Printer Friendly

Using additional analyses to clarify the functions of problem behavior: an analysis of two cases.

Abstract

Functional analyses (FA) have proven useful for identifying contingencies that influence problem behavior. Research has shown that some problem behavior may only occur in specific contexts or be influenced by multiple or idiosyncratic variables. When these contexts or sources of influence are not assessed in an FA, further assessment may be necessary to identify the variables that influence behavior. In the current study, initial FA results suggested a specific source(s) of reinforcement for the problem behavior of two preschool children. Function-based treatments were implemented to increase appropriate behavior and decrease problem behavior; however, treatment did not result in clinically significant reductions of problem behavior in all cases. Additional within-session analyses and experimental analyses were used to clarify the functions of problem behavior, and modified function-based treatments were used to decrease problem behavior and increase appropriate behavior. These effective treatments were taught to the classroom teachers to implement throughout the day.

KEYWORDS: Functional Analysis, Multiple Control, Problem Behavior, Within-Session Analysis

Functional analyses (Iwata, Dorsey, Slifer, Bauman, & Richman, 1982/1994) are often effective in determining the function of problem behavior and for prescribing an effective function-based intervention (Arndorfer & Miltenberger, 1993; Hanley, Iwata, & McCord, 2003; Iwata et al., 1994; Kurtz et al., 2003; Neef & Iwata, 1994; Wacker et al., 1994). Hanley et al. (2003) conducted a literature review of 277 studies that included functional analyses of problem behavior and found that 95.9% of functional analyses produced differentiated outcomes (i.e., showed a clear function for problem behavior). Although these results suggest that a majority of published functional analyses are successful in identifying the function of problem behavior, Hanley et al. suggested that the contingencies for publication might promote a higher level of positive findings in the literature. Thus, these data may not be representative of the actual success rate of functional analyses.

Although functional analyses are often successful in deterrnin-ing the variables influencing problem behavior, in certain situations functional analyses may not produce clear results (Carter, 2009; Kuhn, DeLeon, Fisher, & Wilke, 1999; Vollmer, Marcus, Ringdahl, & Roane, 1995). For instance, functional analyses may yield unclear results when behavior is undifferentiated across conditions (Conners et al., 2000; Smith, Iwata, Vollmer, & Zarcone, 1993) or behavior occurs at low or zero rates during the functional analysis (Call, Wacker, Ring-dahl, & Boelter, 2005; Carr, Yarbrough, & Langdon, 1997; Kahng, Abt, & Schonbachler, 2001). In addition, clear functional analysis results in certain situations may produce a false-positive or false-negative result with respect to the function(s) of problem behavior, possibly resulting in the implementation of treatments that are less than optimally effective. In the case of unclear functional analysis results or less than optimally effective function-based intervention, additional assessments and experimental analyses may be warranted to clarify the function of problem behavior.

Several studies have included an investigation of the possibility of false positives or false negatives with respect to functional analysis results (e.g., Kuhn et al., 1999; Rooker, Iwata, Harper, Fahmie, & Camp, 2011; Shirley, Iwata, Smith, & Kahng, 1999; Vollmer, Iwata, Smith, & Rodgers, 1992). A false-positive result would be indicated when functional analysis results suggest a particular function that is not an actual function of the problem behavior (i.e., the problem behavior is not maintained by an identified variable in the natural environment). A false-negative result would be indicated when functional analysis results do not suggest a particular function that is an actual function of the problem behavior. Both false-positive and false-negative results would suggest an incorrect (or at least incomplete) identification of the function of the problem behavior, thereby resulting in less than optimal treatment effects.

A few studies have proposed methodologies to identify false positives and false negatives in functional analyses. Kuhn et al. (1999) conducted a functional analysis of an adult's self-injurious behavior (SIB) and found that his problem behavior was maintained by automatic reinforcement and negative reinforcement. The experimenters used a multielement design to analyze treatments involving sensory extinction (helmet use) and escape extinction. Only sensory extinction was effective in reducing problem behavior to clinically significant levels. The authors suggested that the functional analysis yielded false-positive results (escape) and suggested that using treatment analyses to clarify ambiguous functional analysis results may be more efficient than conducting extended functional analyses.

Vollmer et al. (1992) conducted a standard functional analysis with one participant and found low levels of SIB across all conditions. In an attempt to clarify the results of the functional analysis, data from interviews and informal observations were used to create additional experimental assessments. Interview results suggested that problem behavior occurred during self-help tasks, and informal data collection in the natural environment suggested that one common consequence for problem behavior was physical prompting to complete the task (physical attention). Using this information, two modified demand conditions were used to assess whether problem behavior was maintained by escape from self-care tasks or physical attention in the form of a simulated struggle. Results showed that higher levels of problem behavior occurred when physical attention was provided contingent upon problem behavior. In summary, the initial functional analysis did not suggest that problem behavior was maintained by attention; however, additional analyses suggested that problem behavior was maintained by attention but only in the context of demands. Therefore, the initial functional analysis may have produced a false negative in that the results suggested that problem behavior was not maintained by attention. The authors suggest, however, that had they conducted indirect or descriptive analyses prior to the initial functional analysis, they may have included more pertinent variables in the initial functional analysis.

Roane, Lerman, Kelley, and Van Camp (1999) conducted several functional analyses of SIB. The authors attempted to then use within-session analyses, specifically calculating the rate of SIB during periods in which the relevant establishing operation (EO) was present or absent in each condition, to either verify the results of clear functional analyses or clarify the results of undifferentiated functional analyses. For example, during the attention condition, the rate of problem behavior was calculated during the period of time when the therapist was ignoring the participant (EO present) and was compared to the rate of problem behavior during the period when attention was being delivered (E0 absent). If a higher rate of problem behavior occurred during the period when the therapist was ignoring the participant, the data suggested an attention function for problem behavior for that participant. However, if a higher rate of problem behavior occurred during the period when attention was being delivered, it might suggest a false-positive result. These within-session analyses supported the results of their functional analyses and also clarified undifferentiated and potentially false-positive functional analysis results. These data suggest that within-session analyses of the EO-present and E0-absent periods of a functional analysis may be useful in clarifying functional analysis results.

The results of these studies suggest that various procedures may be used to determine whether false-positive or false-negative results are shown in a functional analysis. First, results from function-based treatments can be compared with results from an initial functional analysis (Kuhn et al., 1999). Second, direct observation of the problem behavior in the natural environment (e.g., descriptive analyses), as well as interviews may be used to suggest relevant specific environmental events that occur in the natural environment (Shirley et al., 1999; Vollmer et al., 1992). Third, within-session analyses may be used to potentially identify other variables for consideration in the analysis (Fahmie & Hanley, 2008; Roane et al., 1999). Finally, additional experimental manipulations may be used to further clarify the function of problem behavior and determine if outcomes indicate false-positive and false-negative functions (Vollmer et al., 1992). Overall, the results of these studies may suggest a systematic way to determine additional functions of problem behavior that may not be captured through traditional functional analyses (i.e., an attention test condition, an escape test condition, an alone condition, and an omnibus control condition).

In the current study, two preschool children were referred for assessment and treatment of problem behavior. Upon conducting initial functional analyses, clear functions of problem behavior were suggested for both participants. However, treatments based on these analyses did not produce clinically significant changes in problem behavior in all cases. Following suggestions by Roane et al. (1999) and Vollmer et al. (1995), we determined a systematic methodology for determining potential false-positive and false-negative functions. We next conducted within-session analyses, specifically looking at problem behavior during times when the EO was present or absent, to derive hypotheses as to possible variables that may have influenced problem behavior that could be experimentally evaluated. Data from these within-session analyses suggested additional experimental analyses to clarify the functions of problem behavior. Treatments based on the initial and additional functional analyses were implemented to verify the results. Finally, these treatments were successfully transferred to the preschool classroom teachers.

General Method

Participants and Setting

Participants were two children enrolled at a university-based preschool who were referred for assessment and treatment of problem behavior (i.e., aggression and property destruction). Andrew (Case 1) was a typically developing, 4-year-old boy. Samantha (Case 2) was a 4-year-old girl diagnosed with a learning disability. Both participants were able to follow multi-step instructions and communicate using vocal speech.

Sessions were conducted 3-5 days per week, 2-5 times per day in either a session room (Andrew's Functional Analysis) or a preschool classroom (Andrew's treatment and Samantha's functional analysis and treatment). During sessions in the classroom, teachers and peers were present. Trained therapists conducted all functional analyses, initial treatment, additional analyses, and some final treatment sessions. Classroom teachers were taught to implement the final treatment in the classroom. Specific items (such as small blocks and toys) were available during all sessions to provide equal opportunity to engage in property destruction.

Response Measurement and Interobserver Agreement

The dependent variables were instances of problem behavior (aggression and property destruction) and appropriate behavior (mands for relevant reinforcers [e.g., attention, preferred tangibles] and compliance). Aggression was defined as any motion of the participant's limb or grasped object that forcefully contacted or came within 3 in of another person. Property destruction was defined as any forceful motion of the participant's limb that resulted in an object moving more than 1 ft from the point of contact or any forceful motion of a limb or grasped object that came into contact with another object. Mands were defined as vocal requests for relevant reinforcers specific to each condition. Compliance was defined as following a therapist's request within 5 s of a vocal or model prompt during a three-step prompting sequence.

Data were recorded by trained observers using hand-held computers. Data were collected on the frequency of aggression, property destruction, and mands; these data were converted to a rate measure. Data were also collected on the frequency of compliance and the number of demands; compliance data were converted to a percentage by dividing the frequency of compliance by the frequency of demands. To determine interobserver agreement, a second independent observer collected data during at least 30% of all sessions (with the exception of baseline sessions for Samantha's initial treatment analysis). Interobserver agreement coefficients were calculated by dividing the session time into 10-s intervals and comparing observer data on an interval-by-interval basis. If exact agreement occurred, that is, both observers scored the same number within a 10-s interval, a score of 1 was given for that interval. For any disagreements, the smaller score in each interval was divided by the larger. The interval scores were summed, divided by the total number of observation intervals, and multiplied by 100%. Mean interobserver agreement during the functional analyses and additional analyses for Andrew and Samantha were 92.9% (range, 78%-400%) and 93.3% (range, 75%-400%), respectively. Mean interobserver agreement during the treatment for Andrew and Samantha were 95.9% (range, 85%-400%) and 96.0% (range, 80%-400%), respectively. Graphical depictions of interobserver agreement show corresponding patterns of responding for both the primary and secondary observer and are available from the first author.

Case 1 (Andrew)

Functional Analysis

We first conducted functional analyses similar to those conducted by Iwata et al. (1982/1994). Five conditions (ignore, attention, play, escape, and tangible) were conducted. Discriminative stimuli (different color shirts worn by the therapist) were used to enhance discrimination between conditions. Items selected for use in different conditions (i.e, attention, play, and tangible) were determined through a paired-stimulus preference assessment (Fisher et al., 1992). Sessions were 10 min in length and were conducted in a session room. Multielement and pairwise designs were used for experimental control.

Ignore. During the ignore condition, the therapist entered the room with the participant and told him that he could not play with him. There were no programmed consequences for problem behavior.

Attention. During the attention condition, the therapist entered the room with the participant and told him that the therapist had work to do. A few moderately preferred toys were available. Contingent on problem behavior, the therapist provided a brief reprimand and physical attention (e.g., the therapist placed his hand on the participant's shoulder and said, "Stop that").

Play. During the play condition, the therapist entered the room with the participant and told him that they would be able to play together. Multiple highly preferred toys were available, and attention was provided continuously by the therapist. There were no programmed consequences for problem behavior.

Escape. During the escape condition, the therapist entered the room with the participant and told him it was time to work. The therapist presented instructions commonly used in Andrew's preschool classroom (e.g., "pick up the paper" and "wipe the table") continuously using a three-step prompting technique. Contingent on problem behavior, the therapist removed the demand materials and provided 30 s of escape. Following the escape interval, the therapist again provided instructions.

Tangible. Prior to the start of each session in the tangible condition, the therapist entered the room with the participant and told him that he could play with the toys. Highly preferred toys were available for 2 min. The therapist began the session by stating, "It's my turn to play" and restricting access to the toys. Contingent on problem behavior, 30 s of access to the toys was provided. Following the reinforcement interval, the toys were again restricted.

Results

Functional analysis results for Andrew are depicted in Figure 1. The initial multielement functional analysis results for Andrew showed low levels of problem behavior across all conditions; however, when problem behavior did occur, it was in the escape condition. Because these results were unclear, a second three-component phase was conducted using sequential pairwise designs to enhance the discrimination between the functional analysis conditions (Vollmer et al., 1995). This included three test-condition sessions consecutively for every one control condition to enhance the establishing operation present in each test condition. During this three-component phase, high levels of problem behavior occurred in the escape condition as compared to the play condition. During the evaluations of positive reinforcement in the form of attention and tangible items, problem behavior maintained at low, undifferentiated levels. The results of Andrew's functional analysis indicated that his problem behavior was maintained by escape from task demands.

Initial Treatment

The initial treatment was based on the function of problem behavior as suggested by the functional analysis. Treatment was differential reinforcement of alternative behavior (DRA) plus extinction (EXT). That is, Andrew was taught an alternative communicative response (mand) to escape the demand, and problem behavior no longer resulted in escape from demands. Prior to the start of each treatment session, we provided Andrew with rules regarding the contingencies for problem behavior and mands. For example, the therapist told Andrew, "If you ask for a break, I will give you a break; if you hit me or throw things, we will keep working." During the initial phases of treatment, Andrew was prompted every 30 s to mand; however, this prompt was quickly faded when Andrew began engaging in independent mands. Sessions were 10 min in length and conducted in the classroom. An AB design was used to evaluate the effects of the treatment.

Escape Baseline (A Condition). The baseline condition for Andrew was identical to the escape condition of his functional analysis.

DRA+ EXT (ESC) (B Condition). The DRA + EXT (ESC) condition was similar to the escape baseline condition; however, contingent on vocal mands for escape (i.e., "break, please"), the therapist removed demand materials and turned away from the participant for 30 s (removal of the putative establishing operation--EO absent). After 30 s had elapsed, the therapist

again presented demands (E0 present). In addition, problem behavior no longer resulted in escape (EXT).

Results Initial treatment data for Andrew are depicted in Figure 2. During the initial baseline phase, Andrew engaged in high levels of problem behavior and low levels of compliance. During DRA + EXT (ESC), Andrew initially engaged in lower levels of problem behavior, higher levels of compliance (nearly 100%), and low levels of mands. However, as we continued in this phase, problem behavior increased and maintained at moderate levels (less than an 80% reduction from baseline), and compliance decreased to levels similar to baseline. Mands occurred at moderate levels. Overall, these results suggested that the initial functional analysis may have suggested either a false-positive result (i.e., Andrew's problem behavior was not maintained by escape) or a false-negative result (i.e., Andrew's problem behavior was maintained by both escape and an additional function that was not identified in the functional analysis.)

Within-Session Analysis

During the initial treatment phase, we observed that problem behavior was occurring during the escape interval (i.e., the 30 s when the demand had been contingently removed--putative E0 absent). Therefore, we conducted a within-session analysis to determine levels of problem behavior during situations when the E0 for escape was present TO on--demand present) or not present (EO off--demand absent). We specifically looked at the last 5 sessions of each relevant phase because this gave the participant a period of time to contact the contingencies and allowed for the establishment of a stable response rate. Table 1 shows the results of this analysis. During the last 5 sessions of Andrew's functional analysis, there were higher percentages of problem behavior occurring when the putative EO was present (65.5%) compared to when the E0 was absent (35.5%). However, during the initial treatment (DRA + EXT), the majority of Andrew's problem behavior occurred when the putative E0 for escape was absent (85.7%) as compared to when the E0 for escape was present (14.3%). Although a majority of problem behavior occurred when the E0 was absent, the majority of mands for escape occurred when the E0 for escape was present (79.2%) as compared to when the E0 for escape was absent (20.8%). Anecdotally, during these sessions Andrew would immediately begin engaging in aggression towards the therapist when an escape interval began and he continued to engage in problem behavior throughout that interval. Because demands were represented at the end of the escape interval, it is possible that the attention involved in the delivery of demands was adventitiously reinforcing problem behavior on an intermittent schedule. It is also possible, however, that problem behavior continued to occur during the escape interval in order to avoid the re-presentation of demands. Therefore, we hypothesized that this problem behavior was maintained either by access to attention or avoidance of the representation of demands during the escape interval. Therefore, we conducted an additional analysis to assess whether attention (specifically the representation of demands) or the avoidance of demand representation was maintaining problem behavior.

Table 1

Average Percentage of Problem Behavior and Mands

Participant       Condition          EO On (last   E0 Off (last
                                     5 sessions)    5 sessions)

Andrew        FA (ESC) PB               65.5%          35.5%
              DRA + EXT (ESC) PB        14.3%          85.7%
              DRA + EXT (ESC) Mand       792%          20.8%
Samantha      FA (TANG) PB              49.5%          50.5%
              DRA + EXT (TANG) PB       22.9%          77.1%
              DRA + EXT (TANG) Mand     86.3%          13.7%

Table 1. This table depicts the average percentage of problem
behavior and mands during periods when the putative EO for escape
(Andrew) or tangible items (Samantha) was present (EO On) and
absent (EO Off) during the last 5 sessions of each relevant
condition. Data are separated for the initial functional analysis
conditions in which initial treatments based on those conditions
were ineffective (escape for Andrew; tangible for Samantha) and
the initial treatments for escape and tangible for Andrew and
Samantha, respectively.


Additional Functional Analysis

The additional functional analysis (Figure 3) conducted with Andrew involved comparing levels of problem behavior under DRA + EXT (ESC) conditions in which demands were represented contingent upon problem behavior (i.e., a form of contingent attention) during the escape interval (DRA + EXT + Contingent Attention [CA]) or attention was provided noncontingently during the escape interval (DRA + EXT + Non contingent Attention [NCA]).

DRA + EXT (ESC) + Contingent Attention (CA). Following initial treatment with Andrew, this condition was conducted to assess additional contingencies that might have been maintaining problem behavior. This condition was identical to the DRA + EXT (ESC) condition, except that during the escape interval, if the participant engaged in problem behavior, demands were immediately represented and continued to be presented as in DRA + EXT (ESC).

DRA+ EXT (ESC) + Noncontingent Attention (NCA). This condition was used as a control condition for the DRA + EXT (CA) condition. This condition was similar to the DRA + EXT (ESC) condition; however, during the escape interval, the therapist provided continuous vocal and physical attention.

Results

During this comparison phase, Andrew engaged in high levels of problem behavior during the DRA + EXT (CA) condition and zero levels of problem behavior in the DRA + EXT + NCA condition. In addition, Andrew continued to mand for a break during both conditions. Although not depicted graphically, across both of these conditions compliance maintained at zero levels. These results suggested that Andrew's problem behavior during the escape interval was maintained by access to therapist attention in the form of re-presentation of demands. In addition, because the topography of attention in the DRA + EXT + CA condition was in the form of the representation of demands, it is unlikely that Andrew's problem behavior was maintained by the avoidance of additional demands, as responding under this condition should have lead to extinction. Overall, these results suggested that the initial functional analysis produced a false-negative result. Although the initial functional analysis suggested an escape function for Andrew's problem behavior, it failed to identify the additional attention function.

Final Treatment

Given the information from the initial functional analyses and the additional analyses, we created an intervention to treat all of the functions of Andrew's problem behavior. A token program was included in the final treatment to increase compliance with demands, enhance discrimination, and aid in the thinning of the reinforcement schedule. In addition, we used response cost with the token program to make the contingencies for problem behavior salient and to increase the ease of using the treatment in the classroom. Tokens were exchanged for the reinforcers identified in our functional analyses (escape and attention). First, a therapist implemented the interventions in Andrew's classroom until low rates of problem behavior, moderate rates of mands, and high levels of compliance were maintained. Next, we trained the classroom teachers to implement the treatment throughout the day (starting at session 51). Teachers were first given a written protocol of the treatment components. The teachers were then trained by the therapists using in-situ prompting, modeling, and specific feedback on their performance. For example, during the first session, the therapist modeled the steps of the procedures for the teachers, prompted the teachers to demonstrate those steps, and then provided feedback as the teachers implemented the procedures throughout the day. This training started during the first teacher implementation session and was gradually faded. Treatment integrity data were not collected. Observations were 10 min in length and were conducted throughout the school day to assess the effectiveness of the treatment as implemented by the teachers. At the start of each session (not including the sessions conducted by the classroom teachers), rules about the contingencies for problem behavior and mands were stated to the participant. Sessions (and observation periods) were 10 min in length.

DRA + EXT (ESC + ATTN) + Response Cost (RC). Prior to the start of this phase, Andrew was shown the token board and provided rules regarding earning tokens (i.e., "You will get a token when you do what we ask you to do. Once you have 10 tokens, you will be able to have a break and play with me.") During these sessions, the therapist (or teacher) presented demands intermittently. Problem behavior did not result in a break from demands (EXT) and mands for a break resulted in a 30-s break from task demands (DRA). In addition, attention was provided noncontingently throughout the session (i.e., at least once every 3 min). Furthermore, compliance resulted in the delivery of a token, and an instance of problem behavior resulted in the removal of a token (RC). Once Andrew had earned 10 tokens, he was given a longer break (5 min) and an opportunity to exchange his tokens for access to quality attention (i.e., one-on-one attention with the therapist and a preferred peer and access to preferred items and activities). Once treatment effects were observed, the schedule for token delivery was thinned. That is, Andrew was required to comply with more demands to earn each token. Throughout the phase, the schedule was thinned from a fixed-ratio 1 (FR1) to a variable-ratio 5 (VR5) schedule of reinforcement for compliance. Starting at session 51, teachers began implementing the intervention throughout the day in the classroom.

Results

Following implementation of DRA + EXT (ESC + A'TTN) + RC (Figure 4), Andrew's levels of problem behavior decreased to near zero levels and his compliance increased. He also engaged in occasional mands to escape demands. These results maintained as the schedule of reinforcement for compliance was thinned from an FR1 to a VR5 schedule. Next, Andrew's classroom teachers were trained to implement the procedures throughout the day. Problem behavior maintained at low levels (below an 80% reduction from baseline), compliance maintained at moderate to high levels, and mands occurred occasionally. Overall, these data suggest that a treatment based on functions identified through both the initial functional analysis as well as the additional analyses was successful in decreasing Andrew's problem behavior, and that these effects maintained when Andrew's classroom teachers implemented the procedures.

Case 2 (Samantha)

Functional Analysis

The functional analysis for Samantha was identical to Andrew's except that Samantha's functional analysis was conducted in the classroom per caregiver request. A multielement design was used to demonstrate experimental control.

Results

Functional analysis results for Samantha are depicted in Figure 5. The functional analysis for Samantha showed higher levels of problem behavior in the attention, escape, and tangible conditions. The higher levels of problem behavior in these conditions as compared to the play condition suggested that her problem behavior was maintained by multiple sources of reinforcement including access to therapist attention, escape from task demands, and access to tangible items.

Initial Treatment

Initial treatments were based on the function of problem behavior as suggested by the functional analysis. Treatment was differential reinforcement of alternative behavior (DRA) plus extinction (EXT), which was implemented for all functions found in the functional analysis: access to attention (ATTN), escape from demands (ESC), and access to tangibles (TANG). A multiple-baseline-across-functions design was used to demonstrate experimental control.

Baseline. The baseline conditions for Samantha were identical to the attention, escape, and tangible conditions of her functional analysis.

DRA + EXT (ATTN). The DRA + EXT (ATTN) condition was similar to the attention baseline condition; however, contingent on vocal mands for attention (i.e., "play with me" or "talk to me"), the therapist provided brief (5 s) vocal and physical attention to the participant. In addition, problem behavior no longer resulted in therapist attention (EXT).

DRA+ EXT (ESC). The DRA + EXT (ESC) condition was identical to the DRA + EXT (ESC) condition for Andrew in case 1.

DRA + EXT (TANG). The DRA + EXT (TANG) condition was similar to the tangible baseline condition; however, contingent on vocal mands for tangible items (i.e., "toys, please"), the therapist provided 30-s access to the items. After 30 s had elapsed, the therapist again removed the items from the participant. In addition, problem behavior no longer resulted in access to tangible items (EXT).

Results Initial treatment data for Samantha are depicted in Figure 6. Following our functional analysis, we first implemented DRA + EXT (ATTN) for attention-maintained problem behavior. During baseline, Samantha engaged in variable rates of problem behavior. Following the implementation of treatment, rates of problem behavior decreased to low and stable levels, and mands for attention maintained at moderate levels. We next implemented DRA + EXT (ESC) for escape-maintained problem behavior. During baseline, Samantha engaged in moderate but variable levels of problem behavior and compliance. Upon implementing the treatment, problem behavior decreased to low and stable levels, and mands for escape occurred at moderate levels. Percent compliance with task demands, however, steadily decreased throughout this phase. We next implemented DRA + EXT (TANG) for problem behavior maintained by tangible reinforcement.

During baseline, Samantha engaged in moderate and stable levels of problem behavior. Upon implementing treatment, problem behavior did not decrease to clinically significant levels (i.e., below an 80% reduction from baseline); however, mands for tangible items occurred at moderate and stable levels. We observed that problem behavior was occurring during the tangible-access interval (i.e., when the putative EC) for tangible access was absent).

Within-Session Analysis

Similar to Andrew, we conducted a within-session analysis to determine levels of problem behavior during situations when the putative E0 for access to tangible items was present and absent (see Table 1). During the last 5 sessions of Samantha's functional analysis, she engaged in similar percentages of problem behavior when the putative EO was present (49.5%) and when the E0 was absent (50.5%). During DRA + EXT (TANG), Samantha engaged in higher percentages of problem behavior when the E0 was absent (77.1%), compared to when the E0 was present (22.9%). Although a majority of problem behavior occurred when the putative E0 was absent, the majority of mands for tangible items occurred when the E0 for tangible access was present (86.3%) as compared to when the EO for tangible access was absent (13.7%). Anecdotally, during sessions Samantha would immediately begin engaging in aggression towards the therapist as well as property destruction when a tangible-access interval began, and she continued to engage in problem behavior throughout that interval. Because toys were restricted at the end of each tangible-access interval, it is possible that the attention involved in the removal of the toys was adventitiously reinforcing problem behavior on an intermittent schedule. This might be especially likely because we had already determined that attention was a function of Samantha's problem behavior during the initial functional analysis. It is also possible, however, that problem behavior continued to occur during the tangible-access intervals in order to avoid the restriction of tangible items. Based on this information and data, we hypothesized that problem behavior was maintained by either attention or avoidance of the removal of tangible items; therefore, we conducted an additional analysis to assess other variables that may have been influencing problem behavior.

Additional Functional Analysis

The additional functional analysis conducted with Samantha involved comparing levels of problem behavior under DRA + EXT (TANG) conditions in which attention was delivered contingent upon problem behavior (i.e., a form of reprimands) during the tangible-access interval (DRA + EXT + Contingent Attention [CA]) or attention was provided noncontingently during the tangible-access interval (DRA + EXT + Non.contingent Attention [NCA]).

DRA + EXT (TANG) + Contingent Attention (CA). Following initial treatment with Samantha, this condition was conducted to assess additional contingencies that might be maintaining problem behavior. This condition was similar to the DRA + EXT (TANG) condition; however, during the tangible-access interval, when Samantha engaged in problem behavior, brief reprimands (e.g., "Stop doing that!") were immediately presented.

DRA + EXT (TANG) + Noncontingent Attention (NCA). This condition was used as a control condition for the DRA + EXT (TANG) + CA condition. This condition was similar to the DRA + EXT (TANG) condition; however, during the tangible-access interval, the therapist provided continuous attention.

Results

During the additional functional analysis conducted with Samantha (Figure 7) problem behavior occurred at higher levels in the DRA + EXT + CA, compared to the DRA + EXT + NCA condition. In addition, Samantha continued to mand for access to tangible items during both conditions. These results suggested that Samantha's problem behavior during the tangible-access interval was maintained, at least in part, by access to therapist attention. These results overall suggested that the functions identified in the initial functional analysis were accurate, but that the influence of concurrent functions was not taken into account when individually treating problem behavior based on each individual function.

Final Treatment

Given the information from the initial functional analyses and the additional analyses, we created an intervention to treat all of the functions of problem behavior that were determined for Samantha. Similar to Andrew, a token program was included in the final treatment. Tokens were exchanged for reinforcers found to maintain problem behavior in our functional analyses (escape, access to tangible items, and attention). First, a therapist implemented the interventions until low rates of problem behavior, moderate rates of mands, and high levels of compliance were maintained in Samantha's classroom. Similar to Andrew, we trained the classroom teachers to implement the treatment throughout the day (starting at session 48). Rule delivery was identical to Andrew's final treatment. Sessions (and observation periods) were 10 min in length. A BAB reversal design was used to demonstrate experimental control.

Baseline condition (A condition). This condition was conducted with Samantha and was similar to the attention, escape, and tangible functional analysis conditions. At the start of each session, preferred items were restricted and demands were delivered as in the escape functional analysis condition. Contingent on problem behavior, demands were ceased, the preferred items were delivered, and attention in the form of play statements and praise were delivered for 30 s. Demands were then represented, and tangible items were restricted.

DRA + EXT (ESC + ATTN + TANG) + RC (B condition). This condition was conducted with Samantha and was similar to the procedures of Andrew's final treatment, except that in addition to mands for escape, mands for attention resulted in brief (3-5 s) delivery of attention (DRA), and mands for tangible items resulted in access to those tangible items (DRA) at any point in time during the session. Once Samantha had earned 10 tokens, she was given a longer break (5 min) and an opportunity to exchange her tokens for access to attention and tangible items (i.e., one-on-one attention with the therapist and access to preferred items and activities she could pull from a treasure box). Once treatment effects were observed, the schedule for token delivery was thinned from an FR1 to an FR5 schedule. Starting at session 48, teachers began implementing the intervention throughout the day in the classroom.

Results

Following the additional analysis for Samantha, we implemented DRA + EXT (ESC + ATTN + TANG) + RC (Figure 8) in order to treat all of the functions identified in the functional analysis and additional analysis (attention, escape, and tangible). Following implementation, Samantha's level of problem behavior decreased to near zero levels, and compliance increased. She also occasionally engaged in mands for escape, attention, and access to tangible items. These results maintained as the schedule of reinforcement for compliance was increased from an FR1 to an FR5 schedule. We briefly reversed to a baseline condition and Samantha engaged in increased levels of problem behavior and lower levels of mands and compliance. Upon reversal to DRA + EXT (ESC + ATTN + TANG) + RC phase, problem behavior decreased and compliance and mands increased, even as the schedule of reinforcement for compliance again increased from an FRI to an FR5. Following sustained low levels of problem behavior and high levels of compliance, Samantha's classroom teachers were trained to implement the procedures throughout the day. Problem behavior maintained at low levels (below an 800/ reduction from baseline), and compliance and mands maintained at moderate to high levels. Overall, these data suggest that a treatment based on functions identified through both the initial functional analysis as well as the additional analyses was successful in decreasing Samantha's problem behavior, and that these effects maintained when Samantha's classroom teachers implemented the procedures.

General Discussion

In the current study, multiple experimental analyses were necessary to identify the functions of problem behavior for both participants. For both participants, initial treatments for some of the functions initially identified did not result in a clinically significant reduction in problem behavior. Within-session analyses were then used to identify patterns of responding that suggested additional variables that influenced problem behavior. This information was used to conduct additional functional analyses, which allowed us to determine contingencies influencing problem behavior under specific antecedent conditions (i.e., instructional situations for Andrew and tangible-access situations for Samantha). For both participants, treatment based on the results of both functional analyses resulted in a decrease in problem behavior, increase in compliance, and moderate levels of mands. Finally, the treatments were successfully transferred to the teachers in the classroom environment. These results, in addition to the results of other studies, suggest that, although the functional analysis conditions described by Iwata et al (1982/1994) are effective for prescribing treatment for a large number of cases, some cases are more complex and require additional analyses to determine all of the variables relevant to the treatment of problem behavior.

For Andrew, we identified a potential false-negative outcome of the functional analysis. During the initial functional analysis, problem behavior did not occur at high levels in the attention condition. However, with additional analyses, we identified attention as a reinforcer for problem behavior, but only in a demand context. Similar to Vollmer et al. (1992), additional analyses (i.e., direct observation and within-session data analyses) were helpful in identifying variables that may have been influencing problem behavior and allowed us to identify and successfully treat problem behavior whose function was not identified in the initial functional analysis. It should be noted, however, that similar to Vollmer et al., had we conducted descriptive assessments or more in-depth indirect assessments prior to conducting our initial functional analysis, we might have used this information in order to set up the conditions such that a false-negative outcome would not have occurred.

For Samantha, additional analyses were conducted in order to determine if problem behavior occurring in the tangible-treatment condition was maintained by attention. We had previously determined that Samantha's problem behavior was maintained, in part, by attention in the functional analysis. When implementing treatment for each separate function, we could have used this information to control for problem behavior maintained by attention in the tangible context, such as during the DRA + EXT (TANG) + NCA condition. Future researchers may want to evaluate the use of control procedures for alternate functions when treating problem behavior maintained by concurrent sources of reinforcement.

The current study has several limitations. First, for both participants, we conducted one functional analysis on two topographies of problem behavior (aggression and property destruction) rather than conducting separate functional analyses for each topography. These different topographies of problem behavior might have been maintained by different variables (Derby et al., 1994; Richman, Wacker, Asmus, & Casey, 1998). However, both topographies of problem behavior were observed throughout all phases of the study for both participants, so this appears less likely. Second, the tangible condition in the functional analyses was very similar to the attention condition in that attention was restricted and then delivered when tangible items were provided contingent upon problem behavior. Therefore, Samantha's problem behavior might have been solely maintained by access to attention and escape from demands. However, mands for tangible items continued to occur throughout the study, suggesting that tangible items were a reinforcer. Additionally, a majority of mands during the DRA + EXT (TANG) condition occurred during the EO present portions, suggesting that the mands were controlled in part by the E0s in place during the tangible treatment sessions. Third, for both participants, we showed that attention was a reinforcer during rein-forcement intervals for other reinforcers, but we did not rule out the possibility that problem behavior was also occurring as an avoidance response (i.e., avoidance of demands for Andrew; avoidance of tangible restriction for Samantha). To directly test this, we could have conducted an additional condition (similar to the DRA + EXT condition) in which the presentation of demands (Andrew) or the removal of tangible items (Samantha) was delayed or prevented contingent on problem behavior during the EO absent period. Although a direct test for both participants was not conducted, the topography used for the DRA + EXT + CA condition for Andrew was the re-presentation of demands, which would be identical to an avoidance extinction procedure. Andrew's problem behavior did not extinguish during this condition, so it is less likely that his problem behavior was maintained by avoidance. Fourth, the percentage measures we used for the within-session analyses may have been a less accurate way of assessing the relative amount of problem behavior during putative E0 present and EO absent periods. Roane et al. (1999) used a rate measure to compare the levels of problem behavior between these periods. However, the rate of problem behavior and mands in the relevant sessions was very efficient, occurring at a rate of about 2 responses per minute. Because of this efficient rate, the actual E0 present periods during each session lasted only a few seconds, and thus the rate of responding during those time periods would have been artificially high. Therefore, the use of a percentage measure appeared to be necessary. Fifth, tokens were used contingent upon compliance during the final treatment for both participants. It is possible that the introduction of the token contingencies alone could have led to the reductions of problem behavior shown for both participants and that the functional reinforcers manipulated in the interventions were not necessary. However, both participants continued to mand for the reinforcers identified through our assessments (especially during sessions when compliance was low). This suggests that the reinforcers identified were valuable, and under some conditions had more value than the tokens themselves. Sixth, upon conducting within-session analyses of Samantha's functional analysis, we found that problem behavior occurred during the EO absent period in both the tangible and escape conditions. Had we conducted these analyses during the functional analysis, it is possible that this additional information may have been useful in selecting treatments and additional functional analyses for further study. Sixth, we did not collect data on the treatment integrity of teacher implementation of the final treatments. However, the treatments remained effective throughout the evaluation, suggesting that the teachers were implementing the procedures at a level of integrity to maintain low levels of problem behavior and high levels of compliance. Finally, we did not show experimental control with respect to the effects of Andrew's final treatment. He was in his last few weeks at the preschool, and we were unable to introduce a reversal condition prior to this transition. In addition, Samantha's baseline during her final treatment was only two sessions, and the baseline was not repeated. Similar to Andrew, Samantha was leaving the preschool program, so an extended reversal condition was not possible.

The current study included an evaluation of the use of treatments by teachers in the participants' classroom and showed that problem behavior maintained at low levels and appropriate behaviors maintained at high levels. As suggested by Hawkins (1979), when contrived situations are used in an evaluation, it is important to determine whether the performance in these situations is related to performance in the natural environment. Future researchers should continue to evaluate treatments suggested by functional analysis outcomes when being implemented in socially relevant conditions.

One unique feature of the current study was the use of treatments and a baseline designed to control for multiple functions of problem behavior. During Samantha's final treatment, the baseline condition controlled for problem behavior maintained by escape from demands, access to attention, and access to tangible items. Future researchers should continue to identify ways to evaluate problem behavior maintained by multiple sources of reinforcement within a single context.

The results of the current study, as well as the results of previous studies, provide some general guidelines for practice with respect to conducting a functional analysis. First, data from indirect and direct observations can be used to inform a functional analysis (Carr et al, 1997; et al., 1995). In conducting these analyses prior to the functional analysis, a clinician can (a) identify and operationally define the potential target behavior to be used in the functional analysis, (b) develop hypotheses as to the function of the problem behavior, and (c) identify any potential idiosyncratic variables that may influence responding. Following the indirect and descriptive analyses, a functional analysis can be conducted. Based on the data of the current study, as well as those of Roane et al. (1999) and Wilmer et al. (1995), we suggest conducting within-session analysis as an ongoing part of the functional analysis when possible. More specifically, data should be analyzed by looking at the percentage of problem behavior during periods when the putative E0s are present or absent. If a majority of problem behavior occurs when the EO is present, the functional analysis results can be considered conclusive and function-based treatments can be developed. If a majority of problem behavior occurs when the putative EO is absent, however, direct observation of the sessions should be used to develop hypotheses as to the function of problem behavior. Using this information, additional functional analyses should then be conducted to determine an accurate function of problem behavior so that function-based treatments can be developed.

This study is based on a thesis submitted by the first author, under the direction of the second author, to the Department of Applied Behavioral Science at the University of Kansas in partial fulfillment of the requirements of the M.A. degree.

Correspondence to Claudia L. Dozier, Department of Applied Behavioral Science, University of Kansas, Lawrence, KS 66045; e-mail: cdozier@ku.edu

function-based treatments can be developed. If a majority of problem behavior occurs when the putative EO is absent, however, direct observation of the sessions should be used to develop hypotheses as to the function of problem behavior. Using this information, additional functional analyses should then be conducted to determine an accurate function of problem behavior so that function-based treatments can be developed.

Hanley, G. P., Iwata, B. A., & McCord, B. E. (2003). Functional analysis of problem behavior: A review. Journal of Applied Behavior Analysis, 36, 147-186.

Hawkins, R. P. (1979). The functions of assessment: Implications for selection and development of devices for assessing repertoires in clinical, educational, and other settings. Journal of Applied Behavior Analysis, 12, 501-516.

Iwata, B. A., Dorsey, M. F., Slifer, K. J., Bauman, K. E., & Richman, G. S. (1994). Toward a functional analysis of self-injury. Journal of Applied Behavior Analysis, 27, 197-209. (Reprinted from Analysis and Intervention in Developmental Disabilities, 1982, 1983-1920).

Kahng, S., Abt, K. A., & Schonbachler, H. E. (2001) Assessment and treatment of low-rate high-intensity problem behavior. Journal of A.pplied Behavior Analysis, 34,225-228.

Kuhn, D. E., DeLeon, I. G., Fisher, W. W., & Wilke, A. E. (1999). Clarifying an ambiguous functional analysis with matched and mismatched extinction procedures. Journal of Applied Behavior Analysis, 32,99-102.

Kurtz, P. F., Chin, M. D., Huete, J. M., Tarbox, R. S. F., O'Connor, J. T., Paclawskyj, T. R., et al. (2003). Functional analysis and treatment of self-injurious behavior in young children: A summary of 30 cases. Journal of Applied Behavior Analysis, 36, 205-219.

Neef, N. A., & Iwata, B. A. (1994). Current research on functional analysis methodology: An introduction. Journal of Applied Behavior Analysis, 27, 211-214.

Richman, D. M., Wacker, D. P., Asmus, J. M., & Casey, S. D. (1998). Functional analysis and extinction of different behavior problems exhibited by the same individual. Journal of Applied Behavior Analysis, 31,475-478.

Roane, H. S., Lerman, D. C., Kelley, M. E., & Van Camp, C. M. (1999). Within-session patterns of responding during functional analysis: The role of establishing operations in .clarifying behavioral function. Research in Developmental Disabilities, 20, 73-89.

Rooker, G. W., Iwata, B. A., HArper, J. M., Fahmie, T. A., & Camp, E. M. (2011). False-positive tangible outcomes of functional analyses. Journal of Applied Behavior Anaysis, 44,737-745.

Shirley, M. J., Iwata, B. A., Smith? & Kahng, S. (1999). False-positive maintenance of self-injurious behavior by access to tangible reinforcers. journal of Applied Behavior Analyisis, 32, 201-204.

Smith, R. G., Iwata, B. A., Vollmer, T. R., & Zarcone, J. R. (1993). Experimental analysis and treatment of multiply controlled self-injury. Journal of Applied Behavior Analysis, 26,183-196.

Vollmer, T. R., Iwata, B. A., Smith, R. G., & Rodgers, T. A. (1992). Reduction of multiple abberrant behaviors and concurrent development of self-care skills with differential reinforcement. Research in Developmental Disabilities, 13,287-299.

Vollmer, T. R., Marcus, B. A., Ringdahl, J. E., & Roane, H. S. (1995). Progressing from brief assessments to extended experimental analyses in the evaluation of aberrant behavior. Journal of Applied Behavior Analysis, 28,561-576.

Wacker, D. P., Berg, W. K., Cooper, L. J., Derby, M., Steege, M. W., Northup, J., & Sasso, G. (1994). The impact if functional analysis methodology on outpatient clinic services. Journal of Applied Behavior Analysis, 27, 405-407.

Steven W. Payne

Melmark, Inc

Claudia L. Dozier, Pamela L. Neidert, and Erica S. Jowett

University of Kansas

Matthew H. Newquist

Mains'! Services
COPYRIGHT 2014 West Virginia University Press, University of West Virginia
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2014 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Payne, Steven W.; Newquist, Matthew H.
Publication:Education & Treatment of Children
Article Type:Report
Date:May 1, 2014
Words:8224
Previous Article:The Utility of a Brief Experimental Analysis for Problem Behavior Maintained by Escape from Demands.
Next Article:A Meta-Analytic Evaluation of the FRIENDS Program for Preventing Anxiety in Student Populations.
Topics:

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters