AnálisisFuncional-Yoman-APrimerionFnctionalAnalysis-S7

nctional Analysis

e Yomanlls Resource

compts. Tnal aioninrole oorary

tionvariaprovventFun

.g.,theicalionrva-rva-

Available online at www.sciencedirect.com

Cognitive and Behavioral Practice 15 (2008) 325--340www.elsevier.com/locate/cabpceptualization and other complex clinical judgments (forexample, see Haynes, Leisen, & Blaine, 1997; Virus-Ortega& Haynes, 2005). It can help identify sources of resistanceto change (see Edelstein & Yoman, 1991) and strategies topromote persistence of change after intervention ends (seeBaer, Wolf, & Risley, 1968, 1987). Because most practi-

1 Applied behavior analysis is the science in which proceduresderived from the principles of behavior are systematically applied toimprove socially significant behavior to a meaningful degree and todemonstrate experimentally that the procedures employed wereresponsible for the improvement in behavior (Cooper, Heron, &tionanal

1077 20Publ

Ccase. It guides the practitioner to effective interven-s from observations of possible causal or maintainingbles in an individual clients problems and/or goals,iding the critical link between assessment and inter-ion that has been a cornerstone of behavior therapy.ctional analysis provides a framework for case con-

objectively recorded in the form of a number (efrequency, latency, duration, intensity, etc.). (Note thatfrequency, latency, duration, or intensity of categorvariables such as a sad or happy facial expressmight also be recorded.) Objective recording of obsetions means that the reliability of data from the obseA Primer on Fu

JeromLife Ski

This article presents principles and basic steps for practitioners toapplication of functional analysis to adult mental health clienfunctional diagnoses and behavioral interventions, with functiomodel presented incorporates both operant and classical conditbehavioral principles. Finally, the article describes the continuingpotential advantages over other forms of assessment in contemp

FUNCTIONAL analysis emerged early in the history ofbehavior therapy (actually prior to behavior therapyitself) as a way of applying the fledgling science ofbehavior to the practice of behavior change (see Ferster,1972; Kazdin, 1978; Sidman, 1960; Skinner, 1953). Thus, itmight be considered the most primary in this journalsseries of primers in cognitive and behavioral therapy.Yet, the behavioral interventions to which a functionalanalysis can lead the practitioner (e.g., Bootzin & Epstein,2000; Hopko, Lejuez, Ruggiero, & Eifert, 2003; Roth-baum, Meadows, Resick, & Foy, 2000) are arguably themost powerful components of contemporary cognitivebehavioral therapy (Dimidjian et al., 2006; Jacobson et al.,1996; Taylor, 2004).

Functional analysis is a scientific approach to the indivi-dualers are more familiar with the application of functionalysis to children (e.g., Albano & Morris, 1998; Scotti,

-7229/08/325340$1.00/008 Association for Behavioral and Cognitive Therapies.ished by Elsevier Ltd. All rights reserved.

ontinuing Education Quiz located on p. 346347.lete a functional analysis of client behavior. The emphasis is onhe article includes a detailed flow chart containing all majorssessment questions directing the reader through the chart. Theg and a strategy for selecting target behaviors consistent withf functional analysis in effective behavioral intervention and itscognitive-behavioral therapy.

Mullen, & Hawkins, 1998; Sulzer-Azaroff & Pollack, 1982),this article will emphasize application of functional analysisto adults.

Consistent with behavioral practice, we begin thisprimer on functional analysis by defining our terms.Functional analysis is a fundamental tool of appliedbehavior analysis1 for examining the relationship betweenbehavior and the environment. In performing a functionalanalysis, the practitioner describes the behavior of interestand its antecedents and consequences in observablemeasurable terms. Observable terms means that the thingdescribed can be picked up with one of the five humansenses, usually sight or hearing. Sometimes observation isassisted by special instruments (e.g., biofeedback equip-ment).Measurable termsmeans that the observations can beHeward, 1987, p. 14). It usually concerns the relationships betweenbehavior and non-laboratory environments such as workplaces,schools, and clinics. In common usage, Behavior modification refersmore specifically to the implementation of behavior change methodsdiscovered by applied behavior analysis. Behavior therapy tends torefer even more specifically to the implementation of behavior changemethods (only some of which were discovered by applied behavioranalysis), usually with persons suffering from mental health problems.(See Martin & Pear, 1999, for further discussion of these distinctions.)

the practitioner. This is important because reinforcersand punishers often defy what the practitioner expects orintends. For example, praise is often delivered with theexpectation that it will reinforce the behavior it follows. Infact, it often does not.

326 Yomantions can be calculated (e.g., the extent to which one ob-server assigns the same number to the same sample ofbehavior on different occasions, or the extent to which twoor more observers assign the same number upon obser-ving the same sample of behavior).

In this manner, the practitioner of functional analysisuses the scientific method to describe three componentswithin the analysis, form hypotheses about their inter-relationships, and then test those hypotheses. A func-tional analysis is sometimes called an A-B-C analysis afterthese three components: antecedents, behaviors, andconsequences.

Functional Analysis in Operant Conditioning

The component typically described first in functionalanalysis is the behavior of interest. A behavior is theobservable action of a living organism (in the case ofbehavior therapy, a person). For example, the behavior ofinterest for a client with substance abuse problems mightbe lighting and inhaling from a pipe filled with marijuana.For a client entering therapy for help with shyness, thebehavior of interest might be calling a friend to invite himto a social activity.

The next component described is the antecedent (orantecedents) of that behavior. An antecedent is a stimulus(event observable by the person) which precedes thebehavior of interest. Antecedents are also known asdiscriminative stimuli, in that they signal that a particularresponse will be followed by a certain consequence,allowing a person to discriminate when to respond.There are two primary types of discriminative stimuli: SDsand Ss. An SD signals that a certain consequence willfollow the behavior; an S signals that it will not. Forexample, Azrin and Hayes (1984) trained males seekinghelp to improve heterosocial skills to discriminate interestfrom female visual face andbody cues. They found that thistraining alone improved role-played social skills in theirparticipants. While Azrin and Hayes purposely did notspecify discriminative stimuli in their study (successfuldiscrimination does not require the ability to verballyidentify the discriminative stimulus), SDs for initiation ofinteraction might have included the woman maintainingeye contact and showing a felt smile (Ekman & Friesen,1982). Ss for initiation of interactionmight have includedlooking away, frowning, or fidgeting in a particular way.

The third component of functional analysis, a con-sequence, is a stimulus that follows the behavior. There aretwo main types of consequences: a reinforcer and apunisher (see Fig. 1). A reinforcer is a consequence thatincreases the strength (i.e., frequency, duration, intensity)of the behavior it follows. A punisher is a consequence thatdecreases the strength of the behavior it follows. Noticethat these two types of consequences are defined by theireffect on behavior and not the expectation or intention ofFig. 1 depicts the four main types of consequences. Todetermine the type of consequence a practitioner hasobserved, he or she follows five steps: (a) define the behaviorof interest in behavioral measurable terms (see above);(b) observe whether the behavior is increasing or decreasing instrength to determine whether the consequence is areinforcer or a punisher; (c) identify the consequence anddefine it in behavioral measurable terms; (d) determinewhether the identified consequence is being added to or withdrawnfrom the environment (i.e., is it positive or negative,respectively); and (e) reproduce or replicate theincrease or decrease in the behavior by adding andwithdrawing the consequence. The following examplesillustrate these five steps as noted in parentheses:

A depressed clients frequency of completing plannedactivities (behavior) increases (reinforcer) after question-ing in an animated vocal tone how the client accomplishedthis (consequence) is introduced (positive) by the practi-tioner after activity completion. Planned activity comple-tion decreases when the animated questioning is withheldafter several instances of activity completion, thenincreases again when the animated questioning is reintro-duced after activity completion (replication), demonstrat-ing that the animated questioning is a positive reinforcer.

A Generalized Anxiety Disorder clients frequency ofpracticing progressive relaxation (behavior) increases(reinforcer) after she experiences that ongoing muscletension and other uncomfortable bodily cues of stress(consequence) decrease (negative). Replication is morecomplex with internal physiological stimuli which cannotbe systematically introduced or withdrawn. However,negative reinforcement might be demonstrated in thiscase by reproducing the effect with a similar behavior:deep breathing relaxation. If the practitioner taught theclient this technique, and the frequency of practicingbreathing relaxation increased only when the clientreported a decrease in muscle tension and other stresscues, then the practitioner would have demonstrated thatthe decrease in the bodily stress cues is a negative reinforcer.(Note that negative reinforcement involves terminationof an ongoing aversive stimulus.2)

A wifes disclosure of angry feelings (behavior)decreases in frequency (punisher) when her husbands

2 An aversive stimulus is one which a person has a history ofaverting from (turning away) or avoiding. Whether an aversivestimulus is a punisher, whether it's cessation is a negative reinforcer,or neither, depends on its relationship to a specific behavior ofinterest.

quenc

327Functional Analysiscriticizing her and saying she should be ashamed(consequence) starts (positive) after each such disclosure.If the wifes disclosure of angry feelings increases withcessation of the husbands criticism, this supports thehypothesis that criticism is a positive punisher. Due to theinadvisability of reintroducing criticism and shaming afterdisclosure of angry feelings, the practitioner mightdemonstrate the suppressive effect of criticism with asimilar behavior. For example, if criticism and shamingfollow the wifes admitting mistakes, the practitionercould get the husband to agree to cease criticism andshaming after this behavior. If the wifes admittingmistakes now increased, the reversal of the punishingeffects of criticism and shaming would have beenreplicated with a second behavior.

A socially anxious clients asking others out to dinner(behavior) decreases in frequency (punisher) afterseveral co-workers eye contact and conversation (con-sequence) abruptly breaks off (negative) after the clientinvites them to dinner. Again, direct replication of thiseffect would be undesirable. However, with the clientspermission, members of the clients social skills group

Figure 1. Consemight plan to break off eye contact and conversation aftera similar behavior, such as asking questions of groupmembers, without identifying the specific behavior to theclient. If questions by the client decreased during thegroup intervention period (replication), then breakingoff eye contact and conversation are a negative punisher(also called response cost). (Note that it might be quiteinstructional to discuss data from this group experimentto teach members about discriminating and overcomingthe effects of social punishment.)

Reinforcers can also be classified as primary orsecondary, depending on how they were established asreinforcers. Primary or unconditioned reinforcers arethose we respond to instinctively, usually because theymeet a basic need. We are born working to gain mostprimary reinforcers (e.g., food). Their power over ourbehavior doesnt require prior learning. Secondary orconditioned reinforcers, on the other hand, gain theirpower through pairing with primary or other established(secondary) reinforcers. For example, a baby may learn towork for the sight of his mothers smile because that smileis paired with the offer of food.

Adding Classical Conditioning to FunctionalAnalysis

Although functional analysis emerged out of the studyof operant conditioning (i.e., how behavior operates on theenvironment to produce consequences), consideringclassical conditioning (also known as respondent condi-tioning) in a functional analysis facilitates description andchange of a broader range of behaviors. The behaviorsinvolved in classical conditioning differ from those inoperant conditioning, in that they occur within the skinand are elicited by prior stimuli rather than controlled bystimuli occurring after the behavior (see Table 1 for fur-ther comparisons of operant and classical conditioning).Classical conditioning concerns how stimuli come tocontrol such involuntary or instinctive behavior (e.g., sali-

es of behavior.vation, the startle response, heart rate) through theirrelationship with other stimuli which elicit such responses.

As illustrated in Fig. 2, classical conditioning is theprocess by which a neutral stimulus comes to elicit aconditioned response. A neutral stimulus is one withwhich the person of interest has had no prior learning,and to which he or she has no unique instinctive response.For example, to a toddler, the sight of a small black-and-yellow striped flying insect might be a neutral stimulus. Aswith most new stimuli appearing in his environment,before conditioning he would respond to it with anorienting response (e.g., looking toward it, engaging inexploratory visual scanning). Before conditioning, thechild was likely born with an unconditioned (instinctive)fear response to painful stimuli such as bee venom underthe skin. If the toddler explores the sight of the bee

(neutral stimulus) too vigorously, conditioning mightoccur, with the sight of the bee paired with theunconditioned stimulus of the bees venom under theskin. During conditioning the child will exhibit theunconditioned response of crying, having an increasedheart rate, etc. After conditioning, the now conditionedstimulus, the sight of the bee, will have gained the powerto elicit a conditioned response (e.g., crying) similar tothe unconditioned response. Thus, the child will cry at thesight of a bee.

A full description of the state of theory and research onthe relationship between classical and operant condition-

ing is beyond the scope of this article. (See Allan, 1998;Forsyth & Eifert, 1998; and Schwartz & Gamzu, 1977, forsuch discussion.) However, the reader may have notedparallels between the pairing processes involved in creatingsecondary reinforcers and conditioned stimuli. Thus,classical conditioning may shed some light on how stimuliare established as reinforcers or punishers. In addition, theconditioning of a fear stimulusmay precede the escape andavoidance behavior commonly seen in anxiety disorders.In turn, operant avoidance may prevent respondentextinction, the process whereby a conditioned stimulus(e.g., the smell of stagnant water for a jungle combat

Table 1Comparing the components of classical and operant conditioning

Component Classical Operant

Behavior Elicited involuntary, instinctive, or reflexive behaviorwhich mostly occurs within the skin: e.g., salivation,accelerated heart rate, flinching

Emitted voluntary, goal-oriented, motoric behaviorwhich operates on the external environment: e.g.,running, talking, writing

Physiology Primarily involves autonomic nervous system'sregulation of smooth muscles and glands

Involves somatic nervous system's operation ofskeletal muscles

Focus Focus on stimuli Focus on response and its consequencesDirection of control Prior (conditioned) stimuli control behavior

(conditioned responses)Subsequent stimuli (reinforcers or punishers) controlbehavior

328 YomanFigure 2. Classical conditioning.

veteran), through repeated or prolonged presentation inthe absence of an unconditioned stimulus (a gory corpse),loses its power to elicit a conditioned response (fear arousal

329Functional Analysissuch as rapid heartbeat, sweating, etc.).

Types of Functional Analysis

One might describe three types of functional analysisseen in behavioral practice (Martin & Pear, 1999). Thesediffer in their level of rigor and practicality in practice. Allthree types form hypotheses about the controllingantecedents and consequences of the behavior of interest,devise an intervention based on these hypotheses, thenimplement the intervention to test the hypotheses. Ininformant-based assessment3 the practitioner collectsinformation from the client or others familiar with his/her behavior using questionnaires (e.g., Sobell, Toneatto,& Sobell, 1994), interviews, or self-monitoring. Inobservational assessment the practitioner directlyobserves the behavior of interest. Using the data collected,informant-based or observational assessments then testthe hypotheses they generate, employing an A-B design,where A represents the baseline (preexisting) strength ofthe behavior and B is the treatment or intervention phasestrength of the behavior. The gold standard is theexperimental functional analysis. In this approach thepractitioner may gather information and formulatehypotheses about the controlling antecedents and con-sequences of the behavior of interest, similar to informantand observational methods. However, he or she then usesan experimental quality within-participant design such asmultiple baseline (applying the intervention sequentiallyto several behaviors or in several settings) or A-B-A-B(reversal or return to baseline phase, replicating theintervention effect) to either confirm or reject thehypotheses. (See Blampied, 1999; Hersen, 1985; andKazdin, 1982, for further discussion of within-participantdesigns.)

Target Behavior Selection and the Context ofFunctional Analysis

As apparent above, the behavior of interest, or targetbehavior, is the focal point for functional analysis. Yet howand why practitioners and/or clients choose a particularbehavior to target is mostly neglected in writing on thetopic, and in practice. This fundamental conceptual gapin functional analysis may be the primary reason that thereliability of target behavior selection is low (Hay, Hay,Angle, & Nelson, 1979). A related implication of this gapis what Haynes et al. (1997) suggest may be [t]he mostserious threat to a functional analysis: inadequate

3Martin and Pear (1999) use the term questionnaire assessmentfor this type of analysis. However, the current author finds this termoverly restrictive.content validitya model may fail to include important behavior problems or may include irrelevant variables(p. 337).

Without a standard for judging which of the clientsbehaviors are relevant and important to a behavior changeeffort, the determinants of target behavior selectionremain implicit, arbitrary, and/or inconsistent with thefunctional perspective (i.e., the focus on the relationshipbetween behavior and environment) of functional analy-sis. This leaves open the possibility that clinicians will deema behavior change effort successful when it has changedirrelevant or unimportant behaviors. Moreover, targetbehavior selection matters because introducing a newclient behavior into his or her environment createsconsequences. Arbitrarily or erratically choosing targetbehaviors would seem to increase the risk of unintendedconsequences harmful to the client. For example, teach-ing a client how to give negative feedback to others mayactually damage rather than improve his or her relation-ships if it increases the frequency of criticizing others. Thelack of a targeting standard also creates a conceptualdisconnection between target behavior selection and therest of functional analysis. For example, if we performed afunctional analysis of clinician and client behavior inselecting a particular target behavior, it might revealavoidance by both parties (e.g., the practitioner mightnot target behaviors with which he or she has lessexperience) or the influence of short-term reinforcerssuch as approval from each other.

In behavior therapy practice with adults, typically theclients presenting complaint or his or her DSM-IV(American Psychiatric Association, 1994) diagnosis deter-mine the target behaviors. However, many factors shapethe clients presenting complaint, including emotionalarousal associated with discussing certain behaviors andthe clients ability to observe his or her own behavior.Moreover, there are several problems with DSM-IVsymptom remission as a standard for the success of abehavior change effort. First, nomothetic diagnosticcriteria may not adequately represent what is importantto an individual client. Second, and more problematic,symptom remission represents a medical model standardfor success in a behavior change effort. This is incon-sistent with functional analysis (cf. Kazdin, 1978; Wolpe,1989).

In their approach to functional analysis, Haynes et al.(1997) have clients rate the importance of variousbehaviors identified by the client or practitioner. Thisstill begs the question of the basis of those ratings. Haynes,Richard, and Kubany (1995) maintain that contentvalidity in behavioral assessment is judged according tohigher-order variables than the target behaviors them-selves. Such higher-order variables are determined at leastin part by clarifying what confers importance upon a

330 Yomanparticular behavior change targeted by the clinician orrequested by the client (cf. Haynes et al., 1997).

Perhaps the functional perspective, which servesbehavioral practitioners well in identifying other relevantvariables, can suggest higher-order variables that mightconfer importance upon a behavior. A principle inbehavior therapy has been that the most important aspectof behavior is its function, not its topography (appear-ance). For example, in early studies of rat bar pressingbehavior, it was engagement of the bar mechanism thatdispensed food which defined bar pressing, not whetherthe rat pressed with front, hind, left, or right foot (or nose,for that matter). Similarly, in human dating behavior,there are many ways to ask another person out, but whatmatters is if the person says yes. If the consequences of abehavior are central to defining it, a short conceptual leapleads to the assertion that consequences confer impor-tance upon the behavior. It is food that confersimportance on the rats bar pressing, and yeses thatconfer importance on initiation behaviors in dating.Literally, these behaviors exist and persist because of theirconsequences.

How might this functional perspective apply to targetbehavior selection? In adult behavior therapy practice,those client behaviors with the most frequent and/orsignificant undesired consequences often get priority astargets. While avoiding undesired consequences can beimportant, there are several potential limitations of thisapproach. First, it may reinforce avoidance patternsexhibited by the client. Take the example of a clientwho has problematic conflict with his co-workers becausehe frequently interrupts and talks over them. This clientmay prioritize avoiding conflict and criticism rather thanmanaging and learning from these to improve workrelationships. Second, behaviors with undesired conse-quences are best changed by replacing them withbehaviors expected to have desired consequences (i.e.,differential reinforcement of other or incompatiblebehaviors; see Martin & Pear, 1999). Returning to theclient with workplace conflict, increasing behaviors (e.g.,active listening, complimenting) likely to meet with thedesired consequences of approval and cooperation fromhis co-workers has a better chance of success than simplyreducing or eliminating his interrupting and talking over.One might then describe effective behavior therapy forinterpersonal problems as teaching the client to performbehaviors that will serve, in particular contexts, asdiscriminative stimuli for desired reinforcing responsesfrom others. Thus, clarifying the desired consequences ofclient behavior change identifies important target beha-viors that might lead to those consequences, and canincrease the likelihood of change effort success.

Unfortunately, clients may be trapped by short-termdesired consequences that reinforce behavior incompa-tible with their long-term success and happiness. This is acommon conceptualization of client presenting pro-blems. For example, while drug abuse may lead to intensedesired short-term consequences (e.g., euphoric sensa-tions), these consequences powerfully reinforce drug-seeking behavior which often leads to long-term aversivefinancial and relationship consequences. Similarly, meet-ing numerous partners to engage in casual sex may resultin intense desired short-term consequences (e.g., sexualorgasms, flirtation from others) but often leads to aversivelong-term problems (e.g., sexually transmitted diseases,unwanted pregnancies, delay in finding a committedlong-term relationship). Thus, clarifying the long-termdesired consequences of behavior change identifiesimportant target behaviors that can lead to moreprofound and sustainable change.

A variety of behavioral authors (e.g., Evans, 1997;Hawkins, 1986; Hayes, Strosahl, & Wilson, 1999; Kanfer &Schefft, 1988; Yoman & Edelstein, 1994) have advocatedfor targeting behaviors in the context of their desiredlong-term consequences. Similarly, Rosen and Proctor(1981) devised the term ultimate outcome to describesufficient conditions for treatment to be terminatedand considered a success (p. 419). These targetingapproaches take into account the frequent role of short-term consequences in behavior problems and of long-term consequences in defining human successes (e.g.,happy marriage, career advancement). A regular practiceof defining long-term consequences may function as rule-governed behavior for both client and practitioner, in thatit creates a verbal description of long-term contingenciesthat may decrease the control that short term conse-quences have over their behavior (see Malott, 1989).

Yoman and Edelstein (1994) refined Rosen andProctors (1981) ultimate outcome from a behavioralperspective to refer to the valued (usually long-term) re-sults clients obtain from change in the behavior of inte-rest. Yoman and Edelstein encouraged preestablishedultimate outcomes as the touchstone for determining theimportance of potential target behaviors. They describedtwo types of ultimate outcomes commonly encounteredin behavior therapy: interpersonal and intrapersonal.Interpersonal ultimate outcomes involve the reaction to aclients behavior change by a key person in the clients life(e.g., a prospective friend more frequently seeking socialcontact with the client). Intrapersonal ultimate outcomesinvolve the clients own reaction (usually a private eventor involuntary response) to his or her behavior change(e.g., weight loss, increased life satisfaction).

Several authors (Hayes et al., 1999; Yoman & Edelstein,1994) equate long-term consequences with values. Focus-ing on ultimate outcomes that reflect the best interestsand values of the client fulfills important ethical obliga-tions of the helping professions (e.g., American Psycho-

ultim

331Functional Analysislogical Association, 2002; National Association of SocialWorkers, 1999). Identifying ultimate outcomes of beha-vior change, therefore, incorporates into functionalanalysis the prevalent practice of values clarification inprofessional helping relationships (see Kanfer & Schefft,1988). While professionals have no empirical basis forrecommending which values clients should choose, theycan advise clients which behaviors will tend to furtherthose values. For example, our current state of knowledgeallows us to state with some confidence which behaviorscontribute to marital satisfaction, but not whether a clientshould desire marital satisfaction, even though thatchoice may lead to a longer life (which the client may

Figure 3. Defineor may not desire). Similarly, we have some informationon which behaviors a client may want to master toestablish friendships, but not whether the client shoulddesire friends who are fun, friends who are intelligent,friends who share his or her commitment to a particularsocial cause, or none of the above.

As discussed below, defining ultimate outcomes ofintervention may guide not only selection of targetbehaviors, but specification of their optimal topography,frequency, intensity, and timing. Defining with the clientthese hoped for consequences of the new behaviors is ahelpful starting point for functional analysis.

Functional Diagnoses

Hypotheses generated by functional analysis can betermed functional diagnoses. Functional diagnosesprovide names for the types of functional relationshipshypotheses describe in detail. (See Ferster, 1965;Hawkins, 1986; Martin & Pear, 1999; and Tryon, 1996,for overviews of similar classifications of functionalrelationships.) Figs. 3 through 9 together depict a flowchart for functional analysis, with functional diagnosesappearing inside ellipses in each figure. After definingultimate outcomes and the behavior of interest, thepractitioner follows the flow chart in these figures (oftenthrough an increasingly complex series of functionaldiagnoses) to a specific behavioral intervention indi-cated by the functional analysis. Interventions4 appearin rectangles on the right-hand side of each figure inthe flow chart. As suggested above, the eventualeffectiveness of the intervention indicated providesevidence to confirm or disconfirm the functionaldiagnosis (i.e., to convert the diagnosis from provi-sional to firm or final).

ate outcomes.Functional Analysis Step by Step

There is little consensus in the field as to how tointegrate the above approaches into a reliable, validfunctional analysis. This may in part contribute to theunreliability of functional diagnosis in practice (seeAlberts, Edelstein, Yoman, & Breitenstein, 1989; Felton& Nelson, 1984). The protocol below may both assist withmore reliable functional diagnosis (see Cone, 1997) andserve as a point of departure for functional analysistailored to individual clients and settings.

Step 1: Define the Valued Long-term Consequences

(Ultimate Outcomes) of Behavior Change

Given that there is no established procedure in functional analysis for defining ultimate outcomes with a

4 Description of interventions is beyond the scope of this articleFor such description, the reader is referred to Martin and Pear (1999)and Plaud and Eifert (1998).-

.

client, a suggested interviewing technique of follow theresults is described below.5 This technique leads theclient from anticipated desired short-term consequencesof behavior change to valued long-term consequences or

332 Yomanultimate outcomes (see Fig. 3).Functional analysis can begin by the practitioner

asking the client to describe his or her hoped-for resultsof therapy. After each client statement of intended resultsthe practitioner repeats the question, What are theintended results of that?, or something functionallyequivalent, until the client states an ultimate outcome.Arrival at an ultimate outcome is usually signaled by theclient stating that the intended result of the previousconsequence (i.e., the ultimate outcome) is somethingakin to happiness, life satisfaction, or making the world abetter place. Here is an example of how the interviewmight go:

PRACTITIONER: What results do you hope for fromtherapy?

CLIENT: I want to be more assertive.

PRACTITIONER: What benefit do you anticipate frombeing more assertive?

CLIENT: My co-workers wont push me around somuch.

PRACTITIONER: And what would result from that?

CLIENT: My boss might use some of my ideas once in awhile.

PRACTITIONER: And what would be the benefit of that?

CLIENT: He might see that I have some value to thecompany.

PRACTITIONER: And what would be the benefit of that?

CLIENT: I might get that raise Ive been waiting for.

PRACTITIONER: And what would be the benefit of that?

CLIENT: I could afford tomove to a better neighborhood.

PRACTITIONER: And what result do you hope for frommoving?

CLIENT: I dont know. Life would just be better in moreattractive surroundings.

Once the client makes a general statement of a pro-spective ultimate outcome, the practitioner can define

5 One might also integrate into functional analysis the alternatemethods of identifying desired long-term consequences suggested inQuality of Life Therapy (Frisch, 2006) and Acceptance and Commit-ment Therapy (Hayes et al., 1999).the ultimate outcome in behavioral measurable terms.In the example above, the ultimate outcome might bestated, The owner of a home in a neighborhood withat least three young couples, one park, and a below-average crime rate will sign over the title to the clientwithin 12 months.

Further examples of ultimate outcomes follow. A clientwith obsessive-compulsive hand washing might becomeinterested in decreasing the frequency of this targetbehavior because the time it takes away from work tasksthreatens her ultimate outcome of her boss retaining herwith satisfactory performance evaluations in her job as arestaurant hostess. Alternatively, a practitioner for ahusband coming to therapy for marital problems mightbecome interested in the target behavior of emotionalvalidating because increasing these behaviors is likely tocontribute to productive problem discussion by theclients wife. In turn, engagement in productive pro-blem-solving discussion by both partners might result inultimate outcomes such as resolution of specific maritalproblems (e.g., conflicts over budgeting), availability ofmoney for family needs, and increased marital satisfaction(e.g., as measured by the Dyadic Adjustment Scale).Defining the ultimate outcome may suggest related targetbehaviors to accelerate (e.g., requesting a meeting withones boss).

Thus, following the results will uncover chains ofbehaviors and results for discussion by the client andpractitioner. These may include undesired conse-quences to eliminate, desired consequences to pursue,and interpersonal or intrapersonal ultimate outcomes.The clients role is to choose/prioritize which of thesehe or she would like as the focus of therapy. Thepractitioners role is to lend his or her expertise inidentifying to which results the practitioners skills arelikely to contribute (i.e., the extent to which thepractitioner is educated and experienced in interven-tions for identifying, shaping, and reinforcing behaviorsinstrumental to that ultimate outcome). The practi-tioners job is also to be vigilant for avoidance andshort-term reinforcement as obstacles to the clientspursuit of long-term desired consequences. Thisincludes discussing the benefits of long-term desiredconsequences as a focus for the treatment plan.

In this manner, the practitioner can set the stage forfunctional analysis by clarifying with the client the valuedends toward which behavior change will be the means.The practitioner can then examine the relationshipsbetween these ultimate outcomes and prospective targetbehaviors.

Step 2: Define the Behavior of Interest

Defining the valued long-term consequences of beha-vior change informs the process of defining the behavior

of interest or target behavior in observable measurableterms (see Fig. 4). Defining the target behavior beginswith a discussion between client and practitioner aboutwhich behaviors are instrumental in the clients livingenvironment in attainment of a defined ultimate out-come. From that list of behaviors, target behaviors will bethose which the client has difficulty performing with afrequency, intensity, duration, etc., thought necessary forultimate outcome attainment. For example, with theabove ultimate outcome of obtaining a home in a betterneighborhood, the practitioner might explore with theclient other behaviors that might contribute to that result,such as applying for a job with a different company. If theclient maintains his interest in working for the samecompany, the practitioner might also explore alternativesto attaining some of the lower-level results, such as theboss acknowledging his value to the company (e.g., clientmight ask his boss for a discussion of his performance inwhich he presents his accomplishments, or ask his boss fora one-on-one meeting in which the client presents hisideas). Discussion and functional analysis of obstacles tothese alternative strategies might identify behavioralexcesses or deficits that hold the client back in attainingseveral desired results. Ongoing monitoring of the

hand-washing example, hand washing might be definedas applying water, soap, or another cleansing product tothe hands. Because washing is a motoric behavior(involving muscles in the hands and arms) that has animpact on the external environment (removing dirt andoils from the hands), it is diagnosed as an operant (seeFig. 4). In the marital example, the emotionallyvalidating behaviors referenced above can be definedto include (1) summary of partners emotions andpartners stated reasons for them, confirmed by thepartner as accurate and (2) statement that the partnersemotions are understandable, given the circumstances,both stated using an empathic vocal tone. Defining thebehavior of interest allows the practitioner to begin theprocess of functional diagnosis. Fig. 4 contains the startingpoint for functional diagnosis, classifying the topographyof the behavior. Because validating is a motoric behavior(involving muscles in the tongue, throat, and jaw) thataffects the external environment (by creating sound waveswhich cause vibrations in the structures of the listenersinner ear), it is diagnosed as an operant.

Step 3: Collect Data from Informants, and Through

Direct Observation, on the Behavior of Interest, and Its

ds co

333Functional Analysiscontext, frequency, and consequences of the clientsperformance of these behaviors might help further refinetargets for intervention.

Defining the behavior of interest makes more likely thecollection of reliable data regarding the behavior. In the

Figure 4. Define the behavior of interest. Note: Decisional diamondiagnoses.Antecedents and Consequences

Self-monitoring, and/or direct observation during ajob site visit by the practitioner, might establish a baselinefrequency of the hand-washing behavior as well asrevealing its antecedents and consequences. To proceed

ntain functional assessment questions. Ellipses contain functional

with the example of emotionally validating behaviors inmarital therapy, the practitioner might determine thefrequency of the behavior of interest and its antecedentsand consequences by having the husband self-monitorvalidating at home or having his wife complete a behaviorchecklist on his validating. Alternatively, the practitionermight observe and record the frequency of validatingbehaviors and their antecedents and consequences intherapy sessions or in an unstructured problem discussion(cf. Jacobson & Margolin, 1979).

Step 4: Formulate Provisional Functional Diagnoses

(Hypotheses) Regarding the Behavior of Interest

Collection of data in Step 3 enables further refinementof functional diagnoses. In the hand-washing example,the data collected above would likely reveal that thebehavior is too strong to ensure the clients retention ofher job; thus, it is a behavioral excess (see Fig. 5). Observ-ations and discussions with the couple in the maritalexample may suggest that validating is occurring tooinfrequently to enable attainment of the couples ultimateoutcomes described above (e.g., once a week, when thecouple discusses problems several times a week). Thus,validating is too weak and constitutes a behavioral deficit(see Fig. 5).

Step 5: Refine Provisional Functional Diagnoses Based

on Further Data Collection

Fig. 5 directs the practitioner working with the clientshand washing to Fig. 6: Functional Diagnoses andInterventions Involving Consequences and OperantBehavior Excesses, given that hand washing has beenestablished as an operant behavioral excess. The firstdecisional diamond in that figure addresses whether thebehavior interferes with ultimate outcome attainmentonly under specific stimulus conditions. Let us say thepractitioner answers this question affirmatively becausethe client describes the hand washing as a problem only atwork, and hand washing in moderation is a healthful partof personal hygiene and job performance. Fig. 6 thendirects the practitioner to Fig. 7: Functional DiagnosesInvolving Stimulus Control of Operant Behavior. Fig. 7suggests that operant behavioral excesses are a possiblecase of dysfunctional stimulus control.

Continuing the marital case example from the func-tional diagnosis of behavioral deficit in Fig. 5, the prac-titioner might ask the husband and his wife if he has evervalidated her emotional expressions, or simply ask him todo so in a session and see if he is able. For example, thehusband may have actually mastered validating in a pastmarital workshop the couple attended, but he has stopped

al dia

334 YomanFigure 5. Refine functional diagnoses for operants. Note: Decisionfunctional diagnoses. Rectangles contain interventions.monds contain functional assessment questions. Ellipses contain

Figure 6. Functional diagnoses and interventions involving consequences and operant behavioral excesses. Note: Decisional diamondsal dia

335Functional Analysisvalidating since then. This would refine the diagnosis ofvalidating to a performance deficit (see Fig. 5), from which

contain functional assessment questions. Ellipses contain functionthe flow chart directs the practitioner to Fig. 8: Functional

Figure 7. Functional diagnoses and interventions involving stimulus cfunctional assessment questions. Ellipses contain functional diagnoses.Diagnoses Involving Consequences and Operant Perfor-mance Deficits. Continuing to refine the diagnosis of per-

gnoses. Rectangles contain interventions.formance deficit in Fig. 8, the flow chart indicates two

ontrol of operant behavior. Note: Decisional diamonds containRectangles contain interventions.

uencal dia

336 Yomanquestions to answer about the behavior. The first question is:Must the behavior occur under specific stimulus conditions

Figure 8. Functional diagnoses and interventions involving conseqcontain functional assessment questions. Ellipses contain functionto contribute to the ultimate outcome?With no research orcase data available to answer this question, the practitionermay parsimoniously answer no. This would direct thepractitioner to the second question in the flow chart, Areaversive stimuli following the behavior? Perhaps self-monitoring by the husband helps him and the practitionerdiscover that, when he validates, his wife proceeds with along monologue about how she feels on many topicsunrelated to the problem at hand. The husband feelsfrustrated with this, both because the problem does not getsolved and because he gets little or no time to express hisopinion about the problem. The practitioner then hypothe-sizes that the wifes monologues punish the husbandsvalidating: the functional diagnosis is punishment (see Fig.8). Note that the specific punisher heremay bewithdrawal ofopportunities for the husband to express himself and solveproblems (i.e., the wifes cessation of pauses in speech and/or responses to the husbands attempts to speak, as opposedto presentation of her monologues per se), a case of negativepunishment or response cost (see Fig. 1).

Step 6: Collect Data as Needed to Indicate a Specific

Intervention

Self-monitoring reveals that, while the sensation of adoorknob touching her hand is an antecedent to theclients hand washing, it often follows mere thoughts aboutgerms when there has been no doorknob touching. Thus,in response to the next question in Fig. 7, it seems that

es and operant performance deficits. Note: Decisional diamondsgnoses. Rectangles contain interventions.the stimulus cannot be weakened by the practitioner, sostimulus discrimination training is indicated. In the maritalexample, the diagnosis of performance deficit due topunishment leads directly to two interventions: removal ofthe punisher and introduction of reinforcers (see Fig. 8).

Step 7: Implement the Indicated Intervention to

Confirm or Disconfirm the Functional Diagnosis(Test the Hypothesis)

In the hand-washing case, the practitioner mightinstruct the client on the recommended stimulus condi-tions for hand washing (e.g., finishing toilet use, prior toeating or handling food) and have the client monitor theextent to which she washed her hands under only theseconditions. The practitioner could test graphic datapresentation as a reinforcer for improved discrimination.This procedure might reinforce self-monitoring by theclient, which can help her persist in a behavior changeeffort. Self-monitoring can be gradually faded as naturalreinforcers such as improved performance evaluationscome to control the behavior.

The practitioner in the marital example would removethe punisher by interrupting the wifes monologues insessions, coaching the couple to establish cues for endingmonologues, and suggesting the wife pause betweentopics, and teaching her to discriminate her husbands

The practitioner would intervene with a response pre-

337Functional Analysisvention protocol to break this cycle of reinforcement ofexcessive hand washing.

In the example of the husbands validating notmaintaining over time, the practitioner might examinedata from the intervention phase of marital therapy anddiscover that the wife was responding in a sarcastic vocaltone to some of the husbands validating statements. Thehusband in turn eventually decreased the overallfrequency of those statements. The functional diagnosisof operant performance deficit due to punishment wouldlikely remain, but the practitioner might revisit herdecision to answer negatively during Step 5 the firstquestion for performance deficit in Fig. 8: Must thebehavior occur under specific stimulus conditions tocontribute to the ultimate outcome? Exploration mayreveal that the wife responds to the husbands validatingwith a sarcastic tone when the couple discusses anonproblem situation and the wife discloses milder orpositive emotions. Thus the answer to the question maywell be yes, validating only contributes to ultimateoutcome attainment for this couple (and the wife onlyattempts to speak. An interactive interpretation (cf.Jacobson & Margolin, 1979) to the couple about howthe husbands validation is followed by the wifes mono-logues might prove an effective antecedent to these inter-ventions. The practitioner would introduce reinforcers forthe husbands validating by coaching the wife to disclosepositive feelings about the validating, then immediatelyasking the husband for his thoughts and feelings about theproblem each time he validates her feelings. (Note thatthe above process goals and interventions might them-selves be subjected to functional analysis to improve thefocus and efficiency of the therapy process. See Edelstein& Yoman, 1991; Kohlenberg, Tsai, & Dougher, 1993;Lejuez, Hopko, Levine, Gholkar, & Collins, 2006.)

To confirm or disconfirm the functional diagnosis thepractitioner would implement the above interventions andobserve (or have the client self-monitor) whether the hus-bands validating reaches the targeted frequency. Perhapsthe practitioner would discover that validating reached itstargeted frequency, but then did not maintain at that level.

Step 8: Recycle Through Steps 4 to 7 as Needed Until

Discovery of an Effective Intervention

In the hand-washing example, the practitioner wouldrecycle through Steps 4 to 7 to cover the other directionindicated for operant behavioral excess in Fig. 6. Thisconcerns whether ongoing aversive stimuli are terminatedby the behavior. The client may report that hand washingterminates various physiological signs of anxiety (e.g.,sweating palms, rapid heartbeat, restricted breathing).This would identify hand washing as escape/avoidance.reinforces it) when it happens in the context of problemdiscussions and strong negative emotional expressions.

To continue to refine the functional diagnosis for thevalidating, given this situational specificity, the flow chartthen directs the practitioner to Fig. 7: FunctionalDiagnoses Involving Stimulus Control of Operant Beha-vior. Fig. 7 indicates insufficient stimulus control as thefunctional diagnosis when there are antecedent problemsin performance deficits. The practitioner might thenremedy insufficient stimulus control by giving the couplean interactive interpretation of the lack of maintenance ofthe husbands validating, interrupting session discussionsto ask the husband to interpret his wifes nonverbalemotional cues (checking with his wife for accuracy), andteaching the wife to prompt the husband to read heremotional cues (e.g., What is she feeling right now?) ifhe is not validating when she expresses a strong emotion.The practitioner can confirm this new diagnosis ofinsufficient stimulus control if these interventions resultin amaintained change in the husbands rate of validating.

Note that the practitioners job is not finished until he orshe also confirms the hypothesis of the instrumentality ofthe behavior of interest in attaining the ultimate outcome.In the hand-washing example, the practitioner might askthe client to relay her bosss comments about her work, payraises, etc., and, once hand washing has decreased infrequency and come under stimulus control, determinewhether the decrease in hand washing has taken theclients job out of jeopardy. In the marital case the prac-titioner would do this by assessing marital problemresolution, marital satisfaction, and availability of moneyfor family needs (to the extent that the latter can beaffected by the couples communication) once the targetedfrequency of the validating behavior had been reached.

The Continuing Role of Functional Analysis inBehavior Therapy

The above discussion and Figs. 3 through 9 illustrate howfunctional analysis can link client presenting problems toempirically supported interventions. As a scientificapproach to the individual case, it may enable severalimportant improvements to themanualized treatments thatcomprise the current standard for evidence-based practice.Behavior therapists have historically been champions ofspecific interventions for specific problems (e.g., Chamb-less, 2002). Functional analysis is a valuable tool in bothdemonstrating the effectiveness and increasing the powerof those interventions (e.g., Rowan, Holborn, Walker, &Siddiqui, 1984) because it offers more specificity about theindividuals problems than does DSM-IV diagnosis. Forexample, a DSM-IV diagnosis of major depression may berepresented in the individual case as some combination ofoperant extinction of certain behaviors and avoidance ofcertain situations (cf. Dimidjian et al., 2006).

ondituch anal d

338 YomanFunctional analysis is based on the assumption thatcausal factors in the clients problems are contextual andidiosyncratic, and only partially held in common with alarge group of other individuals with whom he or sheshares a diagnosis. In fact, there may be no evidence forthe effectiveness of a manualized treatment for as many as83% of clients whose primary diagnosis would seem toindicate the treatment. Ruscio and Holohan (2006)report that about two-thirds of potential participants areexcluded from psychotherapy outcome studies becausesuch problems as substance abuse or physical illnessaccompany their primary diagnosis. Moreover, among

Figure 9. Interventions for conditioned responses. Note: Most cdepicted in this figure. Conditioned response behavioral deficits stherefore were not depicted in the figure to conserve space. Decisiofunctional diagnoses. Rectangles contain interventions.those not excluded from outcome studies of depression,for example, only about half of patients respond to anygiven treatment (Hollon, Thase, & Markowitz, 2002, p.70). This brings into question the degree to which thecausal models and assessment targets suggested in groupresearch apply to the individual client, and raises thepossibility of greater success with individualizedapproaches based on functional analysis (cf. Haynes etal., 1997). For example, functional analysis may help thepractitioner discover how to disentangle complex pro-blems, improve intervention acceptability, decrease drop-out, and help nonresponders respond. Moreover, it wouldseem an invaluable tool for maintaining a scientificapproach to clinical phenomena for which no evidence-based assessment or intervention methods exist.

The advantages of functional analysis are grounded inits defining qualities. For example, in contrast to cognitivemethods of assessment, functional analysis focuses onobservable behavior change, which not only facilitatesmaintenance of a scientific approach, it directs the prac-titioners efforts where they arguably make the mostdifference. In the hand washing example above, a cognitiveapproach would identify what if thoughts, perhaps aboutcontamination, and then core beliefs which supposedlyunderlie them. A cognitive intervention would then guidethe client to dispute these. However, there is muchstronger evidence for exposure with response preventionfor such a behavior pattern, the intervention to whichfunctional analysis led. In contrast to social learningapproaches to assessment, which rely largely on role-playedsimulation of life situations to promote skill acquisition,functional analysis focuses on naturalistic reinforcers foractual performance of a behavior in a context where it

ioned responses in behavior therapy are behavioral excesses ass functional sexual impotence are much less commonly seen andiamonds contain functional assessment questions. Ellipses containis likely to be effective. In the case of the husbands valid-ating, a social learning approach might suggest repeatedrehearsal with his wife in session or in at-home commu-nication sessions, to increase mastery of the validatingresponse. This approach might still overlook the differ-ences between the arbitrary rehearsal situations and thereal-life situations in which the husband is called upon tovalidate. Moreover, it does not carefully assess the real-lifeconsequences of validating as does functional analysis, andthus may treat a performance deficit as a skills deficit.

Functional analysis has also been described by severalauthors (e.g., Edelstein & Yoman, 1991; Kohlenberg et al.,1993; Lejuez et al., 2006) as a tool for understanding andenhancing the therapeutic alliance. A common analyticalbase may help integrate such work with interventions forspecific target behaviors. While cognitive strategies foranalyzing and enhancing the therapeutic alliance havebeen identified, Castonguay, Goldfried, Wiser, Raue, andHayes (1996) found that treating therapeutic alliancestrains as a manifestation of the clients distortedthoughts led to further alliance strains. The success of

339Functional Analysisseveral therapeutic systems utilizing a functional analyticperspective on the alliance (see Lejuez et al., 2006)suggests a focus on contextual factors in the therapeuticrelationship (e.g., to which client behaviors the therapistattends and responds) may hold more promise for ad-vancing this important aspect of psychotherapy.

When targeted behavior change occurs, the work ofthe practitioner and client may not be complete. Forexample, Hollon et al. (2002) report that only a third ofparticipants in depression outcome studies remainresponders at long-term follow-up. Functional analysisoffers a framework for maintaining and transferring tonew environments the gains made by specific treatments(see Baer et al., 1968, 1987). It can identify not justcurrent maintaining variables for undesired behavior, butpotential maintaining variables for desired behavior. Thisinformation can then be used to educate clients on thereciprocal nature of causation, and prepare them andtheir environments for future challenges. For example, apractitioner suggested that a client who entered therapyto get help with loneliness increase participation in eventsand organizations that reflected the clients interests,where it was anticipated others were more likely to rein-force the clients initiation and invitations. Weekly dis-cussion of the clients social initiation diary includedguiding the client in discriminating encouraging signalsfrom acquaintances, experimenting with timing initia-tions and invitations, and identifying friendly environ-ments based on successes and failures.

Once researchers establish empirically supportedinterventions for groups or individuals, organizationalcontingencies will determine whether those interventionsare disseminated. Functional analysis generates valuablestrategies at an organizational level (e.g., Krapfl &Gasparotto, 1982) to ensure that practitioners effectiveimplementation of those strategies is followed by reinfor-cing consequences in both the short and long-term.

Finally, as a scientific approach to the individual case,functional analysis is a helpful tool for integratingresearch and practice, which presents untapped oppor-tunities for advancing the science of behavior. Withinparticipant designs provide opportunities for clinicians toproduce and publish research, and for research toincorporate more of the concerns of clinical settings.

Functional analysis allows a more fine-grained exam-ination of the effectiveness of treatment manuals withindividual cases, and may thus generate hypotheses forimproving those manuals for nonresponders. Functionalanalysis may be the primary guide for practitioners inuncharted territory, where no evidence-based interven-tion exists for a clients problems.

Hopefully, further research will improve functionalanalysis itself. More study is needed to improve contentvalidity of functional analysis and reliability of target be-havior selection, two overlapping and fundamental con-tributors to the utility of functional analysis. Defining thevalued long-term consequences of a behavior change effortmay help with content validity and target selection re-liability, better connect the target selection process with theresearch upon which functional analysis is founded, shedlight on the interrelationships among long-term conse-quences, and lead to valuable new discoveries about whichbehaviors further which valued long-term consequences.

While some may believe functional analysis to be out-moded or irrelevant, it remains a key to unlock uniqueknowledge for behavioral and cognitive-behavioral practi-tioners and scientists. It is still their most valuable tool forconceptualizing cases, and identifying and demonstratingthe active ingredients of interventions with individuals. Insum, functional analysis seems well-suited to answering Pauls(1969) enduring question: what treatment, by whom, ismost effective for this individual, with that specific problem,under which set of circumstances, and how ? (p. 62).

References

Albano, A. M., & Morris, T. L. (1998). Childhood anxiety, obsessive-compulsive disorder, and depression. In J. J. Plaud, & G. H. Eifert(Eds.), From behavior theory to behavior therapy (pp. 203222).Boston: Allyn & Bacon.

Alberts, G., Edelstein, B., Yoman, J., & Breitenstein, J. (1989). Thebehavioral assessment interview: Interassessor agreement on caseand treatment formulation. Public Service Psychology, 14, 15.

Allan, R.W. (1998).Operant-respondent interactions. InW.ODonohue(Ed.), Learning and behavior therapy (pp. 146168). Boston: Allyn &Bacon.

American Psychiatric Association. (1994). Diagnostic and statisticalmanual of mental disorders, 4th ed. Washington, DC: Author.

American Psychological Association. (2002). Ethical principles of psycho-logists and code of conduct. American Psychologist, 57, 10601073.

Azrin, R. D., & Hayes, S. C. (1984). The discrimination of interestwithin a heterosocial interaction: Training, generalization, andeffects on social skills. Behavior Therapy, 15, 173184.

Baer, D. M., Wolf, M., & Risley, T. R. (1968). Some current dimensionsof applied behavior analysis. Journal of Applied Behavior Analysis, 1,9197.

Baer, D. M., Wolf, M., & Risley, T. R. (1987). Some still currentdimensions of applied behavior analysis. Journal of Applied BehaviorAnalysis, 20, 313327.

Blampied, N. M. (1999). A legacy neglected: Restating the case forsingle-case research in cognitive-behavioral therapy. BehaviorChange, 16, 89104.

Bootzin, R. R., & Epstein, D. R. (2000). Stimulus control. In K. L.Lichstein, & C. M. Morin (Eds.), Treatment of late-life insomnia(pp. 167184). Thousand Oaks, CA: Sage.

Castonguay, L. G., Goldfried, M. R., Wiser, S., Raue, P. J., & Hayes, A. M.(1996). Predicting the effect of cognitive therapy for depression: Astudy of unique and common factors. Journal of Consulting andClinical Psychology, 64, 497504.

Chambless, D. L. (2002). Beware the dodo bird: The dangers ofovergeneralization. Clinical Psychology: Science and Practice, 9, 1316.

Cone, J. D. (1997). Issues in functional analysis in behavioralassessment. Behaviour Research and Therapy, 35, 259275.

Cooper, J. O., Heron, T. E., & Heward, W. L. (1987). Applied behavioranalysis. Columbus, OH: Merrill Publishing.

Dimidjian, S., Hollon, S. D., Dobson, K. D., Schmalling, K. B.,Kohlenberg, R. J., Addis, M. E., Gallop, R., McGlinchey, J. B.,Markley, D. K., Gollan, J. K., Atkins, D. C., Dunner, D. L., &Jacobson, N. S. (2006). Randomized trial of behavioral activation,cognitive therapy, and antidepressant medication in the acute

340 Yomantreatment of adults with major depression. Journal of Consultingand Clinical Psychology, 74, 658670.

Edelstein, B. A., & Yoman, J. (1991). Behavioral interviewing [Laentrevista conductual]. In V. E. Caballo (Ed.),Manual de tecnicas deterapia y modicacion de conducta (pp. 751775). Madrid, Spain: SigloVeintiuno.

Ekman, P., & Friesen, W. V. (1982). Felt, false, and miserable smiles.Journal of Nonverbal Behavior, 6, 238258.

Evans, I. M. (1997). The effect of values on scientific and clinicaljudgment in behavior therapy. Behavior Therapy, 28, 483493.

Felton, J. L., & Nelson, R. O. (1984). Interassessor agreement onhypothesized controlling variables and treatment proposals.Behavioral Assessment, 6, 199208.

Ferster, C. B. (1965). Classification of behavioral pathology. In L.Krasner, & L. P. Ullman (Eds.), Research in behavior modification(pp. 626). New York: Holt, Rinehart, & Winston.

Ferster, C. B. (1972). An experimental analysis of clinical phenomena.Psychological Record, 22, 116.

Forsyth, J. P., & Eifert, G. H. (1998). Phobic anxiety and panic: Anintegrative behavioral account of their origin and treatment. InJ. J. Plaud, & G. H. Eifert (Eds.), From behavior theory to behaviortherapy (pp. 3867). Boston: Allyn & Bacon.

Frisch, M. B. (2006). Quality of Life Therapy: Applying a life satisfactionapproach to positive psychology and cognitive therapy. New York: JohnWiley & Sons.

Hawkins, R. P. (1986). Selection of target behaviors. In R. O. Nelson, &S. C. Hayes (Eds.), Conceptual foundations of behavioral assessment(pp. 331385). New York: Guilford Press.

Hay, W. M., Hay, L. R., Angle, H. V., & Nelson, R. O. (1979). Thereliability of problem identification in the behavioral interview.Behavioral Assessment, 1, 107118.

Hayes, S. C., Strosahl, K. D., & Wilson, K. G. (1999). Acceptance andcommitment therapy: An experiential approach to behavior change. NewYork: Guilford Press.

Haynes, S. N., Leisen, M. B., & Blaine, D. D. (1997). Design ofindividualized behavioral treatment programs using functionalanalytic clinical case models. Psychological Assessment, 9, 334348.

Haynes, S. N., Richard, D. C. S., & Kubany, E. (1995). Content validityin psychological assessment: A functional approach to conceptsand methods. Psychological Assessment, 7, 238247.

Hersen, M. (1985). Single-case experimental designs. In A. S. Bellack,M. Hersen, & A. E. Kazdin (Eds.), International handbook of behaviormodification and therapy (pp. 85121). New York: Plenum.

Hollon, S. D., Thase, M. E., & Markowitz, J. C. (2002). Treatment andprevention of depression. Psychological Science in the Public Interest, 3,3977.

Hopko, D. R., Lejuez, C. W., Ruggiero, K. J., & Eifert, G. H. (2003).Contemporary behavioral activation treatments for depression:Procedures, principles, and progress. Clinical Psychology Review, 23,699717.

Jacobson, N. S., Dobson, K. S., Truax, P. A., Addis, M. E., Koerner, K.,Gollan, J. K., Gortner, E., & Prince, S. E. (1996). A componentanalysis of cognitive-behavioral treatment for depression. Journalof Consulting and Clinical Psychology, 64, 295304.

Jacobson, N. S., &Margolin, G. (1979).Marital therapy: Strategies based onsocial learning and behavior exchange principles. New York: Bruner/Mazel.

Kanfer, F. H., & Schefft, B. K. (1988). Guiding the process of therapeuticchange. Champaign, IL: Research Press.

Kazdin, A. E. (1978). History of behavior modification. Baltimore:University Park Press.

Kazdin, A. E. (1982). Single-case research designs: Methods for clinical andapplied settings. New York: Oxford University Press.

Kohlenberg, R. J., Tsai, M., & Dougher, M. J. (1993). The dimensions ofclinical behavior analysis. Behavior Analyst, 16, 271282.

Krapfl, J. E., & Gasparotto, G. (1982). Behavioral systems analysis.In L. W. Frederiksen (Ed.), Handbook of organizational behaviormanagement (pp. 2138). New York: Wiley & Sons.

Lejuez, C. W., Hopko, D. R., Levine, S., Gholkar, R., & Collins, L. M.(2006). The therapeutic alliance in behavior therapy. Psychotherapy:Theory, Research, Practice, and Training, 42, 456468.

Malott, R. W. (1989). The achievement of evasive goals: Control byrules describing contingencies that are not direct-acting. In S. C.Hayes (Ed.), Rule-governed behavior: Cognition, contingencies, andinstructional control (pp. 269322). New York: Plenum.

Martin, G., & Pear, J. (1999). Behavior modification: What it is and how todo it. Upper Saddle River, NJ: Prentice Hall.

National Association of Social Workers. (1999). NASW code of ethicsWashington, DC: NASW Press.

Paul, G. L. (1969). Behavior modification research: Design andtactics. In C. M. Franks (Ed.), Behavior therapy: Appraisal and status(pp. 2962). New York: McGraw-Hill.

Plaud, J. J., & Eifert, G. H. (Eds.). (1998). From behavior theory to behaviortherapy Boston: Allyn & Bacon.

Rosen, A., & Proctor, E. K. (1981). Distinctions between treatmentoutcomes and their implications for treatment evaluation. Journalof Consulting and Clinical Psychology, 49, 418425.

Rothbaum, B. O., Meadows, E. A., Resick, P., & Foy, D. W. (2000).Cognitive-behavioral therapy. In E. B. Foa, & T. M. Keane (Eds.),Effective treatments for PTSD: Practice guidelines from the International Societyfor Traumatic Stress Studies (pp. 6083). New York: Guilford Press.

Rowan, V. C.,Holborn, S.W.,Walker, J. R., & Siddiqui, A. R. (1984). A rapidmulticomponent treatment for an obsessive compulsive disorder.Journal of Behavior Therapy and Experimental Psychiatry, 15, 347352.

Ruscio, A. M., & Holohan, D. R. (2006). Applying empiricallysupported treatments to complex cases: Ethical, empirical, andpractical considerations. Clinical Psychology: Science and Practice, 13,146162.

Schwartz, B., & Gamzu, E. (1977). Pavlovian control of operantbehavior. In W. K. Honig, & J. E. R. Staddon (Eds.), Handbook ofoperant behavior (pp. 5397). Englewood Cliffs, NJ: Prentice-Hall.

Scotti, J. R., Mullen, K. B., & Hawkins, R. P. (1998). Child conduct anddevelopmental disabilities: From theory to practice in thetreatment of excess behaviors. In J. J. Plaud, & G. H. Eifert(Eds.), From behavior theory to behavior therapy (pp. 172202).Boston: Allyn & Bacon.

Sidman, M. (1960). Tactics of scientific research. New York: Basic Books.Skinner, B. F. (1953). Science and human behavior. New York: Macmillan

Publishing.Sobell, L. C., Toneatto, T., & Sobell, M. B. (1994). Behavioral

assessment and treatment planning for alcohol, tobacco, andother drug problems: Current status with an emphasis on clinicalapplications. Behavior Therapy, 25, 533580.

Sulzer-Azaroff, B., & Pollack, M. J. (1982). The modification of childbehavior problems in the home. In A. S. Bellack, M. Hersen, &A. E. Kazdin (Eds.), International handbook of behavior modification(pp. 917958). New York: Plenum Press.

Taylor, S. (2004). Efficacy and outcome predictors for three PTSDtreatments: exposure therapy, EMDR, and relaxation training. InS. Taylor (Ed.),Advances in the treatment of posttraumatic stress disorder:Cognitive-behavioral perspectives (pp. 1337). New York: Springer.

Tryon, W. W. (1996). Observing contingencies: Taxonomy andmethods. Clinical Psychology Review, 16, 215230.

Virus-Ortega, J., & Haynes, S. N. (2005). Functional analysis inbehavior therapy: Behavioral foundations and clinical application.International Journal of Clinical and Health Psychology, 5, 567587.

Wolpe, J. (1989). The derailment of behavior therapy: A tale ofconceptual misdirection. Journal of Behavior Therapy and Experi-mental Psychiatry, 20, 315.

Yoman, J., & Edelstein, B. A. (1994). Functional assessment inpsychiatric disability. In J. R. Bedell (Ed.), Psychological assessmentand treatment of persons with severe mental disorders (pp. 3156).Washington, DC: Taylor and Francis.

The author would like to acknowledge Keith Massel for hiscomments on a previous draft of this article.

Address correspondence to Jerome Yoman, Life Skills Resource.20449 S.W. Tualatin Valley Highway, #328, Aloha, OR 97006;e-mail: [email protected].

Received: January 12, 2007Accepted: January 12, 2008Available online 21 July 2008

A Primer on Functional AnalysisFunctional Analysis in Operant ConditioningAdding Classical Conditioning to Functional AnalysisTypes of Functional AnalysisTarget Behavior Selection and the Context of Functional AnalysisFunctional DiagnosesFunctional Analysis Step by StepStep 1: Define the Valued Long-term Consequences (Ultimate Outcomes) of Behavior ChangeStep 2: Define the Behavior of InterestStep 3: Collect Data from Informants, and Through Direct Observation, on the Behavior of Inter.....Step 4: Formulate Provisional Functional Diagnoses (Hypotheses) Regarding the Behavior of Inte.....Step 5: Refine Provisional Functional Diagnoses Based on Further Data CollectionStep 6: Collect Data as Needed to Indicate a Specific InterventionStep 7: Implement the Indicated Intervention to Confirm or Disconfirm the Functional Diagnosis.....Step 8: Recycle Through Steps 4 to 7 as Needed Until Discovery of an Effective Intervention

The Continuing Role of Functional Analysis in Behavior TherapyReferences