jaba26

Embed Size (px)

Citation preview

  • 8/13/2019 jaba26

    1/14

    FUNCTIONAL ANALYSIS SCREENING FOR PROBLEM BEHAVIORMAINTAINED BY AUTOMATIC REINFORCEMENT

    ANGIEC. QUERIMUNIVERSITY OF NEBRASKA

    BRIANA. IWATAUNIVERSITY OF FLORIDA

    EILEENM. ROSCOE ANDKEVINJ. SCHLICHENMEYERNEW ENGLAND CENTER FOR CHILDREN

    AND

    JAVIERVIRUSORTEGA ANDKYLEEE. HURL

    UNIVERSITY OF MANITOBA AND ST. AMANT RESEARCH CENTRE

    A common finding in previous research is that problem behavior maintained by automaticreinforcement continues to occur in the alone condition of a functional analysis (FA), whereasbehavior maintained by social reinforcement typically is extinguished. Thus, the alone conditionmay represent an efficient screening procedure when maintenance by automatic reinforcement issuspected. We conducted a series of 5-min alone (or no-interaction) probes for 30 cases of problembehavior and compared initial predictions of maintenance or extinction to outcomes obtained insubsequent FAs. Results indicated that data from the screening procedure accurately predicted thatproblem behavior was maintained by automatic reinforcement in 21 of 22 cases and by socialreinforcement in 7 of 8 cases. Thus, results of the screening accurately predicted the function of

    problem behavior (social vs. automatic reinforcement) in 28 of 30 cases.Key words: functional analysis, automatic reinforcement, screening

    In a typical functional analysis (FA) of problembehavior (Iwata, Dorsey, Slifer, Bauman, &Richman, 1982/1994), an individual is exposedrepeatedly to a series of conditions in whichantecedent and consequent events are manipu-lated to determine which events are responsible

    for behavioral maintenance. The utility of the FAas a basis for intervention has been demonstratedin hundreds of studies; as a result, it is consideredto be the standard throughout the field (Hanley,Iwata, & McCord, 2003).

    Much of the research in the FA literature hasconsisted of systematic replication and extensionacross client population, problem behavior, and

    setting. Other research has focused on methodo-logical refinement or adaptation to accommodateunusual client histories or limiting conditions ofassessment. These modifications have includedvariations in both the content and arrangement ofassessment conditions, and the present study

    focuses on the latter. One limitation in settingssuch as outpatient clinics is the amount of timeavailable for conducting assessments, and a modeldeveloped specifically for use in time-limitedsituations is the brief functional analysis (BFA).As described by Northup et al. (1991), the BFAconsists of single exposures to 5-min assess-ment conditions, with the addition of onereplication (the condition in which behavior

    occurs most frequently) and a treatment probe iftime permits. In a review of data from 79 cases inwhich the BFA was used, Derby et al. (1992)

    Address correspondence to Brian Iwata, PsychologyDepartment, 114 Psychology Building, University ofFlorida, Gainesville, Florida 32611 (e-mail: [email protected]).

    doi: 10.1002/jaba.26

    JOURNAL OF APPLIED BEHAVIOR ANALYSIS 2013, 46,4760 NUMBER1 (SPRING2013)

    47

  • 8/13/2019 jaba26

    2/14

    reported that the BFA yielded interpretableresults in 47% of the cases. Kahng and Iwata(1999) subsequently compared outcomes frombrief (single-session) and more typical (repeated

    measures) FAs for 50 cases and found correspon-dence in 66% of them. Thus, although the BFAmay represent the only option for experimentalassessment under some conditions, the greatlyreduced number and duration of sessions yieldlimited samples of behavior.

    Vollmer, Marcus, Ringdahl, and Roane (1995)proposed a progressive or hierarchical model ofassessment in which lengthier components wereadded as needed. Twenty subjects progressedthrough four phases of assessment, which wasterminated after any phase in which the functionof problem behavior was identified. In Phase 1, 8to 12 sessions of a typical FA were conducted in amultielement design, and data were examined aswithin-session patterns of responding. In Phase 2,FA sessions continued, and data were examined asoverall session means. In Phase 3, additionalsessions from only the alone (or no-interaction)

    condition were conducted. In the final phase, allassessment conditions were repeated but werealternated in a reversal design. With this strategy,Vollmer et al. identified the function of problembehavior in 17 of the 20 cases. Six subjectscompleted the assessment after Phase 1, fouradditional subjects after Phase 2, five subjectsafter Phase 3, and two subjects after Phase 4. Itshould be noted, however, that the assessmentwas not especially brief because even Phase 1

    entailed 8 to 12 10-min sessions, the typicalduration for many FAs.

    Roscoe, Iwata, and Zhou (2013) described analternative assessment model for a specificapplication. They assumed that the targetbehavior in their study, hand mouthing, wasmost likely to be maintained by automaticreinforcement, so they arranged FA conditionsin a 2:1 ratio of alone or no-interaction versus

    attention and demand sessions and eliminatedthe play condition entirely. This resulted in an FAthat consisted mostly of alone or no-interaction

    sessions interspersed with attention and demandprobes. If results were clear after 14 sessions, theFA was terminated; if not, further manipulationswere conducted. Of the 64 subjects, 46 (72%)

    required only the first assessment phase, suggest-ing that exposure to only the alone or no-interaction sessions may have been sufficient for alarge number of individuals.

    The purpose of this study was to evaluate theutility of brief exposure to alone or no-interaction sessions as a screening procedurefor problem behavior maintained by automaticreinforcement. If problem behavior is main-tained during this brief assessment, it may bepossible to forgo other assessment conditionsand proceed directly to intervention. By con-trast, if problem behavior is not maintained,further assessment would be warranted. Inthe present study, we conducted the initialscreening and a subsequent FA for all subjects todetermine whether (a) rates of problem behaviorduring screening sessions were predictive of theoutcome of an FA and (b) behavior maintained

    by automatic reinforcement revealed a moreconsistent pattern of responding during screen-ing sessions relative to behavior that has other(social) functions.

    Because results of several studies have shownthat stereotypy is likely to be maintained byautomatic reinforcement (Piazza, Adelinis, Han-ley, Goh, & Delia, 2000; Rapp, Miltenberger,Galensky, Ellingson, & Long, 1999; Vollmer,Marcus, & LeBlanc, 1994), stereotypy was the

    topography examined most often in this study.We included other topographies of problembehavior (aggression, self-injurious behavior[SIB], etc.) as well because these behaviors havebeen shown to be maintained more often bysocial reinforcement (Iwata et al., 1994; Marcus,Vollmer, Swanson, Roane, & Ringdahl, 2001).The inclusion of multiple topographies ofproblem behavior increased the likelihood that

    screening data would be suggestive of extinctionas well as maintenance, allowing us to verify bothtypes of predictions.

    48 ANGIE C. QUERIM et al.

  • 8/13/2019 jaba26

    3/14

    METHOD

    Subjects and Settings

    Twenty-six individuals who had beenreferred for assessment of problem behavior(stereotypy, SIB, aggression, or property destruc-tion) participated. One individual engaged inthree problem behaviors, and two engaged intwo problem behaviors, yielding a total of 30cases. Table 1 shows demographic informationfor all subjects. The study was conducted at threesites: two school programs and a residentialprogram, all of which served students withintellectual disabilities, autism, or both. All

    sessions were conducted in small rooms thatcontained a desk, two chairs, and other materialsas needed. Screening sessions lasted 5 min; FAsessions lasted 10 min. All sessions were con-

    ducted three to five times per day, 1 to 5 days perweek.

    Response Measurement and Reliability

    The dependent measure was the occurrence ofproblem behavior, which was defined on anindividual basis (see Table 1). Trained observersrecorded data on handheld computers. The datawere converted to rate measures (responses per

    Table 1Subject Characteristics

    Subject Age Diagnosis Topography

    Michele 13 ASD STPY: hand waveBri 19 ASD STPY: finger play Dan 11 ASD STPY: vocalKarl 13 ASD STPY: vocalCor 36 (ID, spastic quadriplegia STPY: hand or object in mouthNiki 47 ID, cerebral atrophy, microcephaly,

    deaf, blind

    STPY: finger play

    Eric 30 ID, left hemiparesis, seizure, retinaldetachment

    STPY: hand in mouth

    Winn 13 ASD STPY: hand waveNatalie 14 ASD STPY: vocalDave 12 ASD STPY: finger tap, spinJake 14 ASD STPY: hand waveNate 14 ASD STPY: hand waveSonia 14 ASD STPY: hand waveHolly 16 ID, SL impaired STPY: rub lipsRon 23 ID, SL impaired, physical impairment STPY: head twirlMark 10 SL impaired, orthopedically impaired STPY: lip rubJude 12 ASD, SL impaired STPY: finger curlEd 36 ID, spastic deplegia, seizure disorder STPY: finger to throat

    Sal 16 ASD SIB: hit headDana 14 ASD SIB: Pinch self Kim 14 ID, SL impaired SIB: bite hand

    AGG: hit, kick, bite, throw objects at personLarry 9 ID, cerebral palsy, hydrocephalus, cor-

    tical blindness, seizure disorderSIB: hit head

    Eve 17 ID, speech and language impaired PD: rip or throw materialsBrad 14 Dandy Walker syndrome, ADHD AGG: hit, kick, biteQueen 13 Other health impaired, SL impaired AGG: hit, kick, pull hair, bite, throw objects at

    personSIB: hit chin

    Pablo 23 ID, SL impaired AGG: pull hair SIB: hit groinPD: rip or throw materials

    Note.ASD autism spectrum disorder; ID intellectual disability; STPY stereotypy; SIB self-injuriousbehavior; PD property destruction; AGG aggression; SL speech-language.

    FUNCTIONAL ANALYSIS SCREENING 49

  • 8/13/2019 jaba26

    4/14

    minute) or the percentage of 10-s intervals duringwhich responding occurred. A rate measure wasused for behaviors that had a discrete beginningand end (e.g., aggression was recorded every time

    the subjects hand made contact with thetherapists body). An interval measure was usedfor behaviors with variable durations (e.g., thesubject pulled the therapists hair). Observers alsorecorded therapist behaviors (initiation of socialinteraction or removal of instructions) to assessprocedural consistency. A second observer simul-taneously but independently collected data for atleast 25% of all sessions. Reliability was calculatedby dividing session time into consecutive 10-sintervals and comparing observers records on aninterval-by-interval basis. Percentage agreementfor frequency measures was calculated by dividingthe smaller number of responses by the largernumber of responses in each interval andaveraging these fractions across the session.Percentage agreement for interval measures wascalculated by dividing the number of agreementintervals (on the occurrence or nonoccurrence of

    behavior) by the total number of intervals in asession. The mean reliability percentage forproblem behavior across all subjects was 94%(range, 69% to 100%). (Reliability results forother measures are available from the firstauthor.)

    Preference Assessment

    A paired-stimulus preference assessment (Fish-er et al., 1992) was conducted to identify items to

    be included in each subjects FA. Nine leisureitems were assessed, and subjects were exposedbriefly to all items prior to the assessment. Oneach assessment trial, the experimenter presentedtwo items and prompted the subject to chooseone. A selection was scored when the subjecttouched one of the two items. The subject wasgiven 10-s access to the selected item while theunselected item was removed, after which the

    selected item was removed and the next pair waspresented. Assessment continued until all possi-ble pairs were presented twice. Items chosen on

    75% or more of trials were designated as highlypreferred (HP) and were used in the playcondition of the FA; items chosen between30% to 60% of trials were designated as

    moderately preferred (MP) and were used inthe attention condition of the FA.

    Sequence

    Screening consisted of a series of 5-min aloneor no-interaction sessions. We selected 5 minbased on the typical session duration of a BFA(Northup et al., 1991) and conducted aminimum of three sessions with every subject,with additional sessions as needed to clarifytrends in the data. All sessions were conducted ina single block with 2-min breaks between sessions(subjects were taken for brief walks or givenbathroom breaks as needed, but no specificprotocol was followed during breaks). We thenconducted a typical FA (Iwata et al., 1982/1994)that included alone or no-interaction, attention,play, and demand conditions in a fixed-sequencemultielement design. Screenings always preceded

    FAs to eliminate subjects prior exposure to FAcontingencies, as would be the case under clinicalconditions.

    Assessment ConditionsAlone or no interaction. The purpose of this

    condition was to determine whether problembehavior was maintained in the absence of anysocial interaction; if so, it was most likelymaintained by automatic reinforcement. The

    subject was in a room that did not contain anyleisure items. In the alone condition, theexperimenter was not in the room; in the no-interaction condition (for subjects whose targetbehavior was aggression), the experimenter waspresent but did not interact with the subject atany time during the session.

    Attention. The purpose of this condition wasto determine whether problem behavior was

    maintained by social-positive reinforcement inthe form of attention. The subject and experi-menter were in the room, and the subject had free

    50 ANGIE C. QUERIM et al.

  • 8/13/2019 jaba26

    5/14

    access to two MP leisure items. The experimenterbegan the session by stating that she had workto do, turned away from the subject, and engagedin a solitary activity (e.g., read a magazine). If

    the subject engaged in problem behavior, theexperimenter delivered 3 to 5 s of verbal andphysical attention. All nontarget behavior wasignored.

    Play. This condition served as the controlcondition. The subject and experimenter were inthe room, and the subject had free access to twoHP leisure items. The experimenter delivered 3 to5 s of verbal and physical attention at least every30 s (or whenever the subject initiated appropri-ate social interaction) and ignored all instances ofproblem behavior.

    Demand. The purpose of this condition wasto determine whether problem behavior wasmaintained by social-negative reinforcementin the form of escape from task demands. Theexperimenter continuously presented academictasks using a three-prompt sequence (vocalinstruction, model prompt, physical prompt),

    terminated the trial (ceased instruction, removed

    all materials, and turned away from the subject)for 30 s contingent on problem behavior, anddelivered brief praise (3 to 5 s) contingent on thesubjects compliance.

    Data Interpretation

    At least three and usually eight Board CertifiedBehavior Analysts examined graphs of screeningand FA data graphed separately and reached aconsensus regarding (a) the prediction made bythe screening assessment and (b) the FA outcome.These judgments were used as the basis fordetermining the extent to which the predictionsmade based on the screening assessment matchedthe FA outcomes (see Figure 1).

    If problem behavior was maintained for threeor more screening sessions, it was predicted to bemaintained by automatic reinforcement. Ifproblem behavior also was either highest inthe alone or no-interaction condition of thesubsequent FA or high in all conditions of theFA, it was determined to be maintained byautomatic reinforcement (hit). If data showed

    a clear downward trend ending in a zero or

    Figure 1. Possible outcomes from the screening assessment and the functional analysis.

    FUNCTIONAL ANALYSIS SCREENING 51

  • 8/13/2019 jaba26

    6/14

    near-zero rate of behavior during screening, it waspredicted to be maintained by social reinforce-ment. If problem behavior during the subsequentFA was highest in the attention or demand

    condition, behavior was determined to bemaintained by either attention or escape fromdemands, respectively (also a hit). Two types oferrors could have occurred. If the screeningassessment predicted that problem behavior wasmaintained by automatic reinforcement but wasobserved to occur at the highest levels in eitherthe attention or demand condition (or at highlevels in both conditions) of the FA, it wasdetermined to be maintained by social reinforce-ment (false alarm). If the screening assessmentpredicted that problem behavior was maintainedby social reinforcement but it occurred at itshighest levels in the alone condition or in allconditions of the subsequent FA, it was deter-mined to be maintained by automatic reinforce-ment (miss).

    RESULTS

    Figure 2 shows results for 10 subjects whoengaged in stereotypy (Jake, Winn, Natalie, Karl,Jude, Dan, Cor, Niki, Eric, and Ron). Allsubjectsstereotypy was maintained during theirscreening assessments, suggesting that theirbehavior was maintained by automatic reinforce-ment. During the subsequent FAs, Ron contin-ued to engage in stereotypy in the alone conditionand also in the demand condition; all other

    subjects engaged in stereotypy in all conditions,confirming all predictions made by the screeningassessment.

    Figure 3 shows results for seven additionalsubjects who engaged in stereotypy. Six subjectsstereotypy was maintained during the screeningassessment (Holly, Ed, Bri, Nate, Dave, andMark) and occurred at its highest levels duringeither the alone condition (Holly, Bri, and Dave)

    or most conditions (Ed, Nate, and Mark) of theFA. Thus, predictions that these six subjectsstereotypy was maintained by automatic rein-

    forcement were confirmed by their subsequentFAs. One subject, Sonia, engaged in stereotypythat decreased markedly during her screeningassessment, suggesting that her behavior was

    maintained by social reinforcement. Results ofher FA confirmed this prediction: Her highestlevel of stereotypy was observed in the demandcondition.

    Figure 4 shows results for five subjects whoengaged in SIB. Pablos, Larrys, Danas, and SalsSIB was maintained in their screening assess-ments; their SIB also occurred in all conditions oftheir FAs, confirming predictions that their SIBwas maintained by automatic reinforcement.Kim did not engage in any SIB during herscreening assessment; she subsequently engagedin the highest levels of SIB during the demandcondition of her FA, confirming the predictionthat her SIB was maintained by socialreinforcement.

    Figure 5 shows results for two subjects whoengaged in property destruction. Pablo did notengage in any property destruction during his

    screening assessment and engaged in propertydestruction exclusively in the demand conditionof his FA. Thus, the prediction based on hisscreening (maintenance by social reinforcement)was confirmed by his FA (maintenance byescape). Eves property destruction showed adifferent pattern; there were high levels ofresponding during the screening and during allconditions of her FA, with highest levels in thealone condition. Thus, her screening assessment

    accurately predicted that property destructionwas maintained by automatic reinforcement.

    Figure 6 shows data for four subjects whoengaged in aggression (Brad, Pablo, Kim, andQueen). All subjects engaged in zero or near-zerolevels of aggression during the screening, suggest-ing maintenance by social reinforcement. Duringthe subsequent FAs, Brad and Pablo engaged inthe highest levels of aggression during the

    attention condition, whereas Kim and Queenengaged in the highest levels of aggressionduring the demand condition, confirming that

    52 ANGIE C. QUERIM et al.

  • 8/13/2019 jaba26

    7/14

    Figure 2. Percentage of intervals of stereotypy across the screening assessment (SA) and FA conditions.

    FUNCTIONAL ANALYSIS SCREENING 53

  • 8/13/2019 jaba26

    8/14

    Figure 3. Percentage of intervals of stereotypy across the SA and FA conditions.

    54 ANGIE C. QUERIM et al.

  • 8/13/2019 jaba26

    9/14

    their aggression was maintained by socialreinforcement (attention for Brad and Pablo,escape for Kim and Queen).

    Figure 7 shows data for the only two subjectswhose screening results were inconsistent withthose obtained in the subsequent FAs. Queens

    SIB decreased during the screening assessment.However, she engaged in the highest levels of SIBduring the alone condition of her FA. These data

    exemplify a miss, in that the prediction based onher screening data (social reinforcement) was notborne out by her FA data (automatic reinforce-ment). Micheles stereotypy occurred at highlevels during her screening assessment; her FAdata showed decreasing trends in all conditions

    initially but gradual emergence in only theattention condition. These data exemplify a falsealarm: Her screening assessment incorrectly

    Figure 4. Responses per minute or percentage of intervals of SIB across the SA and FA conditions.

    FUNCTIONAL ANALYSIS SCREENING 55

  • 8/13/2019 jaba26

    10/14

    Figure 5. Responses per minute of property destruction across the SA and FA conditions.

    Figure 6. Responses per minute or percentage of intervals of aggression across the SA and FA conditions.

    56 ANGIE C. QUERIM et al.

  • 8/13/2019 jaba26

    11/14

    predicted that behavior was maintained byautomatic reinforcement, whereas her FA indi-cated that behavior was maintained by socialreinforcement (attention).

    Table 2 shows a summary of the results.

    Overall, the screening assessment accuratelypredicted that behavior was maintained by eitherautomatic or social reinforcement in 28 of 30cases. Results of the screening assessment for 17of the 18 subjects who engaged in stereotypyaccurately predicted the function of problembehavior (16 automatic reinforcement, one socialreinforcement). Only one error (a false alarm)

    occurred: The screening assessment predictedthat Micheles stereotypy was maintained byautomatic reinforcement, whereas the FA identi-fied attention as the maintaining reinforcer. Thescreening assessment accurately predicted that

    SIB was maintained by automatic reinforcementfor four of the six subjects who engaged in SIBand that SIB was maintained by social reinforce-ment for a fifth subject, whose FA identifiedescape from demands as the source of reinforce-ment. The sixth subjects (Queens) screeningproduced a miss: It predicted that her SIB wasmaintained by social reinforcement, whereas

    Table 2Summary of Results

    Topography Subjects Screening FA Result

    STPY (n 18) 16 Automatic Automatic Hit1 Social Escape Hit1 Automatic Attention False alarm

    SIB (n 6) 4 Automatic Automatic Hit1 Social Escape Hit1 Social Automatic Miss

    PD (n 2) 1 Automatic Automatic Hit1 Social Escape Hit

    AGG (n

    4) 2 Social Attention Hit2 Social Escape Hit

    Note.STPY stereotypy; SIB self-injurious behavior; PD property destruction; AGG aggression.

    Figure 7. Responses per minute of SIB (Queen) and percentage of intervals with stereotypy (Michele) across the SA andFA conditions. Queens data reflect a miss; Micheles data reflect a false alarm (see text).

    FUNCTIONAL ANALYSIS SCREENING 57

  • 8/13/2019 jaba26

    12/14

    results of the FA showed that SIB was maintainedby automatic reinforcement. Two subjects en-gaged in property destruction, one maintained byautomatic reinforcement and the other by social

    reinforcement, and the screening assessmentaccurately predicted both FA outcomes. Finally,the screening assessment accurately predictedthat the aggression exhibited by four of foursubjects was maintained by social reinforcement.The FA data indicated that two of the fourengaged in aggression maintained by attentionand that the other two engaged in aggressionmaintained by escape.

    DISCUSSION

    Thirty subjects with varied problem behaviors(predominantly stereotypy) completed a screen-ing assessment that consisted of several exposuresto a single 5-min test condition that wereconducted in one session. Based on a comparisonwith subsequent FA outcomes, results of thescreening assessment accurately predicted the

    function of problem behavior (automatic orsocial reinforcement) in 28 of 30 cases. In all butone case in which results of the FA indicated thatproblem behavior was maintained by automaticreinforcement (21 of 22), the screening assess-ment provided an accurate prediction. Giventhese results, the screening may be most useful forbehavior suspected to be maintained by auto-matic reinforcement: stereotypy and perhaps SIBor property destruction.

    Because the screening was developed primarilyfor problem behavior maintained by automaticreinforcement, the extent to which it improvedefficiency of assessment in these cases is animportant consideration. We calculated screeningefficiency for behavior maintained by automaticreinforcement by dividing the total duration ofscreening for each subject whose behavior wasmaintained by automatic reinforcement in both

    assessments (hits) by the total duration of eachsubjects FA. Twenty-one subjects engaged inproblem behavior maintained by automatic

    reinforcement, and the mean durations ofassessment were 21.5 min and 170 min for thescreening and FA, respectively. Thus, thescreening was completed in 12.6% of the time

    it would have taken to complete the full FA. Wealso examined the occurrence of problembehavior exhibited by these subjects and foundthat the amount of problem behavior (rates ornumbers of intervals) observed during thescreening was 29.2% of what was observedduring the FAs. The screening was not only quickbut also easy to implement, in that it wascompleted in one continuous time block, and theexperimenter either was not present or did notdeliver any consequences during any session.

    In spite of the high degree of correspondencebetween results of the screening and FAprocedures, it is important to emphasize thatthe screening procedure is not a replacement for afull FA of problem behavior. Data in this studyindicated that predictions based on the screeningresulted in a 93% correspondence with outcomesof a functional analysis, in other words a 7% loss

    of accuracy, which is significant in clinicalresearch. Although our data were based on arelatively large sample (30 cases), we estimated asimilar loss in accuracy for a much largerpopulation. We examined complete sets of FAdata published in theJournal of Applied BehaviorAnalysisand determined the extent to which ratesof problem behavior in the alone or no-interaction condition predicted behavioral func-tion. We selected all articles in which (a) a full FA

    was conducted, (b) an alone or no-interactioncondition was included in the FA, and (c) datawere presented in the article. In 108 of 115published data sets that met these criteria,problem behavior occurred at high rates in thealone or no-interaction condition and wasmaintained by automatic reinforcement; in all222 data sets, problem behavior occurred at lowrates in the alone or no-interaction condition and

    was maintained by social reinforcement. Thus, in330 of 337 data sets, high or low rates of problembehavior during the alone or no-interaction

    58 ANGIE C. QUERIM et al.

  • 8/13/2019 jaba26

    13/14

    condition of an FA were predictive of behavioralfunction, for an overall correspondence of 98%anda2%lossofaccuracy.Thesevendatasetsthatdid not show complete correspondence consisted

    of cases in which problem behavior that occurredat high rates in the alone or no-interactioncondition was maintained by both automaticreinforcement and social reinforcement (multiplecontrol). Maintenance by automatic and socialreinforcement would be missed in our screeningprocedure because it does not contain any test forsocial reinforcement.

    Although the screening assessment predictedthat problem behavior was socially maintainedfor all four subjects whose problem behavior wasaggression, its use with aggression would not bevery helpful because it will not identify whichsocial contingency (positive or negative reinforce-ment) maintains behavior. Nevertheless, trendsobserved during the screening suggested that onesource of social reinforcement was more likelythan another, perhaps making the screeninghelpful as an preliminary test for problem

    behavior maintained by social reinforcement.The two subjects whose aggression was attentionmaintained (Brad and Pablo, Figure 6) exhibitedeither low or decreasing rates of problem behaviorduring the screening, which would be expectedwhen problem behavior maintained by attentionis exposed to extinction. Thus, a decreasing trendduring screening might be followed by a single-function test (Iwata & Dozier, 2008) in which theattention and control conditions are alternated to

    confirm attention-maintained problem behavior.By contrast, the two subjects whose aggressionwas escape maintained (Kim and Queen) neverengaged in problem behavior during screening,which also might be expected because the absenceof interaction initiated by a therapist eliminatesthe motivational basis (establishing operation) forescape. If this pattern of problem behavior isobserved during screening, it might be followed

    by a single-function test for escape-maintainedproblem behavior. Thus, even for cases in whichproblem behavior is maintained by social

    reinforcement, the alone screening might repre-sent an overall gain in assessment efficiency byreducing the number of subsequent testconditions.

    In summary, results obtained from this FAscreening assessment are highly promising andsuggest that it may be an accurate and practicalapproximation to a complete FA in clinicalsituations when (a) the target behavior consists ofstereotypy, (b) preliminary evidence suggests thatproblem behavior may be maintained byautomatic reinforcement, and (c) a full FA cannotbe conducted due to time limitations. Futureresearch might consider the extent to whichanecdotal reports facilitate selection of assessmentstrategies. For example, although verbal reportsby caregivers about specific functions of problembehavior are often unreliable (Kelley, LaRue,Roane, & Gadaire, 2011), they may providereasonable estimates of the extent to whichproblem behavior is more or less likely to occurin social contexts; if so, this information maybe helpful in determining whether to conduct

    an alone screening or a more completeassessment.

    REFERENCES

    Derby, K. M., Wacker, D. P., Sasso, G., Steege, M.,Northup, J., Cigrand, K., & Asmus, J. (1992). Brieffunctional assessment techniques to evaluate aberrantbehavior in an outpatient setting: A summary of 79cases.Journal of Applied Behavior Analysis,25,713721.doi: 10.1901/jaba.1992.25-713

    Fisher, W., Piazza, C. C., Bowman, L. G., Hagopian, L. P.,Owens, J. C., & Slevin, I. (1992). A comparison of twoapproaches for identifying reinforcers for persons withsevere and profound disabilities. Journal of AppliedBehavior Analysis, 25, 491498. doi: 10.1901/jaba.1992.25-491

    Hanley, G. P., Iwata, B. A., & McCord, B. E. (2003).Functional analysis of problem behavior: A review.

    Journal of Applied Behavior Analysis,36, 147185. doi:10.1901/jaba.2003.36-147

    Iwata, B. A., Dorsey, M. F., Slifer, K. J., Bauman, K. E., &Richman, G. S. (1994). Toward a functional analysis ofself-injury. Journal of Applied Behavior Analysis, 27,197209. doi: 10.1901/jaba.1994.27-197 (Reprintedfrom Analysis and Intervention in DevelopmentalDisabilities, 2, 320, 1982)

    FUNCTIONAL ANALYSIS SCREENING 59

  • 8/13/2019 jaba26

    14/14

    Iwata, B. A., & Dozier, C. L. (2008). Clinical application offunctional analysis methodology. Behavior Analysis inPractice,1, 39. PMid: 22477673 PMCid:2846577.

    Iwata, B. A., Pace, G. M., Dorsey, M. F., Zarcone, J. R.,Vollmer, T. R., Smith, R. G., Willis, K. D. (1994).

    The functions of self-injurious behavior: An experi-mental-epidemiological analysis. Journal of AppliedBehavior Analysis, 27, 215240. doi: 10.1901/jaba.1994.27-215

    Kahng, S., & Iwata, B. A. (1999). Correspondence betweenoutcomes of brief and extended functional analyses.

    Journal of Applied Behavior Analysis,32, 149159. doi:10.1901/jaba.1999.32-149

    Kelley, M. E., LaRue, R. H., Roane, H. S., & Gadaire, D. M.(2011). Indirect behavioral assessments: Interviewsand rating scales. In W. W. Fisher, C. C. Piazza, & H.S. Roane (Eds.), Handbook of applied behavior analysis(pp. 182190). New York, NY: Guilford.

    Marcus, B. A., Vollmer, T. R., Swanson, V., Roane, H. R., &Ringdahl, J. E. (2001). An experimental analysis ofaggression. Behavior Modification, 25, 189213. doi:10.1177/0145445501252002

    Northup, J., Wacker, D., Sasso, G., Steege, M., Cigrand, K.,Cook, J., & DeRaad, A. (1991). A brief functionalanalysis of aggressive and alternative behavior in anoutclinic setting. Journal of Applied Behavior Analysis,

    24, 509522. doi: 10.1901/jaba.1991.24-509

    Piazza, C. C., Adelinis, J. D., Hanley, G. P., Goh, H. L., &Delia, M. D. (2000). An evaluation of the effects ofmatched stimuli on behaviors maintained by automaticreinforcement.Journal of Applied Behavior Analysis,33,1327. doi: 10.1901/jaba.2000.33-13

    Rapp, J. T., Miltenberger, R. G., Galensky, T. L., Ellingson,S. A., & Long, E. S. (1999). A functional analysis of hairpulling.Journal of Applied Behavior Analysis,32, 329337. doi: 10.1901/jaba.1999.32-329

    Roscoe, E. M., Iwata, B. A., & Zhou, L. (2013). Assessmentand treatment of chronic hand mouthing. Journal of

    Applied Behavior Analysis,46, 181198Vollmer, T. R., Marcus, B. A., & LeBlanc, L. (1994).

    Treatment of self-injury and hand mouthing followinginconclusive functional analyses. Journal of AppliedBehavior Analysis, 27, 331344. doi: 10.1901/jaba.1994.27-331

    Vollmer, T. R., Marcus, B. A., Ringdahl, J. E., & Roane,

    H. S. (1995). Progressing from brief assessments toextended experimental analyses in the evaluation ofaberrant behavior. Journal of Applied Behavior

    Analysis, 28, 561576. doi: 10.1901/jaba.1995.28-561

    Received July 4, 2012Final acceptance November 27, 2012

    Action Editor, Timothy Vollmer

    60 ANGIE C. QUERIM et al.