SEARCH

SEARCH BY CITATION

Keywords:

  • behavioral skills training;
  • computer-based teaching;
  • dangers;
  • in situ training;
  • lighters;
  • poison;
  • safety skills;
  • stranger

Abstract

  1. Top of page
  2. Abstract
  3. STUDY 1: ABDUCTION PREVENTION
  4. STUDY 2: POISON AND LIGHTER SAFETY
  5. STUDY 3: COMPREHENSIVE SAFETY SKILLS PROGRAM
  6. GENERAL DISCUSSION
  7. REFERENCES

Previous research has demonstrated the efficacy of behavioral skills training (BST) and in situ training (IST) for teaching children to protect themselves. However, BST may be resource intensive and difficult to implement on a large scale. We evaluated a computerized version of BST (CBST) to teach safety skills and determined the extent to which safety skills generalized across different dangers. In Study 1, 11 children learned, via CBST and IST, to respond safely when asked to leave with a stranger. In Studies 2 and 3, IST was implemented with 16 children for 1 or 2 dangers after exposure to CBST for 3 dangers. Participants correctly self-protected from dangers after CBST and IST, and performance generalized to similar dangers for which participants did not receive IST. CBST may be an acceptable substitute for BST when combined with IST to improve efficiency and maintain efficacy in a comprehensive safety skills program.

Unintentional drowning, burns, firearm play, poisoning, and suffocation are among the top 10 leading causes of injury-related deaths for 5- to 9-year-old children (Centers for Disease Control and Prevention, 2011). There are two possible courses of action to prevent unnecessary injuries and deaths. The first is environmental modification such as childproof locks and increased supervision. Although environmental modifications are necessary and are often successful, modifications may sometimes fail. The second possibility is to teach children to protect themselves.

Previous research has demonstrated that children do not protect themselves reliably in dangerous situations. In two studies (Beck & Miltenberger, 2009; Poche, Yoder, & Miltenberger, 1988), 25 of 26 children (5 to 8 years old) did not get away from or tell an adult about a stranger who approached them. In two other studies (Kelso, Miltenberger, Waters, Egemo-Helm, & Bagne, 2007; Miltenberger et al., 2009), 7 of 52 children (6 to 9 years old) touched a gun, and 37 children did not get away or tell an adult about the gun. Children continued to respond incorrectly to dangerous situations even after they received traditional instruction that consisted of videos and lectures. In Poche et al. (1988), 24 of 35 children stayed near a stranger after listening to a standard lecture-type presentation about strangers or watching a videotape, and in Beck and Miltenberger (2009), five of six children continued to stay near strangers after watching a video. In Himle, Miltenberger, Gatheridge, and Flessner (2004), a treatment group composed of 11 children did not have a higher average safety score than a control group composed of 10 children during an in situ assessment after they watched a commercially available gun safety video.

Previous studies have demonstrated the efficacy of behavioral skills training (BST) for teaching safety skills to young children (Johnson et al., 2005, 2006; Poche, Brouwer, & Swearingen, 1981; Poche et al., 1988). During BST, a teacher first provides instructions about the danger and how to respond to the danger. The teacher then reenacts a scenario with the danger and models the correct response. Finally, the teacher arranges a simulation, and the students respond to the danger in a role-play. During the role-play, the teacher provides feedback that may include praise for correct responses and vocal, model, or physical prompts after incorrect responses. BST is different from traditional “stranger danger” programs primarily because children act out dangerous situations and receive feedback on their performance.

Limitations of BST are revealed during in situ assessments that involve reproducing the danger in the natural setting without the child knowing a parent or teacher is watching. Himle, Miltenberger, Gatheridge, and Flessner (2004) found that most children did not leave an area with a gun or report the danger to an adult after they received BST. In a study that evaluated poison safety conducted by Dancho, Thompson, and Rhoades (2008), children continued to consume items from ambiguous containers after two 30-min training sessions that included BST. In a relatively large study with 46 children (6 and 7 years old), 30% did not get away from an abductor or tell an adult during an in situ assessment that occurred after BST (Johnson et al., 2006). An additional 14% moved away from the stranger but did not report the situation to an adult.

BST is one of the most effective interventions currently available, but not all children respond correctly to dangers after only experiencing BST. In many cases, children may also require in situ training (IST) to protect themselves from dangerous situations. During IST, a parent or teacher arranges for the danger to appear in a more natural setting (e.g., a confederate stranger approaches a child at the playground), and the parent or teacher arrives to provide feedback after the child responds to the danger. Previous studies have shown consistent improvement for participants during in situ assessments following BST, only after IST was added (Beck & Miltenberger, 2009; Dancho et al., 2008; Himle, Miltenberger, Flessner, & Gatheridge, 2004; Jostad, Miltenberger, Kelso, & Knudson, 2008; Miltenberger et al., 2009).

Although the combination of BST and IST is effective, Jostad and Miltenberger (2004) suggest that BST and IST, when presented individually to each child as typically described, may be difficult to implement on a large scale (e.g., to all children in a classroom or school), and therefore modifications to the procedures that increase efficiency should be evaluated. In one of the most efficient demonstrations of BST, Carroll-Rowan and Miltenberger (1994) reported that BST was conducted on 3 separate days and lasted approximately 20 min for a classroom of 23 students for a total of 60 min of training time. Due to the individual rehearsal required during BST, as the number of children taught increases, so does the duration of training time. For the average primary school of 446 students (National Center for Education Statistics, 2001), training would consume 20 hr of instructional time for only the BST component and for only one danger. The estimate does not include the time required to train those who implement BST to do so correctly. School administrators and teachers are thus likely to seek programs that reduce the time required to teach safety skills. To that end, knowledge assessments (i.e., measures of vocal responses to questions) are sometimes substituted for measurement of behavior during an in situ assessment. Many studies suggest, however, that performance during knowledge assessments does not predict performance during role-plays or in situ assessments. For example, Himle, Miltenberger, Gatheridge, and Flessner (2004) reported that five children could tell an experimenter how to respond to gun safety scenarios after they had viewed commercially available gun safety videos, but none of the five responded correctly during an in situ assessment. Carroll-Rowan and Miltenberger (1994) found that approximately 30% of participants performed better in knowledge assessments than during in situ assessments after they had received various types of training for abduction-prevention skills. Replacing in situ assessments with knowledge assessments may save time; however, the cost is an increased chance that the correct safety skills will not be used when a child encounters an actual threat.

Some researchers have attempted to package BST into videos and manuals for easier implementation in schools. Carroll-Rowan and Miltenberger (1994) found that teachers implemented BST correctly while following a manual, and that the efficacy of the video and the manual were similar. Nevertheless, most children required IST. Other ways to save time and resources are to present BST to groups of children or include children as trainers. Based on their review of self-protection studies that used individual and group training, Miltenberger and Olsen (1996) asserted that fewer children learned all of the safety skills in the studies using group training compared to those that used individual training. Jostad et al. (2008) trained 6- and 7-year-old children to provide BST for gun safety as peer tutors to 4- and 5-year-old children. The use of peers may decrease the staff resources required for training, but resources are still required to provide training for the peers. Given these challenges, it is not surprising that school administrators and teachers use coloring books, videos, and lectures to teach safety skills, even though these tools are routinely ineffective (Carroll, Miltenberger, & O'Neill, 1992).

Previous research has demonstrated the generality of acquired safety skills across different exemplars of the same danger (e.g., different poison containers or different lures used by an abductor; Dancho et al., 2008; Poche et al., 1981) and across settings (e.g., Poche et al., 1981). Previous self-protection research has not, however, evaluated generalization of safety skills across distinctly different dangers. Children may encounter more than one type of danger and need to know how to respond appropriately to each one. Appropriate responses to different dangers are similar in many ways: Identify that a danger is present, move away from the danger, and report the danger to an adult. However, research has not evaluated procedures for promoting generalization across dangers, and it is unknown whether teaching safety skills for one danger will generalize to others.

The purpose of the current series of studies was to evaluate the effects of a computerized version of BST (CBST) on the acquisition, maintenance, and generalization of safety skills with respect to multiple dangers. A computer game with interactive components was investigated because previous research has demonstrated that video-only treatments may be ineffective (Beck & Miltenberger, 2009). A computer game, if effective, would be a relatively easy way to provide instruction, modeling, and individualized feedback to young children and would have some advantages over BST for school-wide implementation. First, no training of teachers would be required for BST. Teachers would only need access to computers with the computer game to deliver the program effectively. Second, computerized training can present many different situations easily via video. During classroom role-plays, children experience a dangerous situation with only a single adult or a particular gun, poison, lighter, and so on. In a computer game, children can view and role-play a variety of dangers and situations presented via video. Third, a computer program allows multiple children to be taught self-protection skills while maintaining the benefits of individualized instruction and providing each child with immediate and individualized feedback for performance throughout the program.

STUDY 1: ABDUCTION PREVENTION

  1. Top of page
  2. Abstract
  3. STUDY 1: ABDUCTION PREVENTION
  4. STUDY 2: POISON AND LIGHTER SAFETY
  5. STUDY 3: COMPREHENSIVE SAFETY SKILLS PROGRAM
  6. GENERAL DISCUSSION
  7. REFERENCES

The purpose of Study 1 was to evaluate the efficacy of CBST for teaching children to respond safely to a single danger.

Participants and Settings

The authors contacted a local preschool, which primarily served families of low socioeconomic status and which had indicated an interest in teaching safety skills to their students. Consent forms were sent home to the parents of children in a classroom of typically developing 5- and 6-year-olds. Children whose parents consented were selected to participate based on teacher reports of consistent attendance. Twelve children began participation in Study 1. One participant refused to participate in the in situ assessments; therefore, 11 children completed the training and analysis. All in situ assessments, as well as IST, were conducted in a hallway in the school. The computer program was presented in a small room at the school.

Design

We evaluated the efficacy of the CBST for improving self-protection skills in a concurrent multiple baseline design across subjects. We conducted in situ assessments for each participant before and after the participant completed CBST. An IST phase was conducted if a participant did not protect him- or herself following CBST alone. CBST was conducted once before all of the post-CBST in situ assessments and was repeated before the first IST session was conducted. During the IST phase, the experimenter implemented IST whenever the participant made an error during an in situ assessment. Participants were randomly assigned to a short, medium, or long baseline. The actual length of the baseline for each group was determined by visual inspection of the data.

Data Collection and Interobserver Agreement

During in situ assessments, observers scored whether participants interacted with the stranger, got away from the stranger and moved toward a safe adult (e.g., the experimenter or classroom teacher), and reported the stranger to an adult. Table 1 summarizes the possible participant responses. Reports of dangers were scored if the participant made a statement to the safe adult indicating that there was a danger or that he or she had spoken to a stranger. Any response that specifically indicated a stranger or a response that would cause an adult to become concerned and possibly ask follow-up questions was counted as a report. Participants had an opportunity to tell an adult without getting away from the stranger because the experimenter returned to the participant at the end of each assessment.

Table 1. Summary of Child Responses During In Situ Assessments
CategoryDefinition
Get away and tellThe participant began moving away from the danger within 5 s (strangers) or 10 s (lighters and poisons) to a distance arm's length from the experimenter; then reported the danger to the experimenter.
Get awayThe participant moved away from the danger within 5 s to within arm's length of the experimenter, but did not report the danger.
Tell an adultThe participant remained near the danger, but reported the danger when the experimenter returned.
Stay near dangerThe participant remained near the danger for more than 5 s (strangers) or 10 s (lighters and poisons) and did not report the danger.
Interact with dangerThe participant interacted with the danger at any time during the in situ assessment regardless of reporting or getting away.

The experimenter and the confederate stranger independently recorded participant responses during the in situ assessment. Sessions were videotaped using a very small camera hidden inside the experimenter's wristwatch. The stranger recorded audio for the session with a handheld voice recorder. Data collectors used the video and audio recordings if the experimenter and stranger could not see part of a participant's response.

Interobserver agreement data between the experimenter and the confederate stranger were collected on an average of 35% of in situ assessment sessions (range across phases and participants, 25% to 57%). Interobserver agreement was assessed for four responses: interacting with the danger, staying near the danger, getting away, and telling an adult. In each trial, an agreement was scored for each of the four responses if both observers recorded an occurrence or a nonoccurrence. Otherwise, a disagreement was scored. Agreement was calculated as a percentage of agreement for each session and averaged 97% across all sessions and participants (range, 75% to 100%).

Procedure
In situ assessment

Before the first in situ assessment, the participant's classroom teacher introduced the experimenter. At the beginning of each assessment, the experimenter came to the classroom and told the participant that they were going to play a game in a different room in the school. The participant was allowed to choose a game or toy to play with for 10 min when the experimenter and participant reached the small room. Either on the way to the small room or on the way back to the classroom, the experimenter pretended to forget something or need to take a phone call. He moved at least 1.5 m away from the participant and around the corner of another hallway. In some sessions, no stranger approached the participant; the experimenter simply returned and continued to the destination with him or her. Whether the experimenter received a phone call, forgot something, or nothing happened, and whether or not a stranger appeared, was determined randomly before each session, with the restriction that one of every three sessions did not have a stranger.

If a stranger was scheduled for a session, he or she approached the participant shortly after the experimenter left. Strangers were undergraduate students in their early twenties. Participants encountered some strangers in baseline and again in some sessions after CBST; however, other strangers appeared only in baseline or only after CBST. After approaching the participant, the confederate stranger made a comment about something the participant was wearing or had with him or her. The stranger then asked the participant to leave the area with him or her (e.g., “Wow! That is a cool shirt. Let's play over here.”) and extended his or her hand to the participant. The stranger walked with the participant away from the experimenter if the participant agreed to go with him or her. The experimenter casually interrupted the participant and stranger if they walked more than 3 m away from him. If this occurred, the stranger walked away, and the experimenter and participant continued to their original destination. The stranger asked the participant to leave a second time if the participant did not move more than 1 m away in 5 s. The stranger walked away and the experimenter returned if the participant did not move away within 5 s after the second request to leave. If the participant told the experimenter about the stranger after the experimenter returned, the experimenter said, “Thank you for telling me,” in a calm neutral tone. If the participant moved away from the stranger but did not report the stranger, the experimenter did not provide praise. If the participant moved away from the stranger and within arm's reach of the experimenter within 5 s and reported the stranger, the experimenter praised him or her (e.g., “Thank you for telling me. You did a great job getting away from that stranger and telling me about it.”). After the assessment, the participant and the experimenter continued on to the original destination.

Computerized behavioral skills training (CBST)

The first author created a computer program using the Adobe Production Premium Software Suite and the ActionScript 3 programming language. The program was presented using a Windows-based laptop computer with a standard mouse. Participants played the computer game in a single session that lasted approximately 20 min.

Before starting CBST, the experimenter told the participant that they would be playing a game that would teach them to stay safe. Outside the room was a life-sized cardboard cutout of the character who narrated the computer game. The participant was introduced to the cardboard character and the experimenter pointed out a pouch attached to the cutout. The experimenter explained that the participant would be talking to the cardboard cutout later during the program. Participants appeared to enjoy the cardboard cutout and often gave it hugs or high fives. After the participant sat in front of the computer, the experimenter placed 10 tokens on the table nearby. The experimenter told the participant that the tokens would be used later. He then helped the participant to type his or her name in the welcome screen on the computer and started the computer program.

A few components were present throughout the entire computer program. First, a screen with a pool and an animated stick figure indicated the participant's progress through the entire program (see Figure 1, left). The pool filled up a small amount after the participant completed each module. A “Dive” button appeared when the pool was full and the participant had completed all of the modules. Clicking the “Dive” button caused the animated stick figure to dive into the pool with a splash. The stick figure reemerged from the water and waved goodbye before the program closed. Second, a paper airplane at the top of the screen indicated progress through each individual module when the participant was required to respond. Figure 1 (right) shows the paper plane at the top of the screen. The plane moved across the screen when a participant made a correct response. The plane stopped in the goal area, and the module ended when the participant made the total number of correct responses required. Third, a live-action on-screen narrator (the same person as the cardboard cutout) delivered instructions and provided praise and corrective feedback throughout the program. Although some textual stimuli were presented throughout the program, the text was supplemented by pictures and vocal instruction from the narrator, and did not require a reading repertoire. Table 2 provides a summary of the individual modules included in the game.

image

Figure 1. Screen shots of the computer program. The right panel shows the pool screen that tracked participant progress through the game. The left panel shows a module from the version of the game that was used in Studies 2 and 3, in which participants moved the pictures of safe and unsafe items to the two boxes at the bottom of the screen. The paper plane at the top of the screen keeps track of points and progress through the module.

Download figure to PowerPoint

Table 2. Summary of CBST Components
ComponentsDescriptionStudy 1Studies 2 and 3
Note
  1. Included dangers were strangers in Study 1 and strangers, lighters, and poisons in Studies 2 and 3.

Instructions and modelingThe narrator verbally described dangers and appropriate safety responses and included video models of children responding to dangers.XX
OrderParticipants were taught to put the three safety steps (“no,” go, and tell) depicted in short video clips in order.XX
Go!The game stressed the importance of getting away from dangers. Danger appeared on the screen and participants were required to click “go!” within 5 s.XX
Danger sortingThe participant was taught to discriminate safe and unsafe objects. The participant was required to move pictures of safe and unsafe objects into boxes or tell a stick figure to get away or stay near and object using buttons on the screen. X
Act Out gameDanger appeared on the screen, and participants were asked to get away from the computer and tell the cardboard cutout about the danger.XX
Act+Run! gameIdentical to the Act Out game, except the on-screen narrator appeared shortly after the danger and encouraged the participant to get away quickly.X 

The first module introduced participants to the program and provided instruction on how to use the mouse to interact with future modules. The remaining modules consisted of components similar to other BST packages (e.g., Beck & Miltenberger, 2009).

Instructions and modeling

In this module, an on-screen narrator first provided instructions about strangers, including who is a stranger and how to respond. Next, the participants watched videos of similarly aged children responding correctly to a stranger. In one case, a child in the video went with a stranger and a parent approached the child, provided a mild reprimand, and reminded the child of the correct steps. The incorrect model was included to show children the contingencies for behaving appropriately and inappropriately with respect to strangers. Students practiced skills related to the information from the instructions and modeling portion of the program in the other subgames of the CBST program.

Order game

Before this game began, the participant was given the opportunity to change the color of the mouse cursor on the screen. The ability to change the cursor color was added to provide the participant with some choices during the game and was not relevant to the skill being taught. During the game, there were three video clips on the left side of the screen. The participant was required to move the videos into the correct order (say “no,” get away, and tell an adult) to move on. If the participant attempted to move the video to the incorrect position, the video moved back to its original position. The video snapped into place on the right side of the screen, and a green checkmark appeared over the video if the participant moved the video to the correct location. A large blue arrow pointed to the correct location for the video if the participant moved a video to the incorrect location twice during a trial. The participant was required to move all the videos into the correct order, with fewer than three errors in six trials, to move to the next module. Correct trials did not have to be consecutive.

Go! game

Participants from pilot studies had difficulty with getting away from strangers during in situ assessments. The Go! game was added to emphasize getting away from a stranger as a key step for self-protection. Before the game began, the participant was allowed to choose the color of the stick figure for the game. During the game, there was a large button with the word “go!” and a picture of a stick figure running. Above the button was the stick figure chosen by the participant. A video of a stranger asking the participant to leave appeared to the left of the stick figure. A correct response required the participant to click the “go!” button within 3 s of the stranger's appearance on the screen. A correct response caused the stick figure to run away from the video of the stranger. The on-screen narrator provided praise or said, “Too slow; get away faster,” if latency to clicking the “go!” button exceeded 3 s. The narrator said, “That was too fast. Wait for the stranger to ask you to go,” if the participant clicked the “go!” button before the stranger appeared. The participant was required to click the “go!” button within 3 s on six trials to move to the next module.

Act Out game

This game was designed to be similar to rehearsal during BST. Participants were given the opportunity to respond physically to a simulated stranger and to receive individualized feedback on their performance. Before beginning the computer program, participants were given 10 round laminated paper tokens with a picture of a stick figure running on one side and a stick figure saying “stranger!” on the back. During the game, the participants were instructed to give these tokens to the cardboard cutout of the on-screen narrator. Pilot testing of this game indicated that children were confused by the instruction to get away from the stranger on the computer screen until we added these tokens. Tokens were not contingent on any responses. At the beginning of the game, the participants watched a video of the on-screen narrator play the game four times. The first time, after saying “no,” the narrator walked to the cardboard cutout and put one of the tokens in an attached pouch before saying “I saw a stranger.” In the second video, the narrator saw the stranger but did not act out the safety skills. The narrator watched the error message and appeared to be frustrated. The third video was similar to the first but with fewer instructions about each step. During the final portion of the instructions for the Act Out game, the narrator told the participant to complete the safety skills even if there were no tokens left. The narrator modeled the correct response without a token.

After the instructions, a stranger, who the participant had not seen before, appeared in the center of the screen. Below the stranger video was an “OK” button that remained on screen for the duration of the trial. Occasionally, before a stranger appeared, the narrator appeared and said something an adult might say before leaving a child alone, for example, “One second, I need to make a phone call. I'll be right back.” After the narrator disappeared, the stranger appeared and asked the participant to leave. The computer program counted a response as correct if at least 4 s elapsed between the stranger's appearance and the participant clicking the “OK” button to end the trial. If the participant clicked the “OK” button before 4 s elapsed, the narrator appeared and instructed the participant to act out the correct response. Clicking the “OK” button 4 s after the stranger's appearance resulted in praise from the narrator. To move to the next module, the participant was required to click the “OK” button 4 s after the stranger appeared on six trials.

Act+Run! game

The Act+Run! game was identical to the Act Out game, except that the on-screen narrator yelled “Run! Get away from the stranger!” or similar phrases when a stranger appeared. The additional prompt was added to signal the importance of getting away quickly. Six correct responses were required to complete this game. The participant did not have any remaining tokens for the last two trials if he or she had responded correctly during the Act Out game. This part of the program gave participants an opportunity to emit the correct safety response without giving a token to the cardboard cutout if he or she did not have any tokens left.

In situ training (IST)

IST was implemented during in situ assessments that occurred after the last post-CBST session if a participant did not get away from the stranger and tell an adult. The participant replayed the computer game between the last post-CBST session and the first IST session. During IST, the experimenter approached the participant and described the mistake made: “When a stranger asks you to leave, make sure you get away. Let's practice.” After the instructions, the experimenter left the participant, and the stranger again approached the participant. The experimenter used least-to-most prompting (gestures and short vocal instructions) to ensure that the participant made the correct response. Training was repeated each session until the participant correctly got away from the stranger and reported the stranger to the experimenter without any prompts. Typically, one or two trials of IST occurred per session.

Additional consequences

IST was ineffective for one participant (Shannon). Additional consequences were added to her IST. If she correctly protected herself from the stranger on the first trial of the session, a variety of games and stickers were available when she and the experimenter reached their destination. However, if she did not correctly protect herself, the experimenter implemented IST until she responded appropriately and without prompts. When the participant and experimenter reached their destination, the experimenter chose a toy and Shannon did not receive a sticker.

Maintenance

Available participants completed in situ assessments 8 weeks after their previous in situ assessment. If participants responded correctly, the experimenter delivered descriptive praise. Incorrect responses were followed by IST.

Treatment extension

During the treatment extension, the kindergarten class went to a playground. While there, one of the two regular classroom teachers separated the participant from the group for a fitness test (e.g., measuring how far the participant could jump). When the participant and teacher were about 5 m away from other children and adults, the teacher commented on how she forgot an item (e.g., the measuring tape) and left the participant alone to retrieve the forgotten item. At this point, the assessment continued as described previously. One of the confederate strangers casually approached the participant shortly after the teacher left and asked the participant to leave the area (e.g., “Let's play over here.”). The teacher provided enthusiastic descriptive praise if the participant got away and told a safe adult about the stranger. If the participant did not respond correctly, IST was implemented.

Results and Discussion

CBST was completed in an average of 19 min across participants (range, 14 to 23 min). Duration varied across participants because a set number of correct responses per module was required to move through the program.

Figure 2 displays the data from the in situ assessments conducted during Study 1. None of the 11 participants protected themselves in baseline. Although baselines for some participants were very short, previous research suggests that children do not learn to protect themselves in the absence of training (e.g., Beck & Miltenberger, 2009; Dancho et al., 2008; Miltenberger et al., 2004, 2005; Poche et al., 1981). Nine of the 11 participants acquired the safety skills after CBST and IST. One participant responded correctly after CBST alone. One other participant required additional consequences to respond correctly during the in situ assessment. By the end of the study, all participants moved quickly away from the stranger and reported to an adult. Safety skills were maintained for four of five participants who were available for an 8-week follow-up observation. Furthermore, when a stranger approached a participant at a playground instead of in the hallway at school, generalization of the safety skill to the new location was observed for 8 of the 11 participants. Target responses were subsequently acquired in the new location for the two participants for whom effects of IST were evaluated in the second location. Shannon was unavailable for the second day of assessments at the playground; therefore, the effects of IST in the new location could not be determined.

image

Figure 2. Data for the 11 participants in Study 1 (abduction prevention). Circled data points indicate sessions with in situ training. Bsl = baseline; CBST = computerized behavioral skills training; IST = in situ training.

Download figure to PowerPoint

STUDY 2: POISON AND LIGHTER SAFETY

  1. Top of page
  2. Abstract
  3. STUDY 1: ABDUCTION PREVENTION
  4. STUDY 2: POISON AND LIGHTER SAFETY
  5. STUDY 3: COMPREHENSIVE SAFETY SKILLS PROGRAM
  6. GENERAL DISCUSSION
  7. REFERENCES

Study 1 suggested that CBST and IST result in the acquisition of safety skills, maintenance over time, and generalization to new settings, but it included only a single danger. The purpose of Study 2 was to determine whether safety skills would generalize across dangers when IST was provided for only a single danger. Two additional dangers were evaluated in Study 2: poisons and lighters. Study 2 also explored a possible limitation of the CBST program, which is erroneous feedback by the computer during the Act Out component. Because the computer program could not detect the participants' actual behavior of running away and telling an adult, feedback was based on the time between the stranger's appearance on the screen and the click of a button. In Study 2, we compared the correspondence between human observers' data collection and computer-recorded responses during the Act Out game, because these data were not collected in Study 1.

Participants and Setting

The same 12 children who began participation in Study 1 participated in Study 2. All participants had learned to say “no,” get away, and tell an adult when approached by a stranger during Study 1. Samantha, who did not want to participate in Study 1, learned the safety skills regarding strangers outside the experimental analysis. She opted to participate in Study 2.

All in situ assessments occurred in a small room. The experimenter observed participants via a wireless video monitor placed in the room.

Design

The experimental design was similar to that used in Study 1 except that in situ assessments were conducted for poisons and lighters rather than strangers. Both lighter and poison in situ assessments were conducted during baseline so that generalization could be assessed after the computer game or after IST with one of the dangers. CBST included instruction on all three dangers: strangers, poisons, and lighters. IST was implemented only for lighters if the participant made errors during the in situ assessment after playing the computer game. CBST and IST were implemented according to a concurrent multiple baseline design. An extended phase in which no IST was conducted after the computer game was not included in Study 2. Participants played the computer game only once before IST for lighters began.

Data Collection

Data collectors observed the participant via a wireless camera and monitor. They recorded whether a participant touched the danger, left the room, or reported the danger to the experimenter (see Table 1). Interact with the danger was scored any time the participant touched the danger (lighter or poison) with any body part. Stay near included any time the participant did not touch the danger, get away, or tell an adult. Get away was scored if any part of the participant's body crossed the threshold of the room's doorway within 10 s after entering the room with a danger. The same criteria for Study 1 were used for tell an adult.

Human observers scored the participant's responses during the Act Out game. For each trial, a response was scored correct if the participant moved within arm's reach of the cardboard cutout and reported the danger within 5 s of the danger appearing on the computer screen.

Interobserver agreement was calculated for an average of 33% of in situ assessments (range across participants and phases, 20% to 43%) and 50% of Act Out game sessions. Agreement was calculated the same way as in Study 1. Agreement for the in situ assessments was 100% for all sessions. Agreement for the Act Out game sessions was also calculated using trial-by-trial agreement (i.e., an agreement was scored when both observers recorded a correct or an incorrect response on the same trial). Agreement between human observers for the Act Out game averaged 95% (range, 85% to 100%).

Procedure
In situ assessment

The in situ assessment began when the participant was alone in the session room. As in the stranger assessment, the experimenter left the participant alone while he left to get something or take a phone call. Then, the experimenter left the session room and observed the participant via a video monitor. The participant encountered the danger In the room. During the lighter assessment, a disabled lighter (i.e., with gas and flint removed) was placed near some toys on the table. During the poison assessment, before the participant entered the room, the experimenter mentioned that there was a cookie the participant could eat. A cookie was on the table, and fake poisons were nearby. For this assessment, the poisons were pill-shaped candy placed inside a variety of medicine containers, a procedure modeled after that of Dancho et al. (2008). If the participant left the room within 10 s and reported the danger to the experimenter, the experimenter provided enthusiastic descriptive praise. The duration to exit the room was extended to 10 s, compared to 5 s in the stranger assessment, because it was difficult to determine when the participant noticed the danger. If the participant interacted with the danger or did not leave the room within 1 min, the experimenter entered the room and nonchalantly removed the danger. The experimenter entered the room when a participant interacted with the danger to limit the amount of time he or she could practice dangerous behavior (e.g., opening a pill case or playing with a lighter). During the baseline in situ assessment, corrective feedback and rehearsal did not occur for either danger if an error was made (e.g., trying to open the medicine container).

Computerized behavioral skills training (CBST)

The computer game was similar to that described in Study 1 except that participants learned about strangers, poisons, and lighters. Table 2 summarizes the games and indicates differences from the game in Study 1. Participants completed the Act Out component for three correct trials for each danger, and at the end of the program, the Act Out game was repeated for six correct trials with a different danger for each trial. As in Study 1, a total number of correct trials were required to progress through the modules instead of a set number of trials. Each danger was taught separately until the last Act Out game in which a different danger was presented in each trial.

The Act + Run! game was removed from this version of the computer program. Two new click-based games were added to teach the new dangers and to promote discrimination between safe and unsafe situations. In one game, a variety of pictures were displayed on the screen. Some of the pictures contained dangerous items (i.e., poisons or lighters). The participant was required to sort the pictures into a “safe” and an “unsafe” box. In a second game, a single picture was displayed on the screen. If the picture contained a danger, the participant clicked a red button to make an animated stick figure get away from the danger. If the picture displayed a safe item (e.g., toys only), the participant clicked a green button and the animated stick figure approached the picture.

An error screen was added to the Act Out game that prompted the participant to recruit help from a teacher if he or she made three consecutive errors. During the game, a screen appeared with instructions to find a teacher and ask for help. The experimenter was required to hold down the control key and type T on the keyboard to continue the program. On the next trial, the experimenter watched the participant during the trial and provided vocal and model prompts until he or she correctly emitted the safety response.

In situ training (IST)

IST was implemented only for the lighter danger. IST was similar to that in Study 1 except that after receiving feedback, the experimenter and participant left the room and the experimenter then sent him or her back into the room to initiate another trial. Trials were repeated until the participant got away from the danger and reported the danger to the experimenter without any prompts. Usually, one or two trials were conducted per session.

Results and Discussion

After CBST for all dangers and IST for a single danger (lighters), participants protected themselves from both dangers (lighters and poisons) included in the computer game. Figure 3 displays the results of Study 2. Amy and Dana correctly protected themselves from poisons and lighters after CBST without experiencing IST for either. IST was not used for these participants because they responded correctly to the dangers during their first in situ assessment after CBST. There were two other patterns of acquisition for other participants. Four participants' performance generalized across dangers after they received only one session of IST for a single danger. In other words, the generalized effects of IST were apparent during the poison assessment before the participant demonstrated appropriate self-protection for lighters. Five participants' self-protection skills did not generalize to poisons until they protected themselves independently against lighters. Most participants required only one or two sessions of IST on one danger before protecting themselves independently against both dangers. Two participants, Neal and Mike, required three sessions of IST for lighters before correct performance occurred consistently in the presence of lighters and poisons. Thus, safety skills generalized to poisons after IST for lighters for all participants. IST for poisons was not required.

image

Figure 3. Data from poison and lighter in situ assessments for 12 participants in Study 2. Circled data points indicate sessions with in situ training. Bsl = baseline; CBST = computerized behavioral skills training; IST = in situ training.

Download figure to PowerPoint

CBST was brief and required only minimal experimenter interaction. The mean duration for CBST across participants was 20 min (range, 17 to 28 min). In Study 2, only one participant required additional assistance after making three consecutive errors in the first Act Out module with strangers. The assistance occurred in one trial that lasted 28 s. The experimenter vocally prompted the participant to get out of her chair, walk over to the cardboard cutout, and tell the cardboard cutout about a stranger she saw on the screen. After the brief prompt, the participant was praised by the on-screen narrator and continued to emit the correct safety skills in the presence of the on-screen dangers.

Even without motion-detection hardware and cameras, the computer game accurately provided feedback to participants who did and did not act out the safety skills. During CBST, a human observer recorded whether a participant correctly emitted the safety skills (get away and tell an adult). The computer program recorded a correct response if the delay between the video of the danger and the participant clicking the “OK” button to continue to the next trial was greater than 4 s. Trial-by-trial agreement between the computer game and the primary human observer averaged 89% (range, 71% to 100%). Scores between the two types of observers were well correlated for 11 of the 12 participants. In other words, the computer program provided accurate feedback to participants about their Act Out performance. The computer program recorded a much higher percentage correct than the human observer did for Darla. Darla waited 4 s before clicking “OK” without acting out the appropriate skills. Overall, feedback was fairly accurate, and Darla and all the other participants learned to protect themselves against the dangers quickly even after they experienced erroneous feedback during the computer game.

Performance on the computer game was not correlated with success during the in situ assessment. More specifically, the percentage correct during the role-play component of the computer program (based on human observations) and the duration spent playing the computer game were not significantly correlated with the number of IST sessions required before the participant emitted the correct safety skills, r(10) = −.21, p = .51 and r(10) = .26, p = .41, respectively. Like knowledge assessments, performance during role-plays, or in this case, computer simulations, is not a good indicator of performance in other settings and should not be used to make decisions regarding mastery of important safety skills.

These data suggest that the combination of CBST and IST for a single danger may be an effective technology for developing a comprehensive safety skills program in which children can be taught to protect themselves from a variety of dangers. However, the participants in Study 2 had prior experience with the computer program and learning to protect themselves from strangers before Study 2 began. It is therefore unclear how many dangers need to be taught using CBST and IST before performance begins to generalize to other dangers experienced only in the computer game.

STUDY 3: COMPREHENSIVE SAFETY SKILLS PROGRAM

  1. Top of page
  2. Abstract
  3. STUDY 1: ABDUCTION PREVENTION
  4. STUDY 2: POISON AND LIGHTER SAFETY
  5. STUDY 3: COMPREHENSIVE SAFETY SKILLS PROGRAM
  6. GENERAL DISCUSSION
  7. REFERENCES

The purpose of Study 3 was to determine the extent to which performance would generalize across dangers for children who had not learned to protect themselves from strangers.

Participants

Participants were four 5- and 6-year-old children from a kindergarten classroom at a school in Massachusetts. Participants in Study 3 did not have any prior experience with the in situ assessments or the computer game for any danger.

Data Collection and Interobserver Agreement

Data on in situ assessment performance were collected in the same manner as in Studies 1 and 2. Interobserver agreement was obtained as previously described during 27% of in situ assessment sessions across participants and conditions (range, 25% to 28%). Interobserver agreement was 100% for all in situ assessment types (strangers, lighters, and poisons).

Procedure

In situ assessments for strangers, lighters, and poisons were identical to those described in Studies 1 and 2. The computer game was the same game from Study 2, which taught the participant about strangers, lighters, and poisons (see Table 2). As in Study 2, assistance was provided by the experimenter if the participant made three consecutive errors during one of the Act Out games.

CBST was implemented according to a concurrent multiple baseline design. After CBST, IST was implemented for lighters if the participant did not protect him- or herself during in situ assessments. If the participant did not respond correctly in the other in situ assessments (poisons and strangers) after responding correctly to lighters in at least one session, the participant replayed the computer game before IST was initiated for an additional danger.

Results and Discussion

During baseline, none of the participants protected themselves sufficiently (Figure 4). All participants interacted with at least one of the dangers. Andrea reported the stranger after the experimenter returned, but did not get away from the stranger and instead spoke with him until he walked away. Although this was not as dangerous as immediately going with the stranger, staying near allows the stranger more time to convince the child to comply with his request. After CBST, Tammie protected herself from poisons and lighters without any additional training. The remaining three participants protected themselves from both poisons and lighters after they received IST for lighters. Safety skills did not generalize from poisons and lighters to strangers for any of the participants; however, after they experienced IST for one (Tammie, Andrea, and Darcy) or two (Danny) sessions, all participants got away from the stranger and reported him or her to a safe adult.

image

Figure 4. Data from poison, lighter, and stranger in situ assessments for four participants in Study 3. Circled data points indicate sessions with in situ training. CBST = computerized behavioral skills training; IST = in situ training.

Download figure to PowerPoint

Study 3 replicates the findings of Studies 1 and 2: CBST was an efficacious replacement for teacher-conducted BST for safety skills. IST was, however, still necessary for at least some of the dangers for most of the participants.

CBST remained brief, but required slightly more experimenter intervention than in Study 2. Three of the four participants required brief intervention during the computer program after they made three consecutive errors during the first Act Out game with strangers. The average duration of assistance was 35 s (range, 33 to 38 s). The experimenter provided brief vocal prompts for the participants to walk over to the cardboard cutout and tell it about the stranger on the computer screen. Although some intervention may be required for some participants, the amount of assistance is much less in than traditional forms of BST, which requires the teacher to conduct the entire rehearsal and feedback component for each child. During CBST, our data suggest that assistance may be needed for less than 1 min and only for some children.

The overall duration of the computer program was similar to that observed in Study 2. The average duration across participants was 24 min (range, 19 to 27 min) on the first attempt before IST with lighters and 22 min (range, 20 to 23 min) on the second attempt before IST was implemented in the stranger context. Unlike teacher-directed BST, CBST does not increase in duration as the number of participants increases because the role-play component can run simultaneously for multiple participants while individualized feedback is maintained. In addition, multiple dangers can be presented digitally rather than requiring additional physical materials to teach each danger.

Safety skills generalized from lighters to poisons but did not extend to strangers in Study 3. All four participants responded safely during in situ assessments for lighters and poisons after playing the computer game (which included all three dangers) and experiencing IST only for lighters.

Tammie acquired safety skills with respect to poisons and lighters with only computer-based simulations (i.e., without IST); by contrast, IST was required for her to protect herself from strangers. The different features of strangers and inanimate dangers and differences in the children's histories with respect to these stimuli may have led to limited generalization between the two types of dangers.

GENERAL DISCUSSION

  1. Top of page
  2. Abstract
  3. STUDY 1: ABDUCTION PREVENTION
  4. STUDY 2: POISON AND LIGHTER SAFETY
  5. STUDY 3: COMPREHENSIVE SAFETY SKILLS PROGRAM
  6. GENERAL DISCUSSION
  7. REFERENCES

The current studies suggest that computerized safety skills training combined with IST may be a viable alternative to more intensive training (e.g., BST and IST conducted by a teacher) and that safety skills may generalize between specific subsets of dangers. All of the participants acquired poison, lighter, and stranger safety skills after they played a safety skills computer game and experienced IST, and these safety skills generalized across specific subsets of dangers.

Understanding the conditions under which self-protection generalizes to other dangers and the limits of generalization will allow the development of an efficient, efficacious, and comprehensive safety skills program. It may be the case that IST is required for only one of multiple dangers that share functional similarities, whereas additional IST is required for functionally distinct dangers. The similar stimulus conditions and reinforcement histories associated with poisons and lighters may have promoted generalization. In both the poison and the lighter assessments, the participant entered the room alone and discovered the danger on a table near toys or next to something edible. In addition, participants likely had a limited reinforcement history for interacting with poisons and lighters. Parents likely do not encourage their children to touch lighters or open pill cases. Some parents may have punished prior instances of playing with lighters. The clear stimulus conditions and limited learning histories with respect to poisons and lighters may have promoted generalization.

However, generalized responding did not extend to the stranger in situ assessment in Study 3. IST was required for strangers before safety skills in that situation emerged. The lack of generalization from lighters to strangers may not be unexpected given the different stimuli and reinforcement histories associated with each danger. The poison and lighter in situ assessments contained stimuli that children do not experience on a regular basis; however, children frequently interact with a variety of people throughout the day. The discriminative stimuli associated with strangers are more difficult to detect than poisons or lighters. Strangers can have a variety of appearances, and children must detect whether a person who asks them to leave is a known adult; responding must be sensitive to conditional discriminations about receiving or not receiving prior permission to leave with the person from a teacher or parent.

Beyond the difficult conditional discrimination, the reinforcement histories associated with complying with an adult's request may inhibit the development of safety skills. Children are often required to comply with requests from people with whom they are unfamiliar at school (e.g., an adult, who is not the regular classroom teacher, is in charge of transitions from the classroom to the bus). Compliance to these adults' instructions is expected. In addition, children are often praised for being outgoing and social with important, but unfamiliar, visitors. Both of these contingencies probably create a reinforcement history that runs counter to the training self-protection from strangers.

Future research should be directed towards determination of the natural histories that give rise to functional classes of dangers (i.e., those dangers for which training self-protection on one leads to effective performance with respect to the others). Future research should also be directed towards determination of the types of teaching methods that will lead to the formation of larger functional classes of dangers to maximize generalization across dangers and minimize the extent to which IST is required to build an effective repertoire of self-protection. For instance, it is possible that stronger stimulus-equivalence-based programming (Sidman & Tailby, 1982) might yield functional classes of dangers. Teaching a child to protect him- or herself from one of these functionally equivalent dangers via IST may yield improved generalized performance with respect to the other dangers.

Perhaps, instead of considering the different functional classes, researchers might identify similar stimulus conditions across all dangerous situations. Palaez, Virues-Ortega, and Gewirtz (2012) taught infants to attend to their mother's facial expression before they interacted with ambiguous objects. Children might be taught to discern a wide range of ambiguous stimuli (e.g., food or drinks that are novel or unlabeled or a person who the child has not met before) and check with an adult before interacting with ambiguous people or objects.

Compared to previous research on BST, the current series of studies suggest that CBST outcomes may be similar to those for traditional BST, in that BST alone may not be effective for all participants but is effective when used in combination with IST (Dancho et al., 2008; Himle, Miltenberger, Gatheridge, & Flessner; 2004). Future researchers might directly compare the relative efficacy and efficiency of, and child and caregiver preference for, BST and CBST.

In the current analysis, the necessity of CBST prior to IST cannot be determined, because IST was never conducted without prior CBST. However, previous data suggest that IST may not be as effective without BST. For example, Beck and Miltenberger (2009) implemented IST for stranger safety after participants watched a commercially available video program. Two of the six participants required a session of BST before they consistently performed the correct safety skills, even though IST had been in place. Although it was not the primary conclusion of their study, the data suggest that the combination of BST and IST may be more effective than either component separately. In addition, it seems unlikely that IST would be implemented without first teaching children about the dangers they are likely to encounter.

It is important to note that four participants acquired safety skills after they experienced only CBST. Although current research does not suggest what the key repertoires may be, it is possible that repertoires that promote generalization between a computer and another situation are acquired due to the widespread use of computer-based learning in schools and homes. Researchers should attempt to identify variables that influence this sort of generalization. Evaluating the current CBST program with older children, who likely have acquired relevant repertoires for generalization, is also an important next step in determining the efficacy of computer programs to teach safety skills. Safety skills programs for older children are especially important because those children are more likely to be left without direct parent or teacher supervision.

REFERENCES

  1. Top of page
  2. Abstract
  3. STUDY 1: ABDUCTION PREVENTION
  4. STUDY 2: POISON AND LIGHTER SAFETY
  5. STUDY 3: COMPREHENSIVE SAFETY SKILLS PROGRAM
  6. GENERAL DISCUSSION
  7. REFERENCES
  • Action Editor, Anna Petursdottir