In particularly severe or perplexing cases, or when indirect and descriptive functional assessments produce inconclusive results, best practices often call for simpler functional assessments to be supplemented by experimental functional analyses (EFA ; Iwata, Dorsey, Slifer, Bauman, & Richman, 1982). An EFA is a procedure where antecedents and consequences for challenging behavior are intentionally manipulated to determine which antecedents reliably evoke the behavior and which consequences reliably reinforce the behavior. The classic procedure involves randomly alternating five analogue conditions: (1) attention, (2) escape, (3) tangible, (4) alone or no interaction, and (5) a control or play condition. Each of the first four experimental conditions test one putative function of challenging behavior by setting up antecedent conditions that are likely to evoke the behavior, if indeed it has that particular function, and consequences that are likely to reinforce the behavior, if indeed it has that particular function. The fifth condition serves as a control condition, wherein none of the antecedents are in place and none of the consequences are delivered. Table 2.2 depicts the conditions and the antecedents and consequences that are presented in each. Sessions of each condition are repeated in a random order until differentiation in the rate of challenging behavior between conditions is observed or until it becomes apparent that the analysis is not producing interpretable results.

Experimental functional analyses have several advantages. First, substantial research has shown that they produce interpretable results in a large percentage of cases. For example, a large-scale review of research on EFAs found that 95.9 % of EFAs produce differentiated results (Hanley et al., 2003). However, it should be noted that this was a review of EFAs published in research, not a review EFAs actually done in real-life settings, so it is possible that the actual real-life success rate of EFAs is lower. A significant disadvantage of EFAs is that they require specialized training to administer and very few clinicians are available who possess that training. Even among Board Certified Behavior Analysts, the population of clinicians who possess the greatest training and expertise in functional assessment, only a very small minority possess the skills to safely and validly conduct EFAs. The unfortunate result is that EFAs are very rarely done in real clinical practice, despite their being considered the “gold standard” for functional assessment in research.

As has been discussed throughout this chapter, clinicians use a wide variety of tools and procedures when assessing an individual with ASD. Some tools and procedures have come to be referred to as “gold standard” procedures. For example, the ADOS and ADI-R are often referred to as gold standard diagnostic procedures. Similarly, EFAs are often referred to as gold standard functional assessment procedures. However, in both diagnostic and functional assessments, it is worth noting that gold standard procedures tend to be more costly and labor intensive and require specialized training that a very small percentage of the population of clinicians possess. Even when a clinician does possess the resources and expertise required to implement gold standard procedures, it is critical to remember that no one modality or instrument is more valuable than clinical judgment. It is important to remember that results from any one modality (e.g., cognitive evaluation, diagnostic observation, functional assessment, etc.) comprise only a single component of the full evaluation process. For example, results of a single measure may indicate strengths and weaknesses in domains of intellectual functioning, but not account for possible delays in the realm of social development or compensatory adaptive skills. While findings may provide insight to a particular observed behavior, they may be based on a limited sample of time or a novel setting. Parental endorsements may suggest a high or low frequency of a behavior in one setting that is not observed as generalizing to other settings. Similarly, even though an EFA is likely to produce the most reliable and valid functional assessment results, it is, by definition, analogue and contrived and therefore may produce behavior that occurs in reaction to clinician-contrived circumstances, rather than behavior that is representative of the client’s real behavior in everyday life. Although little or no research has demonstrated it, it is hypothetically possible to “shape up” a new function for challenging behavior that was never before present, merely by systematically giving a particular consequence for a behavior during an EFA. Put differently, it is possible that a client may actually learn for the first time that a particular challenging behavior can earn him/her access to preferred items or activities.

Overall, no measure should be considered in isolation for the purpose of diagnosis or determining eligibility for services. In addition, measures should be evaluated and interpreted against one another in the evaluation process. A caregiver report should be evaluated against the clinician’s direct observation and subsequent findings. It is clinical judgment that incorporates the individual modalities of testing together and produces a cohesive evaluation. Clinical opinion is invaluable in the diagnosis of ASD and functional assessment of challenging behavior and cannot be substituted, only strengthened, with carefully considered and administered measures of development, cognition, language, and executive functioning; diagnostic observations and structured interviews; surveys, questionnaires, and inventories related to social skills, behavioral, emotional, and adaptive functioning; review of psychological and medical records; functional assessment tools; and detailed histories obtained by caregivers and teachers. Furthermore, a clinician has the ability to draw from the findings of one measure to inform his/her decision to administer additional measures as he/she attempts to answer the referral question. During the interpretation of data, an individual’s test performance in one domain of functioning can assist in the understanding of another domain. More specifically, a clinician can utilize an individual’s performance in the areas of cognition and language to support his/her interpretation of that individual’s functioning in the areas of social and communication abilities (Lord et al., 2012).