Search tips
Search criteria 


J Appl Behav Anal. 2009 Winter; 42(4): 813–817.
PMCID: PMC2790928


Jennifer Zarcone, Action Editor


This study examined the effectiveness of behavioral skills training in teaching 2 adult women with mild intellectual disabilities to report inappropriate staff-to-resident interactions. The reporting skill included making a self-advocacy response, walking away, and reporting the interaction. Participants' performance was measured during baseline, posttesting, 2- and 4-week follow-up, and generalization probes in new situations. All participants learned reporting skills, maintained these skills at 2- and 4-week follow-up, and generalized the skills to novel stimulus situations.

Keywords: abuse prevention, behavioral skills training, developmental disabilities, intellectual disabilities

Individuals with intellectual disabilities are at a greater risk of sexual abuse than individuals without disabilities (Lumley & Miltenberger, 1997). To address this problem, Lumley, Miltenberger, Long, Rapp, and Roberts (1998) developed a behavioral skills training program to teach 6 women with intellectual disabilities to respond to inappropriate sexual lures by staff members. Participants were asked to describe what they would do in response to a situation in which a staff member engaged in an inappropriate behavior. They also responded to role-play situations in which a male trainer playing the part of a staff member made appropriate and inappropriate sexual requests. Finally, a male confederate acting as a new staff member attempted to lure each participant. During training, participants were taught to react to a variety of lures with a three-component response (i.e., refuse, leave the area, tell someone). All participants improved their performance after training, and some or all of the gains were maintained 1 month after training. Unfortunately, the participants did not demonstrate generalization of the skills during in situ probes. In a follow-up study, when the skills acquired during training did not generalize during naturalistic assessments, in situ training (Gross, Miltenberger, Knudson, Bosch, & Breitwieser, 2007) was provided and resulted in all participants performing acceptably in the natural environment (Miltenberger et al., 1999).

The two studies reviewed above focused on the sexual abuse of individuals with intellectual disabilities. Although this is an important and ongoing problem that can produce harmful consequences, other forms of abuse, including verbal and physical abuse, are more likely to be experienced by individuals who reside in public residential facilities (McCartney & Campbell, 1998) and have not been addressed through behavioral interventions. The purpose of the current study was to extend the research done in the area of abuse prevention to include physical and verbal abuse towards individuals with intellectual disabilities.


Participants and Settings

Two women with mild intellectual disabilities who lived at a state-operated residential facility served as participants. Amy was a 51-year-old woman (IQ  = 67) with bipolar disorder, pica, caffeine disorder (not otherwise specified), personality disorder (not otherwise specified), and nicotine dependence. Cindy was a 49-year-old woman (IQ  = 64) who had been diagnosed with mixed bipolar disorder with psychotic features, pica, borderline personality disorder, and seizure disorder. Both had mild hearing and vision losses and were prescribed psychoactive medications. Immediately prior to this study, both had completed training to discriminate between appropriate and inappropriate staff-to-resident interactions. Testing and training sessions were conducted in a small, private conference room on the participants' living unit. Generalization probes were conducted in various locations around the facility's campus (e.g., cafe on the facility's campus, bedroom, dining room).

Response Measurement and Interobserver Agreement

The target behavior was accurate responding to inappropriate staff behavior in four areas: physical, sexual-physical, verbal, sexual-verbal. Examples of inappropriate staff interactions included hitting or kicking, kissing, or yelling at a participant, and talking about staff sexual activity with a participant. Examples of appropriate interactions included giving a high five, providing assistance in medical care that required touching private areas, giving feedback regarding unacceptable behavior using a quiet voice, and asking what size bra a participant wears before a shopping trip.

The dependent variable was the percentage of steps performed correctly and independently on a 12-step task analysis that included correctly discriminating whether the staff behavior depicted was appropriate or inappropriate, engaging in a refusal or self-advocacy response, walking away, and making a call that included several pieces of required information (e.g., location of incident, witnesses) to report the incident. An independent observer collected data simultaneously with the experimenter during at least 33% of sessions in all conditions. Interobserver agreement was calculated by dividing the number of agreements on occurrence or nonoccurrence of the steps of the task analysis by the total number of agreements plus disagreements and converting this ratio to a percentage. Mean interobserver agreement was 97% (range, 94% to 100%) for baseline, 98% for training (range, 94% to 100%), 95% for posttesting, 98% for generalization (range, 93% to 100%), and 96% for follow-up.

Experimental Design

A multiple probe design across participants was used to evaluate the effectiveness of reporting training.


A pool of 96 scenarios depicting the four categories of appropriate and inappropriate staff-to-resident interactions was developed based on (a) direct observation of interactions between staff and residents; (b) results of a questionnaire completed by a panel of experts, including a representative of the state guardianship agency; and (c) state regulations. Scenarios were balanced across staff gender, interaction types, and interaction nature (appropriate, inappropriate). One set of eight scenarios served as the test set; a second set of eight was used during generalization probes, and 80 additional scenarios, presented in sets of eight, were used for training.

Scenarios were acted out by confederates and were videotaped for presentation during testing and training sessions. They were filmed using a Sony digital camcorder and edited using iMovie software. During assessment and training, they were presented to participants via a Dell Inspiron laptop. An Apple iPod video was used for presentation of scenarios during generalization probes.


During each baseline session, the experimenter presented the test set of eight scenarios to the participant in random order via video. Scenarios depicted four appropriate and four inappropriate interactions. After the presentation of each scenario, the video was stopped, and the experimenter asked the participant to act out what she would do if confronted with that situation or to state if she would do nothing (the correct response to appropriate interactions), resulting in eight trials per session. When the participant demonstrated how to respond to the situation, the observer used a 12-step task analysis to score performance. The experimenter did not give feedback to the participant regarding response accuracy.


At the beginning of each training session, the experimenter explained the steps involved in reporting inappropriate staff behavior and showed the participant a video containing two scenarios (one appropriate staff behavior and one inappropriate staff behavior) in which the confederate actress modeled the appropriate way to respond to an appropriate and inappropriate situation. Next, the experimenter presented eight scenarios in random order, evenly divided among appropriate and inappropriate physical and verbal interactions. After each scenario, the experimenter asked the participant to demonstrate what she should do in response. If she performed 100% of the steps independently, the experimenter provided descriptive praise, and the session continued. For performance below 100%, the experimenter provided descriptive praise for the steps performed correctly and corrective feedback for the steps performed incorrectly, and the participant repeated the role play. This process was repeated for all eight scenarios. Training continued in this fashion until the participant performed 100% of the steps of the reporting task analysis independently on all eight training trials on the first attempt for three consecutive sessions.

Posttests and follow-up

Participants were individually posttested 1 to 3 days following successful completion of reporting training. Assessments also were conducted at 2 and 4 weeks following training to measure skill maintenance. During posttests and follow-up, participants responded to the same set of eight scenarios presented in baseline.


The experimenter conducted three generalization probes with each participant both before and after training. These probes were identical to baseline sessions except that they were conducted in several different locations on the facility's campus, at different times of the day, in the presence of other individuals, and using a different telephone. These variations required the participant to respond differentially to some steps required for accurate reporting (e.g., location, witnesses). No feedback was provided during these probes.


Because both Amy and Cindy correctly discriminated between appropriate and inappropriate staff-to-resident interactions throughout all phases of the study, only performance in response to the four scenarios in each session that required reporting inappropriate interactions are presented in Figure 1. Each data point represents the percentage of steps on the reporting task analysis performed correctly in response to a scenario. Thus, every set of four data points represents one session. Baseline scores indicated that neither Amy nor Cindy was able to report inappropriate interactions accurately. Specifically, Amy independently completed between 22% and 56% of the steps of the reporting task analysis, and Cindy completed between 11% and 22% of the steps. Both participants typically engaged in a self-advocacy response (e.g., “you can't do that”), but indicated that they would take no further action. Performance for Amy and Cindy during pretest generalization probes was similar to performance during baseline probes. Following training, Amy and Cindy scored 100% correct on all reporting skills at the posttest, at 2- and 4-week follow-up, and during generalization probes. Amy required five training sessions to meet criterion; Cindy required 10 sessions.

Figure 1
Percentage of steps of the reporting task analysis performed correctly by Amy (top) and Cindy (bottom). The filled circles represent performance on baseline, posttest, and 2- and 4-week follow-ups. The open circles represent performance during pre- and ...

Although the use of video scenarios allows the participants to view and role play across a variety of settings and situations, the lack of in situ testing is a significant limitation of the study. Without that measure, it is impossible to know if participants could perform the reporting skills in the presence of the natural stimuli (i.e., an actual inappropriate staff-to-resident interaction). Previous research has found, in fact, that performance during role-play assessments does not always predict performance during in situ assessments (Lumley et al., 1998) and that in situ training can result in criterion-level performance needed to show skill mastery (Miltenberger et al., 1999). In those studies, however, confederates engaged in lures that did not require actual abuse to occur. Because of ethical concerns with in situ exposure of participants to most of the inappropriate interactions involved in this study's instructional content (e.g., touching of private body parts, kissing, yelling at or belittling a participant), it was not possible to conduct a test for generalization in the natural environment. Therefore, although reporting skills in the current study did generalize to novel settings (i.e., location, time of day, witnesses, different telephone), the performance of participants in actual situations of abuse cannot be assumed.

Another limitation of this study is that only adult female participants with mild intellectual disabilities served as participants. The effectiveness of this training with adult males or individuals with more severe cognitive impairments cannot be inferred. In addition, most of the scenarios presented involved nonprofessional staff (direct-care staff) rather than professional staff (e.g., doctors, psychologists). An individual's discrimination between appropriate and inappropriate interactions as well as his or her willingness to report inappropriate interactions may be altered by the status of the perpetrator. Finally, this study examined only abuse by caregivers in a state-operated facility. Abuse may also be perpetrated by family members or community caregivers. The relationship of the perpetrator to the individual with disabilities may alter the individual's discrimination between appropriate and inappropriate interactions as well as his or her willingness to report inappropriate interactions. The effectiveness of this training should be evaluated with individuals who live at home or in community living arrangements.

Several of the above-mentioned limitations can be addressed in future research. The effectiveness of a similar behavioral skills training package with men or individuals with more severe cognitive impairments should be conducted. If similar training is effective with different populations, this would increase the utility of the training package. A second area of future research is the incorporation of perpetrators with varying levels of authority in institutional as well as community and home living situations. Future research could examine the effectiveness of a similar behavioral skills training program in the context of group training, which has been shown to be effective in sexual abuse prevention (e.g., Miltenberger et al., 1999). If effective, group training may serve to decrease the time, cost, and effort involved in training a large number of individuals. Finally and most important, future research could examine ways to assess in situ performance of the self-protection response ethically and realistically.


We thank Doug Boe, Joanna Kooistra, and Maranda Trahan for their assistance with data collection and the graph. We also thank the residents and staff of the facility where the study was conducted.


  • Gross A, Miltenberger R, Knudson P, Bosch A, Breitwieser C.B. Preliminary evaluation of a parent training program to prevent gun play. Journal of Applied Behavior Analysis. 2007;40:691–695. [PMC free article] [PubMed]
  • Lumley V.A, Miltenberger R.G. Sexual abuse prevention for persons with mental retardation. American Journal on Mental Retardation. 1997;101:459–472. [PubMed]
  • Lumley V.A, Miltenberger R.G, Long E.S, Rapp J.T, Roberts J.A. Evaluation of a sexual abuse prevention program for adults with mental retardation. Journal of Applied Behavior Analysis. 1998;31:91–101. [PMC free article] [PubMed]
  • McCartney J.R, Campbell V.A. Confirmed abuse cases in public residential facilities for persons with mental retardation: A multi-state study. Mental Retardation. 1998;36:465–473. [PubMed]
  • Miltenberger R.G, Roberts J.A, Ellingson S, Galensky T, Rapp J.T, Long E.S, et al. Training and generalization of sexual abuse prevention skills for women with mental retardation. Journal of Applied Behavior Analysis. 1999;32:385–388. [PMC free article] [PubMed]

Articles from Journal of Applied Behavior Analysis are provided here courtesy of Society for the Experimental Analysis of Behavior