An evaluation of functional ecological psychology

The present research was designed to investigate four issues which pertain to methods as well as results achieved employing a particular multivariate research design-Functional Ecological Psychology (FEP). Comparing ecological psychological methods with those advocated by FEP indicates that observations which are made on molar behavior should be geared to the type of questions investigators are interested in answering. While FEP provides a more detailed, quantifiable set of observations, it is limited in that it focuses upon behavior frequency. Inferences concerning importance, direction, affect, and other attributes of behavior dimensions must be deduced from data regarding frequency of specific behaviors. Ecological psychology attempts to compile records of behavior which include frequency, however, they also include goal direction, emotional tone,and environmental descriptions, all of which tend to promote richness of description of the subjects' actions. Incorporating these two methods into one may involve too great a strain on the capabilities of the observer. However, further research employing these methods might indicate that specimen records compiled according to the methods advocated by Barker could yield specific behavior frequency data such as those derived from FEP, if samples of situations and behaviors were observed rather than all behavior for an entire day. In conclusion, it appears that the FEP model extends the ecological methods developed by Barker for producing specimen records by providing a framework through which an investigator can specifically quantify molar observations of real-life behavior and examine these scores across situations and subjects. This capability increases the utility of ecological research methods by enabling the investigator to objectively identify dimensions of behavior rather than rely on subjective impressions or judgments by other investigators of such dimensions from sentence-type data (e.g., specimen records). Two groups of 12 subjects each were employed. One group of 12 subjects (SR) were spinal cord injury patients observed by a team of investigators who recorded what these subjects did, with whom, for how long, and in what location. These records were then compiled into specimen records. The second group of 12 subjects (DO) was also made up of spinal cord injury patients at the same hospital, participating in the same treatment program as the first group of subjects. Forty-five behaviors were then selected from the specimen records on the basis of molar, objectively identifiable occurrences. Each of the 12 specimen records was then used to establish frequencies of the 45 behaviors for the SR subjects. The other group of 12 subjects was observed directly on the 45 preselected behaviors. The observations for each group were then compiled into a 24 subject X 45 behavior X 6 situation matrix. The results indicated that the two methods differed greatly in terms of frequency of behavior. The DO subjects had frequencies which were four times as high as the SR subjects. Factor analyses of the two sets of 11 variables across situations for the two 12 subject samples, as well as the total 24 subject sample, indicated that the total sample factors were highly related to the DO subject factors and not related to the SR subject factors. Since DO subjects had much higher frequencies than SR subjects, it was hypothesized that small sample results may be comparable to results achieved from large samples of subjects if the frequency of behaviors for the small sample is high. Non-interpretable factor structures are likely to result from small samples of subjects and low frequencies of behavior. Further results were achieved by rescaling the 24 subjects' scores on the 11 most frequently occurring behaviors in order to eliminate discrepancies in frequency due to methods of observation. These results indicated that specific dimensions of behavior frequency could be identified across situations and that several of these dimensions were consistent within situations. Changes in these dimensions between situations were interpreted to reflect specific differences between the situations and their effects on behavior. Finally, it was concluded that these dimensions of behavior frequency bore little relationship to biographical variables such as age, sex, education, income, occupation, and severity of injury.