Using the materials from class, discuss what information will be critical to you in reviewing and evaluating an intervention/instructional program. What factors may be positively or negatively associated with the outcomes of your training/staff development? What data will you need to collect to support your evaluation? How will you base decisions for remediation of the training program?
Evidence-Based Staff Training: A Guide for Practitioners Marsha B Parsons & Jeannia H Rollyson J Iverson Riddle Center, Morganton, North Carolina Dennis H Reid Carolina Behavior Analysis and Support Center
ABSTRACT
Behavior analysts in human service agencies are commonly expected to train support staff as one of their job duties. Traditional staff training is usually didactic in nature and generally has not proven particularly effective. We describe an alternative, evidence-based approach for training performance skills to human service staff. The description includes a specifc means of conducting a behavioral skills training session with a group of staff followed by on-the-job training requirements. A brief case demonstration then il- lustrates application of the training approach and its apparent effectiveness for training staff in two distinct skill sets: use of most-to-least prompting within teaching procedures and use of manual signs. Practical issues associ- ated with applying evidence-based behavioral training are presented with a focus on providing training that is effective, effcient, and acceptable to staff trainees. Keywords: behavioral skills training, evidence-based practices, most-to-least prompting, staff training
Behavior analysts often share the job duty of training support staff in human service agencies to imple-
ment intervention plans for challenging behavior (Macurik, O’Kane, Malanga, & Reid, 2008) or teaching strategies (Catania, Almeida, Liu-Constant, & Reed, 2009; Rosales, Stone, & Rehfeldt, 2009) with consumers. In addition, staff are often trained in general prin- ciples and practices of behavior analysis (Lerman, Tetreault, Hovanetz, Strobel, & Garro, 2008). Disseminating infor- mation about effective practices among caregivers in this regard has become a professionally expected responsibility of behavior analysts (Lerman, 2009).
The importance of training human service staff was recognized early in the history of behavior analysis as it became clear that making a large-scale impact on consumers required effective training of support staff (Frazier, 1972). Behavioral researchers then began investigating staff training procedures (see Miller & Lewin, 1980; Reid & Whitman, 1983, for reviews of the early research on staff training). Researchers have continued to
EVIDENCE-BASED STAFF TRAINING
examine the effects of staff training strat- egies to allow for more effective and ef- fcient use of behavioral procedures with individuals with disabilities. Despite this existing research, many staff in human service agencies often do not acquire the skills that the procedures are intended to train (Casey & McWilliam, 2011; Clark, Cushing, & Kennedy, 2004; Sturmey, 1998). Hence, if behavior analysts are to successfully fulfll their staff-training responsibilities, additional guidance on best-practice implementation of staff training strategies is warranted.
The purpose of this paper is to describe an evidence-based protocol for training human service staff. Although this training technology has been dis- cussed from several perspectives (e.g., Reid, O’Kane, & Macurik, 2011), the focus here is on describing the basic components of the training pro- tocol for behavior analyst practitioners. Suggestions are also provided for effec- tively implementing the protocol based on our training experience. Following a summary of the evidence-based training protocol, a brief case demonstration is
presented to illustrate its application. Practical issues often related to the overall success of staff training are then offered for consideration.
Before describing the evidence- based training protocol, it should be noted that the focus of this training model is on training performance skills. Staff are trained to perform work duties that they previously could not perform prior to training. The model stands in contrast to approaches that focus primarily on enhancing knowledge or verbal skills, which would allow them to answer questions about the target skills. Though knowledge enhancement is clearly an important function of certain training endeavors, the goal of this pro- tocol is improved performance (Parsons & Reid, 2012). The distinction between training performance versus verbal skills is important because of the different outcomes expected as a function of the training process and because different training procedures are required. Early behavioral research demonstrated that staff training programs relying on verbal- skill strategies (e.g., lectures, presentation
Behavior Analysis in Practice, 5(2), 2-11 2
of written and visual material) are effective for enhancing tar- geted knowledge, but often are ineffective for teaching trainees to perform newly targeted job skills (Gardner, 1972). Thus, programs that rely heavily on verbal-skill training approaches typically prove ineffective in creating a meaningful impact on the job performance of human service staff (Alavosius & Sulzer- Azaroff, 1990; Petscher & Bailey, 2006; Phillips, 1998).
A Protocol for Evidence-Based Staff Training
Evidence-based staff training consists of performance- and competency-based strategies (Reid et al., 2003). The phrase per- formance-based refers to what the trainer and trainees do (i.e., actively perform the specifc responses being trained) during the training. The phrase competency-based refers to the practice of continuing training until trainees competently demonstrate the skills of concern (i.e., meet established mastery criteria). Specifcally, the training is data-based; observational data are obtained to document that trainees demonstrate the target skills at established profciency criteria. More recently, this ap- proach to staff training (i.e., instructions, modeling, practice, and feedback until mastery is achieved) has been referred to as behavioral skills training or BST (Miles & Wilder, 2009; Nigro-Bruzzi & Sturmey, 2010; Sarokoff & Sturmey, 2004). The procedures and literature described here are generally consistent with the research and procedures described as BST, though the specifc procedural steps may vary slightly. A basic protocol for conducting a BST session is presented in Table 1. The protocol consists of six steps, each of which is described in subsequent sections. This protocol is designed for training staff using a group format; however, the same basic steps can be used when training an individual staff member though some variations may be needed for individual implementation such as with behavioral coaching (Rodriguez, Loman, & Horner, 2009) and when all training occurs in-vivo or on the job (Miles & Wilder, 2009).
Step 1: Describe the Target Skill
The frst training step involves the trainer providing a rationale for the importance of the skill being trained and a de- scription of the behaviors required to perform the skill (Willner, Braukmann, Kirigin, Fixsen, Phillips, & Wolf, 1977). This step is generally referred to as instructions in the BST model. To adequately complete this step, trainers must behaviorally defne the target skill using a tool such as a performance checklist of necessary staff actions (Lattimore, Stephens, Favell, & Risley, 1984).
Step 2: Provide a Succinct Written Description of the Target Skill
Following a vocal description of the target skill, trainers should provide each trainee with a written description of the target behaviors that constitute the skill. The performance checklist referred to in Step 1 often serves this function. The trainer may also need to provide a written summary of pre- cisely what staff should do in different situations (Macurik et al., 2008), such as when being trained to implement a plan to
Table 1. Behavioral Skills Training Protocol for Conducting a Training Session With a Group of Staff
Training step Trainer action
Step 1 Describe the target skill
Step 2 Provide a succinct, written description of the skill
Step 3 Demonstrate the target skill
Step 4 Require trainee practice of the target skill
Step 5 Provide feedback during practice
Step 6 Repeat Steps 4 and 5 to mastery
reduce challenging behavior. The description should be suc- cinct and focus on exactly what needs to be done to perform the target skill.
Many trainers fail to provide a succinct, written descrip- tion of the target skill (Reid, Parsons, & Green, 2012, Chapter 4). Instead of providing staff trainees with a written summary, they are referred to a lengthier document (e.g., a formal be- havior plan) available in a central location. Our experience suggests that a number of staff typically will not access the plan to review the information when needed. Documents such as plans for challenging behavior frequently contain much more information than what staff need to implement the plan (e.g., background consumer information, assessment processes used to develop the plan), though the information is important for other purposes.
Step 3: Demonstrate the Target Skill
Once trainees have heard and read a description of the ac- tions to perform the target skill, the trainer should demonstrate how to perform the skill. This step, referred to as modeling in BST, can usually be readily accomplished by using a role-play process (Adams, Tallon, & Rimell, 1980), and particularly when two trainers are present. One trainer plays the role of a staff member and the other trainer plays the role of a consumer (if the target skill involves interacting with a consumer). It is critical that role-play demonstrations be well-scripted and rehearsed prior to the training session to ensure an accurate and fuent demonstration of all key components of the target skill. If a second trainer is not available, a trainee can assist in the demonstration. In the latter case, the trainer must provide detailed instructions to the trainee to ensure the trainee knows exactly what should be done during the demonstration. We have also found it helpful for trainer(s) to stop or “freeze” at certain points and describe what is being done and why to help
EVIDENCE-BASED STAFF TRAINING 3
trainees attend to key actions being demonstrated. Alternatively, video models have been effectively incorporated into BST as the demonstration component for teaching staff various skills such as conducting discrete-trial instruction (Catania et al., 2009; Sarakoff & Sturmey, 2004) and use of picture communication systems (Rosales et al., 2009).
Step 4: Require Trainee Practice of the Target Skill
After demonstration of the target skill, trainees rehearse performing the skill in a role play similar to the trainer demon- stration (Adams et al., 1980). Instructions are given to organize trainees such that one can play the role of the consumer (again, if relevant) and one can demonstrate the target skill while other trainees observe. All trainees must practice performing the target skill.
The trainee practice step, referred to as rehearsal in BST, is frequently omitted during staff training (Reid et al., 2012, Chapter 4). In many staff training programs, only vocal and written descriptions of the target skill are provided, perhaps supplemented with a demonstration. This omission likely occurs because the practice component requires signifcant time investment for each trainee to practice the skill. However, practicing the skill is a critical feature for the success of BST and should be required of each trainee to produce effective performance (Nigro-Bruzzi & Sturmey, 2010; Rosales et al., 2009).
Step 5: Provide Performance Feedback During Practice
The ffth step of the training protocol is for trainers to provide feedback to the trainees as they practice performing the target skill. Trainers should circulate among the trainees to ob- serve their performance and provide individualized supportive and corrective feedback (Parsons & Reid, 1995). Supportive feedback entails describing to the trainee exactly what s/he performed correctly and corrective feedback involves specify- ing what was not performed correctly. Corrective feedback also involves providing instruction about exactly how to perform any aspects of the target skill performed incorrectly in order to facilitate profcient future performance of the skill. Generally we recommend providing feedback following completion of a given role play in contrast to interrupting an ongoing role-play activity to provide feedback.
Observing trainees and providing feedback to each trainee requires time and effort on the part of trainers. This is another reason that it is often benefcial to have two trainers present, and especially if the number of trainees exceeds four or fve. Providing individualized feedback is as critical to the training process as the trainee practice component, and must involve each trainee.
Step 6: Repeat Steps 4 and 5 to Mastery
The fnal step in a BST session is to repeat Steps 4 and 5 until each trainee performs the target skill profciently (Nigro- Bruzzi & Sturmey, 2010). Trainers should establish a mastery criterion, such as trainees performing 100% of the target steps
correctly (Miles & Wilder, 2009) or perhaps a lower percent- age but with identifcation of certain critical steps that must be performed at 100% profciency (Neef, Trachtenberg, Loeb, & Sterner, 1991). This fnal step represents the essence of the competency part of BST. A staff training session should not be considered complete until each trainee performs the target skill competently.
On-The-Job Training
The group training protocol is designed to train staff at one time in a situation that differs from the daily work situa- tion. The format is commonly used in human service settings where behavior analysts practice. However, because the training involves a simulated situation (e.g., role plays, no consumers present), the overall training process is not complete. The ses- sion must be followed by on-the-job training.
On-the-job, or in-vivo, training increases the likelihood that performance of the target skill acquired during the train- ing session generalizes to the usual work situation (Clark et al., 2004; Smith, Parker, Taubman, & Lovaas, 1992). On-the-job training involves trainers observing each trainee applying the target skill in the regular work environment and providing supportive and corrective feedback as described in Step 5 of the training protocol. Observations and feedback should continue until each trainee performs the target skill profciently during the typical work routine.
The on-the-job component is another aspect of the train- ing process that can involve a substantial time investment by trainers because they must go to each trainee’s worksite for observation and feedback. In this regard, we have found that the amount of time trainers will have to spend at trainee work sites will be minimized if each trainee has previously demon- strated competence during role plays in the training session; profciency in demonstrating a target skill on the job often parallels the level of profciency demonstrated during previous role plays.
The on-the-job training component completes the training process. However, it should also be emphasized that although completion of training is often a necessary step to promote profcient staff performance on the job, it is rarely a suffcient step (Reid et al., 2012, Chapter 4). Newly acquired job skills must be addressed from a performance management perspective (Austin, 2000) to ensure they maintain, and particularly with continued presentation of feedback by supervisors and related personnel. Describing effective on-the-job performance man- agement is beyond the scope of this paper; however, a number of resources describe evidence-based approaches to managing daily work performance of staff (e.g., Austin; Daniels, 1994; Reid et al., 2012).
Case Demonstration of Evidence-Based Staff Training
To illustrate how BST can be applied to train staff in a group format in a human service setting, the following case demonstration is presented. The demonstration involved train- ing two sets of skills deemed important by the staff supervisor.
EVIDENCE-BASED STAFF TRAINING 4
Method
Setting and participants. The demonstration occurred dur- ing ongoing services at an education program for adults with severe disabilities. The primary locations were classrooms in which instructional services and paid work (e.g., contract work, retail manufacturing) occurred. Seven teachers and one teacher’s assistant served as participants; six of these participants were women. Participant ages ranged from 30 to 53 years (M = 45 years) and their experience ranged from 1 to 30 years (M = 14). Each teacher was responsible for services in a given classroom and the teacher’s assistant worked in one of the classrooms. Each teacher was licensed in special education. Four teachers had a bachelor’s degree and three had a master’s degree.
Behavior defnitions and observation systems. The skill sets targeted for training were selected by the supervisor of the pro- gram (experimenter) based on her view of relevant skill targets. The frst skill set pertained to using a most-to-least (ML) as- sistive prompting strategy (Libby, Weiss, Bancroft, & Ahearn, 2008) while teaching consumers. All participants had previ- ously mastered using a least-to-most assistive (LM) prompting strategy (Parsons & Reid, 1999), which was the most common prompting approach used in the adult education program. The supervisor’s intent was to expand the participants’ teaching skills by training them to also be able to use the alternative, ML prompting strategy. The second targeted skill involved the use of manual signing in interactions with certain adult students. Only seven participants were involved in this training due to a medical leave. Each participant interacted, or potentially could interact, with a student who responded to and/or used manual signs for communication. However, the participants had not received formal training in manual signing for at least several years, if at all.
The ML prompting protocol involved fve teaching com- ponents based on previous research on LM prompting (Parsons & Reid, 1999). First, correct order was defned as teaching the steps of a student program in the exact sequence specifed in the program task analysis. Second, correct reinforcement was defned as providing a consequence after the last correct step in a program and not providing the same consequence for any incorrectly performed step by the student. Reinforcement could be provided for correct student completion of any step but must be provided for the last correctly completed step. Third, correct error correction was defned as the teacher inter- rupting a student’s error and providing increased assistance suffcient such that the student then correctly completed the step. Correct prompting (modifed from prior research to target ML) involved two components. The frst component, full physical guidance on the frst teaching trial, was defned as the teacher physically guiding the student through all steps of the task analysis. The second component, less assistive prompts on subsequent trials, was defned as the teacher beginning at least one step on the target trial by guiding the student through the step, stopping the guidance at a point earlier than on the previ- ous trial for that step, and not providing more assistance on any
step for the target trial relative to the preceding trial. Hence, there were fve overall components constituting correct teach- ing: the three components pertaining to order, reinforcement, error correction, and the two prompting components.
The fve teaching components were observed for each par- ticipant’s teaching session and each component was scored as correct or incorrect for each instructional trial conducted dur- ing the session. To be scored as correct, a component had to be performed correctly for each step of the task analysis with which it was used. If a necessary component was omitted (omission error) or a component was performed incorrectly (commission error), then that component was scored as incorrect. Following a teaching session, the percentages of the fve teaching compo- nents performed correctly were averaged to obtain a percentage correct score for the teaching session. Due to the specifc focus on ML prompting, the percentages of the two prompting com- ponents performed correctly were also calculated and reported separately. Interobserver agreement checks were conducted during 75% of all teaching sessions, involving each participant and experimental condition. Interobserver agreement was cal- culated by dividing the number of agreements on occurrence of a correct teaching component by the number of disagreements plus agreements, multiplied by 100. Agreement averaged 95% (range, 86% to 100%).
The skill set for manual signing involved 35 signs. The signs were selected by the participants’ supervisor based on her familiarity with the signs that were used by or with the adult students and that would likely be applicable within ongoing activities. The signs pertained to items (e.g., coffee, key, soda), actions (e.g., come, stop, work), descriptors (e.g., hot, good, slow), and private events (e.g., hungry, pain, thirsty). A correct production of each sign required three components (Fitzgerald, Reid, Schepis, Faw, Welty, & Pyfer, 1984) including (a) move- ments of the fngers and hand(s), (b) shapes of the fngers and hand(s), and (c) location of the fngers and hand(s) in respect to the body. These components were derived from the pictures and descriptions presented in Sign Language Made Simple (Lewis & Henderson, 1997).
Signs were assessed on a trial-by-trial basis with an observer recording a correct production only if the sign met all three of the adherence criteria. Incorrect production was scored if an error occurred on any one component or if the participant ver- bally indicated s/he did not know the sign. Interobserver agree- ment checks were conducted on a sign-by-sign basis during all trials on 36% of assessment sessions, for each participant and condition. Agreement averaged 89% (range, 67% to 100%).
Baseline procedures. Baseline sessions occurred individually with each participant. The experimenter explained that, as part of the program’s professional development activities, partici- pants would be assessed and trained to use an ML prompting strategy and a sample of manual signs. For the ML prompt- ing baseline sessions, the experimenter further explained the participant would be asked to train three skills during role play with an experimenter playing the part of a student. The three skills were wiping the mouth with a napkin (three task-analyzed
EVIDENCE-BASED STAFF TRAINING 5
steps), activating a CD player (four steps), and placing paper in a paper shredder (three steps). The participant was instructed to teach the “student” all task-analyzed steps of a respective skill using three trials with ML prompting and to fade the prompt- ing across successive trials. No feedback was provided to the participant.
During the baseline, the experimenter playing the role of the student followed a set script. The script specifed that the “student” should: (1) require full physical guidance for all task-analyzed steps on the frst trial of a given skill, (2) require full physical guidance to initiate a step during the second trial and then complete that step independently and subsequently require full physical guidance to complete the other steps on the second trial, and (3) complete the frst step independently on the third trial and then make an error on a subsequent step and require partial physical guidance on the remaining steps.
Baseline sessions for assessing manual signing skills involved the following procedures. First, the experimenter informed the participant that one word would be spoken at a time and the participant should make the sign that represented the word. Second, the experimenter said one word and waited for the participant to make a sign or indicate that s/he did not know how to make the sign. Third, this process was repeated for the remaining 34 signs. The presentation order of the words for signs was altered across sessions according to a set format.
Training and post-training procedures. Training on ML prompting and signing involved the steps of the evidence- based protocol described earlier and occurred in a group format with all participants simultaneously. Each training session lasted a maximum of one hour to accommodate staff ’s typical daily planning time and minimize disruption of delivery of consumer services. Three training sessions were conducted for ML prompting across different days, and three sessions were conducted for signing, also across different days. Two trainers (experimenters) conducted the training sessions.
The frst training session for ML prompting was initiated by a trainer explaining the rationale for the training, followed by a description of the fve components of teaching. A suc- cinct, written handout of the defnitions was also provided (see http://www.abainternational.org/Journals/bap_supplements. asp to download these defnitions). Any questions posed by the trainees were answered and included reference to the written description. Next, a trainer demonstrated teaching a task- analyzed skill in a role play (i.e., the other trainer in the role of “student”) using the script described previously. The target skill (i.e., decorating cookies with sprinkles) was different than the three skills assessed in the baseline assessment. Following each demonstration trial, the trainer paused to explain what she did and answer participant questions. Subsequently, participants each practiced teaching a trainer using the same target skill (again, decorating a cookie) and received feedback while the other trainees observed. Next, participants practiced teaching each other in a role play. The two trainers circulated among the participants during the role-play practice to observe, score, and provide feedback. The observation, trainee practice, and trainer
feedback continued until each trainee demonstrated 100% cor- rect teaching profciency one time with the target student skill. The second training session involved the same process (i.e., brief vocal description, demonstration, and participant practice with feedback) with two new student skills (i.e., removing trash from a table and placing it into a trash can, playing the game of “cornhole” that involved throwing a bean bag through a hole in a board). The third training session involved continued practice with the student skills covered in the frst two sessions until each participant correctly performed 100% of the fve teaching components.
Manual sign training began with 15 signs during the frst training session. The trainer described the importance of sign- ing with students who used signs for communication. Next, the trainer described how to make each of fve signs according to the three criteria noted earlier while the participants followed along with a handout that described making the signs and provided a picture of each sign (see Lewis & Henderson, 1997, for illustra- tions). The trainer then demonstrated each sign. Subsequently, the participants were divided into small groups. Participants were then instructed to have one participant name the fve signs for the others to produce and provide feedback to each other using the handout as a guide, and then to alternate the role of naming the signs. The trainers circulated among the groups to observe, score each sign production, and provide feedback. The trainees continued practicing until each trainee correctly produced 100% of the target signs. Five more signs were then demonstrated by a trainer and subsequently combined with the initial group of fve signs for trainee practice and feedback. This process was then repeated for fve more signs.
During the second training session, the 15 signs introduced in the frst session were described and demonstrated again, along