Introduction

Throughout the United States and most of the world, pedestrian injury is one of the top causes of child deaths. For boys and girls, ages 5–14 in the US, pedestrian injury is the third major cause of injury-related death (Borse et al. 2008). Children with disabilities are at two to three times greater risk than typically developing children to be killed in pedestrian accidents (Strauss et al. 1998). In addition, children with disabilities, ages 5–17, are five times more likely than typically developing children to be struck by a vehicle. Clearly, there is a great need for effective pedestrian safety skills training interventions (Xiang et al. 2006).

Research shows that when teaching safety skills (including pedestrian safety skills) to individuals with developmental disabilities active learning approaches are most effective. Discussion of safety skills alone is not as effective as behavioral skills training (BST, instructions, modeling, rehearsal, and feedback) (Dixon et al. 2010; Wright and Wolery 2011). Similarly, utilizing only some parts of BST is not as effective as using all aspects of BST (Neilson and Bowes 1994).

Numerous studies have shown in situ training (BST conducted in the natural environment) to be effective for teaching safety skills to individuals with developmental disabilities including: abduction-prevention (Gast et al. 1993; Gunby et al. 2010), seeking help when lost (Bergstrom et al. 2012; Taylor et al. 2004), sexual abuse prevention (Miltenberger et al. 1999), and pedestrian skills (Batu et al. 2004; Blew et al. 1985; Collins et al. 1993; Horner et al. 1985; Marchetti et al. 1983). Additionally, studies showed that training in real settings resulted in better acquisition and maintenance of safety skills than training that occurred in contrived or simulated settings (Dixon et al. 2010; Mechling 2008; Wright and Wolery 2011).

Research examining pedestrian safety skills trainings for children with autism has shown that utilizing rehearsal with a doll in a model intersection combined with exposure to video recordings of intersections resulted in acquisition, maintenance, and generalization to novel street settings (Steinborn and Knapp 1982). Similarly, Neilson and Bowes (1994) demonstrated the effectiveness of in situ training and Blew et al. (1985) demonstrated the effectiveness of peer modeling and peer tutoring in situ (with instruction, reinforcement, guiding, and prompting) for teaching pedestrian skills to children with autism. Studies examining the effectiveness of virtual reality pedestrian training programs for children with autism have shown that when combined with BST components (instruction, modeling, rehearsal, and feedback), training resulted in acquisition of skills in the virtual environment, but minimal generalization to natural street settings (Goldsmith 2008; Josman et al. 2008).

As discussed above, safety skills training using active learning approaches improves pedestrian safety skills for individuals with autism. However, very few studies assessed generalization in more than one street setting (Neilson and Bowes 1994; Steinborn and Knapp 1982) and none assessed implementation of training by parents. If parents were involved in training pedestrian safety skills in the natural environment, it might result in better generalization. When teaching a chained skill or a sequence of behaviors in the actual setting, such as training street crossing on an actual street, it is important to examine the type of training that would keep the participant safe and result in rapid and thorough acquisition of skills (Wright and Wolery 2011).

A variety of response prompts (e.g., modeling, physical guidance) have been effective in teaching chained responding to individuals with autism (Libby et al. 2008). Generally, physical prompts are faded using either most-to-least or least-to-most techniques. Most-to-least prompting is an errorless training method that involves providing the most intense prompt for the individual to respond correctly; it involves physically guiding the individual through the entire sequence of a behavior chain and fading the intrusiveness of the prompt as the learner begins to engage in each skill of the behavior chain more independently. On the other hand, least-to-most prompting procedures first provide the individual with the opportunity to engage in the correct behavior independently, and prompt levels are gradually increased with each trial. Libby et al. (2008) found that most-to-least prompting procedures led to fewer errors (incorrectly performed steps of the behavior chain) than least-to-most prompting procedures when teaching new skills to children with autism.

Numerous studies have indicated that most-to-least prompting has been successful in training individuals with intellectual disabilities and autism (Richmond and Lewallen 1983; Vuran 2008). Specifically, most-to-least prompting procedures have been shown to be effective in training chained skills (Kayser et al. 1986; McDonnell and Ferguson 1989; Yilmaz et al. 2010) including pedestrian safety skills (Batu et al. 2004). However, no studies have examined the effectiveness of most-to-least prompting procedures for teaching individuals with autism pedestrian safety skills, particularly, involving parents as interventionists.

The current literature on pedestrian skills training for individuals with autism suggests several implications for future research. First, generalization of skills should be systematically promoted through training by parent or caregiver in multiple settings. Furthermore, research needs to examine generalization of target skills in response to untrained stimuli. In this study, such untrained stimuli included a stop sign, a crosswalk, or a pedestrian signal in novel settings.

Second, it is important to identify a pedestrian safety skills training method that can be readily and easily implemented by parents or caregivers for individuals with autism. Identifying training methods that do not require substantial parent training, time, cost, or effort would have high social validity, which would promote successful implementation. This is an important aspect to examine when considering a readily available intervention, and especially one that can be used long-term by parents or caregivers.

Third, most of the pedestrian safety skills training studies had minimal family participation. Research has shown the importance of family involvement when developing and implementing interventions for individuals with autism and has shown positive results for the entire family when family members implement the treatment or training (Crockett et al. 2007; Dunst and Trivette 2005; Symon 2005). Parents have been found to effectively teach their typically developing children abduction-prevention skills (Beck and Miltenberger 2009), and using parents as interventionists resulted in improvements in safety skill acquisition of children with developmental disabilities (Dixon et al. 2010; Mechling 2008; Wright and Wolery 2011). Using BST procedures to train parents to implement pedestrian safety interventions has been successful in teaching typically developing children safety skills (Limbourg and Gerber 1981; Phillips and Todman 1999; Rivara et al. 1991). However, none of the studies on pedestrian safety skills training of individuals with autism employed BST procedures to train parents to implement intervention procedures. Furthermore, it is not clear from the literature whether the families can implement the pedestrian safety skills training with fidelity.

Fourth, the literature examining pedestrian safety skills interventions for individuals with autism only targeted children and adolescents, ages 5–16. No studies have been conducted to teach pedestrian safety skills to older adolescents or adults with autism, which is an important skill set to learn when preparing to transition to a more independent lifestyle.

This study examined parent implementation of in situ pedestrian skills training that used most-to-least prompting procedures for individuals with autism. The study expands the literature by: using BST for parent training; promoting and assessing parent use of prompting procedures; teaching pedestrian safety skills to individuals with autism in multiple street settings in the community; and evaluating its impact on acquisition, generalization to novel settings with untrained stimuli, and maintenance of the individual’s pedestrian safety skills. Specifically, we evaluated whether: (a) BST would promote parent implementation of pedestrian safety skills training that employed most-to-least prompting procedures; (b) parents could generalize their implementation of intervention to a novel street setting; (c) implementation of in situ training by parents would result in improvement of pedestrian safety skills for individuals with autism; and (d) the individual’s acquired skills would generalize to novel settings with untrained stimuli and maintain at 2-weeks and 1-month follow-ups.

Methods

Participants

Originally, six parent–child families were recruited to participate in the study. Three of the families withdrew from the study for various reasons: one family was evicted unexpectedly and was required to move, one of the mothers became pregnant and did not believe she would have time, and another child participant from another family was temporally moved to a community facility due to his severe problem behavior. The remaining participants included three individuals with autism (two adolescents and one young adult) and their parents. All families were middle class, Caucasian. The participants were recruited from referrals by behavior analysts in local community agencies and family self-referrals. Information about the study was emailed to local behavior analysts working with families of children and adults with autism. Inclusion criteria for individuals with autism included the following: (a) ages 13–25; (b) having difficulty with crossing streets independently and safely; (c) being able to understand and comply with one-step verbal directions; and (d) living with a parent willing to be trained and to implement intervention. Exclusion criteria included: (a) not able to understand or comply with one- to two-step commands; (b) engaged in behavior that would put them in danger in actual street settings (darting, eloping); (c) did not allow their parent to touch them; and (d) parents had prior experience with pedestrian safety skills training.

A.L. was a 14-year-old male with autism and speech and language impairments. He was in the 9th grade at a public high school and received occupational therapy. He was from a two-parent household. A.L.’s recent Individualized Education Plan (IEP) indicated that he could follow three- to five-word directions and verbally approximate single words in response to questions with a verbal prompt but did not initiate communication. He utilized approximations, gestures, and static picture symbol boards to communicate. He demonstrated functional fine motor skills such as writing, cutting, opening packages, and using utensils to eat. He was unable to discern dangerous situations, did not generalize skills to new settings, and could not safely cross streets or be near streets, and could not safely be in new environments without adult supervision. A.L.’s father provided training; he was 58 and worked from home as a computer programmer.

J.M. was a 15-year-old male diagnosed with autism and attention deficit hyper-activity disorder at age 3 by a licensed psychiatrist. He was in the 9th grade at a public high school. His Individualized Transition Plan (ITP) included in his IEP indicated that his word recognition and oral reading levels were at a fourth grade level and his reading and math skills were at a third grade level. His could multiply and divide single digit numbers. He had trouble complying with three-step directions but followed two-step directions. J.M. navigated his way around his school without assistance or monitoring, and independently engaged in self-help and daily skills without assistance. He had good fine motor skills. J.M. lived with his mother who provided training to J.M. for this study; she was 44 years old and working as a senior project analyst.

I.M. was a 23 year-old, young adult male with a high school degree and was working to start up his own business. He was diagnosed with autism at age 3 by a licensed psychiatrist. He could prepare and pack his lunch for his job, maintain his room at home, take care of his personal hygiene, and change his own clothes. His high school ITP indicated that he was at a 3rd grade level for reading and writing. He received language therapy for 90 min per week and was participating in a transition program at the time of the study with a vocational trainer providing support to him in all environments. He could follow three- to five-step directions with visual and verbal prompting, and generally followed a task analysis independently until he completed the steps in the entire task analysis. He was receiving instruction and information on using public transportation to travel to and from his job. I.M. lived with his mother who provided training for I.M. for this study; she was in her 50 s and working as an advocate.

Setting

This study took place in community settings on three types of streets: (a) roads with no stop signs or pedestrian crosswalks or signalization, (b) roads with stop signs for cars and pedestrian crosswalks, and (c) roads with signalized pedestrian crosswalks. To select the target streets where in situ training would be implemented, the participants’ parents and researcher (first author) went to different street crossing locations and chose streets based on (a) the likelihood of using the streets; (b) parents’ level of comfort; and (d) crossing the streets required parental monitoring. All street crossings occurred on street corners, not at midpoints between streets, or jaywalking.

Measurement

All in situ trials that occurred on street settings were video-recorded, with the exception of generalization probes and follow-up, and later scored by the first author and a research assistant. The video camera was positioned using a tri-pod in a location that could capture the entire trial being performed (from sidewalk or from curb to curb). In addition, each trial was audio recorded on a phone by the first author who acted as the safety confederate to monitor any apparent danger to the parent, child, or both. The author was close enough to the participants to audio record the trials. Each in situ trial varied in length depending on the family. It took approximately 1 min on the 1st and 2nd street types and 1-3 min on the 3rd street type depending on how long the pedestrian light required to change from ‘wait’ to ‘walk.’ Data were collected one to three times per week, depending on availability, for approximately 1.5 months for each participant.

Parent Implementation of Safety Skills Training Procedures

Parent implementation of pedestrian safety skills training that used most-to-least prompting procedures was measured as the percentage of correctly implemented steps involved in the hierarchy of prompts (e.g., deliver task direction, physically assist child to perform the task using one or both hands, praise for completion of task, use of higher prompt level that was last successful for an incorrect or no response) for each task-analyzed pedestrian safety skill. It focused on assessing the extent to which the parents implemented the steps in the pedestrian safety skills training with fidelity as designed. Data were collected using a checklist across street settings in all experimental phases.

Pedestrian Safety Skills

Each participant’s independent correct use of pedestrian safety skills was measured. Safety skills were developed by the authors based on the street type and the literature (Wright and Wolery 2011), and steps were developed for crossing the street (e.g., stops at curb; looks left, right, and left; waits at curb until no cars are coming or traffic is stopped; begin crossing street within 5 s; cross in straight line to other side). We measured whether each participant independently used each of the task-analyzed skills correctly with no prompts (5 points), verbal prompts (4 points), gestural prompts (3 points), partial physical prompts (2 points), or full physical prompts (1 point). Total possible points for each street crossing behavior chain for child participants were 25 for independently engaging in five correct behaviors. The level of independent use of skills was measured as a percentage of safety skills performed correctly by dividing scores earned by total possible scores.

Procedural Integrity

The researcher’s procedural integrity of the parent training delivery was assessed using a checklist to ensure that the training procedures were delivered consistently across parents. The integrity checklist consisted of the 10 steps of the parent training process that included BST and in situ feedback. To measure integrity, 34 % of the training sessions were audio recorded and scored by an independent research assistant. Procedural integrity was calculated by dividing the number of steps delivered correctly by the total number of steps and multiplying by 100. The procedural integrity was scored at 100 % across parents indicating that the researcher delivered all BST training and in situ feedback procedures correctly in each session and trial.

Social Validity

Parents were asked to complete a social validity questionnaire, an adapted version of the Treatment and Acceptability Rating Form-Revised (TARF-R; Reimers and Wacker 1988) during follow-up. The questionnaire used a Likert-type scale to rate effectiveness and acceptability of the intervention from 1 to 5 using 15 items, with counterbalanced questions (i.e., for some questions, 1 indicates acceptability and 5 indicated an unacceptable score).

Interobserver Agreement (IOA)

IOA was assessed during 35 % of the sessions across phases and participants by having a second data collector independently watch the video-recorded trials. A point-by-point procedure was used, and agreements were calculated by dividing the number of agreements by the number of agreements plus disagreements, multiplied by 100. Video clips were used to train two research assistants from a master’s program in Applied Behavior Analysis until 90 % agreement was reached. The training video clips were of parents and child participants who had previously dropped out of the study for various reasons. The researcher watched the video with the research assistants to train them to record and score each sample step. The mean IOA scores across streets and phases for each parent and child participant were 92 % (80–100 %) for A.L., 94 % (86–100 %) for J.M., and 92 % (80–100 %) for L.M. IOA for procedural integrity was 100 % for parents across sessions.

Experimental Design and Procedures

This study employed a multiple baseline design across participants. The families were concurrently enrolled into the study, and intervention was staggered across families. The design consisted of four phases: (a) baseline, (b) intervention, (c) fading, and (d) follow-up.

Baseline

Baseline involved taking participants to a street crossing location, recording the use of prompting procedures by parents to train their children to use pedestrian safety skills, and recording the safety skill performance by children. Parents were asked to interact with their children in the target crosswalks on community streets. The researcher maintained close proximity to the participants, but did not provide parents training on how to use the most-to-least prompting procedures nor feedback that was used in intervention. Baseline data began at the same time for all family participants. At minimum, four data points were collected for each type of street.

Intervention

The in situ pedestrian skills training intervention was implemented by parents in the community. The parents were trained in the use of most-to-least prompting procedures to teach their children pedestrian safety skills.

Parent Training

Parents participated in 10- to 15-min individual BST training sessions in their home each day prior to the parent and their child accompanying the researcher to the training sites in the community. During training, the researcher used instruction, modeling, rehearsal, and feedback to help the parents implement most-to-least prompting procedures. After instruction, the researcher modeled the use of the procedures in a role-play context. The researcher then had the parent rehearse the procedures with the researcher as the child and provided praise for correctly performed behaviors and corrective feedback for incorrectly performed behaviors. The parents were trained to use the following steps during each intervention session to teach pedestrian safety skills to their children in a street setting:

  1. 1.

    Full physical prompt Physically assist child to perform a step while providing verbal directions (e.g., placing one hand on the child’s back while pushing him towards the button and placing another hand on the child’s hand to push/pull hand toward button while providing instruction, “push the button”). Praise for completion of the task or skill with prompt.

  2. 2.

    Partial physical prompt Assist child by lightly placing, touching, or shadowing a part of the body with one or both hands to prompt to perform skill (e.g., lightly touching the child’s elbow with the direction, “push the button.”) and be ready to move into full physical prompting. Provide praise for completion of task, or use full physical prompt that was last successful in having the child complete step for an incorrect or no response.

  3. 3.

    Gestural prompt Point to an object or direction and verbally direct child to perform the task; praise for completion of task, or use of higher prompt level that was last successful in having the child complete the step for an incorrect or no response.

  4. 4.

    Verbal prompt Provide verbal direction without physical or gestural prompt; praise for completing the task independently, or use of higher prompt level that was last successful in having the child complete the step for an incorrect or no response.

  5. 5.

    No prompt Allow child to independently engage in the task; praise for completing the task independently, or use of higher prompt level that was last successful in having the child complete the step for an incorrect or no response.

During training, we explained to the parents that each trial in the street would include a full run through of each step in the task analysis of pedestrian safety skills for each street setting type.

Implementation

Following each parent training, parent implementation of in situ pedestrian skills training began with using the full physical prompts in the first two training trials for each step in the task analysis in each street setting. Subsequent trials involved moving down through the levels of prompting, depending on the child’s behavior, and using higher prompt levels when engagement in the correct behavior did not occur. During the first two in situ training trials, the parent provided the most-to-least prompt hierarchy with no delay following the verbal direction or between each prompt in the hierarchy to facilitate high rates of correct responding. During subsequent trials, the parent followed the same procedure, but used a 3-s delay following the verbal direction and also between each prompt in hierarchy. If the child engaged independently in the target response within the 3-s delay, praise was provided. If the child did not engage in the target response within the 3-s delay, the parent provided the next higher level of assistance in the prompting hierarchy. The parent provided praise (e.g., “good”) when the child engaged in the prompted behavior. However, for skills performed correctly, the parent provided specific verbal compliments with higher inflection and greater enthusiasm than for prompted behaviors (e.g., “Awesome! That was great how you pushed the button all by yourself! You are so smart!”). After the first two trials, the parent and researcher collaboratively decided the level of prompting before beginning a new trial. If the parent was having trouble implementing the procedures, the researcher gave corrective feedback at the end of the trial. If the parent implemented the prompting procedures below 80 % accuracy, the researcher provided in situ feedback.

Fading

When parental implementation of the training procedures reached 80 % accuracy in each type of street for three consecutive trials, BST was removed. If parents made a mistake in providing prompts, the researcher provided corrective feedback at the end of the trial.

Generalization and Follow-Up

To examine the parents’ implementation of the prompting procedures in novel street settings, where no BST or feedback was provided, and to examine its collateral effects on the participant’s use of pedestrian safety skills, probe data were collected across baseline and intervention phases. The second and third street served as generalization settings before introducing interventions for the first and the second street, respectively. Follow-up data were collected 2 weeks and 1 month following the intervention with J.M. and 1 month following the intervention with A.L. and I.M. to determine whether the participants’ levels of independent correct use of the safety skills were maintained. The procedures used during follow-up sessions were similar to those of baseline conditions with the exception that the parents were present, but simply acted as chaperone to intervene if their son was in danger. In addition, video recording was not used during 14 of 36 (39 %) follow-up sessions in order to examine participant reactivity to the presence of a video camera. The data showed no signs that participants reacted to the presence of the video cameras. The average number of steps in which the participants were correctly engaged was 77.18 and 79.14 % with and without the video camera present, respectively.

Results

Parent Implementation of Pedestrian Safety Skills Training

Figure 1 presents the percentage of treatment steps implemented correctly by the parents to teach their children pedestrian safety skills in the community during each phase of the experiment. All parents implemented the training procedures with high levels of accuracy across street locations in intervention and fading phases. As shown in Fig. 1, levels of correct implementation of the steps were low across street locations in baseline (0–3 %). However, during intervention when BST was provided, the parents implemented the intervention procedures with over 80 % accuracy in all sessions across street locations, except during the second training trial in Street 3 for A.L.’s parent and the third training trial in Street 1 for I.M.’s parent. Once they received in situ feedback, their levels of correct implementation increased to 90–100 %. During intervention, the mean percentage of correct implementation of the training procedures was 94 % (80–100 %) in Street 1, 95 % (80–100 %) in Street 2, and 92 % (50–100 %) in Street 3 for A.L.’s parent; 91 % in Street 1, 96 % in Street 2, and 92 % in Street 3 for J.M.’s parent; and 98 % in Street 1, 99 % in Street 2, and 97 % in Street 3 for I.M.’s parent.

Fig. 1
figure 1

Percentage of correct implementation of in situ pedestrian safety skills training across parents

Full size image

As shown in Fig. 1, during fading of BST when only verbal feedback was provided at the end of each trial, parents’ correct implementation of the intervention procedures remained at high levels, averaging over 90 % (80–100 %) across streets for all parents. I.M.’s parent consistently implemented the procedures with 98–100 % accuracy in all streets. No in situ feedback was required for any of the parents during the fading phases in any of the street types.

Figure 1 also displays parents’ implementation of intervention for untrained street types during generalization probes. The data indicated that there was little or no improvement in implementation for street 2 or street 3 for any parents until training was implemented for those streets.

Pedestrian Safety Skills

Figure 2 shows the results of pedestrian safety skills performed by the participants. Results are organized to display the percentage of independent, correct use of pedestrian safety skills across experimental conditions and three streets for each participant. A.L., J.M., and I.M. showed, respectively, stable patterns in baseline at 33 % (33–33 %), 44 % (33–50 %), and 57 % (25–80 %) mean independent, correct use of pedestrian safety skills in Street 1.

Fig. 2
figure 2

Percentage of independently performed correct safety skills across participants

Full size image

During intervention there was a substantial increase in the slope and level of all three participants’ independent, correct use of safety skills in Street 1. Intervention data reflected an average of 74 % (46–90 %), 89 % (60–100 %), and 81 % (55–96 %) for A.L., J.M., and I.M., respectively. During fading in Street 1, the participant’s levels of independent, correct use of safety skills continued to increase. The mean independent correct use of skills was 84 % (71–90 %), 98 % (96–100 %), and 86 % (75–96 %) for A.L., J.M., and I.M., respectively, during this phase in Street 1. The same patterns were observed in Streets 2 and 3; A.L.’s independent use of safety skills increased to an average of 67 % (24–89 %) and 72 % (56–80 %) in intervention from 37 % (33–50 %) and 27 % (25–33 %) in baseline in Streets 2 and 3, respectively.

J.M.’s pedestrian safety skills improved to an average of 86 % (58–100 %) and 85 % (62–98 %) in intervention from 43 % (33–50 %) and 45 % (42–50 %) in Streets 2 and 3, respectively. I.M.’s safety skills improved to an average 83 % (60–96 %) and 85 % (60–97 %) from 61 % (50–87 %) and 61 % (50–80 %) in Street 2 and Street 3 during baseline. During fading of BST for parents, except for one session in Street 2 for A.L. and in Street 3 for J.M., the levels of independent correct use of skills remained stable at high levels.

Figure 2 also displays participant use of safety skills during follow-up and generalization probes to street 2 and street 3 before training on those streets. The follow-up data showed that for all three participants, the acquired skills were maintained above baseline levels at 1-month follow-up. However, data during the generalization probes show little or no generalization of the skills to untrained street settings.

Social Validity

The average TARF-R scores across parents indicated that the in situ pedestrian skills intervention was highly acceptable, with a mean score of 89.5 % (86–93 %) of total possible points across parents. The lowest rating was a score of 2 and the highest rating was a score of 5 on a scale of 1–5. The lowest rating score of 2 was from A.L.’s parent who rated his child’s generalization to novel settings as poor. The parents indicated that it was easy to implement the most-to-least prompting procedures (4 out of 5) and children were more cautious (4 out of 5) when crossing streets after the intervention than before intervention.

Discussion

Results from this study indicated that BST helped parents accurately implement the in situ pedestrian safety skills intervention that employed most-to-least prompting procedures. Furthermore, the parents implemented the intervention during fading when BST was not provided. However, intervention implementation did not generalize to Streets 2 and 3 before the implementation of BST for those streets. Results also indicated that training increased the safety skills for all individuals with autism who received training both during and after intervention. Although there was no generalization of the pedestrian safety skills to untrained street settings, it was found that the skills learned during intervention were maintained for all participants.

The current study supports research that examined the effects of training parents using BST procedures to implement street safety training to children (Phillips and Todman 1999; Rivara et al. 1991; Rothengatter 1984). It builds on this research by showing the effectiveness of BST for teaching parents, the effectiveness of the parents as trainers, and the success of the training on the safety skills exhibited by the adolescents and young adult receiving the training.

The lack of generalization of parent implementation of in situ safety skills training across street types is likely due to the differences in street types. Street types 2 and 3 were regarded as dangerous due to their placement in higher-volume traffic areas compared to Street 1 (Zegeer et al. 2005) where trials generally took place in residential neighborhoods with little to no traffic. Therefore, parents may have felt less confident in their child’s ability to engage in the appropriate steps, resulting in higher prompt levels, and parents often performing the steps for their child. The results indicate that generalization promotion support may be needed to facilitate parents’ successful implementation of intervention during non-trained settings (Lucyshyn et al. 2007; Sears et al. 2003). I.M.’s parent, who received parent training in behavioral techniques prior to participating in the study, was familiar with prompting procedures. It was unknown how familiar she was with the specific procedures implemented in the study. However, her performance was significantly better than the other two parents.

Another factor that may have contributed to limited parent implementation performance during generalization probes and high parent performance levels during intervention sessions is reactivity (e.g., Brackett et al. 2007; Mowery et al. 2010). Because all other intervention sessions involved the use of a video camera to measure parent implementation of intervention procedures, parent performance may have been influenced by reactivity to being monitored by a video camera. Parent implementation performance was not video recorded during generalization probes, and parents were unaware that the researcher was measuring their performance, which may have caused a decrease in performance levels during probes. This finding suggests that parent reactivity may have occurred to the presence of a video camera resulting in high performance levels during intervention and low performance levels during generalization probes. However, as indicated in data collected during follow-up, the possibility of a video effect on the participants’ use of safety skills seems minimal in that no specific patterns were observed in the rates of independent, correct use of the pedestrian safety skills across child participants during sessions with and without the video camera.

Results of this investigation suggest that the pedestrian safety skills training implemented by parents was successful in teaching the participating two adolescents and one adult with autism to use safety skills to cross streets. The results indicate that in situ training may be successful in increasing independent use of safety skills not only for high functioning individuals with autism, but also for low functioning individuals with autism as in the case of A.L. This finding adds to the relatively sparse data providing support for the outcomes of pedestrian safety skills training involving the use of in situ training procedures for individuals with developmental disabilities (Batu et al. 2004), including autism (Neilson and Bowes 1994). Furthermore, the finding supports existing research that incorporates most-to-least prompting procedures to teach pedestrian-safety skills to individuals with autism and other disabilities (Batu et al. 2004; Collozi and Pollow 1984; Kayser et al. 1986; McDonnell and Ferguson 1989). As indicated by Batu et al. (2004), using most-to-least prompting could be a viable intervention method when teaching pedestrian safety skills in natural street environments to prevent the occurrence of errors, especially with the initial introduction of an intervention. The decreased number of errors associated with using most-to-least prompting procedures may allow for a safer in situ pedestrian safety skills training.

One limitation of the current study was that to account for safety precautions, most trials took place on relatively calm streets. However, during fading, all three parents requested or suggested going to street settings where there were likely to be more cars. As shown with A.L. and his parent, the parent implemented the intervention procedures with high levels of accuracy, which led to further increases in A.L’s independent use of safety skills. Another limitation is that this study involved only three families and thus the results should be interpreted with caution. A third limitation of the study is the relatively short time participants were involved in this study. Future research should examine maintenance of the in situ pedestrian safety skills over longer time periods. Despite its limitations, this study offers a meaningful contribution to the literature because it is one of the few that implemented in situ pedestrian safety skills training for individuals with autism and involved parents in implementing the training.