Six functional assessments were applied to 121 topographies shown by twenty people with severe learning disabilities. Assessment methods were drawn to include semi-structured clinical interview, two rating scales, staff-use of ABC charts, descriptive and experimental analysis.
Assessment methods were compared on rate of prediction, and the rate at which multi-function complexes were assigned. The informant-based and descriptive methods assigned function for approximately two-thirds of topographies, the ABC chart and experimental methods for approximately one-third. Interview
assigned multi-function complexes most frequently. Function was assigned for 98% of assessed topographies although no method exceeded 74%. Informant-based methods assigned function more often for outer-directed behaviour and the descriptive method for the remainder. The analogue method failed to assign
for people with greater verbal ability. Context impinged on all methods, but in different ways. Staffs' level of training appeared to influence the selection and observation of topographies; sampling was difficult in the natural setting;
there were problems interpreting both descriptive and experimental data. Differential rates of prediction adversely affected the overall rate of convergence between methods. Assessment methods averaged agreement on approximately two topographies in every three assessed, mostly on primary predictions. When non-predictions were included the overall rate of convergence averaged just one-in-four, and agreement on the functions assigned to each topography was equivalent to one-in-three.
Assessment methods varied across topographies and individuals, and three showed a proclivity for ascribing particular functions. The nature of the setting affected all assessment methods, and each presented methodological problems from topography identification to data collection and analysis.
The search for a universal assessment method appears incomplete, and may even be inappropriate. Results support the routine application of complementary assessment combinations suited to prevailing contextual variables . A pre-assessment of person-related variables, environmental context, topography and
hypothesised function, may aid the selection of optimal combinations. More work is required to develop non-empirical assessment methods and extend the present taxonomy of function categories. Staff training and a clear clinical definition of challenging behaviour might improve the clinical utility of
informant-based and indirect descriptive assessment methods.