An evaluation of interrupted and uninterrupted ... - Wiley Online Library

6 downloads 17014 Views 376KB Size Report
The results showed that the interrupted data-analysis procedure ... stereotypy. Specifically, we compared interrupted and uninterrupted data-analysis procedures.
JOURNAL OF APPLIED BEHAVIOR ANALYSIS

2014, 47, 264–276

NUMBER

2 (SUMMER)

AN EVALUATION OF INTERRUPTED AND UNINTERRUPTED MEASUREMENT OF VOCAL STEREOTYPY ON PERCEIVED TREATMENT OUTCOMES REGINA A. CARROLL WEST VIRGINIA UNIVERSITY

AND

TIFFANY KODAK UNIVERSITY OF OREGON

The type of procedure used to measure a target behavior may directly influence the perceived treatment outcomes. In the present study, we examined the influence of different data-analysis procedures on the outcomes of two commonly used treatments on the vocal stereotypy of 2 children with an autism spectrum disorder. In Study 1, we compared an interrupted and uninterrupted dataanalysis procedure to measure vocal stereotypy during the implementation of response interruption and redirection (RIRD). The results showed that the interrupted data-analysis procedure overestimated the effectiveness of RIRD. In Study 2, we examined the influence of different dataanalysis procedures on the interpretation of the relative effects of 2 different treatments for vocal stereotypy. Specifically, we compared interrupted and uninterrupted data-analysis procedures during the implementation of RIRD and noncontingent reinforcement (NCR) as a treatment for vocal stereotypy. The results showed that, as in Study 1, the interrupted data-analysis procedure overestimated the effectiveness of RIRD; however, this effect was not apparent with NCR. These findings suggest that different types of data analysis can influence the perceived success of a treatment. Key words: automatic reinforcement, noncontingent reinforcement, response interruption and redirection, vocal stereotypy, data analysis

One of the defining characteristics of autism spectrum disorder (ASD) is restrictive, repetitive, and stereotyped patterns of behavior (American Psychiatric Association, 2013). Behaviors that are categorized as stereotypic often involve repetitive and invariant movements of one or more body parts that persist in the absence of social consequences (see Rapp & Vollmer, 2005). The design and data-collection portions of this research were conducted while both investigators were at the University of Nebraska Center’s Munroe-Meyer Institute. We thank Amber Paden, Molly Klum, and Nitasha Dickes for assistance with various aspects of data collection and Wayne Fisher for his helpful feedback on our data-collection methods. Address correspondence to Regina A. Carroll, West Virginia University, Department of Psychology, Life Sciences Room 1232, 53 Campus Drive, Morgantown, West Virginia 26506 (e-mail: [email protected]). doi: 10.1002/jaba.118

Children with ASD typically engage in high levels of stereotypy (Bodfish, Symons, Parker, & Lewis, 2000; Lewis & Bodfish, 1998) that can interfere with the acquisition of other adaptive behavior (Koegel & Covert, 1972; Lang et al., 2010). Previous research has evaluated the effectiveness of response interruption and redirection (RIRD) as an intervention to decrease vocal stereotypy (Ahearn, Clark, MacDonald, & Chung, 2007; Ahrens, Lerman, Kodak, Worsdell, & Keegan, 2011; Cassella, Sidener, Sidener, & Progar, 2011; Colón, Ahearn, Clark, & Masalsky, 2012; Liu-Gitz & Banda, 2010; Love, Miguel, Fernand, & LaBrie, 2012; Miguel, Clark, Tereshko, & Ahearn, 2009; Schumacher & Rapp, 2011) and motor stereotypy (Ahrens et al., 2011; Giles, St. Peter, Pence, & Gibson, 2012). Each time the participant engages in stereotypy during RIRD, the therapist interrupts stereotypy by delivering vocal (e.g., asking

264

MEASUREMENT OF VOCAL STEREOTYPY the participant social questions; Ahearn et al., 2007) or motor demands (e.g., “touch your head”; Ahrens et al., 2011) and redirecting behavior to comply with those demands. Ahearn et al. (2007) were the first to evaluate systematically the effectiveness of RIRD as an intervention for vocal stereotypy with four children with ASD. At the start of RIRD sessions, the therapist started a timer that counted down from 5 min. Each time the participant engaged in vocal stereotypy, the therapist stopped the timer and delivered vocal demands. When the participant had complied with three vocal demands without engaging in vocal stereotypy, the therapist resumed the timer. The session continued until 5 min had elapsed, during which the participant was not in RIRD. During RIRD sessions, the experimenters measured vocal stereotypy only when the participant was not in RIRD (i.e., only when the therapist was not delivering demands). The results of Ahearn et al. (2007) showed that vocal stereotypy decreased for all four participants during RIRD, and appropriate vocalizations increased for three of the participants. Although these results suggested that RIRD was effective, it was not clear whether decreased levels of vocal stereotypy were a function of (a) the RIRD procedure, (b) the data-analysis procedure, or (c) some combination of both. The authors likely removed the intervention time (i.e., time in RIRD) from their data analysis because they wanted to hold session time constant across baseline and intervention sessions. However, because vocal stereotypy was measured only when the participant was not in RIRD, it was not clear whether the participant continued to engage in vocal stereotypy during the procedure. Measuring stereotypy in this interrupted manner is potentially problematic for at least three reasons. First, high levels of vocal stereotypy during RIRD may limit the usefulness of the intervention. Second, Ahearn et al. (2007) did not report how frequently the therapist had to implement RIRD during each session. Thus, it

265

was not clear if the intervention required the therapist to implement RIRD frequently, and if the frequency of RIRD implementations decreased over time. Third, discontinuous measurement, such as interrupted data collection, can result in data that are an artifact of the measurement procedure (Cooper, Heron, & Heward, 2007). Other studies on RIRD share these limitations, in that they did not report data on stereotypy during implementation of RIRD (Cassella et al., 2011; Love et al., 2012; Miguel et al., 2009), the frequency and duration of RIRD implementations (Schumacher & Rapp, 2011), or both (Ahrens et al., 2011; Colón et al., 2012; Giles et al., 2012; Liu-Gitz & Banda, 2010). In one notable exception, Schumacher and Rapp (2011) did not remove the RIRD intervention time from their data analysis. Schumacher and Rapp found that RIRD produced an immediate decrease in vocal stereotypy for two children with ASD. These findings suggested that vocal stereotypy did not increase during RIRD. It is possible that RIRD may not be an ideal intervention for individuals who continue to engage in high levels of stereotypy during the implementation of RIRD and also require RIRD to be implemented frequently during sessions. If the goal is to identify an intervention for stereotypy that will be effective across a variety of settings (e.g., home, school), an intervention that requires substantial effort on the part of the caregiver (e.g., parents, teachers) would not be useful. Thus, to determine the practicality of using RIRD as an intervention for stereotypy, it is necessary to measure how much stereotypy occurs during implementation of the procedure and how frequently a therapist has to implement the procedure. Noncontingent reinforcement (NCR), which is another commonly used intervention for stereotypy, consists of providing individuals with noncontingent access to stimuli that are identified through stimulus preference assessments. Previous studies have found that providing individuals

266

REGINA A. CARROLL and TIFFANY KODAK

with noncontingent access to structurally matched stimuli (i.e., stimulation that matches the putative sensory product of stereotypy) effectively decreased automatically reinforced stereotypy (e.g., Lanovaz, Fletcher, & Rapp, 2009; Piazza, Adelinis, Hanley, Goh, & Delia, 2000; Rapp, 2006, 2007). At present, only one study has compared the effectiveness of NCR and RIRD as an intervention for stereotypy (Love et al., 2012). This study compared the effects of RIRD alone, matched stimulation alone (i.e., NCR), and a combination of RIRD and matched stimulation as an intervention for two participants’ vocal stereotypy. The results for both participants showed that all three interventions decreased vocal stereotypy. For one participant, RIRD alone produced the greatest reductions in vocal stereotypy and the highest levels of appropriate vocalizations. A combination of RIRD and matched stimulation produced the best treatment outcome for the second participant. There are several potential benefits and limitations of the use of RIRD or NCR as an intervention for stereotypy. The relative effects of each intervention likely depend on the context in which the intervention is implemented. For example, implementation of RIRD requires a caregiver to monitor the target individual continuously. This arrangement may not always be practical. NCR may be limited if providing access to alternative items competes with other ongoing tasks (e.g., academic instruction). Thus, RIRD might be an appropriate intervention for stereotypy during academic instruction, whereas NCR may be a more appropriate intervention during unstructured leisure times. Additional research is warranted to identify the variables that influence whether a treatment for stereotypy is truly effective and the conditions under which such treatments should be implemented. The purpose of the current study was to evaluate the effect of one potential variable, dataanalysis procedures for vocal stereotypy, on the interpretation of treatment outcomes. In Study 1, we compared interrupted and uninterrupted

measures of vocal stereotypy during the implementation of RIRD. In Study 2, we compared the relative effects of interrupted and uninterrupted measures of vocal stereotypy during the implementation of RIRD and NCR. GENERAL METHOD Participants, Setting, and Materials Two children who had been diagnosed with ASD and displayed vocal stereotypy participated in this study. Will was an 8-year-old boy; he communicated with two- to four-word phrases, mainly to request items (e.g., “I want ball please”). He followed one-step instructions (e.g., touch your head), responded to some fillin-the-blank statements (e.g., “a dog says—”), and imitated most gross-motor movements. Parker was a 5-year-old boy; he communicated using complete sentences. He followed multistep instructions (e.g., “Stand up and get your coat”), answered questions (e.g., “Where do you live?,” “What is your name?,” “When is your birthday?”), and spontaneously labeled items in the natural environment. Both participants received early intervention services at a hospital-based clinic. All sessions took place in a private therapy room that contained two chairs, a video camera, timers, and preferred toys (NCR sessions only). Response Measurement, Data Analysis, and Interobserver Agreement Vocal stereotypy was defined as any nonfunctional or noncontextual vocalization. Topographies of vocal stereotypy included repetitive sounds and phrases and reciting lines from television shows or movies. Trained observers viewed videotaped sessions and collected data on vocal stereotypy using continuous duration recording on a behavior software program loaded onto laptop computers. Data were converted to percentage of time by dividing the number of seconds engaged in vocal stereotypy by the total number of seconds in a session. Observers also

MEASUREMENT OF VOCAL STEREOTYPY measured the duration and frequency of RIRD implementations and the total session duration. For RIRD sessions, in addition to measuring vocal stereotypy during the total session time, we calculated the duration of vocal stereotypy for the time in RIRD (i.e., when the therapist implemented RIRD) and the time outside RIRD (i.e., anytime the therapist was not implementing RIRD). We compared the percentage of time vocal stereotypy occurred during RIRD and nointeraction sessions (Study 1) and RIRD, nointeraction, and NCR sessions (Study 2) using three methods of data analysis: (a) total session time (i.e., time the participant was and was not in RIRD), (b) time outside RIRD, and (c) time in RIRD. For each no-interaction and NCR session, we calculated the percentage of time that vocal stereotypy occurred outside RIRD and in RIRD using the previous RIRD session. First, we transferred the exact time that the participant was in RIRD to a real-time recording data sheet (see Rapp, Carr, Miltenberger, Dozier, & Kellum, 2001). Next, we transferred the exact times that the participant engaged in vocal stereotypy during the no-interaction or NCR session to the same real-time data sheet with the time in RIRD. Finally, we calculated the percentage of time the participant engaged in vocal stereotypy during the time in RIRD and the time outside RIRD for each of the no-interaction and NCR sessions. The purpose of yoking the time in RIRD and time outside RIRD sessions to each of the no-interaction and NCR sessions was to ensure that interrupted measures of vocal stereotypy for the no-interaction and NCR sessions were calculated using the same time intervals as the previous RIRD session. A second observer independently scored an average of 41% (range, 38% to 46%) of the total sessions across conditions for both participants during Studies 1 and 2. Interobserver agreement was calculated using the block-by-block method for continuous measurement, in which the observation period was divided into consecutive

267

10-s blocks (see Mudford, Taylor, & Martin, 2009). For each block, the smaller value was divided by the larger value. The value from each block was then totaled and divided by the total number of blocks, and the ratio was converted to a percentage. If both observers scored a zero in a block, that block was given a value of 1 (i.e., treated as 100% agreement). The mean agreement for Study 1 was 82% (range, 81% to 87%) for Will and 86% (range, 80% to 91%) for Parker. For Study 2, mean agreement was 89% (range, 82% to 97%) for Will and 87% (range, 81% to 95%) for Parker.

STUDY 1: COMPARISON OF INTERRUPTED AND UNINTERRUPTED MEASUREMENT DURING RIRD The purpose of Study 1 was to compare interrupted and uninterrupted measures of vocal stereotypy during implementation of RIRD. METHOD Design and Procedure An alternating treatments design was used to compare the effects of RIRD to no interaction on the duration of stereotypy. Specifically, we compared the percentage of time vocal stereotypy occurred across RIRD and no-interaction sessions during the total session time, time outside RIRD, and time in RIRD. No interaction. Two therapists were present in the room but did not interact with or respond to the participant during the session. One therapist taped the session, and the other therapist sat or stood next to one of the two available chairs. The duration of the no-interaction sessions was yoked to the previous RIRD session (e.g., if the previous RIRD session was 8 min, the subsequent nointeraction session was 8 min). The purposes of the no-interaction condition were (a) to demonstrate the persistence of vocal stereotypy in the absence of social consequences and (b) to

268

REGINA A. CARROLL and TIFFANY KODAK

measure levels of stereotypy in the absence of RIRD. Response interruption and redirection. Two therapists were present in the room during RIRD sessions. One therapist taped the session but did not interact with or respond to the participant. The second therapist sat next to the participant and, contingent on vocal stereotypy, immediately issued a motor demand (e.g., “touch your head”). Demands were randomly presented from a list of 20 previously mastered motor demands. Praise was provided in a neutral tone of voice (e.g., “yes,” “right”), and the next instruction was presented if the child complied with the initial demand. If the child did not comply within 5 s, the therapist repeated the demand while she modeled the correct response (i.e., model prompt). If the child did not comply with the model prompt within 5 s, the therapist repeated the demand while she physically guided the child to engage in the correct response. The therapist continued to issue motor demands until the participant engaged in three consecutive correct prompted or unprompted responses in the absence of vocal stereotypy. For both participants, we chose to present motor demands rather than vocal demands because it was possible to guide a response physically if the child did not comply with the initial demand (previous research has shown that both vocal and motor RIRD effectively decrease vocal stereotypy; see Ahrens et al., 2011). The therapist started a session timer at the beginning of the session, stopped the timer each time she implemented RIRD, and resumed the timer after the child complied with three consecutive demands in the absence of vocal stereotypy. Sessions continued for Parker until he was not in RIRD for a total of 5 min, or a total of 15 min elapsed. For Will, sessions continued until he was not in RIRD for 10 min, or a total of 30 min elapsed. No differential consequences were provided for any nontargeted responses (e.g., requests, aggression) emitted during the session. The procedures for terminating sessions

varied for each participant as a function of their availability for participation. RESULTS AND DISCUSSION Figure 1 depicts the results for Will. When the uninterrupted measurement strategy was used, levels of stereotypy differed across conditions (M ¼ 55% of session time during no interaction and 26% of session time during RIRD). When the interrupted measurement strategy was used to measure stereotypy in these same sessions, levels of stereotypy again differed across conditions, but to a greater extent (M ¼ 53% of session time during no interaction and 5% of session time during time outside RIRD). When only vocal stereotypy during time in RIRD was measured, levels of stereotypy were more similar across conditions (M ¼ 56% of session time during no interaction and 45% of session time during RIRD). Figure 1 (bottom) depicts the frequency of RIRD implementations and the total session duration during RIRD. Will’s average session duration was 23 min (range, 14 to 30 min), and RIRD was implemented an average of 42 times per session (range, 13 to 94). Figure 2 depicts the results for Parker. When the uninterrupted measurement strategy was used, high levels of vocal stereotypy were obtained during no-interaction sessions (M ¼ 70% of session time). Parker exhibited vocal stereotypy at lower levels during implementation of RIRD. However, similar to Will, he continued to engage in vocal stereotypy during RIRD sessions (M ¼ 42% of session time). When vocal stereotypy for the same session was measured only during the time outside RIRD, near-zero levels of vocal stereotypy were obtained during the RIRD condition (M ¼ 6% of session time). When vocal stereotypy during time in RIRD was measured, high levels of vocal stereotypy were obtained across both no-interaction (M ¼ 72% of session time) and RIRD (M ¼ 70% of session time) conditions. The bottom panel shows that Parker’s average session duration was 11 min (range, 7 to 15 min), and

MEASUREMENT OF VOCAL STEREOTYPY 100

269

Total Session Time (Uninterrupted)

Will

80 NI

RIRD

60 40

Percentage of Time with Vocal Stereotypy

20 0 100

Time Outside of RIRD

80 60 40 20 0 100

Time In RIRD

80 60 40 20 0

Frequency of RIRD

35 30

Duraon

80

25

Frequency

60

20 15

40

10 20

5

Session Duraon (Min)

100

0

0 0

1

2

3

4

5

6

7

Session

8

9

10

11

12

13

14

Figure 1. Percentage of time that Will engaged in vocal stereotypy across the no-interaction and RIRD conditions during the total session time (uninterrupted; first panel), the time outside RIRD (second panel), and the time in RIRD (third panel). The fourth panel shows the frequency of RIRD implementations and the total duration of RIRD sessions.

RIRD was implemented an average of 10 times per session (range, 5 to 13). The results of Study 1 showed that for both participants, only modest reductions in vocal stereotypy during RIRD were observed when the total session time data-analysis method was used. In comparison, it appeared that vocal stereotypy decreased to near-zero levels for both participants when vocal stereotypy that occurred only outside RIRD was measured. Furthermore, high levels of vocal stereotypy were obtained for participants across both the RIRD and no-interaction conditions when vocal stereotypy was measured

during the implementation of RIRD. Finally, both participants’ total session duration and the frequency of RIRD implementations did not decrease substantially across the course of the study (similar to Love et al., 2012). Overall, the results of Study 1 suggest the importance of using uninterrupted measures of vocal stereotypy when evaluating RIRD as an intervention for vocal stereotypy. If we had measured only vocal stereotypy outside RIRD, consistent with the data-analysis procedures of the majority of previously published studies, we may have overestimated the effectiveness of RIRD.

REGINA A. CARROLL and TIFFANY KODAK

270 100

Parker

Total Session Time (Uninterrupted)

80 60 40 RIRD

Percentage of Time with Vocal Stereotypy

20

NI

0 100

Time Outside of RIRD

80 60 40 20 0 100

Time In RIRD

80 60 40 20 0 Duraon

20

Frequency 15

15

10

10

5

5

0

Session Duraon (Min)

Frequency of RIRD

20

0 0

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

Session

Figure 2. Percentage of time that Parker engaged in vocal stereotypy across the no-interaction and RIRD conditions during the total session time (uninterrupted; first panel), the time outside RIRD (second panel), and the time in RIRD (third panel). The fourth panel shows the frequency of RIRD implementations and the total duration of RIRD sessions.

STUDY 2: EFFECTS OF INTERRUPTED AND UNINTERRUPTED MEASUREMENT DURING RIRD AND NCR

data-analysis procedures during the implementation of RIRD and NCR interventions.

The results of Study 1 demonstrated that the type of data-analysis procedure affected the interpretation of RIRD as an effective treatment for vocal stereotypy. In Study 2, we extended these findings by evaluating the extent to which dataanalysis procedures influenced the interpretation of treatment success when two different treatments for vocal stereotypy are compared. Specifically, we compared interrupted and uninterrupted

Design and Procedure An alternating treatments design was used to compare the effects of interrupted and uninterrupted data analysis on the duration of vocal stereotypy during RIRD, NCR, and no interaction. We conducted RIRD and no-interaction sessions using the same procedures described in Study 1 with one exception. Because the results of Study 1 suggested that RIRD may not be a practical intervention for

METHOD

MEASUREMENT OF VOCAL STEREOTYPY Will or Parker, we modified the procedures for terminating sessions. Specifically, we decreased the time required outside RIRD to 5 min for Will, and we decreased the maximum session duration to a total of 10 min for both participants. Stimulus preference assessment. To identify preferred items that may have competed with stereotypy, we conducted two to five free-operant preference assessments (Roane, Vollmer, Ringdahl, & Marcus, 1998) with six to nine items that produced sensory stimulation that structurally matched the hypothesized stimulation produced by vocal stereotypy (e.g., auditory stimulation). During each session, participants were given continuous access to five different items, and each session lasted 10 min. For both participants, the three items associated with the highest percentage of object manipulation were also associated with low levels of vocal stereotypy (less than 13% of the session time). We subsequently used these items during the NCR sessions. Will’s top three items were a LeapFrog Alphabet Pal Caterpillar, a LeapFrog Spin and Sing Alphabet Zoo, and a Munchkin Mozart Magic Cube. Parker’s top three items were a VTech Write and Learn Smartboard, a VTech Touch and Discover Alphabet Town, and a VTech Tote and Go Laptop. Noncontingent reinforcement. The participant had continuous and noncontingent access to the top three toys identified during the preference assessment. As in the no-interaction sessions, two therapists were present in the room but did not interact with or respond to the participant during the session. One therapist taped the session, and the other therapist sat or stood next to one of the two available chairs. The duration of NCR sessions was yoked to the previous RIRD session. The purpose of the NCR condition was to evaluate the effects of noncontingent access to items that produced structurally matched stimulation on the duration of vocal stereotypy. RESULTS AND DISCUSSION Figure 3 depicts results for Will. When the uninterrupted measurement strategy was used,

271

high and variable levels of vocal stereotypy were obtained during the no-interaction (M ¼ 48% of session time) and RIRD (M ¼ 30% of session time) conditions. In comparison, low levels of vocal stereotypy were obtained during the NCR condition (M ¼ 8% of session time) with the uninterrupted measurement strategy. When vocal stereotypy was measured during time outside RIRD, similar levels of vocal stereotypy were obtained during the no-interaction (M ¼ 46% of session time) and NCR (M ¼ 7% of session time) conditions to those seen using the uninterrupted measurement. However, low levels of vocal stereotypy were obtained during the RIRD condition (M ¼ 9% of session time). When vocal stereotypy during time in RIRD was measured,high levels of vocal stereotypy were obtained during both no-interaction (M ¼ 49%) and RIRD (M ¼ 49%) conditions, but vocal stereotypy continued to occur at low levels during the NCR condition (M ¼ 7%). Will’s average RIRD session duration was 10 min (range, 8 to 10 min), which is consistent with the maximum session duration, and RIRD was implemented an average of 21 times per session (range, 13 to 34). Figure 4 depicts the results for Parker. When the uninterrupted measurement strategy was used, lower levels of vocal stereotypy were obtained across both the RIRD (M ¼ 18% of session time) and NCR (M ¼ 3% of session time) conditions than in the no-interaction condition (M ¼ 69% of session time). Thus, NCR was associated with larger reductions in vocal stereotypy than RIRD. When vocal stereotypy was measured for the same sessions only during time outside RIRD, similar low levels of vocal stereotypy were obtained during the RIRD (M ¼ 2% of session time) and NCR (M ¼ 2% of session time) conditions compared to the nointeraction condition (M ¼ 70% of session time). When vocal stereotypy was measured during the time in RIRD, high levels of vocal stereotypy were obtained during both no-interaction (M ¼ 67% of session time) and RIRD (M ¼ 66% of session time) conditions, but vocal

REGINA A. CARROLL and TIFFANY KODAK

272 100

Total Session Time (Uninterrupted)

Will

80 NI

RIRD

60 40

NCR

Percentage of Time with Vocal Stereotypy

20 0

Time Outside of RIRD

100 80 60 40 20 0 100

Time In RIRD

80 60 40 20 0 Duraon

12

30

10 Frequency

25

8

20

6

15

4

10

2

5 0

0 0

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

Session Duraon (Min)

Frequency of RIRD

35

16

Session

Figure 3. Percentage of time that Will engaged in vocal stereotypy across the no-interaction, RIRD, and NCR conditions during the total session time (uninterrupted; first panel), the time outside RIRD (second panel), and the time in RIRD (third panel). The fourth panel shows the frequency of RIRD implementations and the total duration of RIRD sessions.

stereotypy continued to occur at low levels during the NCR condition (M ¼ 5% of session time). The average RIRD session duration for Parker was 7 min (range, 6 to 9 min), and RIRD was implemented an average of 6 times per session (range, 2 to 9). As in Study 1, the results of Study 2 showed that the interrupted data-analysis procedure influenced the perceived effectiveness of RIRD as a treatment for vocal stereotypy. Specifically, vocal stereotypy decreased for both participants during RIRD

when stereotypy during the total session time was measured (i.e., uninterrupted data analysis); however, substantial decreases in vocal stereotypy were observed only when stereotypy outside RIRD was measured (interrupted data analysis). This finding was not apparent with the NCR procedure. Compared to RIRD, vocal stereotypy decreased to low levels for both participants during NCR, and the observed reduction in vocal stereotypy was consistent across interrupted and uninterrupted methods of data analysis.

MEASUREMENT OF VOCAL STEREOTYPY 100

273

Total Session Time (Uninterrupted)

Parker NI

80 60 RIRD

40

NCR

Percentage of Time with Vocal Stereotypy

20 0 100

Time Outside of RIRD

80 60 40 20 0 100

Time In RIRD

80 60 40 20 0 10

Duraon

8

8 6

6 Frequency

4

4 2

2

0

0 0

1

2

3

4

5

6

7

8

9

10

11

12

13

Session Duraon (Min)

Frequency of RIRD

10

14

Session

Figure 4. Percentage of time that Parker engaged in vocal stereotypy across the no-interaction, RIRD, and NCR conditions during the total session time (uninterrupted; first panel), the time outside RIRD (second panel), and the time in RIRD (third panel). The fourth panel shows the frequency of RIRD implementations and the total duration of RIRD sessions.

GENERAL DISCUSSION Overall, the results of the present studies show that the type of data-analysis procedure influenced the interpretation of treatment success across different treatment alternatives for vocal stereotypy. In Studies 1 and 2, the interrupted data-analysis procedure led to an exaggerated treatment effect for RIRD compared to the uninterrupted data-analysis procedure. That is,

when vocal stereotypy was measured outside RIRD, vocal stereotypy decreased to low levels for both participants; however, when vocal stereotypy was measured during the entire RIRD session, vocal stereotypy continued to occur for a moderate percentage of the session. In Study 2, the interrupted data analysis did not influence the interpretation of treatment success with NCR. For both participants, vocal stereotypy remained

274

REGINA A. CARROLL and TIFFANY KODAK

at low levels during NCR sessions, irrespective of the type of data analysis. One potential explanation for the discrepancies across data-analysis procedures is that RIRD may have altered the point in a session when vocal stereotypy was most likely to occur. For both participants, the highest levels of vocal stereotypy occurred during the implementation of RIRD. That is, when vocal stereotypy was measured during the implementation of RIRD, vocal stereotypy occurred at levels similar to the nointeraction condition, when no intervention was in place. This finding may be an artifact of the procedures used during RIRD. Specifically, as soon as the participant engaged in vocal stereotypy, the therapist paused the session time and continued to implement RIRD until the participant stopped engaging in vocal stereotypy during three demands. Thus, the procedures used during RIRD limited the opportunity to engage in stereotypy outside RIRD. Compared to RIRD, the results of the NCR condition in Study 2 showed that vocal stereotypy continued to occur at similarly low levels across all three types of data analysis. The results of the present study vary from previous research that has shown that RIRD effectively decreased vocal stereotypy (Ahearn et al., 2007; Ahrens et al., 2011; Liu-Gitz & Banda, 2010; Miguel et al., 2009; Schumacher & Rapp, 2011). Specifically, both participants’ vocal stereotypy continued to occur at moderate levels during RIRD, and the overall number of RIRD implementations and the total session duration did not decrease across sessions. Thus, at least for the participants in the current study, RIRD did not appear to be a practical intervention. There are at least two potential explanations for this discrepancy. First, some previous studies used interrupted measures of stereotypy during RIRD (e.g., Ahearn et al., 2007; Miguel et al., 2009). The results of the current study suggest that this type of interrupted method of data analysis may overestimate the effectiveness of an intervention. This finding is consistent with previous studies that have evaluated different types of discontinu-

ous data-collection measures (e.g., Cummings & Carr, 2009; Rapp, Colby-Dirksen, Michalski, Carroll, & Lindenberg, 2008). It is also possible that RIRD was less effective for the participants in the present study because we used motor RIRD. A number of previous studies demonstrating the effectiveness of RIRD to decrease vocal stereotypy used vocal RIRD (Ahearn et al., 2007; Liu-Gitz & Banda, 2010; Miguel et al., 2009; Schumacher & Rapp, 2011). It is possible that RIRD may be more effective if the form of the demands matches the form of stereotypy. Ahrens et al. (2011) compared the effects of both motor and vocal RIRD on vocal stereotypy for two participants and on vocal and motor stereotypy for two additional participants. The results showed that both types of RIRD effectively reduced vocal and motor stereotypy. However, additional research should compare the effects of vocal and motor RIRD to other interventions for different forms of stereotypy. Some potential limitations of the current study should be noted. First, previous research on RIRD has shown that in addition to decreasing vocal stereotypy, RIRD may also increase appropriate vocalizations (e.g., Ahearn et al., 2007). In the current study, appropriate vocalizations were not reinforced. It is possible that reinforcing appropriate vocalizations may have enhanced the effectiveness of RIRD. However, for the purpose of the present study, we were interested in evaluating the effects of RIRD alone on vocal stereotypy. Previous studies have shown that RIRD can be used to decrease vocal stereotypy without providing reinforcement for appropriate vocalizations (see Schumacher & Rapp, 2011). Second, we did not conduct a functional analysis to rule out potential social functions that might have maintained participants’ stereotypy. However, both participants’ stereotypy persisted in the absence of social consequences during the no-interaction conditions, suggesting that stereotypy was at least in part maintained by automatic reinforcement (see Iwata & Dozier, 2008).

MEASUREMENT OF VOCAL STEREOTYPY Although the results of the present study suggest that NCR was more effective than RIRD for decreasing vocal stereotypy exhibited by our participants, some potential limitations to the use of NCR as an intervention for vocal stereotypy remain. First, it may not always be feasible to provide an individual with noncontingent access to preferred items because they may interfere with other ongoing events in the natural environment (e.g., academic instruction, social skills training; see Lanovaz & Sladeczek, 2012). Second, previous research suggests that NCR may decrease vocal stereotypy but also increase other forms of stereotypy (e.g., Rapp et al., 2012). The results of the present study suggest several areas for future research. First, additional research should evaluate the effects of interrupted and uninterrupted measures of stereotypy during the implementation of RIRD. In the present study, interrupted measures of vocal stereotypy overestimated the effectiveness of RIRD as an intervention for vocal stereotypy. However, that may not be the case for other participants; thus, additional research is needed. Second, additional research could evaluate the effects of data-analysis procedures on intervention outcomes with modifications to RIRD procedures, such as when reinforcement is provided for appropriate vocalizations and the formats of RIRD and stereotypy match (e. g., vocal RIRD and vocal stereotypy). Finally, future research should also evaluate the use of interrupted and uninterrupted data analysis procedures on the perceived effectiveness of additional interventions for stereotypy (e.g., response blocking) or other types of problem behavior. Overall, the results of this study suggested that RIRD led to modest decreases in vocal stereotypy for the two participants in this study. However, it was evident that stereotypy continued to occur at moderate levels during RIRD only when we measured vocal stereotypy during the total session time. In comparison, NCR decreased vocal stereotypy to low levels for both participants in Study 2 independent of the method of data

275

analysis. It may not always be practical to use uninterrupted measures of behavior; however, these findings emphasize the importance of evaluating the sensitivity of different methods of data analysis when the effectiveness of an intervention for stereotypy is evaluated.

REFERENCES Ahearn, W. H., Clark, K. M., MacDonald, R. P. F., & Chung, B. I. (2007). Assessing and treating vocal stereotypy in children with autism. Journal of Applied Behavior Analysis, 40, 263–275. doi: 10.1901/ jaba.2007.30-06 Ahrens, E. N., Lerman, D. C., Kodak, T., Worsdell, A. S., & Keegan, C. (2011). Further evaluation of response interruption and redirection as treatment for stereotypy. Journal of Applied Behavior Analysis, 44, 95–108. doi: 10.1901/jaba.2011.44-95 American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.). Arlington, VA: Author. Bodfish, J. W., Symons, F. J., Parker, D. E., & Lewis, M. H. (2000). Varieties of repetitive behavior in autism: Comparisons to mental retardation. Journal of Autism and Developmental Disorders, 30, 237–243. doi: 10.1023/A:1005596502855 Cassella, M. D., Sidener, T. M., Sidener, D. W., & Progar, P. R. (2011). Response interruption and redirection for vocal stereotypy in children with autism: A systematic replication. Journal of Applied Behavior Analysis, 44, 169–173. doi: 10.1901/jaba.2011.44-169 Colón, C. L., Ahearn, W. H., Clark, K. M., & Masalsky, J. (2012). The effects of verbal operant training and response interruption and redirection on appropriate and inappropriate vocalizations. Journal of Applied Behavior Analysis, 45, 107–120. doi: 10.1901/jaba. 2012.45-107 Cooper, J. O., Heron, T. E., & Heward, W. L. (2007). Applied behavior analysis (2nd ed.). Upper Saddle River, NJ: Pearson. Cummings, A. R., & Carr, J. E. (2009). Evaluating progress in behavioral programs for children with autism spectrum disorders via continuous and discontinuous measurement. Journal of Applied Behavior Analysis, 42, 57–71. doi: 10.1901/jaba.2009.42-57 Giles, A. F., St. Peter, C. C., Pence, S. T., & Gibson, A. B. (2012). Preference for blocking or response redirection during stereotypy treatment. Research in Developmental Disabilities, 33, 1691–1700. doi: 10.1016/j. ridd.2012.05.008 Iwata, B. A., & Dozier, C. L. (2008). Clinical application of functional analysis methodology. Behavior Analysis in Practice, 1, 3–9.

276

REGINA A. CARROLL and TIFFANY KODAK

Koegel, R. L., & Covert, A. (1972). The relationship of selfstimulation to learning in autistic children. Journal of Applied Behavior Analysis, 5, 381–387. doi: 10.1901/ jaba.1972.5-381 Lang, R., O’Reilly, M., Sigafoos, J., Machalicek, W., Rispoli, M., Lancioni, G. E., … Fragale, C. (2010). The effects of an abolishing operation intervention component on play skills, challenging behavior, and stereotypy. Behavior Modification, 34, 267–289. doi: 10.1177/ 0145445510370713 Lanovaz, M. J., Fletcher, S. E., & Rapp, J. T. (2009). Identifying stimuli that alter immediate and subsequent levels of vocal stereotypy. Behavior Modification, 33, 682–704. doi: 10.1177/0145445509344972 Lanovaz, M. J., & Sladeczek, I. E. (2012). Vocal stereotypy in individuals with autism spectrum disorders: A review of behavioral interventions. Behavior Modification, 36, 146–164. doi: 10.1177/0145445511427192 Lewis, M. H., & Bodfish, J. W. (1998). Repetitive behavior disorders in autism. Mental Retardation and Developmental Disabilities Research Reviews, 4, 8–89. doi: 10.1002/(SICI)1098-2779(1998)4:23.0.CO;2-0 Liu-Gitz, L., & Banda, D. R. (2010). A replication of the RIRD strategy to decrease vocal stereotypy in a student with autism. Behavioral Interventions, 25, 77–87. doi: 10.1002/bin.297 Love, J. J., Miguel, C. F., Fernand, J. K., & LaBrie, J. K. (2012). The effects of matched stimulation and response interruption and redirection on vocal stereotypy. Journal of Applied Behavior Analysis, 45, 549–564. doi: 10.1901/jaba.2012.45-549 Miguel, C. F., Clark, K., Tereshko, L., & Ahearn, W. H. (2009). The effects of response interruption and redirection and sertraline on vocal stereotypy. Journal of Applied Behavior Analysis, 42, 883–888. doi: 10.1901/jaba.2009.42-883 Mudford, O. C., Taylor, S. A., & Martin, N. T. (2009). Continuous recording and interobserver agreement algorithms reported in the Journal of Applied Behavior Analysis (1995–2005). Journal of Applied Behavior Analysis, 42, 165–169. doi: 10.1901/jaba.2009.42-165 Piazza, C. C., Adelinis, J. D., Hanley, G. P., Goh, H. L., & Delia, M. D. (2000). An evaluation of the effects of matched stimuli on behaviors maintained by automatic

reinforcement. Journal of Applied Behavior Analysis, 33, 13–27. doi: 10.1901/jaba.2000.33-13 Rapp, J. T. (2006). Toward an empirical method for identifying matched stimulation for automatically reinforced behavior: A preliminary investigation. Journal of Applied Behavior Analysis, 39, 137–140. doi: 10.1901/jaba.2006.37-05 Rapp, J. T. (2007). Further evaluation of methods to identify matched stimulation. Journal of Applied Behavior Analysis, 40, 73–88. doi: 10.1901/jaba.2007.142-05 Rapp, J. T., Carr, J. E., Miltenberger, R. G., Dozier, C. L., & Kellum, K. K. (2001). Using real-time recording to enhance the analysis of within-session functional analysis data. Behavior Modification, 25, 79–93. doi: 10.1177/0145445501251005 Rapp, J. T., Colby-Dirksen, A. M., Michalski, D. N., Carroll, R. A., & Lindenberg, A. M. (2008). Detecting changes in simulated events using partial-interval recording and momentary time sampling. Behavioral Interventions, 23, 237–269. doi: 10.1002/bin.269 Rapp, J. T., Swanson, G., Sheridan, S. M., Enloe, K. A., Maltese, D., Sennott, L. A., … Lanovaz, M. J. (2012). Immediate and subsequent effects of matched and unmatched stimuli on targeted vocal stereotypy and untargeted motor stereotypy. Behavior Modification, 37, 543–567. doi: 10.1177/0145445512461650 Rapp, J. T., & Vollmer, T. R. (2005). Stereotypy I: A review of behavioral assessment and treatment. Research in Developmental Disabilities, 26, 527–547. doi: 10.1016/ j.ridd.2004.11.005 Roane, H. S., Vollmer, T. R., Ringdahl, J. E., & Marcus, B. A. (1998). Evaluation of a brief stimulus preference assessment. Journal of Applied Behavior Analysis, 31, 605–620. doi: 10.1901/jaba.1998.31-605 Schumacher, B. I., & Rapp, J. T. (2011). Evaluation of the immediate and subsequent effects of response interruption and redirection on vocal stereotypy. Journal of Applied Behavior Analysis, 44, 681–685. doi: 10.1901/ jaba.2011.44-681

Received November 10, 2012 Final acceptance February 18, 2014 Action Editor, Joel Ringdahl