It has been suggested that gmphically displayed multivariate data help decision makers better understand information t h y are called on to analyze This study ... traditional tabular displays of financial figures Significant differcnces in task ...
PERFORMANCE DIFFERENCES IN THE USE OF GRAPHIC AND TABULAR DISPLAYS OF MULTIVARIATE DATA David B. MacKay Departments of Marketing and Geography, Indiana University, Bloomington, IN 47405
Angelina Villarreal Senior Market Research Dexelopment Analyst, Miller Bmving Company, Milwaukee, WI 53201
ABSTRACT It has been suggested that gmphically displayed multivariate data help decision makers better understand information t h y are called on to analyze This study compam judgments made from one recently suggested multivariate display technique with judgments made from traditional tabular displays of financial figures Significant differcnces in task performance are found to be related both to differences in the stimulus sets and to individual differences among the subjects Our results suggest that the relative contribution of graphic displays to decision making may vary considerably from situation to situation. Subject Amas: Decision Pmcesses and Management Information Systems.
BACKGROUND Decision making frequently involves the evaluation of complicated sets of multivariate data. Evaluation of tabularly displayed multivariate data can be tedious and time consuming. Furthermore, relationships among the variables may be hard to determine and a comprehensive picture of the data may be difficult to establish. Graphic displays frequently have been suggested as a way of overcoming traditional limitations of tabular displays. Most graphic methods available to managers deal with two- or three-dimensionaldata. Reviews of graphic methods, from a variety of perspectives, may be found in [ 6 ] ,[lo:, [13], [14], [19], [301, and [311. Recently, procedures have been developed for graphically displaying multivariate data (data characterized by five, ten, or even more dimensions). Unlike the long multidisciplinary history of inquiry into simpler graphic methods, multivariate gmphic displays still are in their infancy. Despite this, their potential has been realized by many, and descriptions of multivariate graphic displays now are beginning to be found in general-purpose texts on multivariate analysis [25]. In professional business journals, Huff, Mahajan, and Black 1171 in marketing and Moriarity [24] and Stock and Watson [29] in accounting have proposed that a multivariate graphic display method developed by Chernoff [8] (one that represents a multidimensional set of data as a cartoon drawing of a face) be used to portray data of decision-making interest. They describe and illustrate how such a face could be drawn with different facial features corresponding to different decision-making variables. Interest by the business community in this method has been expressed [16], and reports on the use of facial displays to portray multivariate information have appeared in the Wall Street Journal [32]and elsewhere In the Huff, Mahajan, and Black (171 application, a series of faces was used to display the values of eleven key financial variables for different firms over a period 535
19871
MacKay and Villarreal
545
the generality of facial displays, future studies should consider other stimuli. It also would be advantageous to use subjects with a wider range of business experience Future studies also need to look at other open issues that have been discussed [7] [8] [17] [27] such as alternative means of constructing faces and assigning variables to facial features. A cautionary note also must be raised about generalizing the findings of this study to other types of graphic displays, since the ability of Chernoff s displays to capture multivariate data holistically in a mnemonic way perhaps is unique among the different means of graphically representing multivariate data. [Received: July 22, 1985. Accepted: July 21, 1986.1
REFERENCES Anderson, E. A semigraphical method for the analysis of compla problems lPchnometrin 1960, 2 , 387-391.
Andnws, D. F. Plots of high-dimensional data. Biometrics, 1972, 28, 125-136. Anscombe, E J. Rejection of outliers. Rchnomerrim, 1960. 2, 123-147. Argyle, M., Salter, V., Nicholson. H.. Williams, M., & Burgess, P. Communication of inferior and superior attitudes by verbal and nonverbal symbols. British Journal of Social and Clinical PSyChOlO~, 1970, 99, 221-231. Barnett, V., & Lewis, T. Outliers in statistical dara. New York: Wiley, 1978. BenigerPJ. R., & Robyn, D. L. Quantitative graphics in statistics: A brief history. American Statistician, 1978, 32, 1-1 1. Bruckner, L. A. On Chernoff faces. In P. C. C. Wang (Ed.), Gmphical repmentarion of mulfiuariate data. New York: Academic Press, 1978. Chernoff. H. Using faces to represent points in k-dimensional space graphically. Journal of rhe American Starktical Association, 1973, 68, 361-368. Crouch. W. W., Brindle, B,& Frye, J. K. The Syracuse person perception test: A measure of responsiveness to facial and verbal cues. In P. C. C. Wang (Ed.), Gmphical rrpmentarion of niultivariate data. New York: Academic Press, 1978. DeSanais. G. Computer graphics as decision aids: Directions for research. Decision Scierices, 1984, 15, 463-487. Dwyer, F. M., Jr. Adapting visual illustrations for effective learning. HarvardEducufionalReview, 1967, 37(2), 250-263. Ekman, P., & Friesen, W. V. Unmasking fhe face. Englewood Cliffs, N J PnnticoHall, 1975. Ekman. P., Friesen, W. V., & Ellsworth, P. Emotion in the human face: Guidelinesfor mearch and an integmtion of findings. New York Cambridge University Press, 1982. Fienberg, S E. Graphical methods in statistics. American Statkticicm. 1979, 33, 165-178. Goldwyn, R. M.. Friedman, H. P., and Siegel, T.H. Iteration and interaction in computed data bank analysis: Case study in the psychological classification and assessment of the critically ill. Computers in Biomedical Reseatrh, 1971, 4, 607422. Huff, D. L. Personal communication. March 1986. Huff, D. L., Mahajan, V., & Black, W. C. Facial representation of multivariate data. Journal of Marketing, 1981. 45, 53-59. Kleiner, a,& Hartigan, J. A. Representing points in many dimensions by t r e e and castles. Jawnu1 of the American Statistical Association, 1981, 76, 260-275. Lucas, H. C. An experimental investigation of the use of computer-based graphics in decision making. Management Science, 1981, 27, 757-768. Lusk, E. J., & Kersnick. M. Effects of cognitive style and report format on task performance: The MIS design consequences. Management Science, 1979, 25, 787-798. Mant, M. H. Multiple-choice learning of linodrawn facial features: 11. Sac differences. Bulletin of the Psychonomic Society, 1979, 14. 439441.
5 46
Decision Sciences
[Vol. 18
[22] Mason, R. 1.. & Mitroff, 1. I. A program for m h on management information systems. Management Science. 1973, 19, 475-487. I231 Minton, H. L., & Schneider, F. W. D ~ f e r e n t i a l p ~ c h o l oMonterey, ~. C A BrookdColC 1980. (241 Moriarity, S. Communicating financial information through multidimensional graphics. Journal of Accounting Research, 1979, 17, 205-224. (251 O'Sullivan, M. Measuring the ability to recognize facial expressions of emotion. In P. Eckman (Ed.), Emotion in the human face Cambridge, England Cambridge University Press, 1982. I261 Seber, G. A. F. Multivariate observations. New York: Wiley, 1984. (271 Sharma, S.. & Mahajan. V. Early warning indicators of business failure. Journal of Murketing 1980, 44,80-89. (281 Sherman, J. Sex-related cognitive diffemnces. Springfield, IL: Thomas, 1978. (291 Stock, D., R: Watson, C. J. Human judgment accuracy, multidimensional graphics, and humans versus models. Journal of Accounting Research. 1984, 22, 192-206. [30] Wainer, H., & Thissen, D. Graphical data analysis Annual Review of Psychology. 1981,32, 191-241. [311 Wang, P. C. C. (Ed.). Gmphical representation of multivariate data. New York: Academic Press, 1978. I321 With computer help, marketing professor faces up to corporations' financial data. Wall Srreer .loirnia/, July 22, 1981. p. 31. I)a\ id hlscKay is Professor of Marketing and Geography at Indiana University. His papers have appeared in a variety of journals including Decision Scienms, Marketing Science, Psychometnka, Journal of Murkering Rwwrch, and Grogmphicul Analysis. Dr. MacKay's current research interests include m e a w e m e n ( , scaling. spatial behavior, and graphic displays Angelina Villarreal is Senior Market Research Development Analyst at the Miller Brewing Company, Milwaukee. Dr. Villarreal completed a Ph.D. at the University of Pittsburgh. Her current reseaich focuses on measuring [he effect of commercial and noncommercial communication on consumer behavior. Somc of her reiearch has appcared recently in the proceedings of the Asrwiation for Consumer Research and f h e American hlarketing Association.