of factors that may be associated with employee separation (Mor Barak, Nissly & Levin, 2001). Among these, poor social and supervisory support has been found to be a predictor in turnover or intention to quit in several studies among health and human service workers (Kim & Stoner, 2008; Mittal, Rosen & Leana, 2009; Nissly, Mor Barak & Levin, 2005). In a study of social services workers, perceived supervisor support was important in predicting turnover, because an employee’s perception of support from a supervisor could help to weaken the effects of low perceived organizational support, while perceptions of low supervisory support exacerbated the effects of low organizational support, underscoring the importance of the supervisory relationship in organizational context (Maertz, Griffeth, Campbell & Allen, 2007). Communication patterns may be of particular importance, as positive job-related communication has been shown to relate to reduced turnover intention among social service workers (Kim & Lee, 2009).

High levels of job stress and burnout are also often associated with intention to leave or actual turnover. Specifically, role ambiguity, which occurs when job expectations are ill-defined, was related to turnover intention in mental health social workers (Acker, 2004). Job stress, which can lead to burnout, has also been associated to intention to leave in DSPs working with people with IDD (Gray-Stanley & Muramatsu, 2011), and social workers, particularly when role stress was high and autonomy was low (Kim & Stoner, 2008), and when DSPs have an external locus of control related to work stress (Gray & Muramatsu, 2013).

Having few opportunities for professional development or inadequate training may also be associated with turnover and intent to quit. For instance, Ejaz and colleagues (2008) found lack of continuing education to be among the factors associated with low job satisfaction in DSPs working in long term care settings. In turn, low job satisfaction was associated with intention to leave and eventual turnover, and organizations with high turnover had lower overall worker satisfaction (Ejaz, Noelker, Menne & Bagaka’s, 2008). Acker (2004) found similar results for social workers in mental health settings; Castle and colleagues (2007) observed similar findings among DSPs working in nursing homes, where extent of available professional development activities was among the predictors for low job satisfaction and intention to quit. It should be noted, however, that highly trained child care workers were more susceptible to turnover if they earned low wages in relation to their high level of training (Whitebook & Sakai, 2003).

Finally, it is worth noting that low job satisfaction is often used as a predictor of intention to quit or turnover (e.g. Acker, 2004; Castle et al., 2007; Ejaz et al., 2008), though different variables are used to construct job satisfaction measures from study to study.

### Training to Reduce Turnover

Across sectors of the U.S. labor force, training has been considered as a potential means by which workers may be retained and turnover may be reduced. Because recent studies on the effects of training interventions among DSPs are scarce, evidence in this section is drawn from a number of different industries.

For instance, a 2006 study of technical employees found on-the-job training to be related to increased organizational commitment, while participation in external training (via a tuition reimbursement program) was associated with increased turnover intention, suggesting that in-house training focused on job-specific skills may provide greater dividends than organizational support of external training in more generalized competency (Benson, 2006). Technology firms that established a positive culture of learning also had employees with less intention to leave the organization (Egan, Yang, & Bartlett, 2004).

Examining a sample of customer service employees, Schmidt (2007) found that both satisfaction with job training and the amount of time a worker spent in training predicted overall job satisfaction. Likewise, an international study of nurses found that employees’ perception of their employers’ investment in their professional development was positively associated with their job satisfaction, though in this case the high satisfaction did not translate to intention to remain employed with the organization (Lee & Bruvold, 2003).

Perhaps most pertinent to the present study, Dill and colleagues (2010) found that nursing homes that participated in a program in which their DSPs received a pay increase for participating in a standardized training program were 15% more likely to have below average turnover than facilities that did not participate in the program. While type of ownership (for-profit vs. non-profit) and Medicaid acceptance were also important variables related to turnover, participation in the training intervention was a significant predictor of low turnover (Dill, Morgan & Konrad, 2010). Similarly, a mentorship-based model aimed at increasing retention of newly hired nursing home workers was observed to improve retention by 25%, while a control group saw improvement of only 10% in the same timespan (Hegeman, 2005). The influence of training on the reduction of turnover, however, is not universal; a large Canadian study found that employer-provided training actually related to higher turnover, leading to speculation that increased training may make workers more competitive in the open job market (Haines, Jalette, & Larose, 2010).

### Research Question

The authors conducted the present study to investigate the relationship between participation in a competency-based training intervention on turnover rates in among DSPs in worksites providing residential or day training and habilitation services for individuals with IDD. Specifically, one hypothesis was tested: After accounting for key factors and covariates, does participation in a competency-based training intervention affect annual crude separation among DSPs in community services for people with IDD?

## Method

### Research Design

The research team used a group randomized controlled design in conducting this research, which is part of a larger study on the effects of a systematized competency-based training intervention for DSPs in community services for people with IDD.

#### Ethical considerations

All procedures employed in this study were reviewed and approved by a university-based institutional review board. Two of the study’s authors have an existing relationship involving the development of the online training program used as part of this study’s intervention: the center where they are both employed, has financial arrangements with the publisher of the curriculum. However, because their respective roles in research and curriculum are distinct, and because personnel from the publisher of the online training played no role in any stage of this research, independent observers have assessed that no conflict of interest exists.

### Sampling

A total of 14 organizations with 141 sites volunteered to participate in this study. To be considered for inclusion, human service organizations were required to have multiple service sites that did not share staff, provide Home and Community Based Waiver services to people with IDD, provide executive commitment for the study’s training and technical assistance (TA) implementation activities, and certify that they had not used the study’s competency-based training program in the past.

Purposive sampling was used to reflect the diversity of human service organizations in the state where the study was primarily conducted. Half of the participating organizations provided residential services, and half provided day training and habilitation. The geographic dispersion of the organizations represented the state’s population in urban, suburban, and non-metropolitan regions, and organizational size ranged from small to large.

Upon enrollment, each of an organization’s sites was assigned an identification number, and randomly assigned into either the treatment condition or the control condition using an online random number generator (Random, 2014). Half of each organization’s sites were distributed to each condition.

Within each site, DSPs were also randomly selected to complete surveys. Because sites varied considerably in size (both number of people served and DSPs), a disproportionate random sampling procedure was used to obtain a robust sample from each site, while accounting for the possibility of turnover over the life of the study and in order to most closely represent the population of DSPs. In smaller sites, a greater proportion of DSPs and services users completed surveys than at large sites, where a smaller proportion of DSPs and service users could be selected while still constructing a robust sample.

### Measurement

The data used in this study came from two sources: (a) a survey of site-level workforce development indicators; and (b) a survey completed by DSPs. Site-level surveys were completed by participating sites (treatment and control groups) at baseline and at the conclusion of the one year intervention period. The DSP survey used in this study’s analysis was completed at baseline.

#### Site-Level Workforce Survey

This instrument was a two-page survey eliciting information about a variety of commonly tracked indicators that measure the overall condition of the direct support workforce within a site or organization. The survey, which had been used in previous studies (e.g. Lakin, Polister, & Prouty, 2003; Larson et al, 2008), was completed by a manager or human resource professional with direct knowledge of the DSP workforce at the site-level. Variables collected through this survey included information on DSP crude separation rates in the past year, crude separation in the past six months, the number of vacant positions, and incident reports. All data were collected at the site level, and, when relevant, sought average data for DSPs working in a particular site. Annual crude separation, which is used synonymously with turnover in this article, was calculated as:

Voluntary and involuntary turnover were not differentiated in this study. The authors chose to consider all forms of employee turnover together because a robust training intervention should potentially affect both voluntary and involuntary turnover. A training intervention such as the one used in the study may help to reduce voluntary turnover by giving the support and skills that DSPs need to do their jobs well and to feel like valued employees; the training may also reduce the need for involuntary turnover via terminations by better preparing the workforce to carry out basic tasks. Additionally, whether turnover is voluntary or involuntary, both the effects of a lack of continuity and disruption for people who receive services and the cost to the organizations providing these services are the same.

#### DSP Survey

The researchers randomly selected a portion of DSPs at each site to complete a survey about their experiences as a DSP. Participants were asked about their plans for long-term work as a DSP, their impressions of the support provided by their employer, their attitudes toward their work, their uptake of employer sponsored benefits, and a series of demographic questions. This eight-page survey contained potentially sensitive information about DSPs’ perspectives about their employer; thus, to ensure confidentiality, participants placed their completed surveys in a sealed envelope (provided by the researchers) before returning it to a designated location. Sealed surveys were then mailed back to the research team.

### Interventions

Sites that were randomly assigned into the intervention group were required to participate in a standardized training and site-level technical assistance (TA) intervention lasting one year, in addition to the organization’s usual training routine. Control sites participated in only the organization’s usual training. Said training varied from organization to organization, but typically included an initial orientation to the job of a DSP, annual updates on policies and procedures, and completion of state-mandated CPR, First Aid, and blood borne pathogen trainings. Employees in the control sites were able to access the comprehensive training provided to the intervention group after a one year waiting period. An overview of the entire intervention approach may be viewed in Figure 1.

For intervention sites, the first stage of intervention included TA that was conducted by a member of the research team; the participants were site managers and administrators. This TA lasted roughly ten hours and aimed to familiarize managers with study protocols (including data collection), prepare them to administer the study’s other intervention components, and troubleshoot potential problem areas that were anticipated to potentially interfere with successful implementation. Members of the research team provided additional TA as needed. Common issues addressed in these TA sessions included troubleshooting technical glitches, developing strategies for finding training time with minimal interference in employee work schedules, and preparation for facilitated discussions (described below).

The main intervention for DSPs was a systematized, online competency-based training program. This interactive training program used a multi-modal instructional design that combined audio, video, interactive exercises, and didactic information. The program included pre-and post-measures of knowledge, and embedded stories and real life experiences from people with IDD and their DSPs. DSPs working at all intervention sites participated in a set of 35 standardized lessons—each lasting roughly one hour for a typical learner—of online training over the span of one year. The lessons were organized into six thematically similar modules which were delivered on a timeline pre-determined by the researchers. Online training content aimed at building DSPs’ skills in the competency areas of professionalism, supporting rights and choices of people with IDD, fostering community participation, building self-determination, and other topics related to DSP competence. DSPs were required to score at least 70% on each lesson post-test in order for the lesson to be counted as complete. All training modules were written at the 8th grade reading level or lower in order to remain accessible. Excerpt learning objectives and on-the-job observation skills for each module are identified in Table 1.

The final two intervention components were designed to bring online learning into real-world work scenarios. Roughly every two months, at the conclusion of each training module, managers of each intervention site facilitated a discussion with DSPs on how concepts from online learning might translate to practice in the worksite. These facilitated discussions were semi-structured, and designed to synthesize learning for application in a DSP’s job at a particular site. Additionally, site managers were instructed to complete lessons along with (or slightly ahead of) DSPs, so that they could provide on-the-job mentorship as DSPs sought to apply online learning in their work.

### Participants

Participants in this study are defined as sites where Medicaid funded Home and Community Based Services residential and day training and habilitation services for people with IDD are delivered. A total of 52 sites comprised this sample, including 42 residential services sites and 10 day training and habilitation service locations. Sites represented both for-profit and nonprofit organizations. The sites included in this study comprised 502 DSPs and 989 people with IDD. Although the number of sites was skewed toward residential services, the number of people using supports and number of DSPs was more evenly distributed across the service types; day training and habilitation sites tend to both serve larger numbers of individuals and employ more DSPs. A summary of the number of DSPs and people with IDD represented in the sampled sites is presented in Table 2.

A total of 11 organizations ultimately participated in the robust intervention. Of these, eight organizations with 55 sites participated in the site-level pre and post-baseline surveying process. An additional three sites were removed from the analysis due to the limited data available. Of the final 52 participating sites, 10 were day training and habilitation (eight control and two intervention) and 42 were residential programs (12 control and 30 intervention). Additionally, on average, seven (sd  =  3.2) DSPs provided data at the site-level. Because the analysis for this study required a complete dataset for all sites, some sites that participated in the overall research project were excluded from this analysis, resulting in the unbalanced distribution of control and intervention sites outlined here.

### Analysis

Researchers performed statistical analyses in SPSS Version 22. A series of descriptive statistics were obtained to summarize the nature of DSP turnover within sites in the study in order to gain a general overview of the character of DSP turnover in participating organizations. Following this, a three-step inferential model was constructed using one-way Analysis of Covariance (ANCOVA) to isolate and compare the effects of the competency-based training intervention to the control group in reducing DSP turnover. The model steps isolated the variation explained by site characteristics, the site DSP workforce, and the experimental group assignment. Due to the unbalanced design Type III Sum of Squares was utilized, allowing for as little as one data point per cell.

The dependent variable in this analysis is the difference in annual site turnover from baseline to one year. Two factors representing experimental group was assessed in the model, coded as control (0) or intervention (1) group assignment. An additional seven site-level covariates were included; the average DSP tenure at the site, average wage at the site, and average education level at the site, and the proportion of DSPs access health benefits and in higher education. Using dummy coding, service type include residential (1) and day training and habilitation (0), whereas service area included rural (1) or non-rural (0).

## Results

Table 3 shows the mean differences in site-level turnover from baseline to one year for the experiment groups. Of the 52 sites in this analysis, 81% provided residential services and 62% were located rural settings. Site averages of DSP workforce variables include: 44.1 months (sd  =  21.9 months) tenure at the current worksite, wage earnings of $11.06 per hour (sd =$1.46), and 13.3 years of education (sd  =  .73 years). Additionally, 65.6% of all DSPs received health insurance from their employer, and 17.9% were enrolled in higher education.

### Annual Crude Separation

Our results present a statistical model suggesting that, when important factors and covariates are controlled, differences in site annual turnover from baseline to one year can be assessed based on experimental group assignment. The site characteristics model alone was found to explain 9% of the variance in changes in annual DSP turnover, where residential settings significantly greater reductions (, F (1, 43)  =  4.127, p < .05). Upon the addition of the site DSP workforce model, an additional 4% of variance was explained. However, there were no significant predictors. Finally, the experimental group model results show that experimental grouping was a significant predictor of site-level crude annual separation rates among DSPs in community services for people with IDD, after controls were in place (, F (1, 43)  =  4.251, p < .05). The parameter estimates of this ANCOVA model may be viewed in Table 4. The full model accounted for about 21% of variability in change in site-level annual turnover. Of this explained variance, 8%, or 38% of all explained variance, was attributable to the intervention.

After accounting for important covariates, the control group was found to experience a 10.3% average increase in turnover, whereas the intervention group experienced a 6.0% average reduction, a difference of 16.4%.

## Discussion

Results of this study are among the first to empirically establish the benefits of a comprehensive competency-based training and TA approach on site-level turnover of DSPs working in community IDD services. Though other studies have also revealed the benefits of training in reducing DSP turnover when additional compensation is also involved (Dill, Morgan & Konrad, 2010; Hegeman, 2005), this is the first known evidence that participation in systematized, competency-based training may be related to site-level turnover reduction in community services for people with IDD. Interestingly, this is also among the first studies to demonstrate that training alone, in the absence of additional financial incentives for DSPs who complete training, may influence turnover.

### Training as an Investment

With the cost of per-DSPs turnover estimated between $2413 (Hewitt & Larson, 2007) and$4872 (ANCOR, 2010), the results of this study suggest that training may be considered as a long-term investment in the fiscal stability of an organization. By way of illustration: if an organization was operating ten group homes that each employed seven DSPs with an assumed turnover rate of 38% based on the most recent national estimate (ANCOR, 2010), the organization could expect to replace 27 DSPs per year. At a conservatively estimated cost of $2500 per DSP (based on Hewitt & Larson’s 2007 estimate plus inflation, which included on-boarding training costs, coverage of temporary vacancy, advertising, etc.), the financial toll of turnover would be about$67,500 annually. If the same hypothetical organization reduced turnover by 16.4% (the observed difference between intervention and control sites in this study), they could expect to replace 18.4 DSPs per year, costing roughly $46,000. Based on conservative financial estimates, and typical turnover rates, the organization would save$21,500 annually in DSP replacement costs. This potential savings could very well offset the cost of offering a competency-based training curriculum, such as those accredited by the National Alliance for Direct Support Professionals (NADSP, 2011) and paying DSPs to complete training, while also strengthening continuity and quality of services to people with IDD.

Because of the limited fiscal resources of many providers of community services for people with IDD, many community residential and day service providers are not able to invest up-front dollars for robust training options for DSPs. However, state funding options may be available to assist with investment in high-quality training. At the state level, some economy of scale may be achieved by purchasing training access for organizations within the state; alternately, states may split or share costs with organizations. In either scenario, states that facilitate greater access to competency-based training materials for DSPs may expect to see system-wide payoffs in reduced turnover and increased service quality.

Though not specifically addressed in the current study, state-facilitated access to competency-based training may also be of benefit to individuals who self-direct their own services. Studies have indicated that finding quality training is a particular struggle for people with IDD who self-direct their supports (Bogenschutz, et al., 2010; Heller et al., 2012). Because self-directed service options have been expanding in recent years (Greene, 2007; Simon-Rusinowitz, Loughlin & Mahoney, 2011), finding training solutions that are workable in this support delivery model is of critical importance.

### Establishing Learning Culture

Participation in a robust, yearlong training intervention among DSPs in the intervention sites of this study may be seen as a first step in establishing an overall culture of learning within those study sites. Consistent with other studies (Egan, Yang, & Bartlett, 2004; Schmidt, 2007), results of this study suggest that creation of a workplace that values professional development and expects workers to spend time engaged in professional development activities is likely to experience positive workforce outcomes.

Further, the fact that the training intervention in this study was provided within the DSP’s worksite corroborates previous findings (Benson, 2006) that training on job-specific skills, within the context of the work environment, supports employee retention. Employers may benefit from seeking ways to carve out specific training time for DSPs each month in order to both focus on refining existing skills and developing new skills that are directly relatable to their work. A competency-based approach, where classroom or online learning is brought into the real-life work environment through on-the-job training or mentoring, may be particularly beneficial due to its application of learning in the context of the workplace.

It should be noted, however, that the intervention implemented in this study was complex and multi-layered. Because of this, it is difficult to say which specific parts of the intervention may be most influential in the results, or if it is essential to combine elements of supervisor training, online learning, and on-the-job training in order to achieve reductions in turnover. Future research may seek to better isolate the influence of specific training features on organizational outcomes.

### Limitations & Future Directions

Though the findings from this study may have a potentially significant impact, cautious interpretation is warranted. This work used exclusively site-level data in the analyses, and therefore caution must be observed in the application or use of any of the above conclusions at the level of individual DSPs. Use of site-level data in this study also means that control variables were aggregated as the mean for each site (for example, mean wage of all DSPs at a site was used rather than individually reported DSP wages). This resulted in some loss of discriminatory power compared with what might be expected if an array of DSP wages were reported individually within each site.

It is also important to remember that the sample for this study was drawn from only one state. Although the sample is representative of community service providers for individuals with IDD within that state, there is considerable variation between state IDD service systems, so results may not apply outside of the sampling frame. Variability in DSP wage rates, level of education and other factors also vary from location to location. Because these variables are likely to influence turnover, they must be taken into account in any future models examining the effects of training on turnover.

With the expected rise in demand for direct support workers accompanied by a decrease in supply of workers who typically gain employment as DSPs, there is ample space for future investigation of methods for strengthening and stabilizing this critical portion of the American labor force. Building on the current research, it is suggested that future lines of inquiry focus on the relationship between supervision, training, job satisfaction, and intent to quit. While these constructs are often located in the literature, the relationships between them for DSPs that provide community services for people with IDD are largely unexplored. Further, it may be useful to gain deeper understanding into the differences between trying to achieve a more generalized DSP skill set and training on more specialized topics that relate to specific needs of the individuals being supported by a DSP may relate to turnover and job satisfaction. Additionally, studies that consider the effects of DSP training on personal outcomes for individuals with IDD would help to close the research circle in demonstrating the importance of DSP training.

The findings of this study are among the first to provide empirical evidence for the notion that participation in robust competency-based training can contribute to a reduction in DSP turnover in community services for people with IDD. Although future study is necessary to reinforce this finding, this research indicates that strengthening the direct support workforce can also help to stabilize it, and that training should be seen as a long term investment in the financial and service wellbeing of organizations supporting people with IDD.

## Acknowledgments

This article was supported in part by Grant H133B080005 and Grant H133B130006 from the National Institute on Disability and Rehabilitation Research, U.S. Department of Education to the Research and Training Center on Community Living at the University of Minnesota. All research activities described in this report were reviewed and approved by the author's affiliated Institutional Review Boards.

## References

References
Acker
,
G
.
(
2004
).
The effect of organizational conditions (role conflict, role ambiguity, opportunities for professional development, and social support) on job satisfaction and intention to leave among social workers in mental health care
.
Community Mental Health Journal
,
40
(
1
),
65
72
. Doi: .
Albanese
,
M. A
.,
Mejicano
,
G
.,
Anderson
,
M. W
.,
&
Gruppen
,
L
.
(
2008
).
Building a competency-based curriculum: the agony and the ecstasy
.
Advances in Health Sciences Education
,
15
,
439
454
. .
American Network of Community Options and Resources
. (
2010
).
2009 Direct Support Professionals Wage Study
.
Alexandria, VA
:
Author
.
Benson
,
G
.
(
2006
).
Employee development, commitment and intention to turnover: A test of ‘employability’ policies in action
.
Human Resources Management Journal
,
16
(
2
),
173
192
. .
Bogenschutz
,
M
.,
Hewitt
,
A
.,
Hall-Lande
,
J
.
&
LaLiberte
,
T
.
(
2010
).
Status and trends in the direct support workforce in self-directed supports
.
Intellectual and Developmental Disabilities
,
48
(
5
),
345
360
. .
Bogenschutz
,
M
.,
Hewitt
,
A
.,
Nord
,
D
.,
&
Hepperlen
,
R
.
(
2014
).
Direct support workforce supporting individuals with IDD: Current wages, benefits, and stability
.
Intellectual and Developmental Disabilities
,
52
(
5
),
317
329
. .
,
D
.,
&
Hemp
,
R
.
(
2004
).
Developmental disabilities in North Dakota: A study of the structure, financing and quality of residential and community services
.
Boulder, CO
:
University of Colorado, Department of Psychiatry
.
,
D
.
&
Mitchell
,
D
. (
1992
).
Residential services and developmental disabilities in the United States: A national survey of staff compensation, turnover and related issues
.
Washington, DC
:
American Association on Mental Retardation
.
Castle
,
N
.,
Engberg
,
J
.,
Anderson
,
R
.,
&
Man
,
A
.
(
2007
).
Job satisfaction of nurse aides in nursing homes: Intent to leave and turnover
.
The Gerontologist
,
47
(
2
),
193
204
. .
Dill
,
J
.,
Craft Morgan
,
J
.,
&
,
T
.
(
2010
).
Strengthening the long-term care workforce: The influence of the WIN A STEP UP workforce intervention on turnover of direct care workers
.
Journal of Applied Gerontology
,
29
(
2
),
196
214
. .
Egan
,
T.M
.,
Yang
,
B
.
&
Bartlett
,
K
. (
2004
).
The effects of organizational learning culture and job satisfaction on motivation to transfer learning and turnover intention
.
Human Resource Development Quarterly
,
15
(
3
),
279
301
. .
Ejaz
,
F
.,
Noelker
,
L
.,
Menne
,
H
.
&
Bagaka's
,
J
. (
2008
).
The impact of stress and support on direct care workers’ job satisfaction
.
The Gerontologist
,
48
(
S1
),
60
70
. .
Firth
,
L
.,
Mellor
,
D
.,
&
Moore
,
K
.
&
Loquet
,
C
. (
2004
).
How can managers reduce employee intent to quit?
Journal of Managerial Psychology
,
19
(
2
),
170
187
. .
Gray
,
J.A
.
&
Muramatsu
,
N
. (
2013
).
When the job has lost its appeal: Intentions to quit among direct care workers
.
Journal of Intellectual & Developmental Disabilities
,
38
(
2
),
124
133
. .
Gray-Stanley
,
J.A
.
&
Muramatsu
,
N
. (
2011
).
Work stress, burnout, and social and personal resources among direct care workers
.
Research in Developmental Disabilities
,
32
,
1065
1074
. .
Greene
,
J
.
(
2007
).
Issue Brief: State Approaches to Consumer Direction in Medicaid
.
Hamilton, NJ
:
Center for Health Care Strategies
.
Haines
,
V
.,
Jalette
,
P
.,
&
Larose
,
K
.
(
2010
).
The influence of human resource management practices on employee voluntary turnover rates in Canadian non-governmental sector
.
Industrial & Labor Relations Review
,
63
(
2
),
228
246
. .
Hall
,
P
.
&
Hall
,
N
. (
2002
).
Hiring and retaining direct care staff: After 50 years of research, what do we know?
Mental Retardation
,
40
(
3
),
201
211
. .
Harchik
,
A
.,
Anderson
,
M
.,
Thomson
,
R
.,
Forde
,
K
.,
Feinberg
,
L
.,
Rivest
,
S
.
&
Luiselli
,
J
. (
2001
).
Evaluation of a participatory, competency-based model of staff training in a community habilitative setting
.
Behavioral Interventions
,
16
,
1
13
. .
Hatton
,
C
.,
Emerson
,
E
.,
Rivers
,
M
.,
Mason
,
R
.,
Swarbrick
,
R
.,
Mason
,
L
… &
Alborz
,
A
. (
2001
).
Factors associated with intended staff turnover and job search behavior in services for people with intellectual disability
.
Journal of Intellectual Disability Research
,
45
(
3
),
258
270
. .
Hegeman
,
C
.
(
2005
).
Turnover turnaround: A CNA mentoring program can improve staff retention in long-term care centers
.
Health Progress
,
86
(
6
),
25
30
.
Heller
,
T
.,
Arnold
,
C
.,
van Heumen
,
L
.,
McBride
,
E
.
&
Factor
,
A
. (
2012
).
Self-directed support: Impact of hiring practices on adults with intellectual and developmental disabilities and families
.
American Journal on Intellectual and Developmental Disabilities
,
117
(
6
),
464
477
. .
Hewitt
,
A
.
(
1998
).
Identification of Competencies and Effective Training Practices for Direct Support Staff Working in Community Residential Services for People with Developmental Disabilities (Doctoral dissertation)
.
.
Hewitt
,
A
.
&
Larson
,
S
. (
2007
).
The direct support workforce in community supports to individuals with developmental disabilities: Issues, implications, and promising practices
.
Mental Retardation and Developmental Disabilities Research Reviews
,
13
(
2
),
178
187
. .
Hewitt
,
A
.,
Larson
,
S
.,
&
Lakin
,
K. C
.
(
2000
).
An independent evaluation of the quality of services and system performance of Minnesota's Medicaid Home and Community Based Services for persons with mental retardation and related conditions
.
Minneapolis
:
University of Minnesota, Research and Training Center on Community Living
.
Hoge
,
M.A
.
&
McFaul
,
M
. (
2010
).
Alaskan Core Competencies for Direct Support Workers in Health and Human Services
.
Anchorage, AK
:
The Alaska Mental Health Trust Authority
.
Kim
,
H
.
&
Lee
,
S
. (
2009
).
Supervisory communication, burnout, and turnover intention among social workers in health care settings
.
Social Work in Health Care
,
48
(
4
),
364
385
. .
Kim
,
H
.
&
Stoner
,
M
. (
2008
).
Burnout and turnover intention among social workers: Effects of role stress, job autonomy and social support
.
Administration in Social Work
,
32
(
3
),
5
25
. .
Lakin
,
K
.
&
Bruininks
,
R
. (
1981
).
Occupational stability of direct-care staff of residential facilities for mentally retarded people
.
Minneapolis
:
University of Minnesota: Center on Residential and Community Services
.
Lakin
,
K
.,
Hill
,
B
.,
Bruininks
,
R
.,
Hauber
,
F
.,
&
Krantz
,
G
.
(
1983
).
AID-JCOP2290110306>3.0.CO;2-Z
.
Lakin
,
K. C
.,
Polister
,
B
.,
&
Prouty
,
R. W
.
(
2003
).
Wages of non-state direct-support professionals lag far behind those of public direct support professionals and the general work force
.
Mental Retardation
,
41
(
2
),
141
146
. .
Larson
,
S
.,
Hewitt
,
A
.,
&
Knoblauch
,
B
.
(
2005
).
Recruitment, retention, and training challenges in community human services: A review of the literature
.
In
S
.
Larson
&
A
.
Hewitt
(
Eds
.),
Staff recruitment, retention, & Training Strategies for Community Human Services Organizations
(pp.
1
18
).
Baltimore
:
Paul H. Brookes
.
Larson
,
S
.,
Lakin
,
K
.,
&
Bruininks
,
R
.
(
1998
).
Staff recruitment and retention: Study results and intervention strategies
.
Washington, DC
:
American Association on Mental Retardation
.
Lee
,
C.H
.,
Bruvold
,
N
.
(
2003
).
Creating value for employees: Investment in employee development
.
The International Journal of Human Resource Management
,
14
(
6
),
981
1000
. .
Maertz
,
C
.,
Griffeth
,
R
.,
Campbell
,
N
.
&
Allen
,
D
. (
2007
).
The effects of perceived organizational support and perceived supervisor support on employee turnover
.
Journal of Organizational Behavior
,
28
,
1059
1075
. .
Maciekowich
,
M
.,
Williams
,
B
.,
&
Canterio
,
M
.
(
2012
).
What We Pay: New York State Agencies Serving People with Developmental Disabilities 2011 Salary and Pay Range Data
.
Albany, NY
:
New York State Association of Community Residential Agencies
.
Mittal
,
V
.,
Rosen
,
J
.
&
Leanna
,
C
. (
2009
).
A dual-driver model of retention and turnover in the direct care workforce
.
The Gerontologist
,
49
(
5
),
623
634
. .
Mor Barak
,
M
.,
Nissly
,
J
.
&
Levin
,
A
. (
2001
).
Antecedents to retention and turnover among child welfare, social work, and other human service employees: What can we learn from past research? A review and metanalysis
.
Social Service Review
,
75
(
4
),
625
661
. .
National Alliance for Direct Support Professionals
. (
2008
).
15 NADSP Competency Areas
. .
National Alliance for Direct Support Professionals
. (
2011
).
List of Accredited Curricula
. .
Nissly
,
J
.,
Mor Barak
,
M
.,
&
Levin
,
A
.
(
2005
).
Stress, social support, and workers’ intentions to leave their jobs in public child welfare
.
Administration in Social Work
,
29
(
1
),
79
100
. .
PHI
(
2013
).
Occupational Projections for Direct-Care Workers: 2010-2020. Bronx, NY: Author
. .
Random.org
. (
2014
).
Random.org: True random number generator
.
.
Ricciardi
,
J.N
.
(
2005
).
Achieving human service outcomes through competency-based training
.
Behavior Modification
,
29
(
3
),
488
507
. .
Sedlezky
,
L
.,
Taylor
,
M
.,
Nord
,
D
.,
Hoge
,
M
.,
Robbins
,
R
,
Dilla
,
B
,
&
Flin
,
B
. (
2014
).
Phase III-B: Road Map of Core Competencies for the Direct Service Workforce Project validation
.
Baltimore, MD
:
Centers for Medicaid and Medicare Services
.
Schmidt
,
S
.
(
2007
).
The relationship between satisfaction with workplace training and overall job satisfaction
.
Human Resource Development Quarterly
,
18
(
4
),
481
498
. .
Simon-Rusinowitz
,
L
.,
Loughlin
,
D
.
&
Mahoney
,
K
. (
2011
).
How did cash and counseling participants spend their budgets, and why does that matter for CLASS
.
CLASS Technical Assistance Brief Series, No. 8
.
Long Beach, CA
:
The Scan Foundation
.
Smith
,
S
.
&
Dollace
,
R
. (
1999
).
AMEE Guide No. 14: Outcome-based Education: Part 2-Planning, Implementing, and Evaluating a Competency-based Curriculum
.
Medical Teacher
,
21
(
1
),
15
22
. .
Taylor
M
.,
V
.,
Warren
R
. (
1996
).
The community support skill standards: tools for managing change and achieving outcomes
.
Cambridge, MA
:
Human Services Research Institute
.
The Council on Quality and Leadership
. (
2014
).
Key factors and Success Indicators in Person-Centered Supports
. .
U.S. Bureau of Labor Statistics
(
2014
).
Economic News Release: Table 16. Annual total separations rates by industry and region, not seasonally adjusted
. .
U.S. Department of Labor
(
2011
).
Long-Term Care, Supports, and Services Competency Model
. .
Whitebook
,
M
.
&
Sakai
,
L
. (
2003
).
Turnover Begets Turnover: An examination of job and occupational instability among child care center staff
.
Berkeley, CA
:
Center for Study of Child Care Employment
.
Woltmann
,
E.M
.,
Whitley
,
R
.,
McHugo
,
G.J
.,
Brunette
,
M
.,
Torrey
,
W.C
.,
Coots
,
L
.,
Lynde
,
D
.,
&
Drake
,
R.E
.
(
2008
).
The role of staff turnover in the implementation of evidence-based practices in mental health care
.
Psychiatric Services
,
59
(
7
),
732
737
. .

## Author notes

Matthew Bogenschutz, School of Social Work, Virginia Commonwealth University; Derek Nord, Research and Training Center on Community Living, Institute on Community Integration, University of Minnesota; and Amy Hewitt, Research and Training Center on Community Living, Institute on Community Integration, University of Minnesota.