Functional analyses are the gold standard of confirming maintaining variables of problem behavior. Despite widespread support, many clinical settings instead use anecdotal assessments. These have been shown to have poor reliability when used by non-experts but can be useful for confirming maintaining variables of problem behavior when agreement has reached a certain level. We used behavior skills training to teach new staff member pairs behavior function to increase their reliability on these assessments. We found that although agreement increased slightly, this was not clinically significant. Out of the two pairs of participants one pair of participants was able to identify their client's maintaining variable of problem behavior. Future research should investigate the type of training used for non-experts for example training non-experts to state problem behavior in objective operational terms.

We administered the Motivation Assessment Scale (MAS) and the Questions About Behavioral Function (QABF) to five raters and compared the results with functional analysis outcomes for 12 cases in which agreement was obtained for at least four out of five raters on either anecdotal assessment. The scores for functional categories on the MAS and QABF were ranked by averaging the scores for the raters who agreed on the primary maintaining variable. Functional analysis results were ranked by examining average responding across all conditions. A comparison showed correspondence between the highest category and the highest functional analysis condition for 10 out of 11 cases for the MAS and all 10 cases for the QABF. Correspondence for the category and condition ranked second was found for 2 out of 11 cases for the MAS and 2 out of 10 cases for the QABF. The magnitude of difference between categories on the MAS/QABF did not appear to predict the amount of difference in responding in the corresponding functional analyses.