Robert E. Jensen, [Observations on Jensen's Experimental Design for Study of Effects of Accounting Variations in Decision Making]: A Rejoinder, Journal of Accounting Research, Vol. 5, No. 2 (Autumn, 1967), pp. 230-251
The research reported herein was undertaken to investigate relationships between (a) security evaluation and portfolio selection and (b) alternative inventory valuation and depreciation methods in financial reporting. First, a computer simulation model of a manufacturing firm was developed. In this simulation phase, the effects of alternative methods of financial statements and related measures (earnings per share, working capital, earnings margin, current ratio, inventory turnover, and other financial ratios along with corresponding rates of change and moving averages) were investigated under a wide range of operating conditions. In the second phase, an attempt was made to measure the effects of these accounting variations on evaluations by professional security analysts. Complete prospectuses were developed for two hypothetical companies named ETX Electronic Industries, Inc. and Rayco Electronics Corporation. Financial data for the companies were generated by the computer model. Four sets of financial reports with their related measures were generated for each company for a ten-year span. Two inventory methods (lifo and fifo) and two depreciation methods (straight-line and accelerated) were used. The four different financial reports for each company resulted m sixteen combinations of financial reports for the two companies. Participants in the study received an information packet which included an introductory letter, a return questionnaire, and one prospectus for each of the companies to be evaluated. The only variation in the in-
From the standpoint of financial statement analysis and forecasting, most business firms are also "miserable experimental designs." Accounting researchers, nevertheless, often rely heavily upon the fallible statistical tools of such designs. A major danger is the mechanical use of statistical techniques in overly superficial settings which have little or no relevance to accounting issues. This paper is directed to this danger which seems to be increasingly common in the accounting literature. Particular attention is directed to autocorrelation in accounting data experimental design." Unfortunately, statistical tools in experimental design and analysis are not easily adapted to miserable situations rampant with non-stationarities, interactions, missing variables, measurement error, autocorrelation, multicollinearities, etc. Autocorrelation is also ignored in many behavioral accounting experiments, especially where subjects are asked to make sequential decisions. For example, little attention is devoted to such issues in lens model studies of decision making.
In this article, the author discusses about cost accounting by providing information about one of his friends Robin Stealall, a Congressperson in context of voting on pending Bill 382636, which concerned a ban on the use of plastic in spectacle lenses as part of increased efforts in petroleum resource conservation. Stealall wanted to make a conscientious and responsible choice in this matter. His own spectacles at that time had plastic lenses, so Stealall decided to collect some facts from accounting experts concerning those particular spectacle lenses. Stealall first flew to Houston for consultation with professor Robert Sterling, whose office door, Stealall discovered, bore the letters "TRUTH" in very huge capitals. Stealall curiously approached the bearded man behind the wire cage and asked for "TRUTH" concerning the plastic spectacle lenses. According to Sterling, the truth was that, the lenses have no value to anyone and their exit price was zero. In the same context, Sterling also explained Stealall about historical costs. Sterling also added that, the word "accounting" is only used in archeology now.
The article presents a cluster analysis study of financial performance of selected business firms. Methods of grouping or classifying similar entities based upon measured characteristics possessed by each entity appear in the literature under various synonyms like, cluster analysis, grouping methods, classification theory, numerical taxonomy and clump theory. In clustering research, the objective is to discover natural clusters of the items based upon some internal criterion. The internal criterion is based upon some measure of resemblance or a measure of proximity. Clustering problems cut across various disciplines and seem to be receiving greater attention. Entities to be grouped together may be traits, objects, persons, plants, groups, institutions, structures, fields, stars, cities, companies, variables, etc. Considerable attention is given to this matter in the behavioral and life sciences. A discussion of various approaches for cluster analysis is given in two separate working papers. The method which is used in the preliminary research outcomes has been reported in this paper.
In the article, the author presents an analysis of the paper by John P. Fertakis, which attempts to draw conclusions with respect to accounting on the basis of unrelated or only indirectly related studies in the behavioral sciences. According to the author, his paper merely contains several examples of what may become even more prevalent in accounting. Accounting research frequently has its tentacles extended into the research efforts of scientists in the fields of psychology, sociology, economics, statistics, operations research, management science, etc. In so doing, accountants are looking to others for answers in order to avoid the dirty and painstakingly slow, expensive, and methodical means by which empirical evidence can be conceived and nurtured in their own studies. But this outside empirical evidence often flounders like a fish out of water when plucked from the environment in which it was generated. In addition, accountants must heed the cautions of scientists who generated and/or analyzed the cited empirical evidence. In supplying more information in annual reports, accounts are thereby making it possible for users to make finer distinctions between more companies. Professor Fertakis should note that his suggested "seven items" for a standard report are not one-dimensional at all.
In the article "Sensitivity Analysis in Decision Making," Alfred Rappaport presented an interesting illustration of sensitivity analysis in management decision making. Aside from a general explanation of the concept of sensitivity analysis, his presentation was primarily devoted to a numerical illustration and suggested areas in which sensitivity information may be useful to management. One purpose of this paper is to describe sensitivity techniques using a dual linear programming model. Second purpose is to extend sensitivity analysis into several areas not mentioned by Rappaport. A third purpose of this paper is to discuss the problem of integer stipulations in linear programming. It will be shown that the integer solution obtained by Rappaport is not the optimal integer solution. A fourth purpose is to briefly discuss the formidable problem of sensitivity analysis in an integer model. Rappaport presented an integer solution, but his sensitivity analysis is actually based upon a continuous, noninteger solution.
This article focuses on multiple regression analysis to cost control of decentralized operations in the consumer finance industry. There are potential accounting applications of multiple regression analysis in control of decentralized operations. Moreover, multiple regression can be a useful empirical research tool in other areas of accounting and finance. It is essential, however, to know the hidden limitations and assumptions in the approach and to perform the necessary tests to see that these assumptions are met be- fore plunging head-first into a sea of regression formulae. In cost analysis, one feature of multiple regression is the ability to use dichotomous variables. The advantage herein arises when a given characteristic may or may not exist in decentralized units. Multiple regression may be applied without assuming the disturbance terms are normally distributed. Multiple regression may be used in testing structural relationships between operating costs and various factors which are thought to affect these costs. Analysis of variance procedures may be extended to statistical tests of single coefficients and to statistical tests of the contribution to explained variation of sub-groups of factors included in the model.
C. Edward Arrington, William Hillison, Robert E. Jensen, An Application of Analytical Hierarchy Process to Model Expert Judgments on Analytical Review Procedures, Journal of Accounting Research, Vol. 22, No. 1 (Spring, 1984), pp. 298-312