Keywords

Functional behavior assessment (FBA) involves gathering information about the context(s) during which an individual engages in a particular behavior (Cooper, Heron, & Heward, 2020; Hagopian, Dozier, Rooker, & Jones, 2013). During the process, behavior analysts examine how the environment and behavior interact to determine what environmental events are likely to set the occasion, or evoke the behavior, and what environmental events are likely to follow, or reinforce the behavior, lending to its continuation. The immediate goal of FBA is to identify, or at minimum, hypothesize the function of the behavior. In other words, FBA is used to answer the “why” question, apropos to this chapter, “Why does Jonny hit his sister, Sally”?

Answering the immediate “why” question leads to the terminal goal, the design of an effective and efficient intervention. Research has repeatedly supported the effectiveness and efficiency of function-based intervention when compared to interventions designed without consideration of the function of behavior (e.g., Iwata, Pace, Cowdery, & Miltenberger, 1994; Payne, Scott, & Conroy, 2007; Walker, Chung, & Bonnet, 2018). The identification of the function is essential not only to intervention components targeting decreases in challenging behavior but also functional replacement behaviors. For example, if through the FBA process it is identified that Jonny hits his sister Sally to gain access to Sally’s toys, teaching Jonny to ask for toys, to wait for toys, or to find alternative toys is necessary to promote maintenance of treatment effects over time and across settings. Thus, FBA is a critical step in the treatment of severe challenging behavior such as aggression and violence.

The process of FBA to inform intervention is considered best practice within the field of behavior analysis (e.g., Belva, Hattier, & Matson, 2013; Gresham, Watson, & Skinner, 2001; Hagopian et al., 2013). Empirical support of this process led to its incorporation within Individuals with Disabilities Education Act (IDEA) of 1997 and the Individuals with Disabilities Education Improvement Act (IDEIA) of 2004 (Individuals With Disabilities Education Act, 20 U.S.C. § 1400, 2004). In essence, components of IDEA or IDEIA require FBAs be conducted and inform intervention within the educational system. Although the general process of FBA is considered best practice and mandated by law, the specific procedures conducted are left to the judgement of those responsible for completing the FBA.

This chapter provides an outline of the steps involved in conducting FBA, interpreting FBA outcomes, and integrating outcomes of multiple FBA tools to inform the design of function-based interventions for multiple topographies of challenging behavior including aggression and violence. Relevant research literature is reviewed, clinical recommendations offered, and future research directions proposed.

FBA as Best Practice

Gable, Quinn, Rutherford Jr, Howell, and Hoffman (1999) published a guide for FBA following the initial release of the IDEA (1997). The authors outlined a ten-step process of assessment and treatment of challenging behavior. The outline below provides a summary of the first seven steps as they relate to FBA and the use of FBA results to inform the design of a function-based intervention.

  1. 1.

    Describe and verify the problem. Compare the dimensions of behavior such as topography, frequency, and context to the behavior of peers and consider the influence of cultural variables. Reserve FBA for more severe challenging behavior (e.g., aggression and violence, property destruction, self-injurious behavior) that is not likely to be addressed by general educational or clinical practices (e.g., positive behavior supports).

  2. 2.

    Define the target(s): Consider environmental variables such as time of day, location, and context during which the behavior is and is not likely to occur. Operational definitions should include delineation of the behavior as well as examples and non-examples. For instance, instead of targeting “aggression or violence,” target “hitting, kicking, or verbal threats.”

  3. 3.

    Select FBA methodology and collect data. Selection of methodology should be specific to the individual and setting(s). Multiple methods of FBA will lead to more accurate results.

  4. 4.

    Analyze information gathered. Summarize information to easily identify patterns across environmental and behavior variables.

  5. 5.

    Generate hypothesis statement. Present a concise statement reflecting the observed interaction between behavior and environmental variables. For example, “When Sally has the iPad and is playing within arms-reach of Jonny, Jonny hits Sally in the arm, Sally puts the iPad down and Jonny takes it.”

  6. 6.

    Test hypothesis. The hypothesized function should be tested through the systematic manipulation of relevant environmental variables. This test may be designed as a treatment analysis to decrease the likelihood of the behavior or as a functional analysis (FA) to evoke and reinforce behavior in order to verify the hypothesized function.

  7. 7.

    Develop and implement behavior intervention plan. The results of the FBA process (steps 1–6) inform the intervention plan. Both behaviors for deceleration (e.g., aggression or violence) as well as appropriate replacement behaviors (e.g., communication) are targeted.

Categories of FBA

There are many assessment procedures captured under the umbrella of FBA (Anderson, Rodriguez, & Campbell, 2015; Fryling & Baires, 2016). Different assessment procedures can be classified into three main categories of FBA which include (1) experimental (functional) analysis, (2) indirect assessment, and (3) descriptive assessment. While the goals remain the same, the methods, benefits, and limitations vary. This chapter will only briefly describe the experimental (functional) analysis method of FBA and instead concentrate on indirect and descriptive assessment.

Experimental (Functional) Analysis

One category of FBA is experimental or functional analysis (FA). The term analysis rather than assessment is used to denote the direct, systematic manipulation of environmental events rather than the discussion (indirect assessment) or observation (descriptive assessment) of environmental events (Hanley, 2012; Mayer, Sulzer-Azaroff, & Wallace, 2014). Experimental analysis is a methodology to identify the functional, or cause–effect, relations between behavior and environment through systematic manipulation of environmental events and observation of changes in behavior (Iwata et al. 1982/Iwata, Dorsey, Slifer, Bauman, & Richman, 1994). For example, in the test condition for social-positive reinforcement (attention), the practitioner would withhold attention for a period of time to increase the value of attention as a consequence. Contingent upon the instance of a target response such as aggression, the practitioner would deliver attention for a brief period before again removing attention (test condition). Responding during this condition would be compared to a condition during which attention is delivered independent of the target response (control condition). Differential levels of behavior observed across the test condition (high levels of behavior) and control condition (low levels of behavior) would indicate that the behavior is, at least in part, maintained by social-positive reinforcement in the form of access to attention.

FA is the only FBA method that directly manipulates environmental events, and therefore, the only method that can identity, rather than hypothesize the function of behavior. As such, experimental or functional analysis has long been considered the “gold standard” of FBA methodology within the field of behavior analysis. There has been over 30 years of research conducted and hundreds of publications around FA methodology which has demonstrated its utility and flexibility (Beavers, Iwata, & Lerman, 2013; Hanley, Iwata, & McCord, 2003). Thus, the FA is often used as the “true” function when evaluating the validity of indirect or descriptive assessments.

However, FA has a distinct disadvantage when compared to indirect and descriptive assessment of aggression and violence. That is, the experimental manipulation of possible controlling variables is intended to evoke and reinforce aggressive behavior which can be dangerous and necessitate safety measures during FA sessions (Chok, Harper, Weiss, Bird, & Luiselli, 2020). FBA, on the other hand, relies on hypothesized correlations between behavior and environmental events derived from observations, interviews, and data analysis under natural conditions without changing situations that are associated with the absence and presence of aggression and violence.

Indirect Assessments

A second category of FBA is the indirect assessment, which involves the collection of information through interviews and questionnaires, most often conducted with persons who have directly observed the individual engage in the target behavior such as caregivers, teachers, or therapists, but also may be conducted with the individual themselves (Dufrene, Kazmerski, & Labrot, 2017; Floyd, Phaneuf, & Wilczynski, 2005). Interviews consist of collecting data through verbal exchanges while questionnaires and rating scales may involve collecting information through verbal or written communication.

A number of indirect assessments have been developed over time (Belva et al., 2013; Kelley, LaRue, Roane, & Gadaire, 2011). The most salient component of indirect assessments surrounds the type of questions posed. Open-ended questions, closed-ended questions, or a combination may be asked across indirect assessment procedures. Closed-ended questions are scored dichotomously (yes/no) or on some numerical rating scale (Likert scale). A few common examples of closed-ended indirect assessments are the Motivation Assessment Scale (MAS; Durand & Crimmins, 1988), Questions About Behavior Function (QABF ; Matson & Vollmer, 1995), and the Functional Analysis Screening Tool (FAST ; Iwata, DeLeon, & Roscoe, 2013). Open-ended questions allow for more detailed responses and follow up questions for clarification or additional information (Fryling & Baires, 2016). Common open-ended assessments are the Functional Assessment Interview (FAI ; O’Neil et al., 1997) and the Open-Ended Functional Assessment Interview (Hanley, 2012).

Strengths and Limitations.

Indirect assessments are considered the least-intrusive type of FBA and are often recommended as the first step in the FBA process. Major benefits of indirect assessments include ease of administration and minimization of assessment time. Because indirect assessments do not require direct observation of behavior, they are often quick to complete and can be conducted in-person, on the phone, and via teleconference. Indirect assessments can also assist practitioners in gathering initial information about the individual (e.g., diagnosis) and the frequency, duration, intensity, and topography of the target behavior.

Notwithstanding benefits, practitioners should be cautious when selecting indirect assessments and avoid using these assessments as the sole measure within an FBA (e.g., Dufrene, et al., 2017; Floyd et al., 2005). Whether through interviews or questionnaires, such assessments depend upon the recall of past events which may be inconsistent or unreliable (Belva et al., 2013; Rooker, DeLeon, Borrero, Frank-Crawford, & Roscoe, 2015). The respondent is asked not only to recall information about the behavior itself, but also recall environmental events surrounding the behavior and how these variables interact across time. This reliance on verbal reports has brought into question the reliability (agreement) and validity (accuracy) of indirect assessments (Barton-Arwood, Wehby, Gunter, & Lane, 2003; Dufrene et al., 2017; Kelley et al., 2011).

Reliability and Validity of Closed-Ended Assessments.

Measures of reliability may include comparison of outcomes across time, respondents, or assessments, as well as within assessments to examine consistency of outcomes. The majority of research with respect to indirect assessment has focused on such measures of reliability and reported mixed results (Rooker et al., 2015). For example, initial reliability between respondents for the MAS was reported at moderate to high levels (Durand & Crimmins, 1988) while later studies found reliability across respondents to be in the low range (30–45%) for the MAS (e.g., Zarcone, Rodgers, Iwata, Rourke, & Dorsey, 1991).

Zaja, Moore, Van Ingen, and Rojahn (2011) compared the psychometric properties of three commonly used indirect assessments, the QABF (Matson & Vollmer, 1995), the FAST (Iwata et al., 2013), and the Functional Assessment for Multiple Casualty (FACT ; Matson et al. 2003). The authors reported that across all measures of validity and reliability, the QABF and the FACT were found to have strong psychometric properties as compared to the FAST.

Iwata et al. (2013) reported on both the reliability and validity of the FAST. The FAST was administered for 196 behaviors across dyads consisting of teachers, parents, and direct-care staff. Inter-rater reliability scores ranged from moderate to high for individual items within the FAST, while moderate reliability scores were reported for overall FAST outcomes (64.8%). Overall reliability scores reported by Iwata et al. (2013) are in contrast to those previously reported by Zaja et al. (2011). In addition to reliability, Iwata et al. (2013) compared the outcomes of the FAST to FA outcomes and found moderate levels of agreement.

Within the current literature, reported reliability measures of close-ended indirect assessments are somewhat mixed (e.g., Alter, Conroy, Mancil, & Haydon, 2008; Fee, Schieber, Noble, & Valdovinos, 2016). The complexity of comparison across studies is a direct result of differences in reported measures and the inclusion and exclusion of different assessments during comparisons. However, when closely examined, the most promising close-ended indirect assessments are identified as the QABF and the FAST (e.g., Dracobly, Dozier, Briggs & Juanico 2018; Iwata et al., 2013; Paclawskyj, Matson, Rush, Smalls, & Vollmer, 2000).

Similar to the FAST, the QABF has been found to be a reliable valid measure when compared to FA outcomes. In a comparison of indirect, descriptive, and FA methods, Hall (2005) found the QABF resulted in the same identified function as the FA in three of four cases (75%). In another comparison study, Tarbox et al. (2009) reported that outcomes of the QABF exactly matched the outcomes of the FA in four out of seven cases (57.1%) and partially matched (one of multiple identified functions) in six out of seven cases (85.7%). These two studies highlight the limited validity of even the most reliable closed-ended assessments.

Despite concerns with and inconsistencies in reports of reliability and validity of indirect assessments, they continue as one of the more common forms of FBA implemented in practice (e.g., Oliver, Pratt, & Normand, 2015; Roscoe, Phillips, Kelly, Farber, & Dube, 2015). As such, both researchers and practitioners continue to examine variables that may increase their reliability and validity (Neidert, Rooker, Bayles, & Miller, 2013; Rooker et al., 2015). One factor that may contribute to the variability in reliability and validity is the number and skill set of the respondents being assessed. Emerging research may provide some hopeful direction for both researchers and practitioners alike.

For example, Smith, Smith, Dracobly, and Pace (2012) compared the agreement scores when the MAS and QABF were conducted with five respondents instead of the typical two respondents. Agreement of the function in four out of five respondents was about 50% of cases with the MAS and in almost 60% of cases with the QABF. Similar results were reported when the result of the indirect assessments were compared to FA results with a higher agreement score between the QABF and the FA (almost 85%) as compared to the agreement between the MAS and FA (about 60%). It is important to note that not only did increasing the number of respondents across indirect assessments increase reliability, but that when reliability was at acceptable levels (80%), the validity of the assessment also increased as demonstrated by agreement with FA results (Dracobly et al., 2018).

Dracobly et al. (2018) compared the results of the FAST when either caregivers or experts (board certified behavior analysts) served as respondents. Two caregivers and two experts completed the FAST followed by the completion of a standard FA (Iwata et al. 1982/Iwata, Dorsey, et al., 1994). Experts were not actively involved in the clinical oversight and therefore conducted a short observation (1 h or less) of the individual prior to completing the indirect assessment. Even with limited observation, the values of reliability between expert respondents were higher than that of the caregiver respondents. When compared to the results of the FA, experts correctly agreed on the function of behavior for 80% of cases while caregivers correctly agreed on the function for only 20% of the cases.

Open-Ended Assessments.

While a number of studies have reported on reliability and validity measures across closed-ended assessments, research on open-ended assessments is limited (Fryling & Baires, 2016). In one study, Alter et al. (2008) included one open-ended assessment, the FAI (O’Neil et al., 1997), in a comparison of the reliability and validity across multiple FBA methods including the FAI, MAS, descriptive methods, and FA. Within this study both reliability and validity of the FAI was reported at low levels. The reliability between the FAI and MAS was low (25%) as the two indirect assessments did not report the same outcome. Similarly, the outcomes of the FAI did not match the outcomes of the FAs (25%).

Saini, Ubdegrove, Biran, and Duncan (2019) examined inter-rater reliability and the validity of the Open-Ended Functional Assessment Interview (Hanley, 2012). Two raters conducted the interview with one caregiver across four clinical cases. The two raters agreed on the function of behavior in three out of four cases (75%). The hypotheses were then compared to FA outcomes and verified in two of the four cases (50%). The authors concluded that the interview demonstrated high inter-rater reliability and moderate validity. The high level of inter-rater reliability was not surprising given the recent results of the Dracobly et al. (2018) study and that raters in the current study were credentialed in the field of behavior analysis. That is, one would expect higher levels of reliability across well-trained raters.

Recent advances in FBA has initiated a comprehensive examination of the validity of closed- versus open-ended assessments to inform FA design through the accurate identification of relevant antecedents and consequences (Hanley, 2012). In light of limited comparative research on this topic, Fryling and Baires (2016) discussed some inherent strengths and weakness across closed- and open-ended indirect assessments as they relate to practical application. Closed-ended assessments are more efficient relative to open-ended assessments, in part due to the focus on common environment-behavior contingencies (e.g., positive, negative, and automatic reinforcement) leaving little room to discuss alternative or idiosyncratic sources. However, such a focused assessment may produce either false negative results through missed variables or false positive results due to the forced-choice nature of the procedure. While open-ended assessments provide the opportunity to identify a wider range of environmental variables, the relevancy of such variables to the function of the target behavior must be carefully considered in order to avoid false positive results by including an irrelevant variable in the hypothesized function.

Descriptive Assessments

The final category of FBA is the descriptive assessment. Descriptive assessments involve the direct observation of behavior across routines within settings such as classrooms, homes, and community. During observations, data are collected on the ABCs, or antecedent events preceding the behavior, dimensions of the targeted behavior, and consequences following the behavior. Precise definitions of both the target response and environmental events are necessary to ensure consistent recording across observations. In addition to the ABCs, data may be collected on the time of day, setting, and other people within the environment. Descriptive data are commonly collected by individuals within the daily setting including teachers, paraprofessionals, clinicians, or parents.

Strengths and Limitations.

A general strength across descriptive assessment methods is direct observation of the behavior. Descriptive assessment is improved by real-time data collection around antecedent and consequences as opposed to reliance on verbal reports of past events during indirect assessments (Sloman, 2010). Although there are no standard criteria for termination of descriptive assessments, it is likely that assessment length will vary depending on the frequency of the targeted behavior. For example, assessments may be brief if aggressive behavior occurs several times per day, while extended assessment time may be required to collect sufficient data samples for low-rate aggressive behavior. There are several descriptive assessment procedures outlined in the literature (Belva et al., 2013; Hagopian et al., 2013; Mayer et al., 2014; Sloman, 2010) and described below.

Narrative Assessments.

Similar to open-ended indirect assessments, narrative descriptive assessments (Bijou, Peterson, & Ault, 1968) are conducted when an observer records the events that occur before and after the behavior including as many details as possible. Narrative assessments allow for the recording of idiosyncratic yet relevant variables contiguous to and following the target behavior.

Scatterplot.

A scatterplot (Touchette, MacDonald, & Langer, 1985) involves collecting data on the temporal distribution of the target response. For example, the school day may be split into 15-min intervals and the occurrence/nonoccurrence or frequency of aggression noted during each interval. This type of assessment allows for the detection of patterns of behavior as it relates to time of day, but does not provide information around the specific environmental events that may co-occur during these time periods unless noted.

Antecedent-Behavior-Consequence (ABC) Methods.

Another category of descriptive assessments involves observing and recording antecedent and consequence conditions that precede and follow the target behavior, respectively (Mayer et al., 2014). For example, anytime Jonny hits his sister Sally, the observer would record what happened immediately before and after the aggressive behavior. Data collection on antecedents and consequences may be recorded in a narrative format or may be more structured to include checklists.

Conditional and Background Probabilities.

As first described by Bijou et al. (1968), continuous or interval data collection is often used as a descriptive assessment method within research (Sloman, 2010). During this type of assessment, the observer records occurrence of the target behavior and environmental events separately. Following some predetermined observation period, environment-behavior relations are analyzed. For conditional probabilities, data analysis involves calculating the probability of the environmental events given the occurrence of the behavior (e.g., Lerman & Iwata, 1993). In addition, the conditional and background probability (e.g., Vollmer, Borrero, Wright, Camp, & Lalli, 2001) involves both calculations of the probability of environmental events given behavior and the probability of environmental events across time. The addition of the background, or unconditional probability allows for the comparison of the probability of the correlation between environmental events and behavior to the general occurrence of the environment. For example, to determine if access to the iPad follows aggression, the conditional probability would be derived by dividing the number of times the iPad followed aggression by the total number of times the behavior occurred and comparing this proportion to the number of times the iPad was presented.

Validity of Descriptive Assessments

While there remain some mixed results, the majority of previous research concluded poor validity between descriptive assessments when compared to FA results (Camp, Iwata, Hammond, & Bloom, 2009; Hall, 2005; Lerman & Iwata, 1993; Sloman, 2010; Tarbox et al., 2009). Further, while some studies reported observation of behavioral patterns with a scatterplot analysis (Maas, Didden, Bouts, Smits, & Curfs, 2009; Touchette et al., 1985) others reported no clear patterns across analyses (English & Anderson, 2004; Kahng et al., 1998). The reliability and validity of other common descriptive assessments such as the ABC method have produced similar results.

Pence, Roscoe, Bourret, and Ahearn (2009) compared the results of the ABC method, conditional probability method, and conditional- and background probability methods to each other and to FA outcomes. The three different descriptive methods resulted in similar or identical outcomes while the comparisons between the descriptive assessments and the FA only matched in one of six cases (16.7%). Although not the focus of the study, Pence et al. included an additional analysis during which only antecedents or consequences of the ABC data were compared to FA outcomes. For some cases, antecedent but not consequences aligned with the FA outcomes while in other cases, the consequences but not the antecedent aligned with FA outcomes.

Camp et al. (2009) examined the predictive nature of antecedents alone, consequences alone, and antecedent and consequences identified by descriptive observation and conditional and background probabilities. The results of the combined antecedent and consequence analysis replicated previous studies showing the limited agreement between descriptive assessment and FA. In this case, when both antecedents and consequences were included, outcomes of the descriptive assessment matched outcomes of the FA in four of seven cases (57.1%). In addition, the results of the antecedent only and consequence only descriptive assessments replicated the Pence et al. findings that examining antecedent or consequences alone did not consistently increase the predictive validity of the descriptive assessment.

Both Hall (2005) and Tarbox et al. (2009) included descriptive assessments within a three-way comparison of indirect, descriptive, and FA methods. The results of Hall (2005) replicated Lerman and Iwata (1993) in that conditional probabilities, as a descriptive assessment produced low levels of agreement with FA outcomes (25%). Tarbox et al. used ABC data within their comparison and found agreement with FA outcomes in only one of seven cases (14.3%).

In summary, these recent studies have continued to demonstrate poor validity across descriptive assessment methods even when results of descriptive assessments are compared using antecedents only, consequences only, or both antecedents and consequences. Although descriptive assessments may lack validity with reference to behavior-function hypothesis formulation, practitioners and researchers should not abandon them all together. Rather, a shift in focus around the utility of descriptive assessment within the FBA process seems more appropriate.

Use of FBA Methods by Practitioners

An important area of the FBA literature is descriptive research on the use of FBA methods within applied settings such as clinics, schools, homes, and the community. Notably, there is limited research on implementing FBA methodology within applied settings (Oliver et al., 2015; Roscoe et al., 2015).

Ellingson, Miltenberger, and Long (1999) reported the results of a survey on the use and perceptions of FBA methodology. Thirty-six respondents throughout the state of North Dakota reported that they most often used indirect assessments (interviews and FAST) followed by FA within the natural setting and descriptive assessment (ABC). Commonly reported perceptions of FBA were that closed-ended indirect assessments were easy to use, descriptive assessments provided information about function, while FA in the natural setting informed intervention. In their discussion, the authors qualified that the definition of “functional analysis within the natural setting” used in this study was broad and may have been interpreted as including treatment analysis.

In 2015, two separate surveys on the use of FBA methodology were published by Roscoe et al. and Oliver et al. Roscoe et al. limited their survey to practitioners within the state of Massachusetts (205 respondents) while Oliver et al. used an electronic survey without limits to demographic area (724 initial respondents). Across both studies, practitioners reported most commonly using descriptive assessment methodology (62%–Roscoe et al.; 75%–Oliver et al.) while only about one-third of respondents reported using FA methodology. The survey results were more discrepant with respect to indirect assessment. While indirect assessments were commonly used by just over 75% of respondents in the Oliver et al. study, only 2% of respondents in the Roscoe et al. reported using this assessment method.

Professions outside of behavior analysis are often responsible for conducting FBA, therefore it is important to extend the evaluation of FBA practices to those disciplines. In illustration, Johnson, Goldberg, Hinant, and Couch (2019) conducted a survey across 199 school psychologists and found that indirect assessments were used most often followed by descriptive assessments, while only 25% of respondents had utilized FA methods within the last school year.

From the current surveys, it is evident that indirect and descriptive assessments are used more commonly compared to FA within the FBA process. None of the aforementioned surveys resulted in the exact agreement of FBA implementation practices. Oliver et al. (2015) and Roscoe et al. (2015) both identified descriptive assessments as the most common but differed on the rankings between indirect assessments and FA. Similarly, Ellingson et al. (1999) and Johnson et al. (2019) both identified indirect assessments as the most common but differed on the rakings of descriptive and assessment and FA. Thus, although assessment procedures were found to be most commonly implemented across all surveys, more research is needed to tease out the specific variables that determine what type of FBA will be selected, effective, and efficient.

Practice Recommendations: Evidence Based Practice in FBA

Although the general process of FBA is well outlined (Cooper et al., 2020), practitioners must individualize the specific content of the process to best fit the clinical case at hand. That is, FBA is a multistep process involving many decision points along the way. Decisions based on the framework of evidence-based practice incorporate three variables (1) the best available literature that is most relevant to the particular case, (2) professional judgement, and (3) client/setting characteristics (Slocum et al., 2014; Spencer, Detrich, & Slocum, 2012).

The first seven steps of the FBA process as outlined by Gable et al. (1999) are again outlined below. However, steps three through seven have been modified to now include practice recommendations based on the “best available literature” as presented throughout the previous sections of this chapter. Although these recommendations should be considered during the FBA process, it is important to individualize each step using professional judgement and client/setting characteristics. In the example of aggression and violence, practitioners should follow the steps below.

  1. 1.

    Describe and verify the problem. Compare the dimensions of behavior such as topography, frequency, and context to the behavior of peers and consider the influence of cultural variables. Reserve FBA for more severe challenging behavior (e.g., aggression and violence, property destruction, self-injurious behavior) that is not likely to be addressed by general educational or clinical practices (e.g., positive behavior supports).

  2. 2.

    Define the target(s): Consider environmental variables such as time of day, location, and context during which the behavior is and is not likely to occur. Operational definitions should include delineation of the behavior as well as examples and non-examples. For example, instead of targeting “aggression or violence” target “hitting, kicking, or verbal threats.”

  3. 3.

    Select FBA methodology and collect data. Practitioners should be familiar with the strengths and limitations of different FBA methods. Indirect methods, as they are currently designed, have low reliability and validity measures (Hagopian et al., 2013; Mayer et al., 2014; Rooker et al., 2015). Similarly, hypotheses about the function of behavior derived by descriptive assessments do not often match outcomes of FA (Rooker et al., 2015; Sloman, 2010). Thus, it is recommended that practitioners utilize multiple FBA methods and simultaneously compare outcomes to help guide next steps (Gable et al., 1999). Further, if FBA methods must be restricted to indirect assessment, practitioners should consider conducting both closed-ended and open-ended formats to balance the strengths and limitations of these two sub-categories of indirect assessments (Fryling & Baires, 2016).

    There are several considerations with respect to respondent selection that may increase the reliability and validity of indirect FBA outcomes. First, consider a respondent’s or observer’s level of training and experience around behavior-environment interactions and their understanding of basic learning principles (Dracobly et al., 2018). Second, multiple respondents should be used whenever possible (Smith et al., 2012). Third, reliability of the assessment outcomes is best when respondents are from the same environment and have observed the behavior over time (Borgmeier, Horner, & Koegel, 2006).

  4. 4.

    Analyze information gathered. The analysis of information collected during the FBA process will, in part, be driven by the selected method. However, practitioners should consider several guidelines across methods. First, data analysis methods should be sensitive. Although behavior analysts often prefer visual analysis, additional measures should be considered to better capture the complex, and sometimes subtle interaction between multiple environmental variables. For example, during descriptive assessments, data should be collected in a way to allow computation of conditional and background probabilities (Rooker et al., 2015; Vollmer et al., 2001). Second, data analysis should be easy to interpret. When FBAs involve multiple assessment methods, practitioners should consider supplementing written summaries with data triangulation to highlight consistencies and discrepancies across sources of information (Gable et al., 1999).

  5. 5.

    Generate hypothesis statement. Present a concise statement reflecting the observed interaction between behavior and environmental variables. Include all relevant variables within the hypothesis. If necessary, practitioners should conduct additional FBA to either replicate outcomes from previous respondents, thus increasing the confidence in the hypothesis (Smith et al., 2012), or conduct additional methods of FBA to ensure all potentially relevant variables are examined before making the final hypothesis (Fryling & Baires, 2016; Gable et al., 1999). For example, a scatterplot might be recommended following an ABC descriptive assessment that identified denial of food as a common antecedent to aggression. The addition of the scatterplot might identify a correlation between time of day (e.g., right before meals) and problem behavior.

  6. 6.

    Test hypothesis. The hypothesized function should be tested through the systematic manipulation of variables that precede and/or follow the behavior. This test may be designed as a treatment analysis during which variables are manipulated to decrease the likelihood of the behavior or as an FA during which variables are manipulated to either increase (test conditions) or decrease (control conditions) behavior.

  7. 7.

    Develop and implement behavior intervention plan. The results of the FBA process (steps 1–6) inform the intervention plan. Both behaviors for deceleration (e.g., aggression or violence) as well as appropriate replacement behaviors (e.g., communication) are targeted.

Future Research Directions

Some common themes emerge from recent literature around FBA. Although best practice continues to highlight FA as the gold standard, current reviews of the literature on FBA practices across applied settings coupled with recent surveys of practitioners indicate greater use of indirect and descriptive assessment methods. Future research directions in the area of FBA must focus on ways in which the reliability and validity of indirect and descriptive assessments can be maximized when experimental methods are not preferred or not available (Neidert et al., 2013; Rooker et al., 2015). Some emerging research in this area was reviewed throughout this chapter and an extension of these topics is reviewed here.

The first proposed area of future research surrounds the behavior of those responsible for participating in the FBA process, whether it be behavior analysts, school psychologists, teachers, parents, or therapists. To maximize effectiveness and efficiency of FBA methods, we must first expand the current literature on practitioner behavior as it relates to the FBA process.

Recent surveys examined practitioner’s selection of FBA methods (Ellingson et al., 1999; Johnson et al., 2019; Oliver et al., 2015; Roscoe et al., 2015). When asked about the function of their behavior, respondents across both the Oliver et al. (2015) and Roscoe et al. (2015) studies reported common barriers to the implementation of FA as lack of time and resources. Such responses suggest that setting characteristics affect FBA selection. Future research should examine additional variables that may influence FBA selection such as client characteristics and topography, frequency, magnitude, and duration of the target response. For example, are practitioners more likely to select indirect assessments over descriptive assessments or FA when the target behavior is severe aggression or violence?

It will be important to examine current practices across all aspects of the FBA process including system of data collection (e.g., Lerman, Hovanetz, Strobel, & Tetreault, 2009) and analysis, generation of hypotheses, and integration of FBA information into function-based interventions (e.g., Blood & Neel, 2007; Scott et al., 2005; Sugai, Lewis-Palmer, & Hagan, 1998). Such assessments of practitioner behavior as it relates to FBA will serve as a baseline measure to which comparisons can be made.

For example, Lerman et al. (2009) examined the accuracy with which school teachers and paraprofessionals collected either narrative or structured descriptive data. Participants received a group training (lecture) and then were instructed to score video-taped scenarios using narrative and structured ABC data collection methods. Not only were participants more accurate with the structured format they also reported a preference for this format. It is important to note that Lerman et al. (2009) did not collect baseline data as the purpose of this study was to compare accuracy of data collection across assessment formats.

Other studies have examined the effects of training on the accuracy of data collection during the FBA process (Luna, Petri, Palmier, & Rapp, 2018; Mayer & DiGennaro Reed, 2013). Luna et al. (2018) examined the effects of training on accuracy of data collection across narrative and structured descriptive assessments. The authors reported improved accuracy of data collection on narrative and structured descriptive data following intervention which consisted of a verbal review and group feedback. Ten of 14 teachers/paraprofessionals increased accuracy to mastery levels across both recording formats following the training.

These studies highlight the need for additional research on the necessary and sufficient training practices across different components of the FBA process. Several authors have suggested training, experience, and knowledge of the respondents may lead to more reliable and valid assessment outcomes (Borgmeier et al., 2006; Dracobly et al., 2018). Simple trainings such as those described may reduce the need for more restrictive assessments, therefore reducing the potential risk for all involved, particularly if the target behavior is aggressive or violent in nature.

One other direction of future research should set out to systematically evaluate how aspects of the environment and behavior affect FBA outcomes within the natural setting. The results of several studies have implied that different aspects of the environment (e.g., Thompson & Iwata, 2007), behavior (e.g., Matson & Wilkins, 2008), or interaction between environment and behavior may influence outcomes of indirect and descriptive assessments. For example, what are the effects of rate or intensity of behavior on outcomes of indirect assessments? Does reliability of indirect assessments increase with more opportunities to observe behavior-environment interactions? Such a research agenda might include a combination of controlled, translational experiments followed by more applied, systematic replications. This will certainly be a formidable task that has taken somewhat of a back seat to the continued refinement of FA methodology (Beavers et al., 2013; Hanley et al., 2003).

Finally, acceptance of such research agendas to refine indirect and descriptive assessments does not negate the continued search for adoption of the gold standard, FA. However, data from recent surveys implies that there are contexts for which the FA is not preferred or not available. Research agendas should be driven by such data particularly with reference to persons who have neurodevelopmental disabilities and demonstrate aggressive and violent behavior.