Descriptive analysis involves the direct observation of target behavior in natural (or naturalistic) contexts to gather information on contiguous and potentially relevant environmental events, in the absence of experimental manipulation. That is, descriptive analyses identify events that are correlated with the occurrence of some target response. Descriptive analysis is commonly used as a part of a comprehensive functional assessment of problem behavior, prior to conducting an experimental functional analysis. Several advantages of descriptive analysis have been noted in the literature. First, because descriptive analyses involve direct observation of behavior in the natural environment, they provide a means to gather baseline rates of problem behavior. These rates may be useful to determine treatment efficacy once an intervention is developed and implemented. Second, the direct observation aspect of descriptive analyses may be useful in developing operational definitions of behavior, rather than relying on verbal report of caregivers. Third, descriptive analyses may be helpful in identifying particular antecedents or consequences to incorporate into functional analyses. For example, certain functional analysis conditions may be added or omitted based on the common antecedent and consequent conditions observed in the natural environment. Fourth, it has been argued that descriptive analysis may be used as an assessment procedure when functional analyses are not possible due to the nature of the target behavior (e.g., extremely low rate or dangerous problem behavior). However, the use of descriptive analysis as a sole means of ascribing function to behavior is not recommended. This is because the complexity of the natural environment makes it difficult to determine which events are functionally related to the target behavior. Other potential negative aspects of using descriptive analysis to determine the function of the behavior have also been noted including the amount of time and complex data analysis necessary for a complete assessment, the correlational nature of the data, and poor validity outcomes reported in the literature. The purpose of this paper is to provide a summary of a variety of research areas in descriptive analysis, including suggestions of when descriptive analyses can and should be used. In addition, possible future areas of research using descriptive analysis will be discussed. Descriptive Analysis Techniques Several variations of direct observation techniques have been presented in the literature. The particular technique used affects the data presentation as well as the information that can be gleaned from the data analysis. For this reason, a brief description of the most commonly used techniques will be provided. The first technique is the scatter plot analysis, first described by Touchette, MacDonald and Langer (1985). A scatter plot analysis involves a grid data sheet with time intervals listed vertically (e.g., 30-minute intervals) and days of the week listed horizontally. The time intervals can be individually determined based on the particular individual's schedule. Caregivers record the frequency of behavior throughout the day by appropriately marking the boxes of the grid. For example, the nonoccurrence of behavior may be represented by leaving the box blank, low-rates of behavior may be represented by filling in the box with hatch marks, and high-rates of behavior may be represented by filling in the box. Data from scatter plots are typically expressed in terms of percent of intervals with the target behavior. After several days or weeks of data collection, it may be possible to observe patterns of behavior which can be tied to specific activities or times of the day. The advantages of scatter plot analyses may include ease of data collection and estimates of the occurrence of behavior both before and after intervention. Yet, few research studies have used scatter plot analyses, and when they have been conducted, there have been mixed results. …
Read full abstract