Pre-Decisional Information Processes in Teacher Selection

Abstract

This study examined the behavior which takes place during pre-decisional information search and processing of written information inputs to the teacher selection process. The assumption underlying the research was that this behavior was not random, but rather was relatively specific and reflected a rationalized search process on the part of the decision maker. The experiment involved the .manipulation of two variables, (a) information cost. and (b) risk. in a controlled, simulated teacher selection situation in order to examine their effects on the following dependent variables: (a) information selectivity, (b) time, (c) specificity of search pattern, (d) proportion of available information utilized, and (e) certainty. The first four of these dependent variables were postulated to be major components of the information search strategy used by decision makers. The simulated situation used in this experiment consisted of: (a) the description of a hypothetical community school district which provided the setting for the experimental task; (b) the presentation of information on fictitious applicants; (c) a simplified, computer-based information storage and retrieval system with which subjects interacted to obtain information on the applicants; and d) a set of decisions to be made regarding the applicants. The design of the study was a completely randomized 3 x 3 fixed model treatment arrangement with measures on the five dependent variables. The subjects in the study were eighty-one elementary school principals from the three county Seattle Metropolitan Area. This population selected for sampling since it was seen as being the largest single group which is involved in the selection of teachers on a part-time basis. Analysis of variance was used to test the significance of all main and interaction effects and post analyses of significant differences were conducted using a Newman-Keuls procedure. The data analysis indicated the following results. Cost had a significant main effect on a) five of the six sub-measures of the information selectivity measure; b) the proportion of available information utilized; c) specificity of search pattern; and d) time. For the three levels of the cost treatment, the high cost treatment group consistently contributed the major portion of the variance, except for the specificity measure. On this a significant curvilinear relationship was evidenced with the medium cost treatment contributing most of the variability. The results of the study indicated that distinctive patterns of search behavior existed for each of the three levels of the cost treatment. The major conclusions of the study were: 1) Under these treatment conditions, subjects did evidence distinctive patterns of behavior on the four measures postulated to be key elements of information search strategies. 2) In terms of the single criterion of decision quality used in this study (certainty) decisions in teacher selection can apparently be made with substantially fewer written information inputs than are typically used without loss of decision quality. It appears than an optimum information level exists which is capable of description and measurement for a given situation. 3) Although prescriptive models of information acquisition generally reflect a direct relationship between amount of information and certainty, the results of this study sustantiate that individuals stop short of acquiring all the information available, even though this could have been done at no loss to them. Subjects who tended to exceed the optimum information level for this task (fifty-eight items) did so without significant improvement to the quality of decisions made. The primary implication of these results is that if decisions can be made using far less information than is normally utilized - without a reduction in decision quality - then substantial savings can be realized by reducing the volume of information presented to decision makers. That is, greater concern must be evidenced for the quality of inputs, rather than for quantity. The relationship between the amount of information considered and the probability of making a "correct" decision (as defined in this study) does not appear to be an increasing monotonic function as many models have indicated. Rather, the relationship appears to be curvilinear. Finally, if an optimum level of information exists, as this study suggests, then the range of this level should be precisely measured under varying conditions and this range should be a key determinant in the design of information systems used in making teacher selection decisions.