Toward accurate inferences of response class membership

CHRISTINE A. WARNER

WESTERN NEW ENGLAND UNIVERSITY AND NEW ENGLAND CENTER FOR CHILDREN

GREGORY P. HANLEY, ROBIN K. LANDA, KELSEY W. RUPPEL,ADITHYAN RAJARAMAN AND MAHSHID GHAEMMAGHAMI

WESTERN NEW ENGLAND UNIVERSITY

JESSICA D. SLATON

WESTERN NEW ENGLAND UNIVERSITY AND NASHOBA LEARNING GROUP

HOLLY C. GOVER

WESTERN NEW ENGLAND UNIVERSITY

In their review of synthesis within the functional analysis (FA) literature, Slaton and Hanley(2018) reported that most synthesized contingency analyses have included multiple topographiesof problem behavior in the reinforcement contingency class. This leaves the question of whetherone, some, or all forms of problem behavior are sensitive to the synthesized reinforcement con-tingencies in published analyses. To address this ambiguity, all topographies of problem behav-ior that were reported by caregivers to co-occur with the most concerning problem behaviorwere analyzed for 10 participants. We implemented extinction across one or more forms ofproblem behavior to determine whether all forms reported to co-occur were sensitive to thesame synthesized reinforcement contingency. For nine of 10 participants, the most concerningtopographies were sensitive to the same synthesized reinforcement contingencies as the less con-cerning topographies (results were inconclusive for one). Implications for inferring response classmembership from single analyses are discussed.Key words: consecutive controlled case series, extinction, functional analysis, precursors,

problem behavior, response classes

The prevalence of problem behavior such asaggression, self-injurious behavior (SIB), anddestructive behavior in individuals with intel-lectual disabilities ranges from 10 to 52%(Dworschak, Ratz, & Wagner, 2016; Emersonet al., 2001; Holden & Gitlesen, 2003). Mur-phy, Healy, and Leader (2009) found that 64%

of children with autism exhibited aggression,SIB, or both. Furthermore, individuals withintellectual disabilities (Emerson & Bromley,1995) and autism (Murphy et al., 2009) oftendisplay multiple forms of problem behaviorincluding aggression, SIB, disruption, destruc-tion, tantrum, noncompliance, screaming, andstereotypy.If multiple responses have the same effect on

the environment, they are considered membersof a response class (Catania, 2007). By exten-sion, members of a response class are also likelyto be evoked by the same establishing opera-tions (EOs). The concept of a class of topo-graphically dissimilar behaviors reinforced bysimilar consequent events and evoked by similarEOs has important implications for treatment,

This manuscript was prepared in partial fulfillment of aPh.D. in Behavior Analysis from Western New EnglandUniversity by the first author. We thank JonathanPinkston, Eileen Roscoe, and Rachel Thompson for theirfeedback on earlier versions of this manuscript and AshleeHoule, Juliana Marcus, Shannon Ward, and Ellen Gagefor their assistance with data collection and other aspectsof the project.Address correspondence to: Gregory P. Hanley, Western

New England University, 1215 Wilbraham Road, Spring-field, Massachusetts 01119 (e-mail: ghanley@wne.edu)doi: 10.1002/jaba.598

JOURNAL OF APPLIED BEHAVIOR ANALYSIS 2019, 9999, 1–24 NUMBER 9999 ()

© 2019 Society for the Experimental Analysis of Behavior

1

because a single treatment can be implementedand should be effective for all members of aresponse class. Therefore, when attempting toaddress severe problem behavior, it is importantto understand which topographies share thesame controlling reinforcement contingency.In a review of research relevant to functional

analyses (FAs) of problem behavior, Hanley,Iwata, and McCord (2003) recommended that asingle topography of problem behavior betargeted in an FA to ensure an efficacious treat-ment is developed for each topography of prob-lem behavior. When multiple topographies areincluded in the reinforcement contingency andonly one or some topographies show sensitivityduring the FA, one may make an incorrect infer-ence that the variables controlling the topogra-phies that showed sensitivity are the samevariables that are controlling the topographiesnot observed in the analysis. If these inferencesare wrong, the treatment will be only partiallyeffective; if response topographies are analyzed inaggregate, the treatment will show only marginaleffects or, if response topographies are measuredseparately (Derby et al., 1994), only some topog-raphies will be affected by treatment while otherswill either show no change or may worsen.The results of Iwata, Pace, Cowdery, and

Miltenberger (1994) partly illustrated theseeffects when they showed that extinction proce-dures mismatched to function had no effect onproblem behavior. In addition, Iwata et al.(1994) noted the possibility that certain irrele-vant extinction procedures may be con-traindicated for certain functions. Furthermore,Thompson, Fisher, Piazza, and Kuhn (1998)identified two different functions for differenttopographical sets of aggression and subse-quently showed that distinct treatments wererequired for each topographical set.The general necessity of separately analyzing

each topography of problem behavior is, how-ever, not strongly supported. Thompson et al.(1998) is the only study showing the benefitsof possibly doing so, and it should be noted

that the authors were able to glean the differentcontrolling variables in the initial FA in whichall topographies were included. In addition,functionally analyzing each topography of prob-lem behavior probably requires a prohibitiveamount of time and resources for most behav-ior analysts, who most often report not con-ducting any FAs of problem behavior (Oliver,Pratt, & Normand, 2015; Roscoe, Phillips,Kelly, Farber, & Dube, 2015).The alternative to conducting separate FAs

for each topography of problem behavior is toprogram putative reinforcers for multiple topog-raphies of problem behavior in a single FA(e.g., Hagopian, Bruzek, Bowman, & Jennett,2007; Hanley, Jin, Vanselow, & Hanratty,2014; Lalli, Mace, Wohn, & Livezey, 1995;Mace, Page, Ivancic, & O’Brien, 1986;Magee & Ellis, 2000; Slaton, Hanley, &Raftery, 2017). Including multiple topographiesin single analyses has been fairly common prac-tice in recent years. A review of FAs by Beavers,Iwata, and Lerman (2013) found that 76% ofreviewed FA studies included multiple topogra-phies of problem behavior. Focusing on analysesin which two or more reinforcing contingencieswere arranged in a single test condition, Slatonand Hanley (2018) reported that 81% of syn-thesized contingency analyses included multipletopographies of problem behavior. Researchershave used multiple ways to communicate thedifferences of including one or a few versusmany topographies of problem behavior in thecontingency class in a FA: single versus multipleresponse topographies (Beavers & Iwata, 2011),dangerous problem behavior versus dangerousand nondangerous problem behavior (Jessel,Hanley, & Ghaemmaghami, in press), and iso-lated contingency class versus synthesized con-tingency class (Slaton & Hanley, 2018).1 In an

1Throughout the paper, the terms open- and closed-contingency class indicate the responses available for rein-forcement during the functional analysis. These terms areto be distinguished from open and closed economies whichrefer to operant contexts in which appetitive stimuli are

CHRISTINE A. WARNER et al.2

attempt to depict the continuum on whichresponse topographies are synthesized, we willuse the term open-contingency class to refer toinstances in which all co-occurring topographieswere available for reinforcement and relativelyclosed-contingency class to refer to instances inwhich one to several topographies were availablefor reinforcement during the FA.When researchers have, for various reasons

and with various methods, analyzed whethermultiple topographies of problem behaviordescribed by caregivers are members of the sameresponse class, the answer is overwhelmingly yes(Borlase, Vladescu, Kisamore, Reeve, & Fetzer,2017; Borrero & Borrero, 2008; DeRosa,Roane, Doyle, & McCarthy, 2013; Fritz, Iwata,Hammond, & Bloom, 2013; Harding et al.,2001; Herscovitch, Roscoe, Libby, Bourret, &Ahearn, 2009; Lalli et al., 1995; Langdon,Carr, & Owen-DeSchryver, 2008; Lieving,Hagopian, Long, & O’Connor, 2004; Magee &Ellis, 2000; Richman, Wacker, Asmus, Casey, &Andelman, 1999; Smith & Churchill, 2002).For example, several studies have functionallyanalyzed precursors to problem behavior andresults showed that precursors (i.e., behaviorsreported to precede severe problem behavior)were members of the same response class as themore severe forms of problem behavior (Borlaseet al., 2017; Borrero & Borrero, 2008; Fritzet al., 2013; Herscovitch et al., 2009; Langdonet al., 2008; Smith & Churchill, 2002). Similaroutcomes are evident across several studies thatapplied extinction to analyze potential responseclass hierarchies; they each found that all topog-raphies of problem behavior were members ofthe same response class (DeRosa et al., 2013;Harding et al., 2001; Lalli et al., 1995; Lieving,et al., 2004; Richman et al., 1999). Relying onprocedures from Lalli et al. (1995), Magee and

Ellis (2000) initially implemented FAs with twoparticipants with a relatively open-contingencyclass, and progressively closed the contingencyclass until extinction was applied to andachieved with all topographies of problembehavior. For both participants, all target behav-iors appeared to be sensitive to the same rein-forcement contingency as new topographieswere evoked when extinction was applied to theforms of behavior evincing the highest rate.All studies that have evaluated response class

membership have included isolated reinforce-ment contingencies. Considering that hundredsof synthesized contingency analyses have beenpublished (Jessel, Ingvarsson, Metras, Kirk, &Whipple, 2018; Slaton & Hanley, 2018) andthese analyses usually arrange for reinforcementfor multiple topographies of problem behavior,it seems important to evaluate response classmembership following analyses relying on syn-thesized reinforcement contingencies. Themethod we used in the current study to makeinferences regarding response class membershipin synthesized contingency analyses was largelybased on the procedures of Magee and Ellis(2000) in that procedural extinction was pro-gressively applied to evaluate whether problembehaviors reported to co-occur2 were sensitiveto the same reinforcement contingency. Weevaluated the effects of relatively open- andclosed-contingency classes on rates of individualtopographies and total problem behavior in10 consecutively enrolled study participants, allchildren who were reported to engage in multi-ple topographies of problem behavior. Resultsare reported for all children who experiencedsome form of contingency class manipulation.Therefore, the study conforms to a consecutivecontrolled case series design (CCCSD;Hagopian, Rooker, Jessel, & DeLeon, 2013;Jessel et al., 2018; Slaton et al., 2017). These

2Throughout the paper, co-occurring behavior refers totopographies of problem behavior and precursor behaviorthat were reported to occur in the same time and context.

available as reinforcers within a session and also indepen-dent of behavior outside of session, and in which appeti-tive stimuli are available as reinforcers only with in asession, respectively (Catania, 2007).

3EXTINCTION ANALYSES

methods circumvent the file drawer problem(Rosenthal, 1979; Scargle, 2000) that is inevita-ble in research relying on single-subject experi-mental designs in which positive results(i.e., functional relations) are published andnegative results are not. The generality of func-tional relations between independent anddependent variables can be established via sys-tematic replications (Sidman, 1960), but due tothe inevitable publication bias toward positiveoutcomes, the probability that demonstratedrelations will hold under different conditions isunclear, even when many systematic replicationsare published. By incorporating a CCCSD intothis study, results will be less affected by this tra-ditional bias and may instead convey both thepossibility and probability of the demonstratedfunctional relations. In particular, the probabil-ity that multiple topographies of problembehavior that are reported to co-occur are sensi-tive to the same synthesized reinforcement con-tingency may be revealed in this study.

METHODS

ParticipantsParticipants were consecutively enrolled in the

study across three sites (see Table 1 for participantcharacteristics). Participants would have beenexcluded if (i) caregivers reported only one topog-raphy of problem behavior, (ii) all reported topog-raphies were observed and reinforced in the initialInterview-informed synthesized contingency anal-ysis (IISCA) (Hanley et al., 2014; Jessel,Hanley, & Ghaemmaghami, 2016; Jessel et al.,2018), or (iii) guardian consent and participantassent for the process was not obtained. No sub-ject was excluded based on any of these criteria.Participants included 10 children who engaged inmultiple topographies of problem behavior thatwere reported by caregivers to co-occur.3 The

participants included two females and eight malesbetween the ages of 3 and 18 years. Six were diag-nosed with autism, one was diagnosed withautism and attention-deficit hyperactivity disor-der, one was diagnosed with hydrocephalus, con-duct disorder, and destructive behavior disorder,and two participants did not have any formaldiagnoses. All participants engaged in at least onetopography of severe problem behavior, mostengaged in multiple forms, and most werereported to also engage in some less dangerousforms of problem behavior.

Settings and PersonnelSessions were conducted in three settings.

Sessions with Cole, Raj, Ali, and Caleb wereconducted in session rooms that ranged between3 m × 3 m to 3.5 m × 4 m at a specializedschool for children with autism. Sessions withBill were conducted at a different specializedschool for children with autism. Sessions withMilly, Luke, Jordan, Annie, and Mike were con-ducted in a 4 m × 3 m session room at a uni-versity outpatient clinic. Across all locations,session rooms were equipped with video record-ing equipment, a table and one or two chairs,and participant-specific materials (e.g., toys,activities, electronics, teaching materials).Two analysts were Board Certified Behavior

Analyst Doctorates (BCBA-D), four were BoardCertified Behavior Analysts (BCBAs), and onewas enrolled in behavior analytic coursework tocomplete BCBA certification requirements atthe time of the study. All analysts had con-ducted IISCAs prior to this study. Supervisionwas provided by the second author who was aBCBA-D and experienced in conducting FAs.

Response MeasurementResponses were scored using a laptop com-

puter with a data collection and analysis pro-gram. For each participant, the duration ofeach reinforcement interval and the frequencyof each topography of problem behavior were

3Caregiver interview information and functional analy-sis data for Milly and Luke were previously published inGhaemmaghami, Hanley, Jessel, and Landa (2018).

CHRISTINE A. WARNER et al.4

scored. Response topographies for each partici-pant are shown in Table 2 and are definedindividually for each participant. Specificresponse definitions are available upon request.

The reinforcement interval began when puta-tive reinforcers were delivered to the participantand ended when those reinforcers wereremoved. Session counts of problem behaviorwere converted into a session rate for all ana-lyses. The duration of emotional responding(defined as crying or wailing), which was scoredfor Raj, Jordan, and Mike, was converted intopercentage of session by dividing the durationof the response by the total session duration.

Interobserver AgreementA second observer independently scored at

least 20% of sessions across participants in eachcondition of the analysis for each participant(range: 20–66%) to evaluate interobserveragreement (IOA). Agreement was calculated bydividing sessions into 10-s intervals and divid-ing the number of agreements per interval bythe number of disagreements plus agreementsper interval and multiplying by 100. MeanIOA across participants was 97% (sessionrange: 77–100%).

Experimental DesignsA multielement design was used to compare

rates of problem behavior in the test and con-trol condition during the analysis. A multiple

Table 1Participant Characteristics

Participant Age Sex Program type Diagnosis Communication Mode

Cole 8 M Day program at specialized school Autism Full sentencesRaj 5 M Day program at specialized school Autism Vocal approximations, signs, and

SGDMilly 10 F University-based outpatient clinic Autism Nonvocal, gestures, a few phrasesBill 15 M Day program at specialized school Autism Short, disfluent sentencesCaleb 18 M Residential program at specialized

schoolAutism Vocal approximations, gestures, and

SGDAli 12 M Day program at specialized school Autism Signs and gesturesLuke 5 M University-based outpatient clinic Autism and ADHD Full sentencesJordan 3 M University-based outpatient clinic None Full sentencesAnnie 4 F University-based outpatient clinic Hydrocephalus, conduct disorder,

destructive behavior disorderShort, disfluent sentences

Mike 4 M University-based outpatient clinic None Full sentences

Note. ADHD = attention deficit hyperactivity disorder; SGD = speech generating device.

Table 2Topographies of Problem Behavior

Participant Problem behaviors reported to co-occur

Cole Aggression*, disruption, disruptive vocalizations(screaming, name calling, threatening, swearing),SIB, whining

Raj Aggression, disruption, face and ear covering,flopping, SIB*, stomping

Milly Aggression, chin-grinding*, disruptive vocalizations(screaming and yelling), flopping, SIB, swipingitems

Bill Complaining, disruption*

Caleb Aggression, bolting, face covering, screaming, SIB*

Ali Aggression*, body SIB, disruption, head SIBLuke Aggression, disruption, disruptive sounds (signing,

hissing, growling), disruptive vocalizations(screaming or yelling), throwing*

Jordan Aggression*, disruption, disruptive vocalizations(screaming and yelling), throwing items

Annie Disruption, disruptive vocalizations (screaming,yelling, swearing), hitting and pushing,scratching*, throwing items

Mike Aggression*, disruption, disruptive vocalizations(screaming and yelling), flopping, vocal protests(whining or complaining)

Note. Problem behaviors are listed in alphabetical order.SIB = self-injurious behavior. The asterisk indicates themost concerning topography of problem behavior for eachparticipant as indicated by the caregiver interview.

5EXTINCTION ANALYSES

baseline design across topographies and reversaldesign were used during the extinction analyses.Experimental control of a topography of prob-lem behavior was demonstrated by a change inthe level of responding when other members ofthe contingency class were added or removed.

Participant ProtectionsSeveral measures were taken to protect the

participants. All procedures were reviewed byan Institutional Review Board. Prior to con-ducting any analysis, conservative session termi-nation criteria were determined by the leadauthor and the supervising clinician of all chil-dren who participated in the school. These ter-mination criteria were never met. At least onecaregiver was present for all sessions conductedin the outpatient clinic and at least one teacherwas present for all sessions conducted in theschool. The caregiver or teacher was informedthat they could terminate the analysis at anypoint if they were uncomfortable with proceed-ing for any reason. Analysts and data collectorswere provided with the same authority.

Caregiver Interview and FAAn open-ended interview was conducted

with one or two caregivers to identify potentialvariables contributing to the maintenance ofproblem behavior. The interview questionswere the same as those in the appendix of Han-ley (2012) and additional questions were askedfor clarification as necessary. The open-endedinterview lasted 30–45 min and included ques-tions about the participant’s interests, languageabilities, situations in which problem behavioroccurred, and ways in which caregiversresponded to the child’s problem behavior. Theinterview included questions to determine thespecific forms of problem behavior, whichtopographies of problem behaviors were consid-ered a priority for treatment, which tended tocluster together, and whether the child emittedany precursors to the more dangerous forms.

The interview also included a question to deter-mine whether any of the reported co-occurringforms were suspected to be a form of self-stimu-lation. If a potentially automatically maintainedform of behavior was reported (e.g., stereotypicvocalizations), this form was not included in theanalysis. A brief observation which lasted15 min was also conducted for the childrenserved at the school during which the analystobserved the child’s interactions with peopleand materials and ways in which the teachersinteracted with the child.An IISCA was conducted with each partici-

pant. Information about the child’s problembehavior from the interview (and observationfor some), informed the design of distinct FAsfor each child (see Table 3 for the general andspecific synthesized contingencies evaluated foreach participant). All reported topographies ofproblem behavior were measured throughouteach analysis and reinforced in the test sessionsof the analysis. Sessions alternated between asingle test and control condition. In the controlcondition, which was conducted first, all puta-tive reinforcers were available continuously andnoncontingently throughout the session. In anyanalysis in which adult compliance with mandswas a putative reinforcer, mands were honoredon a continuous reinforcement schedule. In thetest condition, all putative reinforcers wereremoved simultaneously; contingent on anyinstance of problem behavior, all putative rein-forcers were simultaneously delivered for 30 s.Sessions had a duration of 5 min for each par-ticipant, except Bill whose sessions were 3 mindue to efficient responding in the analysis andCaleb whose sessions were 10 min due to initiallow rates of responding. Additional informationon how analyses were designed from interviewsis available Appendix S1.

Extinction AnalysesAfter the initial FA, an extinction analysis

was conducted during which extinction

CHRISTINE A. WARNER et al.6

procedures were progressively applied to thehighest probability topographies of problembehavior observed during the IISCA while allother topographies remained available for rein-forcement. Three versions of the extinctionanalysis were conducted. During the progres-sive extinction analysis (Cole, Raj, Milly, Bill,Caleb), the analyst applied extinction proce-dures to the most probable topography of prob-lem behavior from the IISCA and continuedthe test sessions until another topographyemerged and persisted for at least two consecu-tive sessions. Extinction procedures were thenapplied to the topography that emerged, inaddition to the topography that was previouslyplaced on extinction. This process continueduntil either all reported topographies of prob-lem behavior emerged and persisted, or foursessions occurred without the emergence of anyreported topographies of problem behavior.During the expedited extinction analysis (Ali,Luke, Jordan, Annie), extinction procedureswere applied to all topographies of problem

behavior following the initial analysis, and onlythe most concerning topography of problembehavior was reinforced. The relatively closed-contingency class continued until the mostconcerning form of problem behavior emergedand persisted or four sessions occurred withoutthe emergence of the most concerning form ofproblem behavior (this did not occur for anyparticipant). The procedures for the briefextinction analysis (Mike) were identical to theexpedited extinction analysis, except that alltopographies of problem behavior were placedon extinction and only the most concerningtopography was reinforced for a single session.Across all three versions of the extinction analy-sis, the contingency was reopened such that alltopographies of problem behavior werereinforced in the final phase (as they had beenin the initial analysis). Although the proceduresare described as extinction analyses, it is impor-tant to note that the focus was on the emer-gence of other topographies of problembehavior as procedural extinction was applied

Table 3General and Specific Contingencies for all Participants

Participant General contingency Specific contingency

Cole Escape to tangibles and mand compliance Escape from adult instruction to engage in math, reading, writing,and problem-solving tasks to iPad and toys, and compliance withmands (e.g., to change positions, color in specific ways, or watchvideos together)

Raj Escape to tangibles, attention, and mandcompliance

Escape from adult instruction to engage in sorting, identifyingattributes, and stretching tasks to toys, attention (hugs, tickles,making faces), and compliance with mands to play in specific ways

Milly Access to tangibles and mand compliance Access to tablets, playdoh, soft toys, and mand complianceBill Escape to tangibles, attention, and stereotypy Escape from vocational tasks to figurines, conversation about

preferred topics, and stereotypyCaleb Escape to tangibles and attention Escape from interruption of ongoing activity by adult instructions to

iPad, puzzles, and books, and attention (conversation, praise, headrubs)

Ali Escape to mand compliance Escape from life skills tasks (brushing hair and teeth) to compliancewith mands for walks, ribbons, and snacks

Annie Access to tangibles, attention, and mandcompliance

Access to tangibles, attention from analyst and Grandma, and mandcompliance to play

Luke Escape to tangibles, attention, and mandcompliance

Escape from adult instruction to tangibles, attention, and mandcompliance to play in specific ways

Jordan Escape to tangibles, attention, and mandcompliance

Escape from adult instruction (put on coat, play with different toys,clean up) to toys, attention, and mand compliance to providespecific forms of attention

Mike Escape to tangibles and attention Escape from cleaning up toys to Legos and action figures andattention

7EXTINCTION ANALYSES

across topographies, as opposed to achievingextinction of any particular response form.Concerns for participant and analyst safety andefficiency were factors in this decision.

Contingency Strength AnalysisTo determine whether the persistence of

Cole’s inappropriate vocalizations during theextinction procedures was due to the responsesplaced on extinction adventitiously co-occurring with reinforcement deliveries forother responses, contingency strength valueswere calculated for Cole’s disruptive vocaliza-tions based on the procedures by Luczynskiand Hanley (2009). The measure of contin-gency strength included two relations betweentarget responses placed on extinction and rein-forcement delivery. A response conditional proba-bility was calculated as the number of times thetarget response occurred within 10 s of rein-forcement delivery, divided by the total num-ber of target responses. An event conditionalprobability was also calculated by measuring thenumber of times reinforcement deliveries werenot preceded by the target response within10 s, divided by the total number of reinforce-ment deliveries. The event conditional proba-bility was subtracted from the responseconditional probability, producing a value ofcontingency strength between −1 and 1. Theresponse conditional probability and event con-ditional probability for the target response werecalculated for each session during the applica-tion of extinction only.

RESULTS

Differential responding between the test andcontrol condition in the FA was obtained forall participants (see Figure 1). For four of thefive participants for whom the progressiveextinction analysis was conducted, topographiesof problem behavior reported to co-occur inthe caregiver interview appeared to be evoked

and maintained by the same contingencies (seeFigures 2–5).The progressive extinction analysis for Bill

showed that complaining and disruption wereevoked and maintained by the same synthesizedcontingency (Figure 5). Although aggressionwas not reported in the interview and was notreinforced, aggression emerged whencomplaining was placed on extinction, and ret-urned to zero in the third phase. It is unclearwhether aggression was induced by the applica-tion of extinction of a previously reinforcedresponse (complaining) or we observed theresurgence of aggression.The results of Caleb’s extinction analysis

were ambiguous (Figure 6). For Caleb, scream-ing, which was observed at the highest rate dur-ing the IISCA, decreased to zero during theextinction analysis and no other forms of prob-lem emerged, therefore we were unable todetermine whether the less-concerning topogra-phies of problem behavior were members ofthe same response class as the most concerningtopographies (aggression and SIB). The resultssuggest that either (i) extinction occurred andhad a generalized effect or (ii) the responsesreported to co-occur were not in the sameresponse class as screaming (for Caleb only).One might expect that if the responses werenot in the same response class, screamingwould have occurred again in the third phasewhen it was scheduled for reinforcement. Thefact that screaming remained at or near zerowhen it was available for reinforcement pro-vides some evidence that extinction occurred;however, without further investigation, Caleb’sresults are difficult to interpret.For all participants for whom an expedited

extinction analysis was conducted (Figures 7–9), the most concerning topography wasevoked and maintained by the same contingen-cies as the topography or topographies observedduring the analysis. For Mike (Figure 10), whoparticipated in the brief extinction analysis, themost concerning topography appeared to be

CHRISTINE A. WARNER et al.8

Figure 1. IISCAs across the 10 consecutive participants.

9EXTINCTION ANALYSES

Figure 2. Progressive extinction analysis data for Cole.The top three most concerning behaviors are indicated bythe numbers 1–3 on the y-axis labels, with 1 being themost concerning topography. SIB = self-injuriousbehavior.

Figure 3. Progressive extinction analysis for Raj. Thetop three most concerning behaviors are indicated by thenumbers 1–3 on the y-axis labels, with 1 being the mostconcerning topography. SIB = self-injurious behavior.

CHRISTINE A. WARNER et al.10

evoked and maintained by the same contingen-cies as the topography observed during the FA.Within-session data depictions were used to

resolve some interpretive ambiguity. Forinstance, Cole’s extinction analysis shows an ini-tial increase in whining, disruption, and aggres-sion each time a new topography was placed onextinction, followed by a decrease in thoseresponses. Disruptive vocalizations, however,persisted despite no longer being programmedfor reinforcement. To evaluate why disruptive

Figure 4. Progressive extinction analysis for Milly.The top three most concerning behaviors are indicated bythe numbers 1–3 on the y-axis labels, with 1 being themost concerning topography.

Figure 5. Progressive extinction analysis for Bill. Themost concerning behaviors are indicated by the numberson the y-axis labels, with 1 being the most concerningtopography (aggression was not ranked because it was notreported in the interview).

11EXTINCTION ANALYSES

Figure 6. Progressive extinction analysis for Caleb.The most concerning topography of problem behavior isindicated by the number 1. SIB = self-injurious behavior.

Figure 7. Expedited extinction analysis for Ali. Thetwo most concerning behaviors are indicated by the num-bers on the y-axis labels, with 1 being the most con-cerning topography. SIB = self-injurious behavior.

CHRISTINE A. WARNER et al.12

Figure 8. Expedited extinction analyses for Luke and Jordan. The most concerning topography of problem behaviorfor each participant is indicated by the number 1.

13EXTINCTION ANALYSES

Figure 9. Expedited extinction analysis for Annie.The most concerning topography of problem behavior isindicated by the number 1.

Figure 10. Expedited extinction analysis for Mike.The most concerning topography of problem behavior isindicated by the number 1.

CHRISTINE A. WARNER et al.14

vocalizations persisted under extinction, we con-ducted a second-by-second analysis (based onthe within-session analysis described by Stocco,Thompson, & Rodriguez, 2011) across all ses-sions in which extinction was applied to disrup-tive vocalizations. Figures 11 and 12 show thatalthough reinforcement was not programmedfor disruptive vocalizations in sessions 10–19,reinforcement was contiguous with theseresponses. Figures 11 and 12 also depict thatdisruptive vocalizations were evoked immedi-ately following reinforcement withdrawal anddid not occur during reinforcement. In addi-tion, we conducted a contingency strength anal-ysis for all sessions in which extinction wasapplied to disruptive vocalizations. An increas-ing contingency strength across extinction

phases is evident. Collectively, these results sup-port the notion that disruptive vocalizationswere maintained by the same reinforcementcontingency as that influencing SIB and aggres-sion, even though disruptive vocalizations didnot extinguish when the programed reinforce-ment was discontinued.Because we placed all of Mike’s responses

except for aggression on extinction for a singlesession only (Figure 13), it was unclear whetheraggression was controlled by the synthesizedcontingency or was induced by extinction. Toaddress this question, we conducted a second-by second analysis for aggression and disruptivevocalizations (the topography observed at thehighest rate when the contingency class wasopen) for sessions 3–5. Figure 13 shows that

Figure 11. Second-by-second within-session data for Cole depicting SIB, aggression, disruptive vocalizations (Disrvocs), and reinforcement intervals when reinforcement was programmed for aggression and self-injurious behavior (SIB).The contingency strength values for disruptive vocalizations are noted by CS values for each session. The question beingaddressed here is why disruptive vocalizations persisted when placed on extinction in sessions 10–12 for Cole onFigure 2.

15EXTINCTION ANALYSES

Figure 12. Second-by-second within-session data for Cole depicting SIB, aggression, disruptive vocalizations (Disrvocs), and reinforcement intervals when reinforcement was programmed for only self-injurious behavior (SIB). The con-tingency strength values for disruptive vocalizations are noted by CS values for each session. The question being addressedhere is why disruptive vocalizations persisted when placed on extinction in sessions 13–19 for Cole on Figure 2.

CHRISTINE A. WARNER et al.16

aggression did not occur in sessions 3 and5 when all responses were available for rein-forcement in sessions 3 and 5. During session4, aggression was evoked following reinforce-ment withdrawal and did not occur duringreinforcement. The results of the within-sessionanalysis and the fact that Mike’s parentsreported aggression to co-occur with the otherresponses, support the notion that aggressionwas maintained by the same reinforcement con-tingency as that influencing Mike’s less-concerning topographies.The question being addressed in Figure 14 is

whether open- and closed-contingency classesaffect total problem behavior and emotionalresponding. Figure 14 displays total problembehavior across sessions for all 10 participants.Breaks in the data path within a phase (seepanels for Milly, Cole, and Raj) indicate thepoint at which extinction procedures wereapplied to another topography. For five

participants (Mike, Jordan, Ali, Luke, andMilly), there was an immediate level increase intotal problem behavior as some responses wereremoved from the contingency class. For twoparticipants (Cole and Raj), a gradual levelincrease in total problem behavior was observedas more responses were removed from the rein-forcement contingency class. For Bill andAnnie, no change in the level of total problembehavior was observed as responses wereremoved from the contingency class. For Caleb,there was a gradual decrease in the level of totalproblem behavior as extinction was applied toone topography. We also measured emotionalresponding for three participants (Raj, Jordan,and Mike) and found that for all three partici-pants, levels of emotional responding were highwhen the contingency class was relativelyclosed. It is unclear whether the emotionalresponding observed was a side effect of extinc-tion or if the responses were members of the

Figure 13. Second-by-second within-session data for Mike depicting aggression, disruptive vocalizations, and rein-forcement intervals in that order on each panel. Topographies available for reinforcement are in bold. This within-session analysis shows that aggression did not occur in sessions 3 and 5 when the contingency class was open in sessions3 and 5. During session 4, aggression was evoked following reinforcement withdrawal and did not occur duringreinforcement.

17EXTINCTION ANALYSES

Figure 14. Total problem behavior during the extinction analyses across all participants and the percentage of ses-sion with emotional responding during the extinction analyses for Mike, Jordan, and Raj.

CHRISTINE A. WARNER et al.18

same response class as the other forms of prob-lem behavior. For all participants except Caleb,the forms of problem behavior were moresevere when the reinforcement contingency wasrelatively closed.

DISCUSSION

Systematic extinction procedures revealedthat the most concerning topography was sensi-tive to the same synthesized reinforcement con-tingency as other topographies of problembehavior that were reported to co-occur bycaregivers in nine of 10 participants (one out-come was ambiguous). By extension, whencaregivers reported that the child engaged indifferent forms of problem behavior in similarsituations, these different forms were likely tobe the members of the same response class.Considering the consistency of results obtainedwithin a CCCSD, in which any negative oranomalous effects were reported, our resultssuggest that control of at least one topographyof problem behavior in a single FA is probablysufficient for proceeding to the evaluation of afunction-based treatment for children whoengage in multiple topographies of problembehavior reported to co-occur in similarsituations.Because the study was conducted across

three sites with male and female children ofvarying ages, communication skills, and diagno-ses who exhibited multiple topographies ofproblem behavior, the notion that problembehaviors reported to co-occur in similar situa-tions are members of the same response classappears to have strong generality. This notionis further supported by the fact that theseresults are consistent with 12 of 13 other stud-ies that have analyzed response class member-ship of multiple topographies of problembehavior (Borlase et al., 2017; Borrero &Borrero, 2008; DeRosa et al., 2013; Fritz et al.,2013; Harding et al., 2001; Herscovitch et al.,2009; Lalli et al., 1995; Langdon et al., 2008;

Lieving et al., 2004; Magee & Ellis, 2000;Richman et al., 1999; Smith & Churchill,2002; Thompson et al., 1998).The current study included synthesized

response topographies in addition to synthe-sized EOs and reinforcers, representing anextension of response class membershipresearch. This extension to the synthesizedreinforcement literature is important as webegin to understand the prevalence of synthesisin the FA and treatment literature. In theirreview of synthesis in analyses, Slaton andHanley (2018) found that at least 120 synthe-sized contingency analyses had been publishedas of January 2018 and that open-contingencyclasses were highly prevalent, in that 81% ofsynthesized contingency analyses included anopen-contingency class. One must acknowledgethe possibility that different responses may becontrolled by different reinforcers when anopen-contingency class is programmed, andreinforcers are synthesized. However, given thatthe results of the current CCCSD are consis-tent with the majority of studies that have eval-uated response class membership using isolatedreinforcers (Borlase et al., 2017; Borrero &Borrero, 2008; DeRosa et al. 2013; Fritz,et al., 2013; Harding et al., 2001; Herscovitch,et al., 2009; Lalli, et al., 1995; Langdon et al.,2008; Lieving et al., 2004; Magee & Ellis,2000; Richman et al., 1999; Smith &Churchill, 2002), it seems unlikely that the useof synthesized reinforcers moderated the highlikelihood of response class membership.The validity of conclusions from synthesized

reinforcement contingencies with an open-contingency class is also supported by recenttreatment effectiveness studies. Beaulieu,Clausen, Williams, and Herscovitch (2018),Rose and Beaulieu (2019), Hanley et al. (2014),Herman, Healy, and Lydon (2018), Jessel et al.(2018), Taylor, Phillips, and Gerzog (2018),Santiago, Hanley, Moore, and Jin (2016), andStrand and Eldevik (2017) all described sociallyvalidated outcomes of function-based treatments

19EXTINCTION ANALYSES

designed from single-test analyses that arrangedinterview-informed synthesized reinforcementfor open-contingency classes. Treatments fromthese studies included practical schedules of rein-forcement and were implemented by caregiversin relevant contexts by the close of each study.These same outcomes have not, to our knowl-edge, been reported from approaches involvingmultiple analyses distinguished by topography.Although we referred to them as extinction

analyses, we did not apply extinction forenough sessions to allow responses to extin-guish. Rather, our focus was the emergence ofother forms of problem behavior as the proce-dure of extinction was sequentially applied totopographies of problem behavior. Safety andefficiency were key factors in this decision. Inthis way, Catania’s (2007) definition ofresponse class membership was expanded toinclude evocation and maintenance of aresponse by the same contingency. The failureto extinguish responses should, however, beconsidered when concluding that the resultsdemonstrated that the most concerning topog-raphy showed sensitivity to the same synthe-sized contingency as the less concerningtopographies for nine of 10 participants.It is plausible that some responses that did

not extinguish were automatically reinforced.Including automatically maintained problembehavior in the programmed reinforcementcontingency is a concern because it may resultin an undifferentiated analysis (e.g., Derbyet al., 1994; Thompson et al., 1998). Weattempted to minimize the possibility of out-comes being affected by automaticallyreinforced behavior by relying on the questionin the interview to distinguish potentially self-stimulatory responses from socially mediatedresponses, analyzing each topography separately(Derby et al. 1994), and conducting within-session analyses when a possible pattern of self-stimulatory behavior was detected (Figure 3).Although complete extinction was rarely

observed in our analyses (e.g., Raj’s covering

face and ears), indirect effects of extinction(Lerman & Iwata, 1996) were evident. Weobserved increased variability in responding asextinction was applied (e.g., Cole’s whining,disruption, disruptive vocalizations) consistentwith research on variability produced by extinc-tion (Lattal, St. Peter, & Escobar, 2013). Wealso observed patterns (e.g., Cole’s whining anddisruption) seemingly indicative of resurgence(Bruzek, Thompson, & Peters, 2009; Lattalet al., 2013). It is challenging to distinguishbetween extinction-induced variability andresurgence in our analyses. However, becauseall responses were reported by caregivers, pre-sumably the participants had a history of rein-forcement for all responses making it morelikely that the increases observed were examplesof resurgence.We observed additional patterns of

responding that have important implicationsfor practitioners. Higher rates of total problembehavior (Figure 10) and increases in the mostconcerning topography were typically observedwhen the reinforcement contingency was rela-tively closed. The practical implication is thatan open-contingency class seems optimal giventhat an understanding of the most concerningtopography may be evident without observinghigh rates of problem behavior or the mostconcerning topography in the analysis.We also found that closed-contingency clas-

ses were more likely to yield emotionalresponding (Figure 13) and rarely-witnessedaggression (Figure 6). Regardless of whetheremotional responding was categorized as a sideeffect or as another member of the responseclass and whether aggression was induced oranother member of the response class with ahistory of reinforcement, the implication is thatopen-contingency classes allow for control ofproblem behavior without responding that maysocially invalidate the analysis results.Ali was the only participant for whom we con-

sistently observed the most concerning topogra-phy (aggression) during the FA. However, we also

CHRISTINE A. WARNER et al.20

observed unreinforced responses that producedlonger latencies to the target behavior (see bottompanel of Figure 7). Responses that fall above thethreshold but below criterion have been referredto as subcriterion responses (Pinkston & Libman,2017). The implication for applied work is thatrates of target behaviors may often be an artifactof the criterion. Ali’s analysis demonstrates oneway in which there is an unavoidable degree ofspeculation involved with the selection of targetresponses.Extinction analyses are an interpretation safe-

guard when only one of multiple topographiesof problem behavior is observed during anFA. However, given the results of the currentand previous studies show that topographies ofproblem behavior that are reported to co-occurare most often members of the same responseclass (Borlase et al., 2017; Borrero & Borrero,2008; DeRosa et al., 2013; Fritz et al., 2013;Harding et al., 2001; Herscovitch et al., 2009;Lalli et al., 1995; Langdon et al., 2008; Lievinget al., 2004; Magee & Ellis, 2000; Richmanet al. 1999; Smith & Churchill, 2002), practi-tioners need not be compelled to conductextinction analyses and should instead considermaking the inference that the reinforcementcontingency controlling the observed topogra-phy is the same as that controlling other topog-raphies that interviewees report co-occur insimilar situations. This recommendation to relyon inference is grounded in the empirical evi-dence that an inferential error is not likely andoccasioned by concerns for client and analystsafety when using closed-contingency classes.In the case that caregivers request verification

of response class membership of the most con-cerning topography prior to moving to treat-ment because it was not observed during theFA, we suggest that the brief extinction analysisbe implemented. Furthermore, the expeditedextinction analysis should probably only beused only when results of the brief extinctionanalysis are not clear. Rather than conductingextinction analyses, response to treatment also

may be used as verification of response classmembership. That is, practitioners can measureforms of problem behavior separately during theFA and treatment to affirm the response classinterpretation (e.g., Dracobly & Smith, 2012;Fritz et al. 2013; Hoffman, Sellers, Halversen, &Bloom, 2018; Najdowski, Wallace, Ellsworth,MacAleese, & Cleveland 2008). An inferentialerror would be revealed in treatment if a topog-raphy persists despite treatment effects for othertopographies.

REFERENCES

Beaulieu, L., Clausen, M., Williams, A., &Herscovitch, B. (2018). Incorporating interview-informed functional analyses into practice. BehaviorAnalysis in Practice, 11, 385–389. https://doi.org/10.1007/s40617-018-0247-7.

Beavers, G. A., & Iwata, B. A. (2011). Prevalence of mul-tiply controlled problem behavior. Journal of AppliedBehavior Analysis, 44, 593–597. https://doi.org/10.1901/jaba.2011.44-593.

Beavers, G. A., Iwata, B. A., & Lerman, D. C. (2013).Thirty years of research on the functional analysis ofproblem behavior. Journal of Applied Behavior Analy-sis, 46, 1–21. https://doi.org/10.1002/jaba.30.

Borlase, M. A., Vladescu, J. C., Kisamore, A. N.,Reeve, S. A., & Fetzer, J. L. (2017). Analysis of pre-cursors to multiply controlled problem behavior: Areplication. Journal of Applied Behavior Analysis, 50,668–674. https://doi.org/10.1002/jaba.398.

Borrero, C. S. W., & Borrero, J. C. (2008). Descriptiveand experimental analyses of potential precursors toproblem behavior. Journal of Applied Behavior Analy-sis, 41, 83–96. https://doi.org/10.1901/jaba.2008.41-83.

Bruzek, J. L., Thompson, R. H., & Peters, L. C. (2009).Resurgence of infant caregiving responses. Journal ofthe Experimental Analysis of Behavior, 92, 327–343.https://doi.org/10.1901/jeab.2009-92-327.

Catania, A. C. (2007). Learning. Cornwall-on-Hudson,NY: Sloan Publishing.

Derby, K. M., Wacker, D. P., Peck, S., Sasso, G.,DeRaad, A., Berg, W., … Ulrich, S. (1994). Func-tional analysis of separate topographies of aberrantbehavior. Journal of Applied Behavior Analysis, 27,267–278. https://doi.org/10.1901/jaba.1994.27-267.

DeRosa, N. M., Roane, H. S., Doyle, N. M., &McCarthy, C. A. (2013). Modification of responseclass hierarchies through differential reinforcement offunctionally equivalent communication. Journal ofDevelopmental and Physical Disabilities, 25, 119–133.https://doi.org/10.1007/s10882-012-9312-2.

21EXTINCTION ANALYSES

Dracobly, J. D., & Smith, R. G. (2012). Progressing fromidentification and functional analysis of precursorbehavior to treatment of self-injurious behavior. Jour-nal of Applied Behavior Analysis, 45, 361–374.https://doi.org/10.1901/jaba.2012.45-361.

Dworschak, W., Ratz, C., & Wagner, M. (2016). Preva-lence and putative risk markers of challenging behav-ior in students with intellectual disabilities. Researchin Developmental Disabilities, 58, 94–103. https://doi.org/10.1016/j.ridd.2016.08.006.

Emerson, E., & Bromley, J. (1995). The form and func-tion of challenging behaviours. Journal of IntellectualDisability Research, 39, 388–398. https://doi.org/10.1111/j.1365-2788.1995.tb00543.x.

Emerson, E., Kiernan, C., Alborz, A., Reeves, D.,Mason, H., Swarbrick, R., … Hatton, C. (2001).The prevalence of challenging behaviors: a total pop-ulation study. Research in Developmental Disabilities,22, 77–93. https://doi.org/10.1016/S0891-4222(00)00061-5.

Fritz, J. N., Iwata, B. A., Hammond, J. L., &Bloom, S. E. (2013). Experimental analysis of precur-sors to severe problem behavior. Journal of AppliedBehavior Analysis, 46, 101–129. https://doi.org/10.1002/jaba.27.

Ghaemmaghami, M., Hanley, G. P., Jessel, J., &Landa, R. (2018). Shaping complex functional com-munication responses. Journal of Applied BehaviorAnalysis, 51, 502–520. https://doi.org/10.1002/jaba.468.

Hagopian, L. P., Bruzek, J. L., Bowman, L. G., &Jennett, H. K. (2007). Assessment and treatment ofproblem behavior occasioned by interruption of free-operant behavior. Journal of Applied Behavior Analysis,40, 89–103. https://doi.org/10.1901/jaba.2007.63-05.

Hagopian, L. P., Rooker, G. W., Jessel, J., &DeLeon, I. G. (2013). Initial functional analysis out-comes and modifications in pursuit of differentiation:A summary of 176 inpatient cases. Journal of AppliedBehavior Analysis, 46, 88–100. https://doi.org/10.1002/jaba.25.

Hanley, G. P. (2012). Functional assessment of problembehavior: Dispelling myths, overcoming implementa-tion obstacles, and developing new lore. BehaviorAnalysis in Practice, 5, 54–72. https://doi.org/10.1007/BF03391818.

Hanley, G. P., Iwata, B. A., & McCord, B. E. (2003).Functional analysis of problem behavior: A review.Journal of Applied Behavior Analysis, 36, 147–185.https://doi.org/10.1901/jaba.2003.36-147.

Hanley, G. P., Jin, C. S., Vanselow, N. R., &Hanratty, L. A. (2014). Producing meaningfulimprovements in problem behavior of children withautism via synthesized analyses and treatments. Jour-nal of Applied Behavior Analysis, 47, 16–36. https://doi.org/10.1002/jaba.106.

Harding, J. W., Wacker, D. P., Berg, W. K., Barretto, A.,Winborn, L., & Gardner, A. (2001). Analysis ofresponse class hierarchies with attention-maintainedproblem behaviors. Journal of Applied Behavior Analy-sis, 34, 61–64. https://doi.org/10.1901/jaba.2001.34-61.

Herman, C., Healy, O., & Lydon, S. (2018). Aninterview-informed synthesized contingency analysisto inform the treatment of challenging behavior in ayoung child with autism. Developmental Neuro-rehabilitation, 21, 202–207. https://doi.org/10.1080/17518423.2018.1437839.

Herscovitch, B., Roscoe, E. M., Libby, M. E.,Bourret, J. C., & Ahearn, W. H. (2009). A proce-dure for identifying precursors to problem behavior.Journal of Applied Behavior Analysis, 42, 697–702.https://doi.org/10.1901/jaba.2009.42-697.

Hoffman, A. N., Sellers, T. P., Halversen, H., &Bloom, S. E. (2018). Implementation of interven-tions informed by precursor functional analyses withyoung children: A replication. Journal of AppliedBehavior Analysis, 51, 879–889. https://doi.org/10.1002/jaba.502.

Holden, B., & Gitlesen, J. P. (2003). Prevalence of psy-chiatric symptoms in adults with mental retardationand challenging behaviour. Research in DevelopmentalDisabilities, 24, 323–332. https://doi.org/10.1016/S0891-4222(03)00060-X.

Iwata, B. A., Pace, G. M., Cowdery, G. E., &Miltenberger, R. G. (1994). What makes extinctionwork: An analysis of procedural form and function.Journal of Applied Behavior Analysis, 27, 131–144.https://doi.org/10.1901/jaba.1994.27-131.

Jessel, J., Hanley, G. P., & Ghaemmaghami, M. (2016).Interview-informed synthesized contingency analyses:Thirty replications and reanalysis. Journal of AppliedBehavior Analysis, 49, 576–595. https://doi.org/10.1002/jaba.316.

Jessel, J., Hanley, G. P., & Ghaemmaghami, M.(in press). On the prevalence of the components ofstandard functional analyses in published research.Behavior Analysis in Practice.

Jessel, J., Ingvarsson, E. T., Metras, R., Kirk, H., &Whipple, R. (2018). Achieving socially significantreductions in problem behavior following theinterview-informed synthesized contingency analysis:A summary of 25 outpatient applications. Journal ofApplied Behavior Analysis, 51, 130–157. https://doi.org/10.1002/jaba.436.

Lalli, J. S., Mace, F. C., Wohn, T., & Livezey, K. (1995).Identification and modification of a response-classhierarchy. Journal of Applied Behavior Analysis, 28,551–559. https://doi.org/10.1901/jaba.1995.28-551.

Langdon, N. A., Carr, E. G., & Owen-DeSchryver, J. S.(2008). Functional analysis of precursors for seriousproblem behavior and related intervention. BehaviorModification, 32, 804–827. https://doi.org/10.1177/0145445508317943.

CHRISTINE A. WARNER et al.22

Lattal, K. A., St. Peter, C., & Escobar, C. R. (2013).Operant extinction: Elimination and generation ofbehavior. In G. J. Madden, W. V. Dube,T. D. Hackenberg, G. P. Hanley, & K. A. Lattal(Eds.), APA handbook of behavior analysis(pp. 77–107). Washington, DC: American Psycho-logical Association.

Lerman, D. C., & Iwata, B. A. (1996). Developing atechnology for the use of operant extinction inclinical settings: an examination of basic andapplied research. Journal of Applied Behavior Analy-sis, 29, 345–385. https://doi.org/10.1901/jaba.1996.29-345.

Lieving, G. A., Hagopian, L. P., Long, E. S., &O’Connor, J. (2004). Response-class hierarchies andresurgence of severe problem behavior. The Psycholog-ical Record, 54, 621–634.

Luczynski, K. C., & Hanley, G. P. (2009). Do childrenprefer contingencies? An evaluation of the efficacy ofand preference for contingent versus noncontingentsocial reinforcement during play. Journal of AppliedBehavior Analysis, 42, 511–525. https://doi.org/10.1901/jaba.2009.42-511.

Mace, F., Page, T. J., Ivancic, M. T., & O’Brien, S.(1986). Analysis of environmental determinants ofaggression and disruption in mentally retarded chil-dren. Applied Research in Mental Retardation, 7,203–221. https://doi.org/10.1016/0270-3092(86)90006-8.

Magee, S. K., & Ellis, J. (2000). Extinction effects duringthe assessment of multiple problem behaviors. Journalof Applied Behavior Analysis, 33, 313–316. https://doi.org/10.1901/jaba.2000.33-313.

Murphy, O., Healy, O., & Leader, G. (2009). Risk fac-tors for challenging behaviors among 157 childrenwith autism spectrum disorder in Ireland. Research inAutism Spectrum Disorders, 3, 474–482. https://doi.org/10.1016/j.rasd.2008.09.008.

Najdowski, A. C., Wallace, M. D., Ellsworth, C. L.,MacAleese, A. N., & Cleveland, J. M. (2008).Functional analyses and treatment of precursorbehavior. Journal of Applied Behavior Analysis,41, 97–105. https://doi.org/10.1901/jaba.2008.41-97.

Oliver, A. C., Pratt, L. A., & Normand, M. P. (2015). Asurvey of functional behavior assessment methodsused by behavior analysts in practice. Journal ofApplied Behavior Analysis, 48, 817–829. https://doi.org/10.1002/jaba.256.

Pinkston, J. W., & Libman, B. M. (2017). Aversive func-tions of response effort: Fact or artifact? Journal of theExperimental Analysis of Behavior, 108, 73–96.https://doi.org/10.1002/jeab.264.

Richman, D. M., Wacker, D. P., Asmus, J. M.,Casey, S. D., & Andelman, M. (1999). Further anal-ysis of problem behavior in response class hierarchies.Journal of Applied Behavior Analysis, 32, 269–283.https://doi.org/10.1901/jaba.1999.32-269.

Roscoe, E. M., Phillips, K. M., Kelly, M. A.,Farber, R., & Dube, W. V. (2015). A statewide sur-vey assessing practitioners’ use and perceived utilityof functional assessment. Journal of Applied BehaviorAnalysis, 48, 830–844. https://doi.org/10.1002/jaba.259.

Rose, J. C., & Beaulieu, L. (2019). Assessing the general-ity and durability of interview-informed functionalanalyses and treatment. Journal of Applied BehaviorAnalysis, 52, 271–285. https://doi.org/10.1002/jaba.504.

Rosenthal, R. (1979). The file drawer problem and toler-ance for null results. Psychological Bulletin, 86,638–641.

Santiago, J., Hanley, G., Moore, K., & Jin, C. (2016). Thegenerality of interview-informed functional analyses:Systematic replications in school and home. Journal ofAutism and Developmental Disorders, 46, 797–811.https://doi.org/10.1007/s10803-015-2617-0.

Scargle, J. D. (2000). Publication bias: The "file-drawerproblem" in scientific inference. Journal of ScientificExploration, 14, 94–106.

Sidman, M. (1960). Tactics of scientific research: Evaluatingexperimental data in psychology. New York, NY: BasicBooks.

Slaton, J., & Hanley, G. P. (2018). On the nature andscope of synthesis in functional analysis of problembehavior. Journal of Applied Behavior Analysis, 51,943–973. https://doi.org/10.1002/jaba.498.

Slaton, J. D., Hanley, G. P., & Raftery, K. J. (2017).Interview-informed functional analyses: A comparisonof synthesized and isolated components. Journal ofApplied Behavior Analysis, 50, 252–277. https://doi.org/10.1002/jaba.384.

Smith, R. G., & Churchill, R. M. (2002). Identificationof environmental determinants of behavior disordersthrough functional analysis of precursor behaviors.Journal of Applied Behavior Analysis, 35, 125–136.https://doi.org/10.1901/jaba.2002.35-125.

Stocco, C. S., Thompson, R. H., & Rodriguez, N. M.(2011). Restricted interests and teacher presentationof items. Journal of Applied Behavior Analysis, 44,499–512. https://doi.org/10.1901/jaba.2011.44-499.

Strand, R. C. W., & Eldevik, S. (2018). Improvements inproblem behavior in a child with autism spectrumdiagnosis through synthesized analysis and treatment:A replication in an EIBI home program. BehavioralInterventions, 33, 102–111. https://doi.org/10.1002/bin.1505.

Taylor, S. A., Phillips, K. J., & Gertzog, M. G. (2018).Use of synthesized analysis and informed treatmentto promote school reintegration. Behavioral Interven-tions, 33, 364–379. https://doi.org/10.1002/bin.1640.

Thompson, R. H., Fisher, W. W., Piazza, C. C., &Kuhn, D. E. (1998). The evaluation and treatment ofaggression maintained by attention and automatic

23EXTINCTION ANALYSES

reinforcement. Journal of Applied Behavior Analysis, 31,103–116. https://doi.org/10.1901/jaba.1998.31-103.

Received May 18, 2018Final acceptance March 22, 2019Action Editor, Matthew Normand

SUPPORTING INFORMATION

Additional Supporting Information may befound in the online version of this article at thepublisher’s website.

CHRISTINE A. WARNER et al.24