- EVEN RETAIL PRICING {m DUCT PERCEPTION VALUE DETERMINATION, PRO THE EFFECTS 013001) I ) m3 PROPENSITIES AND BUY! Thesis for the Degree of ,Ph. D. MICHIGAN STATE UNIVERSITY GEORGOFF 1970 DAVID M .... I , :5. _ . YHfing This is to certify that the thesis entitled The Effect of Odd and Even Retail Pricing on Value Determination, Product Perception and Purchase Propensiiies presented by David M. Georgoff has been accepted towards fulfillment of the requirements for Ph.D. degree in Marke‘i i ng _~L K - _ ./ 3/7 /¢7’[9 '// ’7 //‘:,p:V/A:‘.-d/ ./ I ’ I X ' I J“ ”(I’V ‘ ,9 by / /_ .. / "I / x; c- .rxv'v/v‘ - L"/,, C Major professor ' / ' i I ’M. ' I, /) fl) ’ ' /(/) Date //..I a x. 4 a _’_L_¢' 1' (I /’4 0-169 ABSTRACT THE EFFECTS OF ODD—EVEN RETAIL PRICING ON VALUE DETERMINATION, PRODUCT PERCEPTI N, AND BUYING PROPENSITIES By “avid M. Georgoff The study examines the effects of odd and even retail price endings on a heterogeneous group of eleven department store items and attempts to determine, for each item, the direction and intensity of any price illusion. In addition, the influence of price endings on subjects' indicated buying dispositions and product perceptions are analyzed, as are differences in illusion susceptibility for selected demo— graphic classifications; fcr subjects' indicated buyer roles; for past buying experience; and for future buying intentions. The study involves two essential stages. The first is a simulated inquiry in which differences in the mean responses of 2A3 subjects to selected questions about each of the items, which were randomly assigned odd or even endings, were used as a basis for establishing the effects of the alternative price endings. Using an experimental approach, the second stage involves a measure of sales of the eleven items for a David M. Georgoff four-week period within the six stores of a leading depart— ment store group. During this phase the items were also assigned odd or even retail endings for predetermined weekly periods. The total sales at each of the prices were then analyzed to determine the sales impact of the alternative endings, and to provide some measure of the strength of the illusions which were revealed in the previous stage. The results of the study indicate that in the aggre— gate (i.e., summing across all subjects) no evidence of price illusion appears to exist. Furthermore, the alterna- tive price endings do not appear to influence subjects' indicated purchase dispositions or perceptions of products as measured by ten factor analyzed, semantic differential scales. In addition, significant differences in illusion susceptibility do not appear to be associated with the subjects' indicated buyer roles, with past purchase experi- ences, or future buying intentions. Illusion susceptibility, however, did vary within selected demographic strata. Subjects who are more highly educated, whose household heads have white collar Jobs, who earn higher family incomes, or who have female house- hold heads employed full—time, are susceptible. In addi- tion, the direction of the illusion appears mixed; influenced by the type of product. In some cases odd prices enhance the value impressions of an item while in other cases they depress them--and vice versa. A comparison of the results of the two stages of the study shows the David M. Georgoff absence of significant variations in the second phase and a low level of relationship between the results of the two stages. This suggests that while price illusion appears to exist for certain products within selected demographic segments, any net effect on sales is relatively weak or is swamped by situational and intervening variables. THE EFFECTS OF ODD—EVEN RETAIL PRICING ON VALUE DETERMINATION, PRODUCT PERCEPTION, AND BUYING PROPENSITIES By David M. Georgoff A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY College of Business Department of Marketing and Transportation Administration '1970 ’; -'— 2;. 7 .2.” ~12 7» _/ ~ 70 ACKNOWLEDGMENTS A project of any magnitude is seldom the product of any single person. In looking back it becomes apparent that this study also incorporates the direct and indirect assistance of many individuals. Although mentioning each would be overly burdensome and impossible at this point in time, simple gratitude compels me to acknowledge the help of some major contributors. In this light I must first mention the assistance provided by the executives and associates of Burdine's whose selfless cooperation made it possible to conduct the second phase of the study in their stores. I also deeply appreciate the financial support made available by the Department of harketing and Transportation Admin- istration at Michigan State University, which allowed me to undertake a study of this scope, and the computer assistance furnished by the Computer Technology Depart- ment of Florida Atlantic University. In addition, Dr. Francis E. Brown of the Warton School of Finance and Commerce, University of Pennsylvania, provided valuable suggestions which were incorporated into the first phase of the study. ii My greatest debt is owed to the members of the dis- sertation committee, which included Drs. Paul E. Smith and Kenward L. Atkin, for their continued support, interest, and guidance in the development of this study. I would specifically like to acknowledge the generous contribu- tion of Dr. Stanley C. Hollander, who functioned both as my major professor and as chairman of the dissertation ccmmittee. His continued encouragement, unflagging interest, and profound observations heavily shaped the formulation, implementation, and interpretation of the project. By his example, Dr. Hollander has provided me with a positive model of scholarship which has made my association with him easily the most rewarding experi- ence of my academic career. Lastly, but most deeply, I would like to thank Pat and Chris for their support, sacrifice, and warm under- standing which sustained me during this period. iii TABLE OF CONTENTS Page ACKNOWLEDGMENTS . . . . . . . . . . . . 11 LIST OF TABLES. . . . . . . . . . . . . vii LIST OF FIGURES . . . . . . . . . . . . xi Chapter I. INTRODUCTION. . . . . . . . . . . l A Statement of the Problem . . . . . l The Issue. . . . 7 The Significance of the Issue to the Retailer . . . IA The General Approaches to the Problem . 17 The Distinction Between Effect and Illusion . . . . . . . 21 The Approach of the Study . . . . . 25 II. THE PRICE VARIABLEe-ITS RELATIONSHIP TO DEMAND. . . . . . . . . . . . . 28 Historical Antecedents of Odd and Even Prices. . . 28 The Significance and Function of Price . 36 Inverse Price—Volume Relationships . . AC The Classical Approach—-Its Limitations. 51 A Statement of Hypotheses to be Tested . 53 III. A MEASURE OF ILLUSION. . . . . . . . 57 The Structure of the Study . . . . . 57 The Approach of the Test. . . . . . 61 A Measure of Illusion. . . . 61 A Measure of Purchase Propensity . . 6A The Perception of the Product .- . . 66 An Indication of a Subject's Assumed Customer Role, Purchase Experi- ence, and Anticipated Needs. . . 66 Demographic Information . . . . . 67 iv Chapter Page The Selection of Items . . . 67 Establishing a Measure of Product Perception . . . . . 68 Alternative Approaches to the Problem . . . . 68 Use of the Semantic Differential Scale . . 71 The Approach Used in Scale Construction. 71 The Product Concepts . . . 71 The Selection of the Bipolar Terms . 7A Factor Analysis. . . . . 82 The Variable Reduction Process. . . 8A The Factor Analysis Results. . . . 88 Establishing Factor Dimension . . . 91 The Selection of the Final Variable . 9A The Questionnaire Development . . . . 97 Pretesting the Questionnaire . . . 101 The Administration of the Study . . . 102 Analysis of the Results . . . 10A The Results of the Aggregate Ques- tionnaire Responses at Odd and Even Prices . . . 105 The Results of the Analysis of. Subjects' Perceived Customer Roles and Price Illusion. . 106 An Analysis of the Effects of Buying Experience and Purchase Expecta- tions on Illusion Susceptability . 108 An Analysis of Education and Illu— sion Susceptibility . . . . 108 An Analysis of Illusion Suscepti- bility and Occupational Level of Household Head . . . . 110 An Analysis of Illusion Suscepti- bility and Working Female Head of Household. . . . . 111 An Analysis of Illusion Suscepti- bility and Marital Status . . . 113 An Analysis of Illusion Suscepti- bility and Age . . . . 11A An Analysis of Illusion Suscepti- bility and Income . . . . . . 115 IV. A TEST OF EFFECT . . . . . . . . . 116 The General Research Design. . . . . 116 Pre-Conditions to the Study. . . . . 122 The Items. . . . . . . . . . . 124 Chapter Page The Participating Retailer . . . . . 125 The Time of the Study. . . . . . 128 The Administration of the Study . . . 129 The Analysis. . . . . . . . . . 132 V. CONCLUSIONS AND RECOMMENDATIONS FOR FUTURE STUDY . . . . . . . . . . . . . 139 Introduction. . . . . 139 The Aggregate Test .of Illusion. . . . 139 The Test of Effect. . . . . . . . 140 Purchase Propensities. . . . . . 141 Product Perceptions . . . . 142 Analysis of Demographic Illusion . . . 142 Illusion by Demographic Groups. . . . 145 Subject's Perceived Customer Role. . 145 Past Buying Experience and Future Purchase Expectations. . . . . 146 Educational Levels. . . . 146 Occupational Classification of. . Household Head . . . . 148 Working Female Head of Household . . 149 Marital Status . . . . . . . . 151 Age. . . . . . . . . . . . 152 Income. . . . . 153 Illusion Between Price Lines . . 154 The Relative Impact of $. 98 and 98¢ Prices. . . 156 An Assessment of Illusion by Product. . 156 Conclusions . . . 166 Recommendations for Further Research. . 170 SELECTED BIBLIOGRAPHY . . . . . . . . . . 174 APPENDICES . . . . . . . . . . . . . . 184 vi Table 10. 11. l2. l3. l4. 15. LIST OF TABLES Independent variables reflecting no inter- relationships where r > i .600 Symbolic representation of experimental design Program factor values . . . . . . Matrix of variables with r values greater than 1.600. . . . . . . . . . . . Factor-variable-clusters Rotated factor matrix List and alternative price endings for study items. . . . . . . . Net sales and merchandise item exposures (MIE) during the study period. . Item product-moment correlation coeffi- cients . . . . . . . . . . . Item F ratios Number, intensity, and direction of signifi- cant deviations by product. Item mean quality and value estimates Initial list of 42 semantic differential scales . . . . . . . . . . . . Value estimates, buying propensity indica— tions, and product perceptions of carryall bag priced at $2.98 and $3.00. . . . . Value estimates, buying propensity indica- tions, and product perceptions of casual dress priced at $17.98/$18.00. . vii Page 86 87 89 90 92 118 126 133 136 138 157 160 186 206 207 Table 16. 17. 18. 20. 21. 22. 23. 24. 25. 26. 27. Page Value estimates, buying propensity indica- tions, and product perceptions of women's shorts priced at $9.98 and $10.00 . . . . 208 Value estimates, buying propensity indica- tions, and product perceptions of laundry cart priced at $5.98 and $6.00 . . . . . 209 Value estimates, buying propensity indica- tions, and product perceptions of reversible broiler priced at $19.98 and $20.00. . . . 210 Value estimates, buying propensity indica- tions, and product perceptions of cotton blend duster priced at $11.98 and $12.00 . . 211 Value estimates, buying prOpensity indica- tions, and product perceptions of cotton knit tOp prices at $3.98 and $4.00 . . . . 212 Value estimates, buying propensity indica- tions, and product perceptions of cleaning fluid priced at $.98 and $1.00 . . . . . 213 Value estimates, buying propensity indica— tions, and product perceptions of cleaning fluid priced at $.98 and $1.00 . . . . . 214 Value estimates, buying propensity indica— tions, and product perceptions of men's knit shirt priced at $7.98 and $8.00 . . . 215 Value estimates, buying prOpensity indica- tions, and product perceptions of "No Turn" broiler priced at $24.98 and $25.00. . . . 216 Value estimates, buying propensity indica- tions, and product perceptions of chip and dip set priced at $6.98 and $7.00 . . . . 217 Value estimates, buying propensity indica- tions, and product perceptions of chip and dip set priced at $7.98 and $8.00 . . . . 218 Value estimates, buying propensity indica- tions, and product perceptions of chip and dip set priced at $8.98 and $9.00 . . . . 219 viii Table Page 28. Value estimates, buying prOpensity indica— tions, and product perceptions of chip and dip set priced at $7.00 and $9.00 . . . . 220 29. Value estimates, buying propensity indica- tions, and product perceptions of chip and dip set priced at $6.98 and $8.98 . . . . 221 30. Value estimates, buying propensity indica- tions, and product perceptions of chip and dip set priced at $6.98 and $9.00 . . . . 222 31. Analysis of subjects' perceived customer role and price illusion. . . . . . . . 224 32. Analysis of subjects' illusion suspectibil- ity and past buying experience and future purchase expectations . . . . . . . . 226 33. Illusion susceptibility and educational levels . . . . . . . . . . . . . 228 34. Illusion susceptibility and occupational classification of household head. . . . . 230 35. Illusion susceptibility and working female head of household. . . . . . . . . . 232 36. Illusion susceptibility and marital status . 234 37. Illusion susceptibility and age . . . . . 236 38. Illusion susceptibility and income . . . . 238 39. Women's Cotton Knit TOp: Analysis of co- variance of sales (y), adjusting for varia- tions in merchandise available for sale (x) . 258 40. Casual Dress: Analysis of variance of sales. . . . . . . . . . . . . . 259 41. Women's Shorts: Analysis of variance of sales. . . . . . . . . . . . . . 260 42. Cleaning Fluid: Analysis of variance of sales.‘ . . . . . . . . . . . . . 261 ix Table Page 43. Men's Knit Shirt: Analysis of covariance of sales (Y), adjusting for variations in merchandise available for sale (x) . . . . 262 44. Laundry Cart: Analysis of variance of sales. 263 45. "No Turn" Broiler: Analysis of covariance of sales (y), adjusting for variations in merchandise available for sale (x) . . . . 264 46. Reversible Broiler: Analysis of variance of sales. . . . . . . . . . . . . . 265 47. Cotton Blend Duster: Analysis of covariance of sales (y), adjusting for variations in merchandise available for sale (x) . . . . 266 48. Carryall Bag: Analysis of covariance of sales (y), adjusting for variations in merchandise available for sale (x) . . . . 267 Figure [0 U) LIST OF FIGURES Demand curve when odd-even retail pricing is effective . . . . . . . . Demand curve illustrating Veblen effect Demand curve illustrating strong price— quality relationship. . . . . . . Figures and sketches used in Stage I study xi Page 13 44 44 196 CHAPTER I INTRODUCTION A Statement of the Problem Marketing practitioners and theorists have for many years been intrigued by abnormal variations in "cus- tomary" or expected patterns of demand. Conventional wisdom as well as economic theory assume that customers purchase more of a product the lower it is priced. It is reasonable to expect that this assumption holds true across an infinite range of price points as long as the ceteris paribus condition prevails. We therefore expect that the demand curve for any product is a smooth nega— tively sloped function. But empirically we often find significant non-random variations in the normal patterns of demand which violate such assumptions about buyer price behavior. .While economists generally View such variations as atypical, insignificant, and outside their legitimate realm of concern, marketing practitioners and students of consumer behavior consider such phenomena significant for two basic reasons. First, they have pragmatic value because the results suggest that certain price points have either enhancing or deterring effects on demand. Knowledge of the existence, direction, and 1 ... I. .~_"4 _ ., v .ai“ ... ,,,.~ .'I ‘ . Q‘- ..;. . .n ., - .., ... w.-. . '5..- .‘V‘ u 3. t__-. ' A ,- l . *. . " . ~. 7 s . 4 ‘8 -. ‘L . K ‘s ‘- ‘1 o ,‘. impact of such phenomona offers an astute pricemaker a measure of competitive advantage. Such knowledge sug- gests price points that increase the appeal of his offer- ing, or cautions him against the use of other endings that may leave his offering relatively disadvantaged. The second reason is less direct, but in the long run may be more important. Selected price points may be relatively insignificant and justify no more than the passing concern of the businessman establishing price endings, but an understanding of the cause of such de- viations provides insights into the customer decision processes. The point was recognized by the late Gary Steiner who stated that, "No difference is too small to he of scientific value, although it may be too small to be of practical value."1 Although variations are discounted as irrational by the traditional economist, they are a product of the customer's inner rationality2 and a prime concern of the student of consumer behavior. Because the actions of customers can be assumed to involve higher order mental processes, the student of consumer behavior must take 1Gary A. Steiner, "Consumer Behavior: Where Do We Stand?" in On Knowing the Consumer, ed. by Joseph W. Newman (New York: John Wiley & Sons, Inc., 1966), p. 206. 2Robert W. Pratt, Jr., "Consumer Behavior: Some Psychological Aspects," in Science in Marketing, ed. by George Schwartz (New York: John Wiley & Sons, Inc., 1965). p. 110. « - 7-... O .,,. _ _. . _~ . '9‘ . _‘ u ‘l h..- "iA , 'h 1'” . . ‘~_ - -. ". note of external manifestations of behavior in order to construct plausible and predictable explanations con- sistent with the actions of consumers in varying situa- tions.3 Any abnormal increases and decreases in sales which may occur at certain price points are profitable topics of inquiry for these students. These variations are classified under the heading of psychological pricing, which in the broadest sense includes such phenomena as: fictitious pricing, bait pricing, prestige pricing, and odd pricing.” The term psychological pricing has various interpretations. These range from specific rules stating, "Whenever possible price your product so that the customer can pay for it with one coin or bill, and get something back in the way 3James F. Engel, David T. Kollat, and Roger D. Blackwell, Consumer Behavior (New York: Holt, Rinehart and Winston, Inc., 19687, pp. 20-23. “Perry Bliss, "Price Determination at the Department—Store Level," Journal of Marketing, XVII (July, 1952), 44; Edward R. Hawkins, "Price Policies and Theory," Journal of Marketing, XVII (January, 1954); Clare W. Barker and Ira Anderson, Principles of Retail— lgg (New York: McGraw-Hill Book Company, Inc., 1935), p. 166; Andre Gabor and C. W. J. Granger, "Price Sensi- tivity of the Consumer," Journal of Advertising Research, IV (December, 1964); Donald V. Harper, Price Policy and Procedure (New York: Harcourt, Brace & World, Inc., 19667, p. 39; John Cameron Aspley and John Cousty Hark- ness, The Sales Manager's Handbook (9th ed.; Chicago: The Dartnell Corporation, 1962), p. 242; Jules Backman, "Pricing," in Science in Marketing, ed. by George Schwagtz (New York: John Wiley & Sons, Inc., 1965), p. 27 . 5 of change," to more general definitions which include all prices that have great consumer appeal,6 or that maximize profits through margin-volume concerns.7 Although these phenomena have underlying similari- ties which justify broadly grouping them under the psy- chological pricing term, they also represent practices and effects which differ in many essential respects. In addition, because many of the phenomena overlap they do not represent mutually exclusive classifications. For this reason it is necessary to refine the sc0pe of this inquiry still further. Specifically, the thrust of this effort will be directed primarily at a consideration of the effect on sales which arises from the use of odd and even retail price endings. But there is no general agreement as to the exact meaning of odd and even price endings. Some authors have interpreted even prices to include quotations with end- ings which are divisible by five,8 whereas others have 5Louis T. Montant, Jr., "How to Determine Prices," Printers' Ink (September 10, 1948), 30. 6 Harper, Price Policy and Procedure, p. 281. 7Q. Forrest Walker, "Some Principles of Department Store Pricing," Journal of Marketing, XIV (January, 1950), 533. 8Harold J. Rudolph, "Pricing for Today's Market," Printers' Ink, CCXLVII (May 28, 1954), 22. considered even prices to be those divisible by two.9 Generally odd prices are said to include quotations in the immediate price ranges under an even dollar (whether ending in 88, 95, 97, 98 or 99 cents); while even prices are those ending in round dollar amounts.10 Price illusion, or the tendency of a prospective customer to perceive a price variation which cannot be justified by an absolute difference in a price quotation, is assumed to be created by pricing merchandise just under an even dollar. But any attempt to measure this illusion must recognize a complication which arises as a result of customary retailing practices. Because of historical associations and retailing conventions, cer- tain price endings are given different interpretations by prospective buyers.11 Some price endings (notably 9Milton H. Spencer and Louis Siegelman, Managerial Economics (2d ed.; Homewood, Illinois: Richard D. Irwin, Inc., 1964), p. 356. 10Eugene Jerome McCarthy, Basic Marketing: A Mana— gerial Appraoch (3d ed.; Homewood, Illinois: Richard D. Irwin, Inc., 1968), p. 551; Harper, Price Policy and Pro— cedure, p. 282; Hawkins, "Price Policies and Theory," p. 234. llAlfred Oxenfeldt, Executive Action in Marketing (Belmont, California: Wadsworth Publishing Co., Inc., 1966), pp. 54-55; Oswald Knauth, "Considerations in the Setting of Retail Prices," Journal of Marketing, XIV (July, 1949), 8; Edgar A. Pessemier, "A New Way to De- termine Buying Decisions," Journal of Marketing, XXIV (October, 1959), 46; Bliss, "Price Determination at the Department-Store Level,’ p. 44. 88, 97, and 99)12 are generally used with sale, promo- tional, or clearance merchandise. Therefore, an item may be purchased at these endings for two reasons: the sales appeal suggested by the ending, and the illusion created by the use of the odd price. In order to isolate such influences it is necessary to further restrict the range of this inquiry. Only prices ending in 98 cents and even dollars will be considered. This, presumably, would limit the effects of the study to the odd-even price illusion issue. The fact that odd prices are extensively used is well established. This popularity is underscored by surveys made of the incidence of various price endings.l3 These surveys also reveal the widespread association of the figure nine with promotional merchandise. As a result, it is hypothesized that because of this nine- promotional association the figure sould be avoided by quality stores.lu Conversely it is suggested that the 12As an extreme example refer to: "88-Cent Store Does Well in Secondary Location," Chain Store Age, XXXV (December, 1959), E22f-E22g. 13"The Right Way to Price," Super Market Merchandis— igg, XXIII (May, 1958), 163-166; "Set Prices Psychologi— cally," Progressive Grocer, XLI (February, 1962), 178; "Psychology Makes Cents in Pricing," Printers' Ink, CCXLIV (October 29, 1954), 45; Dik Warren Twedt, "Does the '9 Fixation' in Retail Pricing Really Promote Sales?" Journal of Marketing, XXIX (October, 1965), 54-55; Rudolph, “Pricing for Today's Market," pp. 22-24. 1“Fred M. Jones, Principles of Retailing (New York: Pitman Publishing Corporation, 1949), p. 287. a... .n. a ,. -.. u. . - Q. a (.44- . 3.... -.. q " \ ‘ ~ ..~ ,,._ .- --... "A._. ..,v u. - V.‘. i ‘- . o ‘u- .‘.' ., p c ., «A 'r -~ .g- 'A ‘5 . nine figure has an inherent appeal to price-conscious 15 customers. To the extent that the nine-promotional association does exist it would increase the impact of odd price endings, particularly if the price also in- cludes a dollar break (e.g., $9.98 as opposed to $10.00) which results in a so-called "double-nine" effect. This nine-promotional effect is not the same as the sale endings discussed earlier. The essential difference is that even prices presumably create a prestige effect which deters price-conscious buyers while appealing to quality conscious customers; although in both cases the items are assumed to be regular mer- chandise sold at regular prices. The Issue The issue of odd and even price endings has for many years been a concern of retail price setters and manufacturers-—particularly those manufacturers who pre- mark merchandise or in other ways use suggested retail 16 prices. Yet as extensive as interest in the issue has been, surprisingly little has been published in the way 15Delbert J. Duncan and Charles F. Phillips, Eg— tailing: Principles and Methods (5th ed.; Homewood, Illinois: Richard D. Irwin, Inc., 1959), p. 454. 16As an example of this concern refer: Barker and Anderson, Principles of Retailing, p. 166. of rigorous research which would offer price makers reliable guides for resolving their dilemma.l7 The classic study of the effect of odd and even price endings in an actual sales situation was reported in 1936.18 Although few details were made available, a mixed effect was found. Without specifics one cannot establish the range of variation in sales, or determine the extent to which the conditions of the study are applicable to the present, but the very fact that sub— stantial fluctuations in response to alternative price endings did occur is in itself significant. It strongly suggests that the effects of various price endings can be substantial. Within the past few years experimental studies of subject responses in simulated purchase situations have been conducted. One of the studies attempted to measure the probable reaction of subjects to a range of market prices. Because the study was conducted in England (which introduces differing value orientations, cultural influences, economic orientations, and market processes) the findings can only be cautiously 17For examples of some of the earlier studies refer to: Montant, "How to Determine Prices," p. 30; Knauth, "Considerations in the Setting of Retail Prices," p. 10; Aspley and Harkness, The Sales Manager's Handbook, p. 244; Barker and Anderson, Principles of Retailing, p. 167. 18Eli Ginzberg, "Customary Prices," American Economic Review, XXVI (June, 1936), 296. .. .- ..- 5 . .p. , ,,.v . ' s -. ‘ ..,‘ s. 'r. 1.. .. .r . .-~ '7 I u . . . generalized. The results indicated that the impact of odd prices varied among products. In one case odd prices immediately below even price points were found to be significantly more effective as an inducement to purchase than either even prices or prices which fell in the range below the odd price points. In another case no effect attributable to the odd price was found. The conclusion of the researchers was that odd pricing has no apparent effect. The impact in the case of the first product was assumed to be the result of a history of industry-wide odd price promotions and the conditioning effect it had on consumers. Because of this it was felt that prices other than those at the tradi— tional price points were viewed as unreal by the sub- jects.19 An experimental study conducted in the United States by Holloway also involved a simulated purchase situation, and produced results supporting the existence of effects attributable to odd prices. After testing the subjects' "bargain" and "quality" perceptions of twenty-four pro— ducts (ranging in price from one to two hundred dollars) Holloway concluded that, ". . . even prices were per- ceived as being of higher quality than the same product carrying an 'odd' price." The results also indicated 19Gabor and Granger, "Price Sensitivity of the Consumer," pp. 266-268. 10 that, ". . . 'odd' priced items were seen as being greater bargains than the same item carrying an 'even' price."20 The findings of the Holloway study support the existence of some influence arising from price illusion, but it is not possible to determine if the results arose from the illusion created by the odd price, from the nine-promotional association, or from some combined effect. The study does suggest that odd or even price endings affect a consumer's product perception in dis- cernably different ways. Two opposing views exist with regard to the issue. One is that no price illusion is created--the implicit assumption being that the customer is an astute (rational) buyer and is not deceived by the use of odd prices. The opposing view is that the customer perceives a $3.98 price as being significantly lower than the two-cent differential would justify. It is assumed that he trans— lates the price into the deliberative processes as three dollars and something, rather than rounding it to the four dollar figure. Then, in the traditional view, the customer weighs the perceived price against a product's promised stream of satisfactions (which are assumed 20Robert J. Holloway, "Experimental Work in Market— ing: Current Research and New Developments," in Applica- tion of the Sciences in Marketing Management, ed. by Frank M. Bass, Charles W. King, and Edgar A. Pessemier (New York: John Wiley & Sons, Inc., 1968), pp. 393-394. - I . ..r . ..- . ..“ H‘- . .‘r . \‘ .1 ‘ . 11 constant) and the value is enhanced because of the re- sulting higher quality—price ratio.21 In the end, marginal customers are induced to purchase the product-- resulting in greater sales at odd prices.22 In addition to these opposite views some middle positions also exist. Some feel that the issue has not been adequately investigated and that further study is needed,23 or that the effect is probably mixed and is not determinable.”4 Others hold that the effect of odd prices is not a product of price illusion, as tradition— ally viewed, but is the result of the retailer pricing 25 his product as low as possible. It is also suggested that such prices became established, and are subsequently effective simply as a matter of custom.26 Whatever the cause, those that hold odd pricing is effective view the 21Where: Quality/Price = Value. 22As an example of some positions supporting this View refer to: John W. Wingate and Elmer O. Schaller, Techniques of Retail Merchandising (New York: Prentice- Hall, Inc., 1950), p. 162; "Psychology Makes Cents in Pricing," p. 45; Aspely and Harkness, The Sales Manager's Handbook, p. 244; Backman, "Pricing," p. 276. 23Harper, Price Policy and Procedure, p. 283. 2“Wingate and Schaller, Techniques of Retail Merchandising, pp. 162-163. 25Harper, Price Policy and Procedure, p. 283; Barker and Anderson, Principles of Retailing, p. 167. 26Stanley C. Hollander, "Customary Prices," Business Topics, XIV (Summer, 1966), 47; Barker and Anderson, Principles of Retailing, p. 167; Wingate and Schaller, Techniques of Retail Merchandising, p. 162; Gabor and ..--~r- x ,I..‘ ’x' -.u.v-I J . “r rV' - ,.. . .., ‘ .- -O-a-..J _ r 'V‘ . -. -n . - "pal ‘ n 3p.._, .-' no ‘ 4 .. ‘. ‘~. I -'l". -,‘_" .“ h .k‘ - 'ns - a u. h. -,. v}, y -.‘ s . I u... \ "s n. .‘ Q ‘l ‘n .g . " ~. I ‘4- lv l s '. h . .. \"F K... p ‘.-"- (,‘15. -u .‘y. ‘ ~?_ \ . -. r,‘_| '. x n \‘ ., ‘a a . ‘ A .‘ 12 demand curve for a product not as a smooth, but as a kinked curve that is relatively more elastic at the odd- even price points (see Figure l). The original reasons offered for such deviations differed widely, and in present terms appear naive,27 but in time they were formulated into explicit guides to pricing behavior.28 Because marketing has been concerned with interdisciplinary consumer behavior studies for only a relatively short time, one can appreciate why the explanations lacked sophistication. Also explaining the lack of any deeper inquiries into the issue was the essential pragmatic orientation of the businessmen, whose concern was limited to whether the phenomenon in fact existed, and if so, in what way and to what extent it would affect the sale of various products. They felt that little of value was to be found in a more exhaustive study of the issue, and as a result, were satisfied with the somewhat superficial inquiries which were under- taken. Granger, "Price Sensitivity of the Consumer," p. 40; Harper, Price Policy and Procedure, p. 283; Knauth, "Considerations in the Setting of Retail Prices," p. 10; Backman, "Pricing," p. 276. 27As examples refer to: Harper, Price Policy and Procedure, pp. 282-283; Barker and Anderson, Principles of Retailing, p. 167. 28"Psychology Makes Cents in Pricing," p. 45; Aspley and Harkness, The Sales Manager's Handbook, p. 243. 13 Figure I DEMAND CURVE WHEN ODD-EVEN RETAIL PRICING IS EFFECTIVE Price $ 4.75 4.50 4.25 4.00 3.98 5.75 5.50 Quantity v .r' ,..v .A .— .v' v w» .— ..4. ‘r‘ .- -Hu 'vn-§ _. tl’ , 4" \1 14 Even today the issue is far from being resolved. On the one hand widespread adoption and presistent use of odd prices suggests that price illusion is, or was, an effective buying influence. Alternatively, growing adop- tion of even prices by department and specialty stores suggests that the impact of illusion may be so weak as to allow some retailers to forego the advantages of any price illusion with no substantial effect on either volume or customer reaction. Because of the significance of the issue (the fact that it touches directly or in— directly all retailers and manufacturers), the potential impact as suggested by the Ginzberg and Holloway studies, the reasons which prompted the adoption of odd prices in the past, and because of the possible insight further study may provide about the consumer decision processes, it would appear that a further inquiry is justified. The Significance of the Issue to the Retailer Empirically a retailer or manufacturer might con- sider hedging the issue by charging the odd price, the rationale being that if the illusion arising from odd prices is strong its use would allow the retailer to take advantage of the situation. The cost of selecting this alternative, in the extreme case where an impact does not exist, is the foregone revenue between the odd and the even prices. The cost of the alternative in the A ,..;..-- - - . «~.A4 . '-~ g l 4 _ ., ,.~J, ‘- ... I ' 7» u. . .'- 7., ~- - "~ I ‘ n . u- . ‘ - - Vt ' .. ‘, " -5 ‘. . ‘ ‘ -. -‘ Y5. .\ . 'v ‘~ . _ y. “-L ._‘. . a. _ o ‘q -. . . v I. a . ‘s p ‘. .‘. l .f a“ ".A . ~‘ ‘§ ‘»._ ‘h I . x '- . _ . ~ . ‘s _. .‘I s t. . _r .‘ 15 situation of a 98 cent ending is about 2% (more specifi— cally 2.04%) for a dollar item, then it quickly becomes infinitesimal, being about .5% for a four dollar item and about .2% for a ten dollar item. As a result one might argue that the issue is trivial and the lost revenue is well worth the hedge in an area of such uncertainty. If the foregone revenue does not appear impressive in relative terms, the situation changes dramatically when viewed in absolute dollars. In the simplified case of a department store doing $30,000,000 volume per year, 29 and an average gross sale of $6.00, the number of items sold is 5,000,000. Assuming the store uses 98 cent end- ings exclusively, the store has foregone $.02 per item.30 Assuming one item per transaction, the total foregone dollar amount is $100,000. Moreover, because no 29An approximation derived from Average Gross Sales for Year, for main store departments of Department Stores with Annual Sales of $20—$50 million as offered in: National Retail Merchants Association, Controllers Con— gress, Depgrtmental Merchandising and Operating Results of 1964 (New York: National Retail Merchants Associa- tion, 1965). 30The simplified assumptions being used may strike the critical reader as extremely naive in that they assume all sales are made at regular price or that the store em- ploys 98 cent endings for sale merchandise, that small transactions such as greeting cards and candy also use 98 cent endings, the impact of multiple pricing does not exist, and that 98 cent endings are also used on major items. On the other hand, this example is intended for illustrative purposes, and the incorporation of extensive, although realistic, qualifications would not signifi- cantly alter the essential point being raised. .r"hfl;r6 ,— .,.:o ,....-d .*"v '- .. u-t : v {a r-« p r" .— ... 4‘4“ n ' C u ¢.-. ‘ ’“R‘ -.- L. -.A. ' in “"io~_ -.~...._- ‘E‘ Vt' "' ~.A . ""‘a .. M. T‘°‘~U I! . 783‘ F.“ “I." J; I “v r-.. ~. . “n ‘vav ‘ _'r 3“ . , - "“cs-E. ‘ . o "_P‘~-. "‘Vcl i. . ‘ra ' ‘. ~ I- ,- .,_ ‘ c. g. M .\ ‘F'a ~. \. ‘-u .4 . “vs . .‘ r:- “v .4 "“I V - D...‘ ~.' . A ..v ~ § \ 7“ - ‘t. . \ . ’I .§ ._§\ ‘._ \ v V \v . i ‘0 . ~~ ‘ . U .. 1!, ~ ._-',‘u ‘. N ‘< t ‘x. ... .} ‘ _ o 4‘... 16 incremental costs are associated with the incremental revenue, in the short run the impact on the firm's earn- ings before taxes would be direct. Alternatively, if a retailer uses even prices for all products when the illusion created by the use of odd prices is strong, his prices are perceived to be significantly higher and marginal sales are lost in the short run. If his competitors employ odd prices, their prices would appear to be substantially lower, causing a shift of sales in the competitors' direction. In the long run (because these price impressions are probably gen- eralized)--a store's even price levels seem to be more expensive, resulting in a weakening of customer loyalty to the store. Even after making allowances for the simple assump- tions used in the above examples and recognizing that what has been offered is a limiting case, it appears that the matter is an important concern for retailers. The issue is simpler for the manufacturer. Assuming that retail markup percentages are not sensitive enough to reflect the differences in retail price endings (which seems probable for items retailing beyong two dollars per item) the astute manufacturer should pre-mark his mer- chandise at the odd price. Because the retailer's markup calculations (by assumption) are insensitive to the differential, the manufacturer need not reflect the l7 differential in the invoice price. If no odd price illu- sion exists he is unaffected and the retailer absorbs the lost revenue. To the extent that odd price illusion is real, his products are afforded an advantage over competi- tive brands priced in even dollars. Against this advan— tage he has only to weigh the cost of the possible antagonism of retailers who prefer even dollar endings. Although the pull strategies used by many manufacturers have the effect of creating varying degrees of brand preference for their products, and weakening the influence 31 and leverage of the smaller retailers, the astute manu- facturer cannot completely discount any such reaction. The General Approaches to the Problem For most retailers the approach to the problem has often been gross. As will be discussed, the assumption is that odd—even influences operate either strongly or have no effect. But when the range, direction, intensity, and variety of influences affecting purchase decisions within product categories are considered; when the scope of influences affecting the perception of price and price illusion is recognized; and when these price condition- ing effects are aligned by product (resulting in some which are conflicting to produce a weak price effect and 31Michael Halbert, ed., The Meaning and Sources of Marketing Theory (New York: McGraw-Hill Book Company, 1965). p. 163. 18 others which are reinforcing to produce a strong effect), the possibility arises that price impact may vary sub- stantially between products. Furthermore, because there are perceptions, needs, and purchase dispositions common to specific groups, the impact of price illusion also varies between customer types. Moreover, when one considers that the customer mix of each department and specialty store is unique as a result of the location, promotional policies, service mix, and merchandise offerings of the retailer, it is also possible that profit maximizing odd-even price policies will vary by store. Finally, because customer types often purchase different items in a department or specialty store, and because these customer types possess different degrees of illusion susceptibility and are disposed toward products with varying illusion effects, the ultimate blends which occur at the point of purchase produce product-customer combinations which are either mutually reinforcing or conflicting. Because the combinations will vary from synergistic to antagonistic, it is possible that the in- tensity of buying effects would vary by product, customer, store, and cyclical period. In some cases the effects may be very strong and in others they would be weak, but without additional information it is not possible to determine the appropriate circumstances, direction, or l9 intensity. The extent to which these effects exist suggests that a general rule applied to the price end— ings of a firm may prove costly. If such a rule is used the firm may not maximize all incremental revenues from the sale of certain items, while at the same time it might fail to realize all of its potential volume from other product categories. To the typical retailer price endings are a third- order concern. More attention is usually devoted to adequate representation in important merchandise types, to balancing assortments, and to adding items to increase stock interest. Short-run profits are subordinated to reinforcing the proposed store image and to maximizing 32 the long—run attraction of a department. Another con— cern is the establishment of appropriate price zones for merchandise, although pricing is predetermined and rou- tinized because of practices established in the trade, agreements with manufacturers, pre-marking, competitors' 32On the other hand there exists some evidence to suggest that the approach of many merchants is hardly a sophisticated one. Simplified assumptions and objec— tives appear to be the rule with short-run sales maximi- zation their overriding concern. As an example markon in percentage terms is often used as a simple proxy for assessing alternative item profitability [Douglas J. Dalrymple, "Quantitative Methods of Measuring Merchandis- ing Performance in Selected Department Stores," in fig- flections in Progress in Marketipg, ed. by L. George Smith (Chicago: American Marketing Association, 1964), pp. 119—131.] 20 actions and potential reactions, and the policies of the store. Even the alternative price endings available to a merchant are predetermined. It has been established that within a given cultural context certain price points have significant communication value.33 For this reason merchants assume that a consistent price ending policy provides: greater impact through the use of established promotional and sale endings, some measure of control over pricing abuses, and minimum organizational drift. For these reasons most large retailers have policies, expressed or implied, concerning the use of various price endings. However, these policies often originate at the general management level and are therefore made to apply to broad groups of products (with certain exceptions in highly competitive departments such as drugs, toys in season, and traffic appliances).3[4 Additionally, poli- cies are established at points far removed from the sell- ing floor and may tend to be insensitive to product demand variations. For these reasons it would seem that most price ending decisions are often the result of a few 33Pessemier, "A New Way to Determine Buying Deci- sions," p. 46; Bliss, "Price Determination at the Department-Store Level," p. 44; Knauth, "Considerations in the Setting of Retail Prices," p. 8; Oxenfeldt, Execu— tive Action in Marketing, p. 54. 3uKnauth, "Considerations in the Setting of Retail Prices," p. 2. .olr‘ ,-.... 0 Va: \ L»..- o "— .... w . _‘\ - ~' ..'- 21 generalized considerations and some uncritical assump- tions perpetuated from the past. Nevertheless they are applied to a range of products whose demand characteris— tics differ widely with respect to the effects of odd or even prices. If the impact is strong and varies between products, such a generalized approach will not maximize profits. It will prove costly from a sales or profit maximizing point of view regardless of the alternative selected; that is if the impact of odd prices is uni— versally strong while even price endings were used, and vice versa. The Distinction Between Effect and Illusion Because of the diverse influences affecting the purchase decision, customer responsiveness to the price variable differs widely among products. It is necessary to make a distinction between the strength of the illu- sion which arises from an odd price and the ultimate effect that will have on sales at retail. Such a dis— tinction is needed, because with the impacts created by the intervening variables, it does not always follow that a strong illusion will translate iteslf into a strong effect. As an example, in the case of an item with a highly inelastic demand the sales effect of both odd and even price endings will be the same no matter how strong the illusion. Because of this, a measure of retail sales 22 is insufficient because it does not provide a measure of any illusion which may have arisen. Conversely, a simple measure of illusion is insufficient for two reasons. First, because of the elasticities existing among pro— ducts, it is to be expected that product sales will have only an incidental relationship to the product's illusion intensity. Therefore, any measure of illusion cannot be used to accurately predict an item's relative sales effect. Second, as a result of the intervening variables, it can be expected that the sales effect relative to any illusion would be dampened for any product.35’36 But a test of effect coupled with a test of illusion would make it possible to determine roughly how much weaker the effect on sales may be. Moreover, without a measure of effect following one of illusion there is a tendency to overvalue the impact of the illusion. One of the leading factors dampening the effect of illusion arises from convenience and minimization of effort. Generally it is considered a burden for customers 35This would explain such inconsistent situations in which women's expressed preferences—-presumab1y valid-- are for G. E. refrigerators over Frigidaire, although the latter outsells the former. James H. Myers and William H. Reynolds, Consumer Behavior and Marketing Management (Boston: Houghton Mifflin Company, 1965), p. 56. 36As an example of some intervening variables which illusion must overcome are: product design, quality dis- tinctions, brand preferences, promotional effects, avail- ability of alternatives, past product experiences, and others which will be developed later in the paper. 23 to compare prices between all retailers because of the time and effort required. As a result, customers esti- mate the probable payoff from additional search against the secondary purchase costs of making the comparisons.37 Probable payoffs would include, in addition to the actual savings in product price, improved terms of sale, as well as the psychic value provided by the assurance of a price comparison.38 We can assume that shopping is a burdensome chore for many customers and competes with other demands on limited time.39 Consistent with this view, one marketing theorist comments: The initial state of the consumer is one in which she hopes to visit a single store, get everything she wants, and get it at a favorable price. Once she is in the store, pressure of 37Wesley C. Bender, "Consumer Purchase Costs--Do Retailers Recognize Them?" Journal of Retailing, XL (Spring, 1964), 1—8; Donald L. Shawver, The Development of Theories of Retail Price Determination in England and the United States, Illinois Studies in the Social Sciences, Vol. XXXIX (Urbana, Illinois: The University of Illinois Press, 1956), p. 79. 38Alfred Oxenfeldt, et al., Insights Into Pricing from Operations Research and Behavioral Science (Belmont, California: Wadsworth Publishing Company, 1961), pp. 98-101. 39For an elaboration of the sensitivity of the time issue refer to: Charles J. Cellazzo, Jr., Consumer Atti— tudes and Frustrations in Shopping (New York: National Retail Merchants Association, 1963), p. 116; William P. Dommermuth and Edward W. Cundiff, "Shopping Goods, Shop— ping Centers, and Selling Strategies," Journal of Market- ;pg, XXXI (October, 1967), 32—36; Bender, "Consumer Pur- chase Costs-—Do Retailers Recognize Them?" pp. 1-8. 2n time may cause her to buy what she wants even though the prices are higher that she expected, and accept substitutes for preferred items, or go without temporarily. O The issue is that the concept of customer conven- ience has many dimensions which are very real in determin- 41 ing purchases. These dimensions are recognized both by the astute retailer in formulating retailing mixes,142 and by customers in selecting patronage alternatives}43 While we expect the impact of this concept of convenience to vary among products, its net effect, especially in the short run, is to dampen significantly the intensity of any illusion. It is possible to speculate about the intensity of any given illusion. Using two goods as an example, one with elastic and one with inelastic demand, and assuming all other factors are equal, the following would likely uoWroe Alderson, Dynamic Marketing Behavior (Home- wood, Illinois: Richard D. Irwin, Inc., 1965), p. 228. ulEugene J. Kelley, "The Importance of Convenience in Consumer Purchasing," Journal of Marketing, XXIII (JUly’ 1958), 32-38- u2Seymour Baranoff, "Retailing as an Operating System," in Theory in Marketing, ed. by Reavis Cox, Wroe Alderson, and Stanley J. Shapio (Homewood, Illinois: Richard D. Irwin, Inc., 1964), pp. 160-161; Bender, Consumer Purchase Costs-—Do Retailers Recognize Them?" pp. 4-7. l'3Jon G. Udell, "Prepurchase Behavior of Buyers of Small Electric Appliances," Journal of Marketing, XXX (October, 1966), 52. .4.‘ .,‘0 Vi u‘.‘ , ‘A n..~ ' . _v\. '( u- an. n '- .f‘ ' v '. V. Q (7'4 r (U - r. o 25 occur when the illusion produced by each of the items is the same: 1. If the illusion is weak, the ultimate effect on sales will be weak. In this case the profit maximizing retailer would price his merchandise in even dollars. 2. If the illusion is strong, and assuming that no substantial dampening effect arises from the intervening variables, the sales effect will be strong for items with elastic demand. In this case the item should be given an odd price. On the other hand, if the demand is elastic, but the dampening effect is strong, even price endings can probably be used with no substantial deterrent effect on sales. 3. If the illusion is strong and the demand for the item is inelastic, at least in the short run, the use of even prices is justified. Additionally, the use of even price endings is appropriate regardless of whether the intervening variables are strong or weak. The Approach of the Study As a result of the considerations previously dis— cussed, a two-stage approach to the problem was under- taken. The first phase involved a specific test of illusion. Using a simulated situation an attempt was made to determine if the respondents' value estimates for a heterogeneous group of products (which were 26 randomly assigned odd or even prices) differed signifi- cantly. The study design also made it possible to determine: whether the subject's perception of the item was affected by the alternative endings; whether selected demographic characteristics were related to illusion susceptibility; whether illusion was affected by the subject's past or proposed purchases of the item or by his perceived buyer role; and whether the use of odd or even prices influenced the expressed purchase disposi— tion of the subject. The second phase attempted to determine the sales effect of odd or even retail price endings. This phase involved a measure of actual sales at retail for the same items used in the first stage. Using an experi- mental design which incorporated controlled rotation and change-over, a four-week measure of sales at randomly determined odd and even prices in six stores of a leading department store group was conducted. Because the re- search design provided control against the influences of confounding variables, it provided a precise and un— ambigious measure of the effect of odd and even price endings on sales for each of the items. The availability of measures of illusion and effect for identical items permitted some rough basis for determining the amount of dampening created by inter- vening variables. While recognizing that such dampening 27 would also include effects arising from sample differ- ences, differences attributable to the structure of the smniies, error variance, as well as the effects of the hfixarvening variables, it was still felt that some amnessment of the treatment effects for each of the items tmtvneen the two phases would provide a basis for determin- ing ishe extent to which the illusion effects translate thenmselves into buying determinants. CHAPTER II THE PRICE VARIABLE--ITS RELATIONSHIP TO DEMAND Historical Antecedents of Odd and Even Prices The issue of odd or even prices could not arise until the one price policy became prevalent in retailing. The Quakers were said to have experimented briefly and unsuccessfully with the one-price system in the 17th Century,1 but its institution was discouraged because of the widespread barter system which dominated trade at the time and because there existed no standardized cur- rency.2 The establishment of the one-price system has beerl attributed, at one time or another, to A. T. Stewart C“ Thaw York3 and John Wanamaker of Philadelphia.“ It is gene'l'flally agreed that institution of the one price policy becEime firmly established shortly after the close of the CiVifll War. At that time it was the practice of merchants to FHFice merchandise in even dollars so as to conform to \ lShawver, Theories of Retail Price, p. 47. 21510., p. 67. 31bid., pp. 71—72. N “The Golden Book of Wanamaker Stores (Philadelphia: "P- . 1911), p. 28. 28 29 the existing coinage system.5 Captain Macy, founder of the R. H. Macy Company, is claimed by some authors to have been the first to introduce odd prices as an out- growth of the application of a fixed markup percentage.6 Other authors have stated that odd pricing arose from several sources; one of the earliest was said to have developed from the conversion of the English pound sterling into dollars. As a result of the quality attributed to English goods, odd prices in time became associated with superior products. Because of this relationship, in the middle of the nineteenth century odd prices were used by retailers to lend a quality image 7 Paradoxically, in the 1920's and 30's, to their wares. because of their promotional link, odd prices were often associated with products of inferior quality;8 a factor which probably weakened their later impact. It was only after even or "round" prices became established that odd pricing as we know it emerged as a 5John Freeman Pyle, Marketing Principles: Organi- zation and Policies (New York: McGraw-Hill Book Company, Inc., 1931), p. 356. 6Parch C. Kelley and Norris B. Brisco, Retailing: Basic Principles (Englewood Cliffs, New Jersey: Prentice— Hall, Inc., 1957), pp. 27-28. 7History of Macy's of New York, 1858—1919 (Cam- bridge, Massachusetts: Harvard University Press, 1943), Pp. 52-53- 8Delbert J. Duncan and Charles F. Phillips, 52‘ tailing: Principles and Methods (Chicago: Richard D. Irwin, Inc., 1941), p. 501. 30 common retail pricing practice. A survey of the New York IEQEE reveals that occasional odd prices appeared in the 1850's and 1860's although the practice did not become Widespread until the turn of the century. The incidence Of odd pricing then rose rapidly until 1910, during which time about 40% of advertised products were quoted in odd endings. The next decade saw a 25% drop in odd price quo- tations followed by another sharp rise in the next five years. It has been suggested that this last rise in odd price quotations was, in part, attributable to the depression.9 Odd prices then stabilized at about 55% or 60% of the quotations for the following thirty years.10 Writers offer various reasons for the success of odd prices. Some feel that the initial impact was because of their uniqueness;ll because customers like to get change back;12 because they give the impression of 9G. Henry Richert, Retailing: Principles and Practice (3d ed.; New York: Gregg Publishing Division of McGraw-Hill Book Company, Inc., 1959), p. 307. lo"Odd-pricing in the Retailing of Consumer Goods Throughout the Past 100 Years" (unpublished study, Florida Atlantic University, 1966). 11Walter F. Kohn, "98 Cents or $l--Which is Better Price Strategy?" Printers' Ink, CCL (January 7, 1955), 30. 12John W. Wingate, Technigues of Retail Merchandis— ing (2d ed.; Englewood Cliffs, New Jersey: Prentice— Hall, Inc., 1956), pp. 171—172. 31 reductions from higher prices;13 or because they suggest that the merchandise was marked at the lowest possible price.lu The three common reasons offered for the rise of odd prices are: (l) the creation of an illusion that reduces resistance to purchase, (2) employee honesty is encouraged because sales must be rung in making change, and (3) the customer, waiting for change, is more in— 15 clined to purchase other items. In a more exhaustive inquiry into the arguments supporting the evolution of odd prices one author has found that the practice has been attributed to a compelling belief that ". . . circles attract the eye . . ." and ". . . transient, foreign born, and scatter-brained people are attracted by odd prices . . ."16 Some writers hold that variations in illusion are related to differences in sex in that women are more 13Pyle, Marketing Principles: Organization and Policies, p. 358. luHarold H. Maynard, Walter c. Widler, and Theodore N. Beckman, Principles of Marketing (New York: Ronald Press Company, 1927), p. 545. 15Harold A. Baker, Principles of Retail Merchandis- ipg_(New York: McGraw-Hill Book Company, Inc., 1939), p. 187; Duncan and Phillips, Retailing: Principles and Methods, 5th ed., p. 453; Richert, Retailing: Principles and Practices, p. 306; Harper, Price Policy and Procedure, pp. 282-283. 16 Hollander, "Customary Prices," p. 49. x'h .l-uwi‘ . ‘ , .. ’r-r‘ .. ...-u . ,...: fl ...‘-¢ .4. .0004 A 32 susceptible to price illusion than are men.17 The extent of this belief is supported by the survey of odd and even retail prices in the New York Times. The incidence of odd and even price quotations reveals that the percentage of advertised odd prices of women's products was con- sistently and significantly higher than men's.18 Looking back through the origins, establishment, and rise of odd pricing, and relating these movements to the parallel comments available in the leading texts of the time (which presumably had some influence in shaping the views of the practitioners and teachers of retailing) several conclusions seem justified. Retail pricing be— gan with the use of even or "round" prices, and given the low level of real buying power common at that time, one can assume that a high degree of price sensitivity existed. Recognizing also that this sensitivity occurred in a market melieu in which the influence of heavily advertised national brands and new product introductions were rela- tively unimportant factors (two influences which have presumably contributed to reducing price sensitivity and the creation of price inelasticity), the initial impact and subsequent spread of odd pricing can be understood. 17Paul H. Nystrom, Economics of Retailipg, Vol. I (3d ed.; New York: Ronald Press Company, 19307, p. 28; Kohn,"98 Cents or $l--Which is Better Price Strategy?" p. 32. 18"Odd-pricing in the Retailing of Consumer Goods Throughout the Past 100 Years," pp. 2-12. 4 (2) .~..- q -‘...4. vv-‘ga .- | r .5 . v "‘r u~ .. _ ‘Vilu h- a .. . n . , I»‘ ‘I‘.. A. ...-..__- | “.5 PW .._ M. Q .'"‘\r :2 .4,“ ‘\ I !“~:V~ -.-g-h; "— 1.. t Q-."v .v‘, ~.,. .. '- ‘u-v 'l_ ’t .30 -‘. \ I1 N..- “‘- -. . ._ ’ "r -. aft - -v..‘ :-.,, ‘ ‘.._r~. I‘M, '-.1 Fa _‘ f .."~ ‘ :- ‘E- . ' r ._h ’ s p‘q "- . ~. Cat. 8"). 1" .v 33 This is particularly true if it is recognized that price explicitly enters into almost every consumer transaction. Given the prevailing values of the times it seems that price would often serve as the determining variable. Therefore, any departure from customary pricing prac- tices would probably have been noticed and would have been given some interpretation by the customer. With this in mind, it would seem that although the initial value appeal was probably quite effective, in time prices began to stabilize at common price points; such as 95 or 98 cent endings. Thereafter any effect arising from odd prices was probably a result of price illusion or some nine-value association, as these end- ings were invariably considered to be regular merchan- dise at regular prices. Throughout this time one finds two general ap- proaches in literature. One echos and reechos the same essential justifications for odd pricing found in most books on the subject. The same views exist in unaltered 19 form in many texts even today. Possibly a confirming effect was created by the widespread statements of belief offered in these (presumably authoritative) publications. 19William R. Davidson and Alton F. Doody, Retailing Management (3d ed.; New York: Ronald Press Company, 1966), p. 469; Robert D. Entenberg, Effective Retail and Market Distribution (Cleveland: World Publishing Company, 19667, p. 233; Delbert J. Duncan and Charles F. Phillips, Retailing: Principles and Methods (7th ed.; Homewood, Illinois: Richard D. Irwin, Inc., 1967), p. 484. 34 On the other hand, many authors questioned the existence of any illusion. Such challenges generally took the form of simple assertions or statements of doubt, as the issue was seldom considered to be significant enough to warrant an exhaustive inquiry which would have questioned the effects of changing market influences or altered buyer dispositions. While several attempts to measure the strength of the effect were undertaken, these efforts were conducted by retailers who had little knowledge of research methodology or experimental designs. Because of this, most of the studies used insensitive before- after designs. The findings were inconclusive because of the confounding effects caused by uncontrolled vari— ables and in the end these findings failed to offer a convincing challenge to prevailing views. There is the possibility that the perpetuation of these views in literature served to sanction the pre- vailing beliefs which, in time, became self confirming, although the original conditions which made odd pricing effective had changed.20 The current incidence of even prices suggests that any price effect is possibly weaker now than in the past. 20For additional discussion and support for this view refer to: Hollander, "Customary Prices," p. 49; Kohn, "98 Cents or $1--Which is Better Price Strategy?" p. 32. r ‘r- v... 5 , . h..- -"“‘C § .' ‘,‘r-~' U.“ ._§ - ' ‘ .‘ 2r ‘. i ‘L‘i . - ~A' a_._‘ ~e ~ '._~,‘n i. . :‘u‘ I %‘ -.‘_ ‘— 7" ‘ - .‘ “' . -.l . \ ~* . \._ y- . ‘. .,: -. at - J 35 One author has noted that there are objective and subjective aspects inherent in the structure of prices which help explain the perpetuation of the odd pricing practice. The objective aspect is defined as the ". . . way in which the buyers and sellers often adjust their actions to the prevailing custom, so that in time the custom . . . merges into . . . the hard realities "21 The subjective is viewed as, of the economic world. ". . . the way in which whatever is traditional or uni— versal becomes so much a part of ourselves that we 22 Recognizing seldom are really aware of its existence." the provocative possibilities suggested, it may be that the practice was established as a result of a widespread planned defensive reaction on the part of early mer- chants.23 Although the beliefs of merchants are subject to shifts, it is possible that as the practice became established the use of odd pricing was viewed as the common, the expected, and the "right" price by the cus- tomer. In time customers might be expected to react to deviations from such "established" price points. This would further perpetuate the use of odd prices as an 21Hollander, "Customary Prices," p. 46. 22Ibid. 23Walker, "Some Principles of Department Store Pricing," p. 534; Twedt, "Does the '9 Fixation' in Retail Pricing Really Promote Sales?" p. 55. 36 institution,2u particularly so when changes to even prices undoubtedly would be interpreted as a price change on the upside, removing any assurance that the item was being priced as low as possible. The Significance and Function of Price Price is a summary figure which often becomes a critical issue in the purchase process. One statement of the function of price notes that: Price is a monetary expression of value and is the focal point of the entire exchange process. . . . Price quantitatively expresses a large number of subjective evaluations made by the con- sumer and by the supplier concerning the value of the money exchanged for goods sold. Any change in these evaluations will result in a change in the quantity exchanged or a change in prices.2 In addition, it is said that: "Prices are at bottom, "26 psychological phenomena. With this in mind it seems that a consideration of price perception and any effort to measure the relative significance of price focuses on a central aspect of the purchase process.27 2”As an example of the strength of this presumed effect refer to: Andre Gabor and C. W. J. Granger, "Price as an Indicator of Quality: Report on an Enquiry,‘ Economica, XXXIII (February, 1966), 60; Gabor and Granger, "Price Sensitivity of the Consumer," pp. 40-44. V 25Thomas A. Staudt and Donald A. Taylor, A Managerial Introduction to Marketing (Englewood Cliffs, New Jersey: Prentice-Hall, Inc., 1965), p. 451. 26Halbert, The Meaning and Sources of Marketing Theory, p. 123. 27Pierre Martineau, Motivation in Advertising (New York: McGraw-Hill Book Company, Inc., 1957), p. 5; Alderson, Dynamic Marketing Behavior, p. 31. 1 \ ..e.~v 4 n». \ .chv <01.» .n 1 _ -.' . . 37 Prices as purchase determinants are not always assumed to be specific points. Some believe that the customer approaches the purchase decision, and an evalua— tion of its primary elements--the product and the price-- in an open ended fashion.28 He is assumed to have a price range or limit within which he expects to make the pur- chase.29 Others offer an even more precise description of the process, holding that customers attribute an upper and lower price to every commodity. When prices are within the range, they submit that price is a latent factor. Only when the price is above or below these limits does it become manifest. A price exceeding the upper ranges would be considered expensive, and any item priced below the lower range would have its quality questioned.3O Others hold that retail prices need not be exactly the same to be "competitively right." They feel that customers will grant a retailer a "zone of tolerance" conditioned by the services and atmosphere the retailer 28Alderson, Dynamic Marketing Behavior, p. 111. 29Ibid., p. 31. 30J. Stoetzel, "Le Prix Comme Limite," in Lg Psychologie E'Conomique, ed. by P. L. Reynaud (Paris: n.p., 1954); D. Adam, Les Reactions du Consummateur Devant 1e Prix (Paris: n.p., 1958), cited by Gabor and Granger, "Price as an Indicator of Quality: Report on an Enquiry," p. 45. as, ~vnu ‘c. 38 31 offers. This zone is thought to be subjective, varying in individuals and retailers, but is relatively constant 32 with variations in price. Some economists also see this "zone of tolerance" extending among competitive items, creating a demand curve which is not a line but is a band.33 Others believe the price range is sequentially determined. First a "normal" or "fair" price (which is the historic price) becomes associated with a product. In the purchase process this "normal" price is compared with an item's current price. If no difference exists, the customer is not conscious of the product as either expensive or cheap. With new products or brands, no traditional price or past reputation functions as a 31Bliss, "Price Determination at the Department- Store Level," p. 43; Walker, "Some Principles of Depart— ment Store Pricing," p. 529; Knauth, "Considerations in the Setting of Retail Prices," pp. 2-3. 32The fact that this "zone of tolerance" does not apply to all customers is continually underscored by the instances most department store buyers readily recite of the customer complaints they receive that are prompted by two or three cent differentials in identical items between competitive stores. For an insightful discussion of reasons causing this phenomenon refer to: Oxenfeldt et al., Insights Into Pricipg from Operations Research and Behavioral Science, pp. 85-101. 33Peter J. D. Wiles, Price Cost and Oupput (Oxford, England: Basil Blackwell, 1956) cited by Gabor and Granger, "Price as an Indicator of Quality: Report on an Enquiry," p. 45. 39 standard, and the product is evaluated on the basis of 3“ along with other cues. its current price In the case of established products the assumption would be that the customer has a high degree of price awareness; if this were not the case the evaluative pro- cedure would be subjective and capricious. Two widely cited studies provide some insight into the degree of price awareness which exists in the case of selected convenience items. Although the studies used different procedures, and while one study was conducted in England and the other in the United States, both found that price awareness varied greatly among items and that the differ— ence could not be accounted for by the rate of purchase. Only when given a range of error of 5% to 10% was the rate of recall considered to be relatively high.35 On the basis of these studies, it seems that even when generalizing from the situation of convenience goods (whose relatively limited absolute range of prices and rapid repurchase cycle characteristics would tend to bias the results in favor of more accurate recall) a correct 3uTibor Scitovsky, "Some Consequences of the Habit of Judging Quality by Price," in Marketingfiand the Be— havioral Sciences, ed. by Perry Bliss (2d ed.; Boston: Allyn and Bacon, Inc., 1967), pp. 444-445. 35Andre Gabor and C. W. J. Granger, "On the Price Consciousness of Consumers," Applied Statistics, X (November, 1961), 263-265; "How Much Do Customers Know about Retail Prices?" Proggessive Grocer, XLII (February, 1964), Gabor and Granger, "Price Sensitivity of the Consumer," pp. 40-44. 40 specific price may not be brought to the evaluation pro- 36 One might cess for the formulation of value judgments. further conclude, because price awareness appears to vary among products, that when price awareness is high the tolerable range of price variation is narrower-—resulting in a greater effect for any given amount of illusion. Conversely, to the extent that price awareness encom— passes a range as opposed to a point (the more realistic assumption for our purposes) the effect of any illusion is relatively weak. This is suggested because of the assumption that buyers are less responsive to differ- entials in price (whether actual or illusionary) because of the "slack" created by this range. Inverse Price-Volume Relationships It has been stated previously that while odd pric- ing results in a demand curve which is highly elastic in the affected area, the general shape of the function is assumed to be negatively sloped. As a result, we con— clude that a lower perceived price falling outside this immediate range would still increase product demand. 36An exception would, no doubt, be found in the case of well established price lines which in the past would have included such items as $2.50 neckties, and $5.00 shirts. For an elaboration refer to: Hollander, "Customary Prices," pp. 45—56; Harper, Price Policy and Procedure, pp. 280-281. 41 There has been some discussion, some incidents noted, and some studies conducted that attack this assumption.37 The positively sloped demand function, or the so called "Veblen effect," is well established in economics.38 An extended form of the function which incorporates this phenomenon is presented in Figure 2. While there is no evidence to support the fact that goods demonstrating this Veblen effect would behave in the manner depicted in the lower price regions of Figure 2, incorporating such an assumption highlights the Veblen 37Harold J. Leavitt, "A Note on Some Experimental Findings about the Meanings of Price," Journal of Busi- ness, XXVII (July, 1954), 205-210; Donald S. Tull, R. A. Boring, and M. H. Gonsior, "A Note on the Rela- tionship of Price and Imputed Quality," Journal of Business, XXXVII (April, 1964), 186—191; J. Douglas McConnell, "The Price-Quality Rleationship in an Ex- perimental Setting," Journal of Marketing Research, V (August, 1968), 300-303; Ben M. Enis and James E. Stafford, "The Influence of Price and Store Information upon Product Quality Perception," Journal of Southern Business, IV (April, 1969), 90-94; Gabor and Granger, "Price as an Indicator of Quality: Report on an En— quiry," pp. 43—70; Edward M. Smith and Charles L. Broome, "Experimental Determination of the Effect of Price and Market-Standing Information on Consumers' Brand Prefer— ences," in Science, Technologyy & Marketing, ed. by Raymond M. Haas (Chicago: American Marketing Associa- tion, 1966), pp. 520-531; Folke Olander, "The Influence of Price on the Consumer's Evaluation of Products and Purchases," (mimeograph; The Economic Research Institute at the Stockholm School of Economics, n.d.), pp. 1-43. 38Robert A. Lynn, Price Policies and Marketing Managgment (Homewood, Illinois: Richard D. Irwin, Inc., 1967), p. 10. 42 phenomenon, and, at the same time, distinguishes the pattern of demand in this price region from that which will be offered in Figure 3. Under these assumptions the quantity OD is at a minimum at price OB. But the market for the item can be broadened in two ways: either by increasing the price to OC or decreasing it to 0A. In the latter case, traditional theory holds that marginal customers are attracted to the product and/or existing customers are induced to purchase greater quan— tities because of lower prices—-creating greater value because of the higher quality—price ratio. On the other hand, quantity OE could also be sold by increasing the price to OC. In this case the basis of appeal is the "prestige value" attached to the product as a result of the higher price. This exists with products which have a great deal of social visibility and evaluative content. Examples of this effect have been extensively documented.39 Empirically it is also noted that the same effect occurs in the case of health and cosmetic products.“0 In this case the effect is the result of the assumed relationship between effectiveness and cost 39Walker, "Some Principles of Department Store Pricing," p. 534; Gabor and Granger, "Price Sensitivity of the Consumer," p. 40. uoHerbert W. Warden, "Making Sales Factors Work for You," Printers' Ink, CCXXCII (June 28, 1963), 24-25. 43 in products which are important to a person's well being, rather than the result of any prestige value. The second case arises when a product is presumed to have a strong price-quality relationship, when the basis for the evaluation of alternative products is not readily apparent, when the products are considered to be different, when the product performs an important func- tion, when the product involves the expenditure of substantial effort or cost to function (e.g., the classic case of automobile wax), or when the item involves a high degree of perceived risk to the buyer. Such a demand function is illustrated in Figure 3, in which demand is maximum at price OB, resulting in the sale of quantity OE. Consistent with the general assumption, if prices were increased to OC, the demand would then shrink to OD. But in this case the market for the item would not be broadened (as in generally true) if the price was re- duced to OA. It would also shrink to quantity OD“l because price is used as an indicator of quality.“2 “1A statement of this effect is offered in Gabor and Granger, "Price Sensitivity of the Consumer," pp. 40- 44; Harper, Price Policy and Procedure, p. 39; Warden, "Making Sales Factors Work for You," p. 26; Knauth, "Considerations in the Setting of Retail Prices," p. 10; McConnell, "The Price-Quality Relationship in an Experi- mental Setting," pp. 300-303; Enis and Stafford, "The Influence of Price and Store Information Upon Product Quality Perception," pp. 90-94. u2For a discussion of the problems associated with product quality definition refer to: Oxenfeldt, Executive Action in Marketing, p. 330. 44 Figure 2 DEMAND CURVE iLLUSTRATING VEBLEN EFFECT Price O D E Quantity Figure 3 DEMAND CURVE ILLUSTRATING STRONG PRICE-QUALITY RELATIONSHIP Price C A__l__i___. O D E Quantity 45 When the price of the item is reduced the customer is left with the impression that the quality and expected performance of the item may be unsatisfactory. Price functions as the overriding indicator of quality either because of the employment of simplified decision pro- cesses or, in the absence of other sign-expectancies, it serves as the only differentiating variable to 43 relatively uninformed customers. It has been observed that this effect is particularly strong when such devia— tions occur from established price points. In a simulated experiment Gabor and Granger found a consistent pattern across a range of products (two categories of food products, two nondurable household items, nylon stockings, and a particular type of Wilton carpet identified by a sample). Their findings support the existence of such a backward bending slope, in the 45 short run at least. They concluded that the price effect is particularly strong for items whose product u3For further discussion refer to: Edward M. Smith, "Comments on: The Influence of Price and Store Information Upon Product Quality Perception, The Southern Journal of Business, IV (April, 1969), 107-109; Scitovsky, "Some Consequences of the Habit of Judging Quality by Price," pp. 442-450. 44p. W. S. Andrews, "Some Aspects of Competition in Retail Trade," Oxford Economic Papers, II (June, 1950), 154-155; Gabor and Granger, "Price Sensitivity on the Consumer," pp. 30-44. uSGabor and Granger, "Price as an Indicator of Quality: Report on an Enquiry," pp. 43-70. 46 quality can not be established by sight and when con— stant changes in technology and fashion minimize the carryover of previous buying experience]46 Other studies show that customers feel certain products are subject to significant inter-brand quality differences while others are not. A strong positive price—quality relationship is found among products in the former category.147 The existence of this inverse demand curve is also supported by two word association experiments revealing that the term "expensive," in the vast majority of instances, was associated with "high quality" or "superior.”8 Because the customer establishes the quality of a product as an outgrowth of ". . . a variety of informa— tional inputs concerning a set of criteria he (the cus— tomer) has established for judging the product,“49 price is but one factor in the formulation of his total im- pressions. On the other hand these sign—expectancies u6Ibid. u7Leavitt, "A Note on Some Experimental Findings about the Meanings of Price," pp. 205—210; Myers and Reynolds, Consumer Behavior and Marketing Management, p. 48. “BScitovsky, "Some Consequences of the Habit of Judging Quality by Price," p. 443; Meyers and Reynolds, Consumer Behavior and Marketing Managgment, p. 47. ugMyers and Reynolds, Consumer Behavior and Marketing Management, p. 46. 47 may not be consistent and mutually reinforcing.50 It is also reasonable to assume that sign-expectancies such as a brand's market position (although the results of a )51 study indicated it did not have an effect and a manu— 52 facturer's reputation would minimize the backward bend— ing price curve influence by operating as a standard in the quality-price evaluation.53 This seems plausible in view of the frequent situa— tions in which private brands (which have less quality assurance associated with them) often experience large declines in volume whenever they are priced substantially below leading national brands.5u This situation differs between new and established products or brands. In the 50Leavitt, "A Note on Some Experimental Fidnings about the Meanings of Price," p. 205. 51Edward M. Smith and Charles L. Broome, "A Labora- tory Experiment for Establishing Indifference Prices Be— tween Brands of Consumer Products," in Science, Technology, and Marketing, ed. by Raymond M. Haas (Chicago: American Marketing Association, 1966), p. 513. 52Meyers and Reynolds, Consumer Behavior and Market- ing Management, pp. 47ff. 53This view is indirectly suggested in Gabor and Granger, "Price as an Indicator of Quality: Report on an Enquiry," p. 50. 5“Stanley C. Hollander, "Retail Price Policies," in Readings in Marketing, ed. by Charles J. Dirksen, Arthur Kroeger, and Lawrence C. Lockley (Homewood, Illinois: Richard D. Irwin, Inc., 1963), p. 318; Walker, "Some Principles of Department Store Pricing," p. 534; McConnell, "The Price-Quality Relationship in an Experimental Set- tingé" p. 302; Davidson and Doody, Retailing Management, ID- 0. 48 former case, because the customer lacks a precise standard, he has little past experience to guide him and, as a result, is more inclined to use the price of the item for drawing inferences about its quality. But the quality of an established product or brand has al- ready been determined. In this case a price reduction would not affect the item's perceived quality level; but would result in an increase in the value estimate of the item and a subsequent stimulus to sales because of the higher quality-price ratio.55 The critical issue appears to focus on the relationship which exists between the price of the item and its perceived quality as well as the extent to which the perceived quality level is established independent of the item's price. This situa- tion also suggests that it is not what you pay, nor what others necessarily pay, but what you think others are, or have been, paying for an item that is determining. The case of the price—quality relationship, seem— ingly irrational from a typical economics perspective, does have some basis in logic. To the extent that the customer assumes price levels are established by the interplay of market mechanisms, a lower price may be interpreted as a reflection of the quality of the inputs 55Scitovsky, "Some Consequences of the Habit of Judging Quality by Price," p. 445; Gabor and Granger, "Price as an Indicator of Quality: Report on an Enquiry,‘ p. 48. i 49 on the cost side (the assumption being that competition in the long run eliminates extraordinary profits), or as a price compensation for relative inferiority on the demand side.56 A distinction has recently been made between the effect of the positive price—quality relationship on the perception of the product versus the product offer.57 Because customers are inclined to use price as a cue in forming quality impressions of a product, and recognizing that these two variables may be positively related, it is generally accepted that higher prices produce a corresponding increase in the evaluation of the product. But an enhanced evaluation does not necessarily mean that the customer is more inclined to purchase the pro- duct. Rather, the prospective customer is confronted with a dilemma. On the one hand he must balance his enhanced perception of the product (resulting from the positive price-quality relationship) against the deterr- ing influence of the higher price. Preliminary studies were undertaken to determine the extent to which the customer trades off the two variables.58 Although conclusive results are not 56A related argument is offered in Scitovsky, "Some Consequences of the Habit of Judging Quality by Price," p. 443. 57Folke Olander, "The Influence of Price on the Con— sumer's Evaluation of Products and Purchases," p. 44. 581bid., p. 46. 0‘" u" t" -~\ ~ - “ n». . . _‘ 4: v .‘r ‘ A '|. .- ..._ .‘ . _ u - ' ”v. v‘_. ‘ .“ ,‘ I \n' b . 'ls - t . \ ‘Q “ ‘ ‘1' L .y .. ‘v ‘I '- ~25 ._‘ 50 available, it is imagined that influences prompted by such intervening variables as perceived product differ- ences, brand loyalty, subjective determinants, and exigency factors would result in a substantial variation among products. The essential point is that higher quality evaluations prompted by higher perceived prices do not necessarily result in increased sales and a posi— tively sloped segment of the demand curve, although the effect tends to make the demand curve somewhat more in— elastic. Because most studies of the price-quality rela- tionship are solely concerned with the effect on product evaluations, it seems that there is some value in making explicit the effect of higher prices on the product and on the product offer. Nevertheless, it appears that (given all other factors as constant) lower prices do increase sales for most products. Even in a situation suggesting the possi— bility of an inverse demand function, such a result occurs only when substantial numbers of prospective cus- tomers are discouraged from purchasing a product because of the deterrent effect of its lower price. But enough customers would have to be discouraged to more than off— §g£_customers switching to a product because of the attractiveness of its lower price. Given these circum- stances, plus the limiting preconditions necessary to make it effective, one must conclude that the occurrence 51 of such a situation is unlikely although the effects of any such influence would weaken any price illusion. The Classical Approach-—Its Limitations Traditional explanations about why the consumer behaves the way he does have been derived from eco— nomics, but for the past few years the assumptions of the classical model have been subjected to searching 59 attacks by students of consumer behavior. Because the classical approach employs a reduced form model, simplifying assumptions are made which reduce its sensitivity and predictability. Its purpose is often more prescriptive than descriptive.6O For this reason, the traditional economic theory of buyer behavior, individually or in the aggregate, offers little promise of developing depth of understand— ing about the impact of price illusion or the consumer decision process. Because of its simplified structure it is mildly predictive only when behavioral extensions 59George Katona, Psychological Analysis of Economic Behavior (New York: McGraw—Hill Book Company, 1951), chap. 4; George Katona, "Rational Behavior and Economic Behavior," Psychological Review, LX (September, 1953), 307-318; Harold H. Kassarajian and Thomas S. Robertson, eds., Perspectives in Consumer Behavior (Glenview, Illinois: Scott, Foresman and Company, 1968), pp. 2—3; Francisco M. Nicosia, Consumer Decision Processes (Engle- wood Cliffs, New Jersey: Prentice-Hall, Inc., 1966), chap. 5. 60Alfred R. Oxenfeldt, Pricing for Marketing Execu— tives (San Francisco: Wadsworth Publishing Company, Inc., 1961), pp. 144-145. :n; Y -.- 0‘ - Ply-(I‘ “.r‘ v. .e..~ A~ u-u-A‘. . ......-‘ '0 a -. ...,.,__‘ ~-. 4, 9" V“. \y."‘ -‘I ‘v Q‘A -. ...,_, .:-. L ‘r “a \ 11‘ I 52 are incorporated and the variables of the model are elaborated. The theory is more adequate in describing consumer actions in an economy of scarcity where basic homogeneous products, the lack of promotion, and re— stricted outlets create a situation of equivilency and reduce the number of salient determinants. In addition the limited buying power of consumers and the value structure in the past may further justify the emphasis given the price variable. But in terms of contemporary market realities alternative, and more realistic, schemes should be explored. It has been suggested that in the future "internal" or emotional (as opposed to rational) buying determinants will become more significant in influencing consumer behavior. The feeling being that customers are no longer motivated by "rational economic objectives" but are almost entirely influenced by "social and psychological "6l’62 Although such a view is perhaps satisfactions. somewhat extreme, emotional factors do weigh heavily in the purchase of some items, just as rational determinants 61Cited by Myers and Reynolds, Consumer Behavior and Marketing Management, pp. 97-98. 62It is stated that such views are a result of Freud's pervasive influence which contributed signifi- cantly to increased emphasis on the relative significance of extrinsic motivational or emotional factors while lessening the effect of the cognitive or rational. J. McVey Hunt, "Traditional Personality Theory in the Light of Recent Evidence," American Scientist, LIII (March, 1955), 87. 53 dominate in others. But because of the limited insights they provide, and the occasional situations when such determinants are influencing, the traditional economic schemes will be cited sparingly. A Statement of Hypotheses to be Tested The subsequent studies are designed to determine whether the following hypotheses (offered in null form) will be accepted or rejected.63 There are two deriva— tives of the primary hypothesis which will be tested. They are: (la) No significant differences in the illusion, measured by the mean value estimates of products priced at odd or even retail end- ings, will be found for any of the items. (lb) No significant differences in effect, measured by sales of each of the items priced at odd or even retail endings, will be found for any of the items. 63While recognizing that statistical conventions would hold that a hypothesis is either rejected or not re— jected unless a discussion of the beta error is introduced, it would appear that sufficient authoritative precedent exists to allow acceptance of the hypotheses when they are not rejected. As examples in support of this position refer to: Paul E. Green and Donald S. Tull, Research for Marketing Decisions (Englewood Cliffs, New Jersey: Prentice-Hall, Inc., 1966), pp. 309-312; Wilfred J. Dixon and Frank J. Massey, Jr., Introduction to Statistical Analysis (2d ed.; New York: McGraw-Hill Book Company, Inc., 1957), pp. 88-89. 54 Secondary hypotheses which will also be tested are: (2) (3) (4) (5) (6) (7) No significant differences in purchase pro- pensities, measured by statements of buying intent for goods priced at odd or even retail endings, will be found for any of the items. No significant difference in the perceptions of any of the items priced at odd or even retail endings will be found. The perceived customer roles of the subjects will produce no significant differences in illusion susceptibility for any of the items priced at odd or even retail endings. The subjects' past buying experience with the product will produce no significant differ- ences in illusion susceptibility for any of items priced at odd or even retail endings. Indicated future purchase expectations with regard to each product will produce no significant differences in illusion suscepti— bility for any of the items priced at odd or even retail endings. Educational levels will produce no signifi- cant differences in illusion susceptibility for any of the items priced at odd or even retail endings. (8) (9) (10) (ll) (12) (13) 55 Head of the household's occupational level will produce no significant differences in illusion susceptibility for any of the items priced at odd or even retail endings. Whether the wife works or not will produce no significant differences in illusion suscepti— bility for any of the items priced at odd or even retail endings. Marital status will produce no significant differences in illusion susceptibility for any of the items priced at odd or even retail endings. Responses of various age groups will produce no significant differences in illusion sus- ceptibility for any of the items priced at odd or even retail endings. Responses of various income groups will pro- duce no significant differences in illusion susceptibility for any of the items priced at odd or even retail endings. No significant differences in illusion, pur- chase prOpensity, or product perception will be found within various price levels for an odd and evenly priced item. 56 (14) No significant differences in illusion, buying propensity, or product perception for an item which has been priced at $1.00, $.98, or 98¢ will be found. CHAPTER III A MEASURE OF ILLUSION The Structure of the Study The purchase behavior process usually appears to be a single integrated act. In reality purchase behavior is the result of a complex of determinants. In order to achieve a greater understanding of the process it becomes necessary to separate the phenomenon into several discrete sub-processes. The problem then becomes one of establish— ing the relationships between the processes and determin— ing their relative values. Only then can this information be restructured, combined with past information, and new knowledge and insights developed.1 The purpose of a study does a great deal to deter— mine the approach and the structure of each phase. But within the constraints imposed by the study's aims, some latitude is available. Recognizing that each alternative has its own advantages and disadvantages, the problem becomes one of selecting the most feasible. While the final structure of any study is often far from perfect, it should represent the best possible compromise. lNicosia, Consumer Decision Processes, p. 74. 57 58 Within this context, controlled or simulated ex— periments offer real promise for consumer behavior studies. The most significant advantage of such studies is that the approach is relatively efficient. Such an approach does allow the study to be conducted within the usual financial constraints with a large enough sample for statistical stability. Another important consideration is that the ap- proach offers a high degree of control over the influenc- ing variables. The study can be structured so that randomization is used to cancel out confounding influ— ences that might bias or otherwise disturb the results, while maintaining close control over stimulus exposure and psychological sets. Lastly, the approach allows the generation of more information than is possible with most survey or observational approaches. This includes the substantive material relating to the study as well as the demographic information used in the classification of responses. Artificial tests, however, are not without their limitations. Decisions made under such conditions seldom incorporate such effects as personal selling, promotions, displays, and (in our case) a "sensing" of the merchandise. On the other hand, the manipulative possibilities and the 59 high level of control such designs allow offer compensat- ing advantages.2 The simulation phase of the current study involved a group-administered test. Subjects from several reli— gious, civic, social, and fraternal organizations were gathered together and were given a questionnaire on which each item in the study was randomly priced at an odd or even ending. Subjects were immediately instructed as to the procedures necessary to satisfy the study's require- ments and were asked not to discuss their responses among themselves. During the test a black and white 35 mm. photograph or sketch of each of the eleven items included in the study was projected onto a screen while some of the significant features of the product were read. After the features were described, each subject was asked to com— plete a series of seven questions relating to the product. The same procedure was then followed for every other item. After this portion of the study was completed, the sub— jects were asked to fill out classification data. The order of the presentation of the items was randomized between groups in an effort to control for any sequence bias. In addition, to eliminate any set bias, 2Smith and Broome, "A Laboratory Experiment for Establishing Indifference Prices Between Brands of Con— sumer Products," p. 159. 60 the price ending of each item was randomized between individuals. Furthermore, in an effort to maximize the number of useful responses, after the instructions were read a sample item (a clock radio) was projected onto the screen and the item's essential features were read to the subjects. The subjects were then asked to complete the first sequence of questions as it pertained to the sample item. Any problems, ambiguities, or misunder- standings were discussed during this time. These re— sponses were not included in the results of the study. Two price departures were also introduced in this phase. In the second stage of the study, one of the items (a chip and dip set) failed to produce sufficient sales for meaningful results. In an effort to gather additional information about the level of value estimates and the illusion which may exist at various price zones, this item was assigned six endings at three price levels which included: $6.98/$7.00; $7.98/$8.00; and $8.98/ $9.00. Lastly, the lowest priced item in the study (a pint of cleaning fluid) was initially priced at: $.98/ $1.00. Because a 98¢ expression of price may have been perceived differently, this price was included in the study, and tests across each of the endings were under- taken. 61 The Approach of the Test One of the first problems to be resolved was the determination of the kinds of information that should be gathered. After considering such factors as the structure of the study, the hypotheses that were to be formulated, possible value beyond the scope of the immediate study, and the implications imposed by such factors as time, fatigue, and resources, it was decided that the basic data needed could be grouped into the following five informa- tional areas: 1. A measure of illusion. 2. A measure of purchase propensity. 3. A measure of the perception of the product. 4. The subject's perceived buyer role, purchase experience, and anticipated needs. 5. Demographic data. A Measure of Illusion The responses in this area were to provide the critical measures for this phase of the study. The pur- pose of this portion of the inquiry was to establish whether an item with an odd price represented a better buy than the same item at an even price. While such a measure was critical to the study, the issue could not be ap- proached directly because customers assume they are rational buyers who are not easily deceived. Any direct 62 inquiry would have been heavily biased in favor of re- sponses which suggested extensive deliberation and no illusion. For this reason it was necessary to approach the problem indirectly. After some thought it was decided that a valid indirect measure would be provided by having subjects furnish an estimate of the value represented by each of the products in the study. This was considered to be an appropriate approach in that a product's value is pre— sumed to be a function of its perceived quality over its perceived price. More specifically, the general formula for the value of a product is assumed to be: Vii ie-o Where: V'i = the value estimate for product i, Q'i = the perceived quality for product 1, Pi = the perceived price for product i, while Pi can be priced at an even (Pie) or at an odd ending (P10). If quality and price are independent, and the odd price is perceived as being substantially lower than the even its value estimate should be significantly higher-— and vice versa. Tests of significance between the mean 63 odd and even priced value estimates should then establish the direction and intensity of any illusion. Illusion can be viewed as a ratio between the per- ceived values at the alternative prices or: V! 1e Ii : v: 1 0 Where: I1 = the illusion for product 1 Vi = the value estimate for product i at the o odd price Vi = the value estimate for product e at the e even price. We would then observe that when Ii was significantly less than 1, illusion apparently exists. On the other hand, when Ii was approximately equal to 1, price illu- sion was not a factor. Lastly, in the event that Ii was significantly greater than 1 the assumption was that subjects' viewed quality as a positive and direct function of price; while also recognizing that illusion was strong. We might also note that the extent of subjects' variations in value estimates for each of the items was in itself significant. When the range of dispersion was narrow, we assumed that the perceived quality estimates of the item were viewed similarly by most respondents. ... -P ‘t 64 But when the range was broad, it suggested that quality determinants were ambigious, subjective, or dissimilar in other ways. A Measure of Purchase Propensipy In addition to affording a basis for determining price illusion the subjects' value estimates provide additional information. In the first place they offer a crude indication of the item's perceived value level. While we can, by definition, assume that the higher the value estimate the more highly the item is regarded, we cannot assume that a higher estimate necessarily indi— cates stronger propensity to purchase. To begin with, a difference exists between the psychological committments reflected in artificial purchase expressions and actual purchases. Researchers have recognized and cautioned against such assumptions.3 In addition, some individuals tend to be approving and supportive in such studies—- which often produces a disproportionate number of higher value estimates. One can also assume that such value estimates are weakly related to relative deprivation and might, therefore, conclude that lower income and 3Howard Trier, Henry C. Smith, and James Shaffer, "Differences in Food Buying Attitudes of Housewives," Journal of Marketing, XXV (July, 1960), 69. 65 social class levels will assign higher values to the items. Moreover, customers are inclined to use price as a cue for establishing a product's value, so we would expect some positive covariation to occur between the two. On the other hand, we also recognize that the price of an item functions as a significant purchase determinant and is thus normally inversely related to purchase pro— pensity. For this reason we accepted the distinction made by Folke Olander between the subject's perception of a product and a product offer, and recognize that the actual purchase decision involves a trade-off between the higher perceived value and higher price. For these reasons it was decided that the inclusion of some direct measure of purchase propensity would be worthwhile. While recognizing that such responses would be biased upward, the relative difference would still provide a valid measure of buying preference. ”A plausible argument might also be made for the reverse situation. The assumption being that their lack of product experiences, and differing value orientations may influence perspectives to the extent that the items are viewed as over-priced and thus represent poor values. 66 The Perception of the Product Because individuals perceive reality in different ways,5 it was deemed worthwhile to determine how each product was perceived by the subject, and if such per- ceptions were related to their value estimates. A measure of the subject's perception of a product is especially important in that the products used in the study were predetermined by constraints detailed in Chapter IV. An Indication of a Subject's Assumed Customer Role, Pur— chase Experience, and Anticipated Needs The buyer's role, purchase experience, and future needs may affect both the deliberative processes and his perception of a product.6 To establish a basis for analyzing this prospective influence, some indication of a subject's perceived buyer role becomes necessary. While this information is primarily intended for analysis 5Bernard Berelson and Gary A. Steiner, Human Be— havior (New York: Harcourt, Brace & World, Inc., 1964), pp. 96-121. 6John S. Coulson, "Buying Decisions Within the Family and the Consumer-Brand Relationship," in On Knowing the Consumer, ed. by Joseph W. Newman (New York: John Wiley & Sons, Inc., 1966), pp. 61-64; Edmund D. McGarry, "The Merchandising Function," in Theory in Marketipg, ed. by Revis Cox, Wroe Alderson, and Stanley J. Shapiro (Homewood, Illinois: Richard D. Irwin, Inc., 1964), p. 241. 67 of issues beyond the present study, it also provides a basis for measuring illusion susceptibility. Demographic Information Demographic information is an important part of almost any study. Information on such common demographic variables as age, education, and income are needed to provide a basis for the subsequent classification and analysis of results. The Selection of Items Details which entered into the final determination of the items to be incorporated in the study are offered in Chapter IV, but it is necessary to outline briefly the major considerations affecting the final choices. Several dimensions exist upon which an item can be classified. In order to maximize the value of this study, it was desirable to incorporate as many of these dimensions as possible in the items selected. While the results would be more valuable if these items repre- sented only one characteristic so as to provide an unambiguous reading along each dimension, this was not possible in this case because each of the items had to satisfy a list of preconditions in order to qualify as a candidate in the second stage of the study. The final items selected reflect a compromise, therefore they do not represent "pure" forms, nor are 68 they are broadly representative as we would have pre— ferred. A further complication occurs because no two individuals perceive products in the same light. As an example, a luxury item to one person may be a necessity to another. In addition, there exists some overlap between common product concepts. For example, what may be defined as a luxury item may, to some extent, also qualify as a shopping item. Because of this, defi— nition and classification problems arise. The issue involves establishing the product concepts which should be used for identification and determining the extent to which each of the items reflects the selected char- acteristics of these concepts. Without some objective basis for determining the classification of each item the procedure may become arbitrary. Additionally, if we knew how an item was perceived by a customer, some worthwhile insights into his actions could be gained.7 Establishing a Measure of Product Perception Alternative Approaches to the Problem Attitude scaling offers many alternatives for effectively measuring a customer's perception of a 7A somewhat related statement in this regard is made in: Nocosia, Consumer Decision Processes, p. 139. 69 product.8 Each alternative, in turn, has its relative 9 advantages and disadvantages. After some consideration (primarily assessing Thurstone's differential scale and Likert's summated scale--each of which appear to be highly feasible) a decision was made to select the semantic differential scale for this purpose.10 Some of the more important considerations influenc- ing this decision also account for the growing pOpularity of the approach.11 These advantages have been detailed by Mindak and include: 1. It is quick, efficient, and indicates the direction and intensity of the various opinions and attitudes. 2. It is comprehensive in the picture it provides. 3. It is a standardized approach for getting at the multitude of influencing factors. 8As an example of these alternatives refer to: Green and Tull, Research for Marketing Decisions, pp. 193-211. 9For an extended discussion of such advantages and disadvantages refer to: Claire Selltiz, et al., Research Methods in Social Relations (2d ed.; New York: Holt, Rinehart and Winston, 1959), pp. 362-382. 10For an excellent listing and discussion of the criteria necessary for a measuring instrument refer to: Charles E. Osgood, George J. Suci, and Percy H. Tannen- bau, The Measurement of Meaning (Urbana, Illinois: University of Illinois Press, 1957), pp. llff. 11 p. 202. Green and Tull, Research for Marketing Decisions, 70 4. It has proven to be reliable and is repeatable. 5. It avoids stereotyped responses and allows individual frames of reference. 6. It eliminates the problem of question phras- ing.l2 For our purposes there were three overriding con- siderations which strongly influenced the final decision. First was its proven reliability.l3 In addition, the semantic differential is relatively easy (as compared with the Thurstone scale) to construct. But perhaps the most significant factor is its flexibility.lu Because the scales can accommodate almost any concept,15 the semantic differential is approrpiate for use across a range of products and product concepts. After analysis and interpretation, the results can be used in several ways; values of the individual scales can be added and a summated score derived; profiles can be established for each of the concepts and alternative profiles compared, l2william A. Mindak, "Fitting the Semantic Dif— ferential to the Marketing Problem," Journal of Marketing, XXV (April, 1961), p. 29. l3Mindak, "Fitting the Semantic Differential to the Marketing Problem," p. 28; Selltiz, et al., Research Methods in Social Relations, pp. 382-383. l“Osgood, Suci, and Tannenbaum, The Measurement of Meaning, p. 76. 15Selltiz, et al., Research Methods in Social Relations, p. 380. 71 or summated scores can be calculated for the factor 16 subgroups. Use of the Semantic Differential Scale In an effort to balance and extend the SCOpe of the study it was proposed that a list of ten universal, bi- polar terms that effectively measures the significant characteristics of the most common concepts used in marketing should be established. During the administra- tion of the test, each of the subjects would then be asked to indicate his perception of each of the products on this list of scales. Each of the scales would then be used to establish the perceived dimensions of products which may have been affected by odd or even retail price endings. The Approach Used in Scale Construction The Product Concgpts The term concept as it is used by Osgood is very loosely interpreted to include any object or abstraction measured by the scale. For our purposes a more restric— tive definition is needed. While Osgood officially de- fines a concept as any ". . . 'stimulus' to which the 16Selltiz, et al., Research Methods in Social Rela- tions, pp. 380-382; Osgood, Suci, and Tannenbaum, The Measurement of Meanipg, pp. 88-93. 72 17 checking operation is a terminal 'response,'" our defi- nition of the term views a concept in the more traditional sense as a basis upon which groups of products may be classified. As a result, we are able to distinguish be— tween a product and a product concept. The product concepts used in the development of these ten universal dimensions are by nature vague. A survey of marketing tests in fact indicated that these concepts are defined in various ways. In addition, the definitions of some of the concepts have been topics of a rather extended discussion in marketing literature.18’19 The introduction of new terms, definitions, and qualifi— cations further contributes to the confusion. Unfortu— nately, most of the concepts used are not included in the terms submitted by the American Marketing Association's 20 committee on definitions, which would have helped crystalize their meaning. As a result, many of the l7Osgood, Suci, and Tannenbaum, The Measurement of Meaning, p. 77. l8As an example refer to: Louis P. Bucklin, "Re- tail Strategy and the Classification of Consumer's Goods, Journal of Marketing, XXVII (January, 1963), 50—55; David J. Luck, "On the Nature of Specialty Goods," Journal of Marketing, XXIV (July, 1959), 361—364. H 19For an additional insightful discussion into the issue refer to: Oxenfeldt, Executive Action in Market- ing, pp. 408-409. 20Committee on Definitions, Marketing Definitions: A Glossary of Marketing Terms (Chicago: American Marketing Association, 1960). 73 proposed concepts failed to provide precise, authorita- tive, and widely accepted definitions. This is par- ticularly acute because many of the more meaningful con- cepts evolved from terms developed in related areas-- particularly economics and sociology. The transition caused vague and ambiguous interpretations. In approaching the problem the first step was to gather from the leading marketing texts a list of all possible product concepts. These were listed hori- zontally across the top of a page. The products included in the study were listed vertically on the same page so as to create a product-concept matrix. The products in the study which were subjectively assumed to be embraced by any of the concepts were then checked. Those con- cepts (with the exception of "specialty good,’ which was included because of its academic value) which did not apply to any of the products or those which were signifi— cantly overlapped by another concept were eliminated. Some exceptions were also made, for example, a fashion good often falls into the shopping goods classification, but because the former term has a more restrictive mean— ing (i.e., all shopping goods are not fashion goods) it was included. As a result, the final list of product concepts was eventually reduced to: a. convenience good b. fashion good 74 c. impulse item d. luxury good e. necessity f. novelty item g. shopping good h. specialty good 1. status good j. utilitarian good/item The Selection of the Bipolar Terms Underscoring the importance of this part of the study is Osgood's statement that, the crux of the method . . . lies in selecting the sample of descriptive polar terms. Ideally, the samples should be as representative as pos- sible of all the ways in which meaningful judg— ments can vary, and yet be small enough in size to be effective in practice. In treating this subject, Osgood offers three criteria for the selection of terms. His most signifi- cant points are: factorial content, relevance to the Concepts which are to be judged, and the semantic Stability of the terms to be measured across product 22 concepts and individuals. These criteria can be easily .Iustified. Failure to include the significant dimensions 21Osgood, Suci, and Tannenbaum, The Measurement WEE: p- 20- 22Ihid., pp. 78-80. 75 of a concept could either distort perceptions or furnish a partial item profile. On the other hand, an irrelevant scale would result in a substantial number of neutral 23 judgments which would limit the amount of information furnished by a given number of scales.2u Lastly, if the meanings attached to the scales varied significantly among concepts, products, or individuals, the result would be a range of responses which bear little relation- ship to one another. The next step was to select the number of scales to be used in the study. Osgood says that as each judg— ment serves to localize a concept in semantic space, the larger the number of scales (and the more representative they are) the greater the validity with which the con— 25 cept is represented. On the other hand, there were constraints imposed by the physical limitations demanded by the study. Because the semantic differential scale must be completed for each item in the study, the addi— tion of one scale would extend the total number of re- sponses required of each subject by a factor of twelve; therefore, it was felt that approximately ten carefully 23Fred N. Kerlinger, Foundations of Behavioral Re- search (New York: Holt, Rinehart and Winston, Inc., 19655, P- 570. 2“Osgood, Suci, and Tannenbaum, The Measurement of Meaning, pp. 78-79. 25Ibid., p. 26. 76 selected scales would offer an appropriate compromise between the physical demands of the study and yet produce an adequate item profile. Furthermore, this constraint imposes strong pressures to produce a balanced list of meaningful, commonly understood, and unambiguous scales. In selecting possible scales, Osgood offers a list of fifty bipolar adjective pairs whose factor loading and purity across the range of concepts has already been established.26’27 While it would be tempting to draw from these, we recognized that, above all, the terms in the study must be appropriate to the concepts. First, we recognized that the product concepts we intended to use were vague, poorly defined, and ambiguous to even the most informed students of marketing. Furthermore, the subjects of the study would be asked to apply the final scales across a range of disparate products. Lastly, these adjectival pairs, particularly those offered by Osgood, were developed to apply across the widest pos— sible range of concepts. As a result, the adjectives offered by him are more basic and generalized--adding an 26Osgood, Suci, and Tannenbaum, The Measurement of Meaning, p. 37. 27In addition to this list offered by Osgood, there are other pretested lists of adjective pairs which are intended to use in any number of measurement situations. As an example refer to: William D. Wells, et al., "An Adjective Check List for the Study of Product Personality," Journal of Applied Psychology, XLI (October, 1957), 317-319. 77 additional degree of abstraction. Although we incor— porated several varied concepts in this study, the range is still more limited than that visualized by Osgood. Therefore, given the specific scope of the study, the value of meaningful scales, and the limits imposed by the number of scales that could be used, it would appear that the construction of scales appropriate to the unique needs of the study was justified. The next step was to determine which of the product dimensions would furnish possible bases for classifica- tion. This was accomplished by first going to the cur- rent general marketing texts. These texts were checked to determine which product dimensions were thought to be significant by the respective authors.28 These characteristics were converted into two opposing phrases which, it was subjectively felt, adequately spanned each of the product dimensions. This approach initially yielded 68 pairs of bipolar terms to be used in the next phase of the study. Two comments might be made concerning the changes which were made. As constructed by Osgood, the semantic differential scale utilizes bipolar adjectival pairs. 28Texts that were particularly helpful in this re- gard were: McCarthy, Basic Marketing: A Managerial Approach, chp. 14; William J. Stanton, Fundamentals of Marketing (New York: McGraw-Hill Book Company, 1964), pp. l32ff. 78 But adjectives are often more general than compound terms or phrases. Descriptive nouns and qualifiers were in- corporated in the scales because they add precision and develop shades of meaning which are not available from 29 the use of adjectives alone. Furthermore, the use of extreme bipolar terms would cause a clustering of responses in the center of the scale. The effect would be to reduce discrimination and in this way limit the sensitivity of the scale and the ultimate value of the study. Therefore, to achieve a greater measure of discrimination, polar terms were avoided when possible, and an attempt was made to in- corporate terms that suggested relevant segments of the product dimensions. The next step involved translating the formally stated dimensions into more generally understood state— ments having the same denotation. To accomplish this, six housewives were assembled. Care was taken to draw candidates with varied backgrounds, educational levels, and subjectively imputed intelligence. In addition, the author, who conducted the session, was acquainted with all of the housewives and each participant knew one or two of the other members. This was considered 29Mindak, "Fitting the Semantic Differential to the Marketing Problem," pp. 28-33; C. David Hughes, "Select— ing Scales to Measure Attitude Changes," Journal of Marketing Research, IV (February, 1967), 85-87. 79 important in order to encourage broad participation and to facilitate involvement by the more reticent and less secure participants. During the session, the purpose of this study was outlined. Each pair of terms as originally phrased was offered to the members, who were asked to translate these statements into equivalent vernacular terms. In the process, additional pairs of phrases which covered the same dimensions were generated and some retained, while other overlapping or otherwise inapprOpriate terms were eliminated. The end product of this effort was to compress the original sixty—eight paris of bipolar terms into forty—two scales. Next, approximately half of these terms were ran— domly reversed and the original list was randomly re— ordered. This revised order of scales was reproduced on a sheet using a standard seven—interval semantic 30 differential format. These sheets were headed with one of the ten product concepts selected earlier. Each of these concept sheets was then combined in a question- naire containing one of each of the ten product concepts in a unique random order. Twenty-seven such question- naires were assembled. 30A copy of this initial list is offered in the Appendix. 80 These questionnaires were administered to a group of twenty-seven junior, senior, and graduate marketing research students. The purpose was to isolate those scales which were most applicable across the ten product concepts, and to establish some measure of consensus among the subjects. In a previous discussion we touched upon some indi- cators which would reveal terms that most effectively describe each of the product concepts. Kerlinger states that one cannot always be certain which terms are meaning— ful to a concept, therefore any selection of scales re— quires an objective approach.31 An additional criterion added to our selection of alternatives was the factorial composition of the scales. "Factor structure," (or "factor content," as used by Osgood) is intended to represent related clusters which compose independent subsets of the product concepts.32 Osgood established that by summing over concepts the majority of meaningful judgments can be embraced by activity, evaluative, and potency dimensions.33 Al- though it is recognized that other factors contribute to meaningful judgments, the three listed above were 31Kerlinger, Foundations of Behavioral Research, p. 570. 321bid., pp. 570-571. 33Osgood, Suci, and Tannenbaum, The Measurement of Meaning, p. 51, passim. 81 established as the most significant. While the relative significance of each of these factors varies among con— 3“ Osgood's cepts, a general order appears to exist. studies indicate that the evaluative factor generally accounts for one-half to three—quarters of the extract- able variance, the potency factor typically accounts for about half that of the first, and the activity factor is generally equal to or is slightly smaller than the second.35 It appeared that factors offered by Osgood were too general to assure appropriate balance for the pro— duct concept dimensions which we revealed. For example, he found that scales defined as evaluative included more specific sub—factors such as those which are morally, 36 esthetically, socially, or emotionally evaluative. For this reason, the absence of more definitive factors could result in overweighing some specific dimensions and still not assure one of having all concept dimensions repre- sented. These considerations led to the use of an analysis of the factor content across the product con- cepts in determining the make-up of the final set of scales. Bulbid., pp. 71—72ff. 35Ihid., p. 71, passim. 36Ibid., pp. 70—71. 82 Factor Analysis Factor analysis has many possible applications,37 but in this study it was undertaken for four essential purposes: 1. To determine objectively the number of factors that underlie the range of product concepts selected. 2. To determine the scales which are loaded on these factors. 3. To establish the magnitudes of these loadings. 4. To establish the relative importance of each of the factors. Several possible approaches to factor analysis exist. Although Osgood elected to use the popular cen— troid rotation method, this did not prove to be the best approach for our purposes. Commenting on the limi- tations of the method Kerlinger said: The increasing accessibility of high speed com- puters and computing programs for factor analysis is making some of the methods obsolete. Thurs- tone's well-known centroid method, for example, will in a few years probably be little used. It is a computational compromise, as Thurstone said, to avoid the excessive computational labor of more satisfactory solutions. It will no doubt be replaced in time by the principal factors and 37As an example refer to: William F. Massy, "Applying Factor Analysis to a Specific Marketing Prob- lem," in Toward Scientific Marketing, ed. by Stephen A. Greyser (Chicago: American Marketing Association, 1963). pp. 291-307. "T1 '1‘: '7‘. 83 other mathematically and statistically more satis- fying methods.38 Thurstone, in fact, strongly endorses the use of the principal factors approach and offers the following 7' o as a guide: the principle-axis (factors or com- ponents) solution is the ideal solution for the problem of statistical compensation of the test (item) scores into the smallest possible number of factors. . . ." but adds the additional qualification that ". . . this solu- tion (the product of principal-axis method) is inadequate for the analysis of the underlying processes and their identification."39’uo On the basis of such statements it was concluded that the principal factors method promised the most satisfying results, and should be the approach used. An apprOpriate principal factors program was subse- quently found in the IBM scientific sub-routine package, 38Kerlinger, Foundations of Behavioral Research, p. 659. 39L. L. Thurstone, Multiple-Factor Analysis (Chicago: University of Chicago Press, 1947), p. 509 and pp. 503-506. quor a brief discussion of the differences be- tween the centroid and principal factors approaches refer to: Philip H. DuBois, An Introduction to Psy- chological Statistics (New York: Harper and Row, 1956), pp. 464-465; and Thurstone, Multiple-Factor Anaiysis, chps. 5, 7. - 84 although the computer limitations were such that it could accommodate only twenty-five variables}41 The Variable Reduc- tion Process Because of this limitation, it was necessary to establish an objective procedure for reducing the initial forty—two variables to twenty—five. In approaching this problem, all responses to the product concept pre—test which used the forty-two variables were quantified by assigning each response a number from one to seven to represent the indicated scale position. Each of these response values for the forty-two variables were then fed into the computer by subject and concept. Then summing across subject and concept, Pearson product moment correlations of each scale with every other scale were computed.”2 The result produced a forty—two by forty—two R- matrix which was subsequently analyzed for scales evi— dencing a high degree of intercorrelation. To start, an arbitrary level of r = i .600 was considered sufficiently high to qualify any variables for elimination. Con— versely, any scales which were not correlated with other ulThis program is detailed in: BDM Computer Pro— grams Manual, ed. by William J. Dickson (Los Angeles: University of California Press, 1964). u2This is essentially the same approach offered in Osgood, Suci, and Tannenbaum, The Measurement of Meaning, pp. 35—36. 85 variables at the level of r = i .600 were considered sufficiently independent to be included in the list of factor analysis variables. There were sixteen such inde- pendent variables. The list of scales which satisfied the independence criterion is offered in Table 1. In the case of the remaining twenty-six variables, forty—two intercorrelations were produced in which r > i .600. These results are offered in matrix form in Table 2. From this matrix, intercorrelations between the variables were charted to develop networks of inter- correlations which would delineate any existing clusters. After the networks were charted, the first cri- terion in the selection of the final variables was direct intercorrelation between selected variables and all non— selected variables. As an additional condition, to maximize independence it was necessary that no two final variables selected should be intercorrelated with any other cluster at a level greater than i .600. Because the objective was to achieve the lowest possible inter— correlations between the selected variables, it might be noted that of the 600 possible intercorrelations pro- duced by the final twenty-five selected variables, 573 had r—values of less than i .500. When a choice had to be made between intercorrelated variables, the one with the smaller standard deviation was selected. It was I ——: 86 TABLE l.--Independent variables reflecting no inter- relationships where r > t .600. Variable Highest r 1. Recommended by friends--Own opinion most important -.361 2. Specific use—-No need in mind -.530 3. Practical-Stylish .574 4. Reduces worries-—No effect .391 5. Seldom bought--Often bought .556 6. Other's opinion unimportant--Other's opinion important -.501 7. Price no object--Price buyer .481 8. Sold in few stores--Sold in many stores .595 9. Item important——Price important .329 10. Not influenced by salesman-—Influenced by salesman -.366 11. Price unimportant--Shop price .481 12. Urgent purchase--Postponed purchase -.476 13. Brand important--Features important .507 14. Seen by few--Seen by many -.161 15. Popular opinion important--Expert's opinion important .454 16. Looks just like me--No personality .544 TABLE 2.-—Symbolic representation of experimental design (treatments are shown in subscripts and parentheses). Weeks Store Store Store treatment treatment Store total total total 1 2 i A B (A)XA11 (“X512 (”Km “Ann. $(leb1. 2 2 X. 1 (A)YA11 (B)YA12 (B)YA1j S-(A)Ya1. 20313431. ‘22 Y. (B)X£21 (B)XA22 (A)XA2j 2453X32_ ‘2(B)Xb2, 1.2.x. 2 (B)YA21 (B)Y522 (A)YA2j 2(A)Ya2. 2(B)Yb2. Z?— Y. (B)XAil (M19312 . . . (A)xAij Z(A)Xa1. z (B)Xbi. z Z x, i (B)YA11 (A)YA12 o c o (A)YA ij 2 (A)Yal . 2(B)ij_ . Z ELY, ' weekly Z(A)Xa _1 1‘(A)xa_2 . . . 12(A)xa .j 22mm,” total A [(A)Ya.1 2(A)Ya.2 . . . [(A)Ya,j §Z(A)Yau Weekly 203)be zanxbj . .. Z(B)Xb _J- 22(3)bu total B {(B)Yb.1 Z(B)Yb.2 . . . 2(B)Yb.j 22mm,“ Weekly ZZX..1 22 X.,2 22XHJ- ZiZXm total zzrnl ZiYnz .u szuj 212i“ Where: A = Merchandise marked at B = Merchandise X = Merchandise Y = Net sales for week. D II odd retail price endings. marked at even retail price endings. available for sale for week. j/2 or i/2 is reached by either treatment A or B. Individual treatment assignments which are randomly determined until -fi L m»_—a)a..lsh:\~dn 88 felt that the criterion suggested by Osgood of rejecting mean values approaching the midpoint, was inappropriate when summing across concepts. Nine additional scales were selected which satisfied these criteria. These were then added to the original sixteen variables to produce the necessary twenty-five. (The nine additional variables selected by the use of these criteria are indicated in Table 2). The Factor Analysis Results In the factor analysis procedure, responses to the twenty—five variables were again summed across subject and concept. While it is recognized that the factorial composition of any variable might vary substantially 43 among concepts, our purpose was to isolate factors across the range of selected product concepts. For this reason the approach was deemed to be appropriate.uu The purpose at this juncture was to extract the common factor variance among the variables and to determine the components underlying this variance. Other points of interest arising from this analysis were: how much of the variance was explained by the program, u3Osgood, Suci, and Tannenbaum, The Measurement of Meaning, p. 74, p. 100. qubid., p. 35. 89 the number of factors produced, the relative variance accounted for by each factor, and the loadings of the variables on each factor. The results of the analysis revealed the existence of five factors with the results offered in Table 3. TABLE 3.--Program factor values. Factor Factor Factor Factor Factor l 2 3 4 5 Eigenvalues 5.4935 4.1403 2.0778 1.5923 1.1521 Cumulative percentage of eigenvalues 21.97% 38.54% 46.85% 53.22% 57.82% Percentage of total explained variance 21.97% 16.56% 8.31% 6.37% 4.81% Percentage of ex- plained common factor variance 38.00% 28.64% 14.37% 11.02% 7.97% The resulting rotated matrix of the program pro— duced the variable loadings for each of the factors which are listed in Table 4. Because the proportion of variance explained by the factors was relatively high (n2 = .5782), the results were gratifying. This left a relatively small propor- tion of unique and error variance (u2 = .4218). It would seem that these values justified pursuing the approach further. .Iauoun 3.3!.- ..308 I. .3 «3 ‘39:.- 0—3...» 0 35.. 3.. l 2.93.3228... .2 35.. an; :0. sous-auburn .nn 2... a... 3..- 2:. I... a. I: 32.1.8121 all-raid?! .8 8... to... 8.2%. 8 a! flu... an .2 8.. 2.. b2... cult-3.32:0 Ill .2 3.. 2.. 3.. 18.2.. :- 3. 3:368 3 2!. o. 3...... I... .8 .2 2... 8... 2.. 3.. 2.. 3.. n2. .2. 3... ‘38. u 82. 2.: a... .68. 72.38.. 881 It :- 12!» a .2 3.. .2 2... 8121 nus-.1812! 588- .2 2.. 3.. till... 2332...: 2.38 .2 a... 2.. ale!!! .3»...- ..u8-8..I.UI. 8.38 3.2.8 .2 .2. 2..- co..- 23.82.! 1.38-1.3 I:- .2 2... 2... $342!. lend-canal so! .2 «3.. 8..- 30. canon. 3.331.. .0 38.09.3- 103 u. 3.. .3 «a..- 3.. In..." 08.8.02 .83. .2 2.. . 12.8... 81.8.31... .1...- .2 2.. 218.. 18.3.8118. 328.8 .2 O 9 2..- .z. 2..- 3... «oz-cl...- Ioaélui I... .2 2.. 2... 2.. 2... a... $0.32.. 8. .1... 24...: on. 0.8 i... :4 .o a... a... 2.. .3. 2.. - $.38... 3.213.318! III: .. 8... .2. n2. 8.. .2: 8181 3.2.3.8253 083.8 4. 2.. .2. n... 8.. no... 112 u 32. as... 1..3..I§§1u‘2.§. . .o 8.. S..- 2... .2; no..- 81...... 33.58.8321 32.-In .. .2. 0.8.2. act-48.8.8- 282. .. 2.. 8... .3. 8.. Pill-5.05.0!- .n 2.. 1.98.38.33.38. 2 I. .~ 3.. no..- .3. 3... a... 3.. n In... :8 £89..- 3 18. Morgan.“ .— 2 2 2 2 2 2 2 2 .. 2 2 2 2 2 2 2 2 o . a . n . a a . l 2.32. i833. .oow. .1... can» noummnw madam». .H can; moanmfiums mo wanna—zit... mqmafi. 91 Establishing Factor Dimension An analysis of the individual factors proved to be of some interest in its own right. Bearing in mind that the results of any factor analysis are tentative (be— cause the variables included in the study are seldom exhaustive and as a result any factors that are derived would differ between samples and conditions and that the principle axis method does not ideally lend itself to the identification of component factors,“5 it still seemed worthwhile to label each of the factors which were isolated. In approaching this problem, another arbitrary cut off of variables having loadings of approximately i .600 on any factor was established. Using this criterion, the variables offered in Table 5 proved to be most heavily loaded on each of the five factors. The next problem was to establish a denominator for each factor. It appears that the factor common to the variables in the first group was in some way related to the shOpping effort. In the two cases of the "expensive-inexpensive" and the "durable item-frequent Iweplacement" variables, although the relationship is iruiirect, they are positively related to shopping effort. uSKerlinger, Foundations of Behavioral Research, p. 683. 92 TABLE 5.--Factor-variable-clusters. Factor Loadings Variables FACTOR 1 .606a Expensive-Inexpensive .7614a Planned purchase—Impulsive purchase .763a Calculated purchase—Habitual purchase .7663 Durable item-Frequent replacement .719 It doesn't pay to compare-Comparison shop FACTOR 2 .693a Practical-Stylish .590a Seldom bought-Often bought .820 Unnecessary purchase-Urgent purchase .592a Urgent purchase-Postponed purchase .739 Useful—Pleasurable FACTOR 3 .689a Recommended by friends-Own opinion most important .680 Others opinion unimportant-Other's opinion important .6U2 Not influenced by salesman-Influenced by salesmen FACTOR u .761a Price no object—Price buyer .683 Item important-Price important .590 Price unimportant-Shop price FACTOR 5 .726a Widely advertised-Never promoted -.653 Seen by few-Seen by many a = Final variables selected .. ”fl ‘.": 1": ..-v u . . . «a ‘1 'V‘hl .can‘dd . I ,'\ ‘.-.~ ‘3'!» ” :.. .... k. . l A ”D “H *2 .v- ,. Pr V.- ’-,. q., ()1 b . .. . ‘.'_A nay-c \ 4-..'..~- ' Q "v...” n x. “ ‘~.. A ‘ ~' 3‘ ‘ x 1-, a”, -. .. ‘ ”hr; 5T .."‘~ ~P .V .erh‘ ‘VL‘\ . r“. “a “a. L .h; . 1 I 'Q .‘: ..‘ r" “¥‘ 3’ “ “A, ..._:4 t? \4 .K‘.‘ '» 93 This seems plausible in that the subjects would be in- clined to believe that the more expensive the item, the greater the deliberation in the purchase decision. The relatively low intercorrelation of these variables with the price factor (i.e. factor four) serves to further support this View. Although the relationship between shopping effort and the "durable item—frequent replacement variable,” is also indirect, it is well accepted and established in marketing literature. With such considerations in mind, it was felt that this fac- tor was embraced by what could be viewed as the search- evaluation dimension. We might also note that this dimension furnishes the most important basis for dis- tinguishing various product concepts; accounting for almost “0% of the explained common factor variance. Following through in essentially the same way, it was decided that the next factor (accounting for about 29% of the common factor variance) incorporated a need: exigency dimension. Factor three was felt to embrace an influence-authority dimension. Factor four was highly correlated with price variables, and factor five appeared to represent a visibility-exposure dimension. A point in this regard was that the price dimension appeared to be a relatively unimportant factor in distinguishing product concepts-—explaining slightly more than one-tenth the common factor variance. 9A The Selection of the Final Variable Because factor loading is a measure of the corre- lation between a variable and a factor,“6 the selection of the final variables was relatively simple since the scales with the highest loadings are the variables which best represent each factor. But some consideration also had to be given to those variables which are factorially “7 "pure" and factorially ”complex." It would appear that the selection of factorially "pure" variables was par— ticularly important in our situation where an attempt was being made to represent each factor in a rough rela— tionship to its proportion of the explained common factor variance. To disregard this point would, to some degree, disturb this balance. Therefore, the criteria for the selection of the final list of variables was two—fold; that the variables should have either a high positive or negative loading on the factor, and that they have rela- tively low loadings on the other four factors.“8 The next problem was to determine how many variables should be used to represent each factor. In the u6DuBois, An Introduction to Psychological Sta- tistics, p. “58. u7For a discussion of the distinction between the two refer to: Kerlinger, Foundations of Behavioral Re— search, p. 652. u8The use of this factorially "pure" criterion is supported in a discussion by Osgood, Suci, and Tannenbaum, The Measurement of Meaning, p. 78. 95 construction of a test, Osgood suggests that about three sets of scales should be used for each factor, and further Justifies this by claiming that: What we do is to provide the subject with a balanced space which he may actually use as he sees fit; if he makes more discriminating use of the evaluative factor relative to others this will show up in this data (in the elongation of his space along this dimension)but he is not forced by the sample of scales to do this.“9 A decision was made to depart from the use of balanced factors as suggested by Osgood. This was done for two reasons. First, we are not interested in factor scores, as was Osgood.50 While the reliability and the represen— tativeness of such scores is enhanced by increasing the number of variables representing each factor, in our case we had a large number of factors which would sub- stantially increase the number of scales required. Addi- tionally, if a meaningful summated score is wanted which uses a D-statistic derived across all variables, a balanced number of scales allows this to be done without undertaking an elaborate series of additional computa- tions. The second reason for the departure arises from the tentative nature of the factors. These factors were developed from a single homogeneous group of subjects ugIbid. SOIbid. 96 who were responding to an enlarged list of variables.51 In effect, the number, composition, and relative signifi- cance of these five factors is subject to change with the inclusion of a broader and more exhaustive list of vari- ables or a larger and more heterogeneous group of re- spondents. Because these results do offer the best approximation we have in an area of great uncertainty, to deviate in any way from them would arbitrarily com- pound any error which might exist. With these considerations in mind it was decided that the number of variables selected to represent each factor should bear some relationship to the factor's percentage of explained common factor variance. As a result, four factors were selected to represent factor 1, three for factor 2, and one each for factors 3, A, and 5. The percentage of the explained common factor variance was 38.0%, 28.6%, lA.3%, 11.0%, and 8.0% respectively (refer to Table 3). With the three criteria of high factor loading, factor "purity," and representation of factors in rela— tionship to percentage of explained common factor vari- ance in mind, variables 7, ll, 16, and 21 were selected 51The implicit assumption being that the fatigue effect resulting from the demands of “20 individual responses (10 concepts x “2 scales) required of each sub- Ject would increase the amount of error variance and thus minimize the common factor variance. 97 to represent factor 1 in the final list; variables 5, 15, and 25 to represent factor 2; variable 2 for factor 3; variable 12 for factor A; and variable 22 for factor 5 (refer to Table A). After selecting the variables they were rearranged in random order. This final list was then incorporated as the final question in the questionnaire for use in this phase of the study. The Questionnaire Development The final form of any questionnaire is the end pro- duct of several revisions. Among the tOpics considered in the early formulation of the questionnaire, which would provide bases for subsequent analysis, were responses indicating the subject's: 1. Value estimates of the product. 2. Purchase experience and anticipated needs with regard to the product. 3. An indication of each subject's perceived buyer role. A. An indication of each subject's purchase pro— pensity. 5. Indications of each subject's product percep- tions as measured on the semantic differential scales. 6. Demographic data. 98 Several possible scales suggested themselves for measuring value and buying propensity. Some considera— tions in the final selection were the scale's ability to discriminate and the ease with which the scale could be understood by subjects with diverse backgrounds. Further— more, an equal interval scale was preferred because it would eliminate a midpoint. This would force choices, which would minimize any halo effect and result in a ggreater measure of dispersion. After considering several pn:ssibilities, a ten-point, poor—to—excellent scale sueemed to offer the greatest promise for measuring value exatimates.52 In addition, a five—interval descriptive sc21le was initially devised to measure estimates of the sutxject's buying propensity because the response re— QUiJPes a great deal of conjecture and subjects' views WOLtld be relatively imprecise. It was thought that the ”SE? of a scale with a greater number of response alterna— txivfes would simply confuse any valid expression of pur— cnjafise propensities. It was also felt that direct questions would pro— vifiei valid responses to topics two, three, and six. In tr“? case of topic five, no reason could be found for \ 52William H. Reynolds, "Some Empirical Observations CW] 8 'Ten-Point Poor—to—Excellent Scale,” Journal of MEiflESEEgng Research, III (November, 1966), 388—390. 99 departing from the standard seven-interval scale recom- mended by Osgood. A dilemma confronts the researcher involved in establishing demographic classifications for a study. On the one hand, the value of using a standard basis of classification was recognized;55 particularly in making research results cumulative. On the other hand, the unique subject, purpose, and sample of each study often :requires that adjustments in standard classifications be Inade.5u Because of the limitations, requirements and pnirpose of the study, some adjustments were made. While tflie demographic breaks generally follow the revised Iwecommendations of the Research Committee of the American Asssociation of Advertising Agencies,55 the suggested Cmuzupational classifications were rejected in favor of a mOdification of the factor analyzed group established by Carwnan.56 In the modification process, Carman's basic 53For a discussion of the issue refer to: "Standard Br'eakdowns for Population Data," Journal of Marketing, XV (AIDI‘il, 1951), u76—u78; William M. Weilbacher, "Standard ClaSsification of Consumer Characteristics," Journal of Marketing, XXXI (January, 1967), 27-31- and Ralaph Westfall, Marketing Richard D. Irwin, 5uHarper w. Boyd, Jr. R<353638th (rev. ed; Homewood Illinois: Inc. , 19614), p. 2146. C 55Weilbacher, "Standard Classification of Consumer h<"-11"a.cteristics," pp. 29-31. I ' 56James M. Carman, The Application of Social Class $¥~£fi§£flket Segmentation (Berkley, California: University Cétlifornia, Berkley Institute of Business and Economic Research, 1965), pp. 143—118. .- ‘. __—-_—‘-n- in. bx..- .-.. .A 100 16 occupational groups were reduced to seven (which simply involved aggregating groups of related classifications), while separate retiree and student categories were also included. Early questionnaires were presented to selected 57 and a preliminary form individuals for suggestions, was produced for the pretest. At this time one basic addition was made. Another ten-point scale, such as that used for gathering data on value estimates, was included to solicit item quality estimates. The data generated by the question are primarily intended for the study of issues independent of the price—illusion question. Lastly, it was felt that more valid responses to topics two and three and the quality estimate could be gathered by requiring a response before the subject was exposed to the price variable. As a result, a two- page questionnaire was devised and questions pertaining to topics two, three, and the quality estimate were included on the first page, while questions relating to the value estimates, buying propensities, and pro— duct perceptions were listed with the price variable on the second page. 57The author would especially like to acknowledge the worthwhile comments on the questionnaire format offered by Dr. F. E. Brown of the Wharton School of Finance and Commerce. lOl Pretesting the Questionnaire The initial questionnaire was then administered to 22 subjects. They included six housewives and sixteen marketing research students. The responses of the sub- jects were then hand-tabulated to determine what changes should be made in the phrasing of the questions, in procedures for the administration of the study, and in the structure of the questionnaire. On the basis of their comments and an analysis of responses, minor changes were made in both the instructions to the par— ticipants and in the format of the questionnaire. The most significant change required altering the question concerning buying propensities. An analysis of replies revealed little dispersion from the original statements. As a result, the number of possible responses was ex- panded from five to seven, ranked in decreasing order. A consensus with regard to the ordinal ranking of the statements was confirmed by first having a group of forty-two junior and senior retailing students indi- vidually sort a randomly ordered list of revised state- ments into expressions of decreasing preference. There were thirty-nine consistent rankings, and the three which were not consistent did not substantially deviate from the dominant order. 102 The Administration of the Study The initial intent was to gain the voluntary cooper- ation of various civic, fraternal, social, and religious groups in the south Florida area. But early attempts to solicit subjects at times other than regular meetings of organizations produced disappointing turnouts. As a result, it was necessary to accommodate the needs of the various groups by administering the study before, after, or in conjunction with regular meetings. Addi— tional attempts to solicit the c00peration of activity groups participating in programs sponsored by municipal recreational departments were discouraged by the recrea- tional directors of the three municipalities initially contacted. This refusal was based on experiences others have had in similar circumstances. Such participation, it appears, is strongly resented as an intrusion, when the time demands required by the study would displace the scheduled activity which attracts the participants to the recreational center. Shortly after group solicitation efforts were undertaken (in February, 1968), a family illness forced postponement of progress for several months. When efforts were reinstituted, time constraints (pri— marily those imposed by group inactivity during the summer months), required that group solicitation be 103 accelerated. Furthermore, in an attempt to establish some measure of control over the make-up of the sample group, a contribution of $1.50 per subject was offered to selected groups as an inducement to encourage partici— pation.58 To further encourage involvement, the groups were also promised a copy of the results of the study. By and large, the subjects willingly followed instruc— tions and appeared to take the study seriously. In the solicitation of the groups, an effort was made to gather responses from diverse organizations. Initially, it was proposed that a sample of 300 subjects should be used, but this was reduced because of a lag in solicitation (brought about by the fact that many of the groups met monthly), some late turnouts (which proved to be disappointing), limits on the number of subjects that could be accommodated in a group admin- istered test, two late season cancellations, and re- duced levels of group activity during the summer months. Time constraints were also imposed by limited access to the computer for processing the input. A total of 2U3 subjects was finally gathered. The sample of subjects produced 23“ useable, although not always complete, questionnaires. 58Money to cover these costs were made possible through financial support for the study contributed by the Department of Marketing and Transportation Adminis— tration, Michigan State University. 10“ Analysis of the Results The basic statistic used to determine significant differences in the mean response values of groups assigned odd and even retail price endings was the simple test of the standard error of a difference between uncorrelated means.59 This statistic was used with large samples (i.e., N > 30) and the results were converted to 2 ratios. Following conventional practice, when the samples in either the odd or even price group were small (i.e. N < 30) Fisher's E formula for testing the difference between uncorrelated means was used,60 and g ratios were then computed. Because the data were derived from what are narrowly defined as ordinal scales,61 it was recognized that the analysis should be limited to posi- tional measures. On the other hand, the statistics which are used are intended for use on interval scales.62 Because such abuses are the rule rather than the excep- tion, the use of such a statistic involves no great departure from what has been established procedure. 59For a description refer to Green and Tull, Re— search for Marketing Decisions, pp. 307—312. 60For a discussion of the application of the statis- tic refer to: J. P. Guilford, Fundamental Statistics in Psychology and Education (Uth ed.; New York: McGraw-Hill Book Company, 1965), pp. 183-18“. 61DuBois, An Introduction to Psychological Sta— tistics, pp. 99-100. 62Green and Tull, Research for MarketingiDeci- sions, pp. 187-189. 105 The Results of the Aggregate Questionnaire Responses at Odd and Even Prices A comparison of the responses of all subjects to the value estimates, stated purchase propensities, and perceptions of each of the products at the alternative odd and even prices is offered in the Appendix. The columns for each of the tables indicate the number of subjects exposed to the item at the price ending indicated in the caption as well as the mean value and standard deviation of subjects' responses. The last column offers, as a standard measure, the deviation of each sample dif- ference in terms of a 2 value. The sign preceding the 2 ratio is used to indicate the direction of the deviation. Some of the values (for example, the estimates of value and the intentions of buying prOpensity) appear to be inversely related. This is because responses in the latter case were assigned values in increasing order. In addition, it was thought that the use of a constant measure of direction for each of the scales would be less confusing. As a result, the negative sign arbitrarily indicates that the even price mean value was greater than the mean value of the odd. The results of the aggregate analysis reveal six deviations in mean scores which are significant beyond the .05 level. The first significant deviation suggests that the subjects were significantly more disposed to 106 purchase the woman's cotton knit top when it was priced at $U.00 than at $3.98. In addition, they considered their own opinion relatively more important that the recommendations of their friends when the item was priced at the even as opposed to the odd price. The same per- ceptions with regard to their opinions and recommenda- tions of friends were also found in the case of the cleaning fluid priced at $1.00 versus $.98. Two signifi- cant deviations in the perception of the "No Turn" Broiler were found. Jhen the broiler was priced at $25.00 it appeared to be more durable and required less frequent replacement than it was when priced at $2u.98. Further- more, it was presumed to have been seen by more peOple when priced at the even ending. Lastly, the indication was that the chip and dip set represents a significantly greater value at the odd versus the even ending when priced at the $7.98/$8.00 level. The Results of the Analysis of Subjects' Perceived Customer Roles and Price Illusion For each item in the questionnaire the subjects were asked if they were to purchase a product such as the one being shown what their most probable reason would be for making such a purchase? They were offered four alternatives from which to select. The alternatives in- cluded: (a) for my own use, (b) for the use of my entire 107 family, (c) for the use of someone else in my immediate family, and (d) as a gift (for either a family member, relative, or friend). The subjects could indicate more than one alternative if they felt that it was appropriate. For each item the value estimates for the alternative odd and even priced groups were then compared for the various assumed customer roles. The purpose of the analysis was to determine whether the perceived purchase role of the subject was associated with differing degrees of illusion susceptibility for any of the items in the study. These data, and the results of all subsequent analyses, are also included in the Appendix. Three significant deviations were revealed. In the first case (p < .05) the laundry cart appeared to represent a better value at the even price to individuals who were purchasing the item for someone else in their immediate family. Conversely, the cotton blend duster appeared to represent a much better value (p < .02) at the odd price to subjects who viewed their buying role in the same light. Lastly, when viewing the purchase of the chip and dip set as a possible gift purchase at $7.98/ $8.00, it appeared to be a much better value (p < .01) at the odd price. .fi..: __—_'.—_~e..‘~‘. ‘— 108 An Analysis of the Effects of BuyingiExperience and Purchase Expectations on Illusion Susceptibility In the questionnaire the subjects were also asked whether they had purchased a product such as the item illustrated within the past six months. They were next asked if they planned to purchase such a product within the next six months. The mean value estimates for each item for the odd and evenly priced groups were then analyzed. The purpose of the analysis was to determine if past purchase ex- perience or anticipated future needs are related to illusion susceptibility. The results of the analysis revealed three signifi— cant deviations. In the first instance those who did not plan to purchase a duster within the next six months viewed the item as representing a significantly better value (p < .05) at the odd as opposed to the even price. The odd price also represented a much better value in the case of the chip and dip set priced at $7.98/$8.00. This included both those who had not purchased the item within the past six months and those who plan to purchase the item within the coming six months. An Analysis of Education and Illusion Susceptibility In the analysis of the classification data, the questionnaires were first broken down by the highest 109 formal education grade completed by the respondents. These groups were then combined into two general educa- tional levels. The first group consisted of those sub- jects whose formal education extended through high school. The second included respondents whose formal education included at least one complete year of college. An analysis was then undertaken for each of the items. The mean scores of the value estimates for odd and evenly priced groups within each of the educational levels were then compared. The purpose of the analysis was to determine whether significant variations in illu- sion susceptibility exist among educational levels. The analysis offered three significant deviations beyond the .05 level. In each case the deviations were found at the higher educational levels. In one instance subjects felt that the women's shorts represented a significantly greater value when priced at the even as opposed to the odd price; just the reverse of the per- ceptions of subjects with less education-~which approaches significance at the .10 level. In another instance the subjects with more edu- cation felt that the reversible broiler represented a significantly greater value (p < .02) at the odd as opposed to the even price. Again these estimates were the opposite of subjects with less education-—which approached significance at the .25 level. 110 Lastly subjects with more education estimated that the chip and dip set represented a significantly greater value when priced at $7.98 as opposed to $8.00. In this case the subject group with less education shared the same perceptions. An Analysis of Illusion Susceptibility and Occupational Level of Fousehold Head As indicated before, a modification of the occupa— tional categories developed by Carman was used as a basis for classifying the reported occupation of the head of the household. Retirees and student household heads were eliminated from the analysis. The remaining categories were broken into two general groupings with occupations such as professionals, managers, teachers, salesmen, clerical workers, and other white collar occupations included in the first classification, and occupations such as foreman, protective workers, skilled tradesmen, laborers, and other blue collar occupations included in the second. Within each of these groupings the mean value estimates of the subjects exposed to the items at even prices were then compared with the estimates of those exposed to the items at odd prices. The purpose of this analysis was to determine whether differences in 111 illusion susceptibility could be found within the various occupational classifications of household heads. Again three significant deviations were disclosed in the analysis of the data. Interestingly, all occur within the first occupational category. Two of the de- viations were significant at the .05 and one at the .01 level. The white collar occupational category felt that r- the casual dress represented a significantly greater value at the even as opposed to the odd price. Conversely, the same group felt that the cotton blend duster represented -Aa-n- a significantly greater value at the odd as opposed to the even price. Lastly this group indicated a signifi— cant difference (p < .01) between the value estimates of a chip and dip set priced at $7.98 and $8.00. Consistent with prior results, the item seemed to represent a much greater value when priced at the odd as opposed to the even ending. An Analysis of Illusion Susceptibilityiand Working Female Head of Household Each of the subjects was asked to indicate in the questionnaire if the female head of household was em- ployed. If so, these subjects were asked to indicate whether the female head of household was employed on a full or part-time basis. 112 The questionnaires were sorted by response alterna- tive and the relative value estimates for each item at the odd and even prices were then established. The estimates were analyzed to determine if any of the three categories—-the female head of the household is not employed; the female head of the household is employed; and those cases where the female head of the household . ”u... is employed full time--offered any difference in illu- sion susceptibility. -nL- "Tl'n- _ o<< While recognizing that a working wife is a function of such factors as family life cycle, education and unique occupational skills, number of children in the family, and husband's earning power, it can be assumed that families having working wives possess different values. This may be, in part, a product of the additional buying power of the family unit, or the underlying reason which prompts the wife to seek employment. Therefore, such an analysis was deemed worthwhile. Analysis of the various groups' responses reveals five significant deviations. In the first, the subject group in which the wife was not employed, the carryall bag represented a significantly greater value at the odd as opposed to the even price. The subject group with the wife employed full time felt that the blend duster and the men's knit shirt also represented significantly greater values (p < .02) at the odd as opposed to the 113 even prices. But, this same group felt that the cleaning fluid priced at $.98 and $1.00 represented a signifi- cantly greater value (p < .02) at the even as Opposed to the odd price. Finally, the working wife group felt that the chip and dip set represented a substantially greater value (p < .01) at $7.98 than at $8.00. An Analysis of Illusion Susceptibility and Marital Status The subjects were then divided by marital status. The single and married respondents were sorted out and the responses of these groups to the value estimates for each item at both the odd and even prices were analyzed to determine whether illusion susceptibility is related to marital status. Three significant deviations were recorded-—two at the .01 level and one at .05. Married respondents felt that the carryall bag represented a greater value at the odd as opposed to the even price ending. Single subjects felt that the casual dress represented a much greater value (p < .01) at the even versus the odd price. Married respondents felt that the chip and dip set represented a substantially greater value (p < .01) at the $7.98 as Opposed to the $8.00 price. 11“ An Analysis of Illusion Susceptibility and Age The respondents were classified by various age groups. These classifications were recombined into three general age groups; 20 to 3A, 35 to 49, and over 50. The mean value estimates of each were analyzed to discover if variations in illusion could be found within the age strata for any of the products. These data reveal six significant deviations. In the first, the 20 to 3A age group felt that the casual dress represented a significantly greater value at the even as opposed to the odd price. The cotton blend duster appeared to represent a significantly greater value at the odd as opposed to the even price for the 35 to “9 age group. This same age group also felt that the cleaning fluid represented a significantly greater value at $.98 than at the $1.00 price, while the over 50 age group saw a significantly greater difference in the value at the $1.00 price. The older group also saw the evenly priced men's knit shirt as representing a significantly greater value. Lastly, the 20 to 34 age group felt that the chip and dip set represented a greater value at $7.98 as opposed to $8.00. 115 An Analysis of Illusion Susceptibility and Income Each of the subjects was asked to indicate his total family income for the previous year. Those who reported incomes were then grouped into three general classifications. These included: (a) families with incomes to $5,999, (b) families with incomes between {4} '6,000 and $9.999, and (c) families with incomes over $10,000. The mean value estimates at the odd and even prices for each of the products were then analyzed to determine if differences in income susceptibility could be isolated by the income strata. The results revealed three significant variations, all within the over $10,000 income group. These subjects indicated that the carryall bag represented a greater value at the odd as opposed to the even price. On the other hand the value estimates of this income group revealed that the women's shorts represented a signifi— cantly greater value (p < .02) at the even as opposed to the odd price. Lastly this same income group felt the chip and dip set was a significantly greater value at $7.98 as opposed to $8.00. CHAPTER IV A TEST OF EFFECT The General Research Design As previously indicated, most of the earlier attempts to measure the impact of odd or even retail prices employed relatively simple before-after research designs.1 Because of the relative insensitivity of such an approach and because of the confounding influences of uncontrollable variables,2 the results were generally mixed and the conclusions indeterminate. With these limitations in mind it was decided that a more conclusive inquiry should incorporate some of the direct and statistical controls offered by experimental designs. The problem then became one of selecting or developing an approach which would provide maximum statistical precision and an unambiguous measure of the effects arising from the alternative endings. Several established possibilities suggested themselves for lSee supra, p. 3“. 2More specifically these influences could include: variations attributable to time, store traffic, related promotions, test awareness among salespeople, stock levels, and order effects. 116 117 use,3 but each was rejected becasue it failed to provide either adequate statistical precision (e.g., randomized block designs), or because of the excessive constraints demanded (e.g., Latin square designs). Finally, a design was devised which incorporated the general characteris— tics offered in Table 6. One constraint imposed upon the design was that an equal number of weekly exposures at the odd and even prices should occur within both the rows and columns. The purpose, aside from ease of computation, was to in— sure that sufficient sales were achieved at each ending so that more reliable statistical inferences could be drawn. In addition, this constraint presumably offered a better balance of the cancelling effects provided by controlled rotation. As a result, the initial assign- ments of the odd or even treatments was made on a random basis, while the final week (or weeks) were predetermined by the constraint conditions. Some justification should be offered for the use of the design. Certainly one of the basic requirements of a good experiment is the absence of systematic 3A8 an example refer to: Seymour Banks, Experimen— tation in Marketing (New York: McGraw-Hill Book Company, 1965), chaps. iii-viii; D. R. Cox, Planning of Experi- ments (New York: John Wiley & Sons, Inc., 1958), chaps. vi-xiii; Russell L. Ackoff, Scientific Method (New York: John Wiley & Sons, Inc., 1962), chap. x; B. J. Winer, Statistical Principles in Experimental Design (New York: McGraw-Hill Book Company, 1962), chaps. iii—xi. :1].8 TABLE 6.-—Rotated factor matrix. Variable Factor loadings Commune 1— ‘ ities No. Term Factor 1 Factor 2 Factor 3 Factor 4 Factor 5 h2 1. Expensive-Inexpensive .6059 -.1138 .4011 .3603 .1550 .6949 2. Recommended by friends-Own opinion most important -.0033 -.O483 .6692a -.0558 .0265 .4812 3. Widely advertised-Never promoted .0010 .1317 .2884 -.1076 .7261 .6393 4. Specific use-No need in mind .4027 .6290 .0948 .0455 .1915 .6055 5. Practical-Stylish -.0854 .6928a .3023 -.2318 -.0706 .6373 6. Reduces worries-No effect .2979 .5151 .2363 .1704 -.l424 .4592 7. Planned purchase-Impulsive purchase .76373 .3627 .0816 -.0259 .0159 .7224 8. All brands are the same-A11 brands are different -.5179 -.0136 .3895 -.2470 -.2923 .5665 9. Seldom bought-Often bought .4892 -.5895 .0769 -.0270 -.0658 .5977 10. Other's opinion unimportant-Other's opinion important -.2108 .1680 .6793 '.0014 -.0205 .5348 11. Calculated purchase-Habitual purchase .76293 -.2503 .2303 -.0646 -.0356 .7032 12. Price no object-Price buyer - 0795 -.1875 .0816 .7611a .1076 .6390 13. Sold in few stores-Sold in many stores .5105 -.3414 .2940 .4235 -.2792 .7210 14. Item important-Price important .0669 .0676 .1917 .6829 -.1209 .5268 15. Unnecessary purchase-Urgent purchase -.0873 -.8197a .0190 .0967 .0004 .6892 16. Durable purchase-Frequent replacement .7663a -.0304 .1333 .0496 -.0715 .6134 17. Not influenced by salesman- Influenced by salesman -.1542 .1300 .6423 .0479 .0335 .4566 18. Price unimportant - Shop price -.4943 -.1377 .1144 .5899 .0313 .6253 19. Urgent purchase-Postponed purchase -.4513 .5915 .0361 .1033 -.O913 .5739 20. Brand important-Features important .1904 .1624 .4831 .2697 .1362 .3872 21. It doesn't pay to compare- Comparison shop -.7190a -.1310 .1571 .1601 -.0130 .5846 22. Seen by few-Seen by many .0120 .1613 .1152 -.0702 -.6526a .4702 23. POpular Opinion important-Expert's opinion important -.4052 -.2993 .1251 .2875 .2668 .4232 24. Looks just like me-No personality .3092 -.4699 -.2401 .3508 .0527 .4998 25. Useful-Pleasurable .0870 .7388a .1306 -_1763 -_0499 .6041 8 Selected as final variable. 119 error.L4 Although co-variance analysis is relatively robust with respect to the violation of such assumptions as normality, linearity, and residual variance homo- geneity, powerful statistics cannot tolerate a signifi— cant amount of systematic error.5 0n the other hand, one would expect that any attempt to measure the treatment effects would be confounded by such factors as: store C volume, market composition and store customer mix, store size, size of product displays, size of department, com- petitive stores within center, variations in sales sup— port activities, center promotions, assortments in dis- plays of substitute and complementary products, as well as related time factors. The fact that, a priori, we expected the treatment effects to be relatively weak, suggests that one or more of these influences could easily distort or swamp any observed effects unless a large number of stores were incorporated into the study, or the external factors were controlled. For this reason the design uses randomization through frequent change— over in an attempt to maximize the number of replicates within each store and weekly period to insure that the “Cox, Planning of Experiments, p. 5. 5Green and Tull, Research for Marketing Decisions, pp. 382-383; Winer, Statistical Principles in Experimental Design, p. 386. 120 extraneous variables favored neither treatment.6 In this way greater assurance was provided that maximum direct control over the extraneous variables had been achieved. The incorporation of these replicates into the design was deemed to be feasible. Any cross—over effect would probably be minor because of the small absolute 7 On the other hand, while any change in the prices. carry-over effect is controlled by having an equal number of changes from even to odd prices as from odd to even, failure to comply with the demands of the study would tend to dampen any sales differences attributed to the alternative endings. For example, under the assumption that price illusion is strong, if a delay in changing from an even to an odd price were to occur, fewer sales would be recorded at the odd price and presumably any differences between the alternative endings would be narrowed. 0n the other hand, if a delay were made in changing from an odd to an even price (and illusion was also strong), greater than actual sales would be accorded the even price--again narrowing any recorded differences at the alternative endings. Furthermore, the same dampening effect, although in the reverse, would also 6For a discussion of the value of randomization in experimentation refer to: Cox, Planningpof Experiments, chap. v. ‘ 7The cross—over effect in this case is viewed as any tendency to postpone purchases until the lower prices occur. 121 Operate in the case of strong illusion and a direct price- quality relationship. Recognizing that this dampening bias did exist, it was accepted as a tolerable cost as long as reasonable efforts were made to insure compliance with the changeover requirements of the study. To achieve a comparable amount of control over the extraneous vari- ables by using a less restrictive design would have re- quired a substantially larger sample of stores which would have created further problems in the solicitation of COOperating retailers and in control over the demands of the study. Furthermore, an analysis of the results of the design revealed that variance attributable to differ- ences between stores and weeks could be isolated and subtracted by blocking along these dimensions. This capability allows a substantial statistical reduction in the size Of the experimental error sum of the squares and in the end provides a greater degree of statistical precision. Lastly, through the use of co—variance analysis a measure of statistical control could also be achieved.9 While merchandise available for sale (which consists of ending stock and net sales for the exposure period), was 8Banks, Experimentation in Marketing, pp. 87—88. 9Winer, Statistical Principles in Experimental Design, p. 358. 122 finally used as a concomitant variate, ending stock and some measure of departmental sales by store were also considered as possibilities. The former was subsequently rejected when simple correlation analysis indicated that it often proved to be a result rather than a cause of sales.10 The latter possibility was rejected because the experimental design provided some control for sales vari- ations over time. In addition, because a single store group was used, any promotions creating variations in store or departmental volume would, in most instances, have a proportionate effect on all stores within the department store group for any given period of time. As a result, merchandise available for sale was selected as the co—variate, while at the same time recognizing that because each merchandise available for sale value contains a segment which is comprised of net sales for the week, it is not independent Of the variate,ll but rather is biased in favor of a positive correlation between the two. Pre—Conditions to the Study In order to achieve a greater measure of direct control over factors which might confound the treatment 10Several instances of significant negative correla— tions were Observed, particularly in situations of rela- tively low stock levels, which suggested that the direc- tion of causation was inverse. 11Banks, Experimentation in Marketing, p. 128. 123 effects, a list of item conditions necessary for a re- liable and accurate study was developed. These included the following: 1. 2. an adequate back-up stock during the period; sufficient rate of sale to Offer statistical stability; an item that is carried in all stores (pre- supposed under number 2); an item which is relatively insensitive to seasonal demand; an item having a relatively broad base of de- mand (e.g., that the item have no special tourist or gift appeal); the item cannot be advertised during the period; the item cannot be marked down during the period for either clearance or promotional purposes (the former is pre-supposed under number 2); the display of the item should be held rela- tively constant throughout the study period; that the personal selling factor with regard to the item be controlled (e.g., that the item will not be singularly emphasized by the sales force). The purpose and need for complying with the list of pre—conditions is evident, but the important point is that satisfying these conditions restricts the product 124 possibilities which could be included in the study. For example, most products that present attractive volume possibilities would probably be advertised by a retail buyer within a reasonable period of time. Additionally, astute retailers continually attempt to create interest and freshness, and try to capitalize on changing patterns and periodic peaks of demand by altering displays within their departments. The Items Because of the pre—conditions imposed, possible items for inclusion in the study were immediately limited. In addition, several factors outside the demands of the study served to reduce still further the candidate items. Some of the more important limitations were: pre- ticketed merchandise, distributor concern and competi- tive sensitivity over minor departures from established price points, the firm's policy of immediately meeting competitive promotional prices, and the general in- stability Of established retail price points——the last primarily involved drug items and national brand appli- ances. One Of the basic considerations in the selection Of items was that they be as heterogeneous as possible. It was intended that both male and female market items be included as well as hard and soft goods, staples, hedonic 125 items, fashion goods, merchandise appealing to a range of age groups, highly visible items, new and established products, and merchandise embracing a broad range of price levels. Additionally merchandise which measured the "double nine" effect (preferably at two price levels), and different quality levels of a similar item selling at different prices were other factors brought to the selection of the candidate items. The constraints imposed by the preconditions and influences outside the demands of the study forced many compromises with an ideal balance of candidate items. After soliciting the help and suggestions Offered by the various merchandising executives of the participating retailer, the items included in Table 7 were finally selected for incorporation into the study. Furthermore, after making an assessment of the expected sales levels for the items and considering the organizational demands required of the study, it was agreed that the effort would extend for a four-week period. The Participatinngetailer Any agreement to participate in a study of these dimensions involves a substantial Opportunity cost to the retailer. Without detailing these costs explicitly, one can readily appreciate the burden such cooperation imposes upon the sales people who are directly involved with 126 TABLE 7.--List and alternative price endings for study items. Item Price Cotton knit basic women's sportswear top $ 3.98/ 4.00 Chip and dip set with salad tongs 7.98/ 8.00 Women's man—tailored tweed shorts by Loomtogs 9.98/10.00 Ever Blum cleaning fluid .98/ 1.00 Cotton blend duster by Loungees ll.98/l2.00 Casual dress by Serbin 17.98/18.00 Men's Ban-lon knit mock collar shirt by Trend 7.98/ 8.00 Women's vinyl carry-all bag 2.98/ 3.00 Laundry cart 5.98/ 6.00 Retailer's private brand reversible broiler 19.98/20.00 Retailer's private brand "No Turn" broiler 24.98/25.00 changing tickets, department managers who are required to determine sales and stock levels, department buyers, and divisional and general merchandise executives. Because of seasonal peaks, which typify most retailing demand pat— terns, such organizations must Operate with some degree of organizational slack in the off seasons. In spite of this, and because many tasks are postponed until these periods, the opportunity cost of participation remains real. 127 The participating retailer was a leading south Florida department store group. During the firm realized annual sales of approximately at a main store and five branches extending sixty miles along the south Florida eastern study the $80 million approximately coast. The firm has an excellent reputation in the area and appeals to a broad spectrum of the market--ranging, it is esti- mated, from the lower—middle to the lower—upper classes. Although differences in volume levels within the firm do exist, significantly, no dominates the group. There is, in fact, an narrow range of volume generated by the top of stores single store amazingly four stores, which should minimize internal variance, and in this way achieve a more sensitive measure of the treatment effects. The firm is generally committed to an even price policy. The general exceptions are children' s apparel, major and traffic appliances, and those departments (such as drugs) in which presumed customer price sensitivity, and subsequently company competitive policy, requires that all price concessions by competitors be met. It is, of course, difficult to assess, a priori, the strength and direction Of any net effect arising from the company's price ending policy. Because of the contrast effect we might expect that there would be a heightened awareness of the odd price. Given that there is no effect beyond 128 this, and that illusion is strong, we would expect a more significant difference in treatments to result. On the other hand, assuming that this heightened aware- ness increases consideration of the price variable (as it would probably be given some interpretation by the customer), and given that cognition and illusion are inversely related, we might also assume a less signifi- cant difference in treatment effects would occur. But on the whole the assumption of any substantial increase in awareness is probably heroic, so the presumption is that the net effect is minor. The Time Of the Study In the south Florida area, typical retail demand fluctuations are further compounded by a tourist season factor. [Because of the additional stress this season imposes on the existing retail facilities and organiza- tion, the demands of the study required that the effort be undertaken at a time other than at this seasonal peak. It is reasonable to assume that the demand patterns of tourists are atypical (the usual assumption being that such demand is more inelastic), and this would presumably dampen any measured illusion. Additionally, one cannot expect that the same level of compliance to the demands of the study would occur during the time 129 that such acute demand pressures (which must take first priority) are imposed upon the organization. During much of the "Off season" most south Florida retail firms experience a substantial decrease in retail activity. This severely limited the candidate item possibilities and the amount of sales activity each item could generate so as to provide a reliable measure of price effect. Furthermore, retail activity during much of the "Off season" is characterized by sales and price promotions which, it is presumed, would also make demand for regularly priced items less sensitive to any price illusion. As a compromise the period be— tween May 1 and May 27, 1967 (although it included a Mother's Day weekend), was finally selected as a feasible alternative. The Administration Of the Study Because the items in the study were controlled by several merchandise departments in several merchandise divisions of the organization, and these items were to be made available to customers through a group of stores which spanned an extensive geographic area, a problem of control and coordination was presented. It was deter- mined that the basic unit of coordination should be established at the main store department buyer level. At this point the buyers and their assistants were 130 instructed in the purpose and procedures of the study. Detailed written instructions and summary forms were made available to them and assignments and schedules for com- piling the data were established.12 Prior to the start of the study each of the affected department managers at each of the stores was called and informed about the study by the respective department buyers or their assistants. They were then sent a packet containing in- structions detailing their responsibilities, a schedule of price changes as it affected them, and a form for the derivation Of net sales and ending stock for items within their realm Of responsibility. The sales and stock levels were to be reported to the main store each Monday in order to assure compliance with the demands of the study and to serve as a reminder when changes in price were to be made. Actual checks for compliance with the schedule of price exposures was incorporated into the branch store visits of main store personnel and visits expressly undertaken for this purpose by the author. In addition, weekly visits were made each Monday evening to each of the main department store buyers to insure that the re- sults were being reported, and to answer any questions that may have arisen. 12Examples of the forms and instructions used in the study are included in the Appendix. 131 During the actual time of the study, the co0pera— tion of all participants was very good. For this period only three violations within any department for any item ever came to the attention of the author. It might be noted that any violations would have a leveling effect, although the structure of the design would help to insure that neither treatment would be favored. During the time the study was being conducted two significant complications arose. In the first instance it was proposed that a woman's sportswear tOp, which was sold in the budget sportswear department at $3.98/$4.00, be incorporated into the study as a test of the differ- ence in illusion between the Budget and Junior depart— ments, as each draw heavily from different age segments and use different degrees of price emphasis. But during the time that the test was being undertaken there was a change of buyers in the budget sportswear department and conditions necessary for the control of sales and ending stock for the item were violated. Consequently the women's shorts, which were available in the better sports— wear department, were added as a substitute item two weeks after the start of the test. Therefore, the study period for this item extended two weeks beyond the normal expiration date in order to achieve the necessary twenty— four exposure periods. 132 Secondly, at the time Of selection of the items, the housewares buyer suggested incorporating the chip and dip set into the study. The style Of the item was new and the buyer was quite enthusiastic about the prospective sales it would generate. But when the item was introduced, the Old style (which was essentially the same item with a different design) was marked down sub- stantially in order to clear it from stock—-a violation Of one of the pre—conditions. Because of the markdown of the substitute item, and the acceptance of the new style was disappointing, insufficient sales were generated during this period to justify any analysis. The Analysis Table 8 summarizes total sales and merchandise l3 totally at the odd and even prices for item exposures the study period. After compiling the data the next step was to determine if co—variance analysis could be used for each of the items. But co-variance analysis is justified only if one can establish that sales were a function of merchandise available for sale. Although, at first blush, one might assume that the relationship is strong and self-evident, further con— sideration leads one to qualify this assumption. In the l3A merchandise item exposure is defined as the exposure and availability of an item, within a store, during a week, within the test period. 133 TABLE 8.--Net sales and merchandise item exposures (MIE) during the study period. Item Odd Even Total Sales/MIE Sa1eS/MIE Sales/MIE Women's cotton knit top Chip and dip seta Women's shorts Cleaning fluid Cotton blend duster Casual dress Men's knit shirt Carryall bag Laundry cart Reversible broiler "No Turn" broiler 60/514 3/62 51/693 106/881 57/167 18/44 337/1736 180/730 5/90 25/124 27/122 57/538 l/6l 60/690 138/832 83/207 19/59 198/1605 110/623 7/93 22/101 23/114 117/1052 4/123 114/1383 244/1713 140/374 37/103 435/3341 290/1353 l2/183 47/225 50/236 aItem not included in study. situation where all merchandise is displayed on the sell- ing floor, we could expect that with a given flow of traffic through the department, the greater would be the chance of the merchandise display being noticed and, assuming the item has some appeal, an increase in sales would result. One might also expect that where lost sales were an important factor, a positive sales—stock relationship might also exist. Such a situation would be 134 particularly true in the case of fashion items where cus- tomers may have strong color preferences and/or size needs. But it is apparent that this relationship is not linear. We would expect (making some assumptions about stock balance) that sales would be more sensitive to assortments at lower stock levels, whereas, after a point,further increases in stock would have little effect. In fact, with some fashion items, additional increases in stock levels may act as a deterrent to sales because some women value unique style offerings. Although the curvi- linear relationship discussed was cited with reference tO fashion items, we would expect, although to a lesser extent, the same general relationship would be true of most products. While the effects described may be consistent with the assumptions held by most individuals, in the case of items such as traffic appliances, which are sold from floor samples, any such relationship would be difficult to support as long as the merchandise was never out of stock. In such cases if a significant sales-stock rela- tionship were evident, we would assume that the measured association was spurious and probably should be attributed to expected sales which were subsequently realized. In the case of one of the two broilers such a relationship did in fact occur. Although in one case the association between the net sales and the merchandise available for 135 sale was insignificant, the other broiler revealed a relatively high degree of association. Although the assumptions about the relationship were not logically justified, co-variance analysis was used simply to conform to the prior decision rule, and because it had no signifi- cant effect on the outcome of the study. Correlation analysis was used as a measure of the association between net sales and merchandise avail- able for sale for the item during each of the twenty- four periods.114 If it was found that the measured degree of association was significant at the .05 level, the analysis of co-variance was deemed to be justified. The product moment correlation coefficients for each of the items in the study are Offered in Table 9. Correlation analysis is generally intended to be used as a large group statistic, while our sample was limited to twenty-four exposure periods for each item. As a result, we find that the significance levels are 1”An exception to the twenty-four exposure periods was made in the case of the casual dress. Because some of the areas in which the participating retailer has located stores draw heavily from tourists vacationing in the south Florida area, and because it has been found that their clothes tastes and needs differ significantly from those of regular south Florida residents, a separate buying staff was used in selected departments to control the offerings in two of the outlets. The casual dress was made available in four of these stores, but was not purchased by the buyer of the so-called "northern con- cept" stores. As a result, sales Of the casual dress were studied for the four-week period within four stores—— resulting in only sixteen exposure periods. 136 TABLE 9.--Item product-moment correlation coefficients. Item Coefficient of Correlation Women's cotton knit top .495a Chip and dip setC Women's shorts .266 Cleaning fluid .173 Cotton blend duster .475b 8— Casual dress -.189 1 Men's knit shirt .576a l Carryall bag .616a Laundry cart .101 “' Reversible broiler .211 "No Turn" broiler .639a aSignificant at the .01 level. bSignificant at the .05 level. cInsufficient sales. relatively high (i.e., r OS(2,24) = .388 and r 01(2,24) = .496). On the other hand, because we are using the con- cept merchandise available for sale which has a substantial positive bias (it might also be observed that this bias is particularly significant when sales-stock ratios are relatively low), it has been felt that the use of the .05 level of significance was not overly conservative. Addi- tionally, when one considers that the traditional minimum 137 validity coefficient for a test of practical usefulness, as suggested by c. L. Hull, was felt to be about .450,15 it would appear that the standard established in this case was not particularly out of line. Using variance or co-variance analysis, as estab— lished by prior decision rules, the F-ratios indicated in Table 10 were obtained. 15Guilford, Fundamental Statistics in Psychology and Education, p. 104. 138 TABLE 10.-—Item F ratios. Item Type of Analysis F ratiosc Women's cotton knit top covariance .59 Women's shorts variance -l.2Oa’d Cleaning fluid variance -l.36a Cotton blend duster covariance -4.l4b Casual dress variance - .01 Men's knit shirt covariance .52 Carryall bag covariance .74 Laundry cart variance - .19 Reversible broiler variance .67 "No Turn" broiler covariance .17 aThis approaches the .25 level of significance where F 25(1,14) = 1.44. bThis approaches the .05 level of significance where F OS(1,13) = CAnalysis of ing the derivation in the Appendix. dMinus signs greater sales were 4.67 and F 10(1,13) = 3.14. variance or covariance tables detail- of the F ratios above are included preceding the F ratios indicates that realized at even versus odd prices. CHAPTER V CONCLUSIONS AND RECOMMENDATIONS FOR FUTURE STUDY Introduction In the analysis of any study, rarely will the re— sults provide clear-cut answers to the questions under investigation. At the expense of pursuing a discussion of alpha and beta errors, we also recognize that the occurrance of any significant deviation, particularly at the .05 level, does not per se suggest the existence of any effect. As a result, in developing our conclusions we first note the incidence Of significant deviations, explain their meaning on the basis of the information at hand, attain some perspective by looking across these significant deviations, and then draw inferences about the issues we are pursuing. The Aggregate Test of Illusion In testing the first hypothesis we note from the aggregate analysis in Appendix C that fourteen separate measures of illusion were undertaken. Of these fourteen measures one significant deviation in illusion did emerge, and that occurred at the .02 level. On the other 139 140 hand the probability is about .3 that a significant de- viation beyond the .02 level would have occurred on the basis of chance within one Of the fourteen tests. Furthermore, we find no other illusion deviations beyond the .2 level, and eight of the fourteen do not extend beyond the .5 level. loreover, six of the fourteen de- viations are negative (while the remaining eight are positive); suggesting further that the only significant variation may have occurred by chance rather than having been caused by any impact arising from the price endings. Nevertheless we will defer a decision on primary Hypo- thesis la until the subsequent demographic evaluation has been completed. The Test of Effect The most immediate conclusion from the analysis of sales in the second phase of the study is that any measured effect is relatively weak. The results indicate that in only one case did differences in sales approach the .07 level. On the other hand, with ten items in the study the probability of a difference this large occurring by chance is .7; so a deviation of this magnitude becomes highly probable. This conclusion seems even more justi- fied because the variation was again found to occur equally in the direction of odd as well as even prices. As a result it would seem that Hypothesis lb should be accepted. But one should also consider the possibility 141 that an effect does exist,but insufficient sales activity was generated for the effect to be established at the conventional levels of significance. Along this line, one might point out that the casual dress and the laundry cart failed to stimulate sufficient sales activity to provide meaningful results (refer to Table 8). On the other hand,their justification for inclusion in the final results lies not so much in the significance that might be attached to them individually, but in what they con- tribute in terms of a total pattern. It would seem that several of the items together should furnish sufficient sales so that if any reasonable effect did exist a pattern across the ten items would have been evident. But because of the ancipital results (with five items producing sales in the direction of Odd prices and five in favor of even), the assumption of a weak or nonexistent effect appears to be tenable. Additionally no apparent consistent direction with regard to either product types, price levels, or price breaks is indicated by the results; further suggesting that the variations are random and no substantial price illusion is apparent. Purchase Propensities Of the fourteen measures of purchase prOpensity we again find one significant deviation, and that occurs at the .05 level. We also note that eleven of the remaining thirteen tests of purchase propensity lie below the .4 142 level. Furthermore, we again find ancipital variance. While eleven of the fourteen measures are negative, five of those negative deviations lie below the .8 level of significance. These factors suggest that we should accept Hypothesis 2 and conclude at this time that odd and even retail endings appear to have no effect on the subject's indicated buying intentions. Product Perceptions Of the 140 tests of product perception we would have expected seven to occur beyond the .05 level on the basis of chance alone. But in looking across the scales for each Of the products, we find four significant deviations, three Of which are at the .05 level. We must, as a result, accept Hypothesis 4 and assume that in the aggregate Odd-even price endings have had no apparent impact on the perception of the items as measured by the semantic differential scales. Analysis of Demographic Illusion Although our aggregate analysis indicates that price illusion, as we have attempted to measure it, does not exist for any Of the products in the study, the possi— bility still exists that selected subsets within the subject group may still be susceptible to illusion for some of the products. This illusion may not have been revealed in the aggregate analysis either because the 143 illusion susceptibility of the subgroup is relatively weak or the illusion susceptibility of the subgroup may be Offest or swamped by the responses Of subjects in the other cells. In attempting to assess whether any illusion was revealed through the demographic analysis Of the compara- tive value estimates of the subjects, we note that of the 272 individual evaluations we would have expected that approximately 14 significant deviations would occur on the basis of chance beyond the .05 level; about 5 beyond the .02 level; and 3 beyond the .01 level. But analysis of the data reveals 29 significant deviations, with 5 beyond the .01, and 12 beyond the .02 level. The initial conclusion appears to be that illusion, as measured by the study, does in fact exist for certain products within selected subgroups. Of the 29 significant deviations, we note that nine Of these can be attributed to the comparative evaluation of the chip and dip set at $7.98/$8.00. Furthermore, we note that the item at this price line produced the only significant deviation (p < .02) in the aggregate test Of comparative value estimates. In addition, relatively few subjects were exposed to the item at either the Odd or even price (i.e., 35 at $7.98 and 39 at $8.00), which would tend to increase the effect of any extreme deviations. We also find that, while -'-:r--*.f"- «' ‘-‘~“- 144 the direction is consistent, the most extreme deviation at either the $6.98/$7.00 or the $8.98/$9.00 price line just exceeds the .5 level of significance--further suggesting that the $7.98/$8.00 deviation may be random. Lastly, although we indicated that the relationship is not direct,1 we find no indication that the mean values Of the purchase propensities between two price endings have been affected (i.e., E = -.ll6)--further indicating that the deviation is random. The point at issue is that this possible random variation between the mean values of these subsets compounded the incidence Of significant deviations in the demographic analysis by measuring the extreme value perceptions of relatively few subjects several times--particularly so in that a significant deviation for this item at the $7.98/$8.00 price level was found in each demographic classification. Assuming this is the case, and subtracting out all instances of the illusion measures of the chip and dip set at $7.98/$8.00, we find that the number of illusion measures becomes 253, which should, on the basis of chance, produce about 13 deviations beyond the .05 level. But we still find 20 significant deviations where p < -05. Furthermore, to avoid being ultraconservative, if ‘ 1For the chip and dip set, at all price levels, a Subsequent measure Of the relationship between value estinmtes and purchase propensities found r - -.47l. 145 we subtract out all chip and dip set measures (for the same reason we used to justify their elimination at the $7.98/$8.00 price level), we are left with a total of 215 measures which should produce about 11 signifi- cant deviations at the .05 level, while the number Of actual deviations remains at 20. Therefore, we con— clude that for certain customer segments price illusion is a factor in the evaluation of certain products. But while the demographic analysis indicates that illusion appears to be a factor, the results also suggest that the direction, intensity, and extent of its impact is complex and subtle. Illusion by Demographic Groups For the reasons detailed above, we will recognize the bias produced by the chip and dip set, and will exclude it from our subsequent analysis. Furthermore, although we discussed the significant deviations in Chapter III, it is necessary to consider each again in an attempt to integrate them in a broader frame. Subject's Perceived Customer Role Excluding the chip and dip set, the analysis of responses by perceived customer buying role produced 31 measures for the remaining ten items. With this number, random variation alone would have produced one to two deviations beyond the .05 level. Two were in fact 146 produced, with one beyond the .02 level. While the deci- sion is marginal we will not reject Hypothesis 4 and will tentatively accept the statement that perceived customer role does not appear to be related to price illusion. Past Buying Egperience and Future Purchase Expectations The analysis of past buying experience and future purchase expectations produced 34 illusion test pos- sibilities. While this number should have produced about two deviations which are significant beyond the .05 level, only one at the .05 level was revealed. In this instance we again will not reject Hypotheses 5 and 6, and will tentatively accept the position that past buying experience and future purchase expectations are not related to illusion susceptibility. Education Levels In the case of educational levels we found that of the 20 possibilities, one chance variation would have been expected. In this instance two significant devia- tions (one beyond the .02 level) were actually recorded. We will, as a result, reject Hypothesis 7. In attempting to assess the nature Of this illusion we find that the effect is mixed, although both occurred within the higher educational strata. In the case of the women's shorts we find that the illusion is a combined result of lower 147 value at the odd price and higher value at the even. To a lesser extent, this also appears to be the pattern within this stratum in the case of the other fashion item-—the casual dress. Moreover, with this item we note that the value impressions of the other edu— cational stratum are reversed, causing extreme and significant value estimates within the price endings between strata.2 On the other hand, these between strata differences are not found in the case of the casual dress. With one exception, a reasonable measure of be- tween strata agreement exists for all other items. The exception involves the other significant deviation—-the reversible broiler. In this case the Odd price impact on the two strata are reversed. The odd price ending resulted in a much higher value estimate in the case of the upper educational stratum, while it resulted in a lower mean estimate with subjects who did not attend college.3 No substantial difference between the strata arises from the use of even endings. One last point should be noted; in both instances the items involved tests of the "double nine" effect. 2The differences in the value estimates approach the .02 level in the case of the odd prices although the even is at the .16 level. 3The difference between the strata is significant at p < .001. 148 In general we conclude that subjects with more education appear to be more susceptible to price illusion--but the nature of this illusion is mixed. In the case of fashion goods a strong price-quality relationship appears to exist. It would seem that the use of even prices acts as a slight (although not significant) inducement to sales, while the use Of odd prices acts as a signifi— cant deterrent. The reverse appears to exist in the case of subjects with less education. Conversely, with appliances significant evidence of illusion is also found, although in this case the use of Odd prices substantially enhances the appeal Of the item among those with more education. In sum it appears that the "double nine" endings have a strong and mixed impact on various educational strata, which differs among pro- duct types. Occupational Classifi- cation Of Household In this case we have 20 tests of illusion which should produce one significant deviation on the basis of chance. But we again find two significant deviations which suggests that we should reject Hypothesis 8. As with educational levels, the two significant deviations are also mixed, although in this case, they froth involve ready-to-wear items. Both deviations 149 occur within the white collar stratum, which suggests that this group seems to be more susceptible to illusion that subjects from blue collar homes. In exploring the nature of the illusion we find, in the case of the casual dress, the illusion also results from higher values being assigned to the item at the even price. Thus it appears that with fashion items the positive price—value relationship is an effective influence and suggests that even price endings may, in some situations, 1 function as a purchase inducement. : Conversely we find that with the duster this same g group significantly undervalues the item at the even price-~suggesting that in this case even prices would serve as a purchase deterrent. This is the Opposite Of the illusion direction Of blue collar subjects, which suggests that while even prices may act as a strong deterrent in the former situation, the price endings are interpreted differently between occupational strata in the case of this item. Working Female Head of Household In this case the analysis has produced thirty tests of illusion. But we also recognize that the three classifications are not discrete. More specifically, tile "wife does work" category is composed of wives who 150 work both part and full time. Therefore, the "wife works full time" classification is really a redefinition of the former. Given that the "wife works full time" group is especially susceptible to illusion and repre— sents a substantial portion of the "wife does work" classification, it would tend to bias the results. But because the former classification produced no signifi- cant deviations and the final conclusions were not af- fected, the issue becomes moot. To begin with, and under the conservative assumption that the categories are dis— crete, we would have expected one to two random devia— tions at the .05 level. But we find four, all of which are significant beyond the .02 level.“ As a result Hypothesis 9 must be rejected. In looking at the significant deviations more closely we find that the group with wives employed full time accounted for three of the deviations and the wives not working group accounted for the one which remained. This would lead us to conclude that illusion suscepti— bility does differ between the classifications, and that families in which the female head of the household is employed full time are more susceptible to illusion. “If we were to compare the first and third cate- gories, we would have had twenty tests, which on the average, would have produced one significant deviation by change beyond the .05 level while the number of significant deviations would have remained at four. 151 In cases where the wives are employed full time we find that the effects are mixed. The illusion in the case of the duster and the knit shirt was a combined product of higher valuations at the odd price, and lower valuations at the even, with the reverse occurring in the case of the cleaning fluid. We might conclude that for this group (i.e., in the case of the ready—tO-wear items) odd prices appear to encourage sales of items and even prices to discourage them, while the opposite would hold true in the case of low priced instrumental items. In the "wife does not work" group, the illusion found in the case of the carryall bag is a product of substantially lower valuations at the even prices. Lastly, we note a general consistency in the direction of the deviations between these strata for all items. Marital Status In this case we again have twenty measures of illu— sion and two significant deviations (one at the .01 level). We would, as a result, also reject Hypothesis 10. In looking at the deviations we find that both married and single subjects appear to be susceptible, although the types of products and the direction of the illusion differed between the strata. The single group valued the dress substantially lower at the odd end- ing while valuing it significantly higher at the even-- 152 suggesting that, in addition to illusion, a strong posi- tive price-value relationship existed. Interestingly neither the illusion nor evidence of the positive price— value relationship is found in the case of the women's shorts. In the case of the dress the even price ap- parently served as a positive inducement to purchase. Married subjects assigned a higher overall value to the carryall bag at both price endings, and at the same time valued the item at the odd price more highly still. As a result we find the item highly regarded by the group, with the Odd price serving as an even greater inducement to purchase. Ag: Again we have thirty illusion possibilities among the various age strata which would have been expected to produce one or two deviations by chance. Because five were actually revealed, we must then reject Hypothesis 11. In this case we find one deviation at the 20-34 age group, two in the 35-49 group, and two in the group over 50. We also observe that the nature of the devia- tion appears to be somewhat consistent within the strata. First of all, the 20-34 age group is the only one revealing significant valuation deviation for the fashion items. Furthermore, the nature of this 153 deviation was prompted by lower valuations for the item at the odd price. It would, as a result, appear that in this case Odd prices Operate as a purchase deterrent. The 35—49 age group, on the other hand, valued the duster substantially higher at the Odd price and somewhat lower at the even, which are about the same valuations assigned to the cleaning fluid. But the over 50 age group assigned a substantially lower value to the clean- ing fluid at the Odd price and essentially the same valuation at the even. Therefore, it appears that while odd prices act as a strong purchase inducement for the 35-49 age group, it acts as a deterrent to the over 50 group. Similarly the over fifty group valued the men's knit shirt much lower at the Odd price, but also valued the item much higher at the even. It seems appropriate, as a result, to conclude that the impact Of odd and even price endings on selected items are interpreted dif- ferently by the various age strata. Income With three income strata we are Offered thirty tests of illusion. These tests produced two significant deviations; one at the .02 level. Because 1.5 deviations beyond the .05 level would have been expected on the basis of chance, this creates another marginal situation. 154 As a result of some consistencies which will be dis- cussed later, we will tentatively reject Hypothesis 12. In this case we find the instances of illusion con- fined to the over $10,000 income group, although the nature of the illusion is mixed. The significant devia- tion in the case Of the carryall bag is brought about by substantially higher valuations at the odd price and substantially lower valuations at the even. But we find the situation just reversed in the case Of the women's shorts. In this instance the item was assigned a very high valuation at the even ending. As a result we find that with one product the odd price appears to offer a strong inducement to purchase, and in the case of the other it acts as a strong deterrent--and vice versa for the even price. Therefore, it seems that the nature and direction of illusion for this income strata is peculiar to the product or item. Illusion Between Price Lines In the case Of the chip and dip set we find one instance of illusion beyond the .02 level at the $7.98/ $8.00 price line. While such evidence would be suffi- cient for us to reject Hypothesis l3, observations appear to be appropriate in this regard. We might first note the relatively small differences in valuations between the price lines. As a result we might first 155 question the ability of the study to make discriminations in valuations. But when we consider the broad range of mean values between products (refer to Table 12) and the differences in the value levels between subject groups (refer to Appendix D) the internal question, if not the external issue, appears to be satisfied. We must, therefore, conclude that a strong positive price-quality relationship has had a substantial influence on the subjects' value estimates. This would indicate, at least in the case of this item, that values are determined in a relative manner and further suggests that perceived quality is in large part a function of price. This makes the number of significant deviations revealed by the analysis even more meaningful. If, as suggested, the perceived quality level of an item were to covary positively with the perceived price, it would dampen changes in value estimates between endings even when illusion is strong. Moreover, the illusion measures suggests that illu- sion susceptibility is curvilinear; with susceptibility being a partial function of what the customer may con— sider to be the appropriate price range for the item. Lastly, in looking across the mean values at the alterna- tive prices we see that the illusion occurs as the com— bined result of a higher value at the odd price plus a lower value at the even. 156 While the statements offered above are admittedly precarious, the results support little in the way of alternative conclusions, unless one is to discount the deviation as random--which would be even more difficult to justify in view of its .02 significance level. The Relative Impact of $398 and 98¢ Prices In an analysis of mean value levels between the $.98 and 98¢ prices, we find an infinitesimal differ- ence between them. Therefore we have no reason to reject Hypothesis 14. An Assessment Of Illusion by Product The number of significant deviations by product, as well as their intensity, direction, and the demo- graphic breaks with which each is associated, are listed in Table 11. We note that no significant deviations were found in the case of two of the items--the cotton knit top and the "NO Turn" broiler. Furthermore, we note that two additional items, the reversible broiler and the laundry cart, produced just one significant deviation each, although in the former case the deviation ex— ceeded the .02 level. In the former case the subjects felt that the item represented a much greater value at the Odd price, while in the latter the item was viewed 157 TABLE ll.--Number, intensity, and direction Of signifi- cant deviations by product. Direction and Demographic Break Significance Level Carryall Bag Wife does not work .05 Married .05 Family income $10,000 + .05 Women's Dress White collar occupation -.05 Single -.01 Age 20—34 -.05 Women's Shorts Education includes some college -.05 Family income $10,000 + -.02 Laundry Cart Purchased for someone else in immediate family -.05 Reversible Broiler Education includes some college .02 Cotton Blend Duster Purchased for someone else in immediate family .02 Do not plan to purchase the item within next six months .05 White collar occupation .05 Wife employed full time .02 Age 35-49 .05 Cleaning Fluid Wife employed full time -.02 Age 35-49 .05 Age 50 + -.05 Men's Knit Shirt Wife employed full time .02 Age 50 + -.05 158 at a lesser value at the Odd price. In the first case the odd price functioned as a sales stimulus and in the latter as a deterrent. On the other hand we find that the blend duster accounted for five Of the significant deviations—-all in favor of the odd price. We might also note that the groups revealing significant illusion susceptibility in the case Of this item appear to be heterogeneous, having few common characteristics among them. If anything, they seem to be groups not strongly involved with the product--but, we also found that few subjects were closely involved with the laundry cart, yet it reveals but one significant deviation in the Opposite direction. Nevertheless, the duster has prO— duced five instances of significant deviation, and in each case the item appears to represent a significantly greater value at the Odd price. As we assess the nature Of the illusion with re- gard to the blend duster more specifically we find that in the first four cases the significant deviation is more the result of substantially less value being attributed to the item at the even price rather than a greater value being attributed to it at the odd. In the last case the deviation is a combined product of greater value being attributed to the item at the odd price and lesser value at the even. As a result, we might conclude that in this 159 case, rather than odd prices functioning as a sales stimulus, even prices operated as a deterrent. In passing we might also note that the subjects felt that the duster was by far the worst value of the eleven items in the study. The mean value estimate for the duster was 4.16 as compared with 4.60 for the casual dress, which was the next lowest item, and 5.74 for the FEE" reversible broiler which was the highest rated item in 3 the group (refer to Table 12). Furthermore, we find ( : that the difference between the mean quality estimate (which was made before exposure to the price variable) and the mean value estimate was greater for the duster than for any other item in the study. The implication is that the duster was also perceived as being the study's most overpriced item. We might also note that the two fashion items in the study, the women's shorts and casual dress, produced five groups who appeared to be susceptible to price illu- sion. Interestingly, the direction of the illusion was inverse (i.e., the items were seen as representing significantly greater values at the even prices) in each case. Furthermore, it seems that the subjects compris- ing each of the five demographic cells which accounted for the significant illusions would be considered prime prospects for the items at the price. 160 omo.a oem.m omo.m omm.e pom ewe one oweo .HH ems. :em.m soo.m eme.m nofiaonm eenae oz= .OH moo. omm.m :ma.m omm.m obwrm pace m.eoz .0 arm. Hem.m meo.m sem.m pagan meaemofio .m mmo. omm.e eHo.m amm.e ooe page coopoo .e om:.H Hmm.m meo.m Hoa.e noonso eeoam cooooo .0 mos. mom.o mme.m mee.m noflaobm oaofinno>om .m ems. moe.m mmfi.m mmm.: obmo sneezed .: mmo.H omo.m omm.m moo.s nonoem n.eoEo3 .m mmH.H mee.m mmm.m 500.: among Hmanoo .m Hem. :mm.m eme.m mmm.m mom Hamsnbmo .H .oe.t.eeae .NMWMMMW emwwmwmwm mammfiwme e..H .oz awe: cmoz .mopmefipmo ozam> one mafiamzo came EopHan.mH mqmuld definitely be interested in looking at this item. I 1~c>uld consider this item as a purchase possibility. I rni.ght be interested in this item as a purchase possibility. I ubt that I would consider this item as a purchase 1 DOS 8 ibility . I VVC>uld not consider this item as a purchase possibility. I ‘VCDUIO never consider this item as a purchase possibility. 7° Anfs'ésuming that a customer was to purchase this illustrated ‘team, indicate with a check what you believe are the most lJilcely circumstances surrounding the item's purchase, or 'tliee probable feelings about the item. Unnecessary purchase: : : : : : : :Urgent purchase IPIFICG no object: : : : : : : :Price buyer Useful: : : : : : : :Pleasurable IIt doesn't pay Comparison to compare: : : : : : : :shop Planned purchase: : : : : : : :Impulsive purchase Durable item: : : : : : : :Frequent replacement Practical: : : : : : : :Stylish Recommended Own Opinion by friends: : : : : : : :most important Seen by few: : : : : : : :Seen by many cachllated purchase: : : : : : : :Habitual purchase O\ 195 l 2 3 CLASSIFICATION DATA lyre you: 7. For statistical purposes, Male 5 please indicate your total -——- family income for the past Female year. Arms you: Under $4,000 18 Single ———_—— ——— $4,000—5,999 Married ——— $6’OOO-7’999 Divorced ———-$8’OOO—9’999 Widowed __— $10,000-14,999 $15,000-or more Your age: 9 8. The current occupation of under 20 the head of the household. 20-34 20 _____ 35-49 50—64 9. Is the female head of the 65-over household currently employed? Please cir01e the Yes highest grade in school ———- 22 llett you have completed. No ll 12 10. If yes, does she: Work art-time GIFEide School 5 6 7 8 "" p Work full-time 18 High School 1 2 3 u -— r11Annber of dependent 3111.1dren in your family. 14 NOne].231156+ InCiicate below the age of he youngest dependent child. No dependent children now living at home. —____ Under six. 16 -~1_ Six to twelve. -—__ Thirteen to eighteen. ——~__ Over eighteen. \ 196 STAGE I STUDY MATERIALS FIGURE 4 FICTURES AND SKETCHES USED IN STAGE I STUDY Women's Cotton Knit Top Chip and Dip Set Women's Shorts Cleaning Fluid 197 FIGURE 4e—CONTINUED Menfs Knit Shirt Carryall Bag 198 FIGURE 4——CONTINUED Reversible Broiler Laundry Cart . . . Vol 7 r . ($228!? ., .SPWVFHE 1..., fir? '1 Broiler "No Turn" a. Pass OLIt ques "3 i Onnaires STAGE I STUDY MATERIALS STAGE I STUDY SCRIPT Good afternoon (evening) ladies (gentlemen). First off I would like to thank you for your participation in this study. Hopefully the results will be of some value in providing information about the reactions of customers like yourselves to selected products. The general procedure that we will follow for this study is that I will first .5 pass out a pencil and questionnaire to each of you. Please do not open the questionnaire, but wait until we read together the instructions included in the first three pages. This should help to answer some of the questions that you may have. Don't be overwhelmed by the question- naire's size. In spite of the number of pages, you will find that completing the study will only take a little over 30 minutes of your time. If you will, follow along with me as I read aloud the instructions included in these first three pages. 199 ...... '1 Read iudstructions PPOJ e c t AM clock- I"adio 200 Assuming that you would find the final results of the study to be of some interest, within the next few months, after the results have been analyzed, I will be happy to make a copy of the general findings available to you. 57” Unless there are any problems we will 1 now begin with the first item. As we move i into the study do not read ahead, complete 1 each page before moving on to the next, ;~_ and stay, if you will, within the item section which is currently being presented. As each item is being projected, please be sure that the same item is also typed at the top of the page of each question~ naire item section. As you will notice, this first item is an AM Westinghouse clock-radio. This compact unit incorporates a five tube radio, including a four inch speaker and a built-in antenna, into a sleek upright Space-saving plastic cabinet. The item has an easy—to—read contrast clock face, and a single button buzzar alarm adjustment ital-J Read aloud questions 1-7 PPOJeCt items two, three. . .etc. 201 The exterior is available in soft shades of beige, blue or green to blend with any room decor. Now follow along with me as we complete the first series of questions that relate to this item. This concludes the instructions. Unless there are any questions we will now begin the actual study. You will find as , 1 you proceed with the study that the 1; questions asked of each of the following items will be the same as those which we have just discussed. Complete, if you will, the two pages relating to the projected item. Cotton Knit Top @ $3.98/4.00 This is a 100% cotton knit sportswear top. It is a basic favorite that can be used to makeup a wide range of casual wardrobe combinations. The top comes in solid colors, is sleeveless, with a popular boat neckline. 202 Chip and Dip Set @ $9.98, 7.98/8.98, 7.0L 8.00, 9.00 This chip and dip set is hand crafted. Although light in weight, it is unbreakable and can be safely put into a diswasher. The item is stain resistant, and is unaffected by alcohol or citrus jucies. “a” The new, bright daisy pattern is molded into the set to provide long lasting color. As you see, the set also comes with a pair -.._-..-.._.._1_A 1--.. 2.1, of salad tongs for converting the set to a broad range of additionsl uses. Women's Shorts @ $9.98/10.00 These Tara tweed shorts are by Lomtogs. They are man-tailored with side pockets and pleat detail. They are available in a nubby, textured fabric with matching belt. They have been carefully tailored for a smooth and lasting fit. Ever Blum Cleaning_F1uid @ $.98/1.00 This is an all purpose cleaning fluid that can be used on clothes, furni— ture, rugs, and drapes as well as other delicate nonwashable fabrics. This broad guage mixture offers effective non—ring 203 cleaning for most common spots and stains. This item is an established favorite used extensively by professional dry cleaners. It is available in pint sizes. Cotton Blend Duster @ $11.98/l2.00 This duster comes in a Kodel and cotton, carefree fabric. It launders easily, and requires little or no ironing. It has a snap front closing and sleeves, collar, and front trimmed in matching lace. This robe is by Loungees and is available in a range of soft pastel shades. Casual Dress @ $17.98/18.00 This basic white casual cotton, waffle-pique dress is styled and manufac- tured by Serbin. It is sleeveless, with a popular A-line silhouette. The pockets and neckline are stich detailed and edged with cording. Moc-neck Men's Knit Sport Shirt 0 $7.9878.00 This is a Ban-lon knit nock turtle collar men's knit pullover manufactured by 204 by Trend. Made of 100% Antron Nylon, this shirt has fully fashioned ragland sleeves for greater wear, freedom, and comfort. It is available in a range of long-lasting, non-fade, lustrous colors. Go-Go Bag @ $2.98/3.00 This item is an all-around carry— all bag in easy—to—clean, long lasting vinyl. The dimensions are: 12x15x3 inches. The bag has a three sided zipper closure for wide opening convenience. The bag is available in many colors and should prove to be ideal for carrying books, papers, or for many other possible USES. Laundry Cart @ $5.98/6.00 This laundry cart has four large, clear, lucite casters which allow heavy wet loads to be moved easily. It is built with a strong tubular frame and the hamper is of damproof, heavy vinyl linen. For additional convenience, the unit also has a Clothespin pocket. The cart is available in a bright "snowflake" pattern. APPENDIX C THE AGGREGATE STAGE RESULTS OF SELECTED I RESPONSES 4205 206 sss. e::.s mse.m ess mee.s ees.m mes oneroteo smeosoem Iommcopzd UmpmHSOHmo see. 1 eme.s mse.m sms mem.s mee.m mes see: so coomnzoc so comm eee. eme.s emm.m ess sme.s mmm.m ses oceobooss onoe dosesdo czolmocoHsm so ooocoEEooom mes. . eme.m mes.m ems ems.s see.m ees ensssomusmosoomcd Nee. mee.s ees.m ess mee.s ssm.m ees oeoEoomsdob pcosoosmIEOOH osompso eee. . mem.m msm.m ems mme.s eem.m mes nonmrotao o>smssdeslommmossd ooccmsm ees.sn sms.s mee.m ems ess.s sse.m ees dorm consmeEoo IOLmOSOO on son p.Cmooo OH mes. sem.s see.m sms mme.s ems.m ees osomtammosdusseono eee.s- mee.s mmm.m ems mme.s mee.e mes posse obsceneoosoo oe bosbi ess.s ems.s eem.m ess sem.s eee.m mes masseuse ucowmolommsopsm spmmwooocco mesmom smspCOLOMMsc osocmEom eee. sse.s emm.m mes emh.s emm.m ees sosnebdobe messem ese.s esm.m eem.e ems ees.m mee.e ess oomssomm oasm> .>oo .>mo osomm n a,bm coo: z o.om some 2 cosemoao a: :e :77 -7 .llbbll ecosoeootod ee.me ace me.me be eoesse wee ssesneme so posooed new .mCOHBmosUCH spsmcmaopa wasssp .mopMEspmo ossm>nn.:s mdfldge 207 mms.s 0mm.s wmm.m :os 3mm.s mmw.m mss ommrossa swapsom: nommcosso oOpMHSOswo mms.sl ©m©.s msm.m cos smw.s mmo.m mss scmE so commuzoo so comm mmm. ess.s eee.m mes mmm.s sme.m ess oeeoonEs onos coscseo czoumocosso so ooocoEEooom mmo. smm.s sss.e wos moo.m sms.e mss cmssspmnsmospomsm mmz. sss.s ooz.m mos :mm.s mom.m wss pcoEoomsQop pcosoosmIEmps osbmmso smm. mms.m ssm.m :os woo.m mzo.e mss sommcossd O>smssoEHlomwcoszo coccmsm wee. smw.s mmm.m mos msw.s swm.m mss dorm cowsmmdeoo lopmoeoo Op smd p.:mooo as man. I mmo.m smo.: mos msm.s mmm.m sss OsnmLSmmosmlszoomo mmm. 1 mms.m omw.e mos oms.m omm.: oss posse mossmlooonoo on women :sm. I som.s eds.m mos 0mm.s m:s.m ess ommconsm pcomsouommcossm ssmmmooocco mosmom sespcosoosso ospcmeom msm. I mmm.s mmm.m mos mom.s mms.m sms susmccdosm wesssm mem.su msm.m cms.m mos mem.m mms.z mms mmesOmm onsm> .>oo .>oo ospmm m o.om com: 2 o.om com: 7 COHOmoso eeee see IF .oc.1sm\ a. O mco goo - o m ssm on noose motto smsmmo so so Loo posconm rem .mCOspmosmCs sosmCOOosd mCHsso .moomEspmo ossm>||.mfi mumqs 208 mCOsoQooLoa emm. ssm.s mmm.m sss sme.s eme.m mes onmoobeo smoosomr nommooszo ooomssOsmo mom. mmm.s mms.m ess mme.s mem.m mes some so ooomnzoc so comm 3mm. 1 oms.s sw@.m ems mmm.s :sm.m mos osmoLOQEs omoE oososqo osoumocomsw so ooocoEEooom sow. I oso.m mms.: ass mwo.m mom.: mos omsssomnsmosoomsm wsw. mmw.s ssm.m wss wmo.s swo.m :os ocoEoomsQoo ocoooosmIEoos osomsso ese. 1 ese.m emm.m sms eee.m mem.m ses bonmooneo o>smssoEslommoop2Q cooomsm eem. mee.s ssm.m ems emm.s mme.m ees doom constmoeoe loomQEoo Op sad o.Cmmoo pH mmm.s1 moo.m osm.m sms oso.m sm:.m mos osowosmwOHLIHBMOm: mee. sms.m mme.e mss ses.m mss.e mes boseo oosbduooosoo o: moses sos. 1 mmm.s smm.m mss mom.s oom.m mos ommooosm ocowoolommoopsm summmooocco mOsm m smsocosoomHo osocmEom sms. I m:m.s ems.m mms mee.s see.m ees sosncooobm messam 0mm. 1 m::.m sss.m mms smo.m ooo.m mos oomEsomm onsm> .>oo .>oo Osomm N o.om zoos z o.om one: 2 COHomoso co>m one {it}! 11-410.111.11 1.1 1!... 91191 1.1:» II“ .oo.osm com mm.mm om UOOHLQ momoon m.:oEOZ MO oosoooo pom .mCOsomOHUCH sosmCOQOLQ wessso .mOomEsomo osam>ll.wd mqmqs 209 mes. . ems.s eme.m ess eem.s ese.m ees ommoocoo seaosoec nommoosso ooomHSOsmo mms.s see.m mes.m mss mee.m mmm.m mes scmz so coomnzoc so comm eem. . mse.s ese.m sss sme.s ssm.m mes oceocoocs oboe coscsoo czolmocosom so ooocoEEOoom mse. I mmm.s ems.s ems mme.s ssm.s ses onsssomnsmosoooci eme. . ems.s mme.m mss ems.s ese.m mes ococoomsooc ocosoomeEoos osowoso emm. . eme.s ses.m ess sse.s ese.m mes homecocco o>smssQEHlommoosBQ ooccmsm sms. 1 eme.s ese.m sss mse.s mmm.m mes coon conscecsoo noomQEoo on sea o.Cmooo us mse.s- smm.s mmm.s ems eem.s eee.s mes osomcsnmosmusccomo mes. mes.s ssd.m mss eee.s sem.m ses coseo oosccuooosoo oc bosom mse. eme.s esm.m mss ess.s eem.m mes onmcoccm ocowsonommoossm ssmmmooocco mesmom smsocoLOQMso osocmEom smm. mee.s mee.m mms mms.s mme.e ses sosncococd mcssom eem. mes.m ess.m mms ems.m mmm.m ess ooecsomm oases .>oo .>oo osoem n e.om coo: z o.om coo: s cosomoae cosm ope ill," a If. I." ‘flllu .I' 8'1} ill 0.! .07-0 0. '1... O ' 'lll'llal ll .- . fl, lil‘fl' .1 I i lul“. .oo.mm ocm mm.mw om ooospo oomo shoczms so wCOHoQOoL . l on coorOLQ com .mCOHomosocH sosmCOQOLQ messso .mooMEHomo osfim>ll.sa mqmsmssQEHnommoopso ooocmsm mes.s- mmm.s eme.m ess eee.s eem.m ses coon conscmocoe IOLMQEOO oo smo o.:mooo oH mam. 2mm.s mmo.m mms osz.s mos.m mos osomoSmmosmlssmomo oom.sn mos.s mmz.m sss msw.s m:s.m mos Losso mosoEIoOOnoo o: bosom mmm.s ssm.s ses.m mss mm©.s mmm.m mos omwoonsm ocowoolommooozm ssmmmooocoo mesmom scsocoooooso osocmEom mse. esm.s emm.m mms mem.s mms.m mes sosncodocm mcssem mmm.s mmm.m omm.m mms omm.s mmm.m mos oomEsomm ossm> .>oo .>oo Osomm N o.om one: 2 o.om coo: z cosomoso co>m who EEK 5.8111111.“ 11.19-...” ”H.811.” Us 1" mCOsoooo oo.omm not mm.msm om ooosod obssoso ososmso>os so new .mCOHomosocH sosmcoooso messzo .mOomEsomo 63Hm>ll.mm mqqu 211 ha...- eme. I :em.s mme.m mss mmm.s mme.m mes omecocco seeosooz zommooo2d oooms3osmo mmm. eme.s mee.m ess mme.s mes.m ess sco: so coomnzoc so coom emm. . mmm.s mme.m sss eme.s ems.m ses ocoocooEs oooc coscsoo ozonmocossm so oooooEEooom esm. eee.s mem.m mss eme.s esm.m ees cosssomusoosoomcd mem. . smm.s ssm.m ess see.s mee.m ses ococooosooc ocosooomIEoos osomoso sem. eee.s mme.m mss ems.m mmm.m mes coomcoceo o>sms2dEHnommooosa ooccmsm ses. emm.s eme.m ess esm.s sss.m ses coco conscoQEoo noomoEoo oo sad o.:mooo us smm. : sss.s mem.m ess eme.s sem.m mes osooceooosousccooo esm.s eme.m see.e ess eee.s emm.m ses coseo ooscmaooosoo oc oosco mem. ess.s see.m mss mee.s mem.m mes oomcocco powwoouommooosm spmmmooocco mosmom smsocosoooso osocmEom ems.s- eee.s mmm.e ess ese.s ese.m sss sosocoooco messem ems.s mmm.m eem.e ens eme.s esm.e sss ooecsoom oases .>oo .>oo osoem n o.om coo: z o.om coo: z cosoooeo co>m woo I !.|Il1. l 1"" .1. 1111.1 . .oo.msm mom mm.ssm ow ooosoq Loomso ccoso COoooo mo mCOsooooooq oosoooo com .mCOsomosocH sosmooooso mossso .moomEsomo mosm>ln.ms mommfi mo. pw acm0fl@ficmflmm mHH.H How.H mom.m @HH mmm.a mmm.m moa mmmsopsa Hmzpfipm: Immmnopza Umuwazoamo 5:5.H- ”Hm.fi mmfi.® omfi mom.a mom.m wed mam: an cmmmnzmm hp cmmm maom.mn :m:.a ~mo.w omH mom.a om©.m mofi pcmppoaefi pmoE coflcfiqo czoumccwfipm mp UmocmEEoomm omm. moo.m Hmfi.: mHH mmo.m mom.: moa nmfifizpmnfimOHpompm owe. omw.a mma.: mHH moo.m mom.: 50H acmEmomHQmp unmadmpmnsmpfi manmpso gum. I mHo.m m:m.m mHH Hzo.m m:m.m mofi pmmmzogsa m>HmH5QEHnmmmcopsa Umccmam 0mm. Hmw.fi m::.m cmH mmw.H mum.m moa mocm comfimeEoo umgmasoo on mm: p.2mmoc pH Hmm. mom.a mmm.m Hmfi n:w.a ~H2.m mod mHQmLSmmmHmuasmmmD own. I mmw.fi 00H.m mHH mom.a 000.: MOH Amman mofipmuuomhpo oc mofipm mmm.an mm:.a mmz.m mHH mmm.a on.m :oH wmmnopsm pcmwpblmmm£opzm mpmmmmomCCD mmflmom Hmfipcmgmgmflo oapcmEmm memo.m H:N.H mam.m :mH ozm.fi oom.m OHH hpfimcmaopm wcfizsm mwo. :mo.a wmm.m :ma w:o.m m:m.m QHH mpmefiumm mzam> .>ma .>ma ofipmm N n.pm cam: z n.pm saw: 7 cofiummsa cm>u v60 'l‘l" Illl’lllll‘llll'l'l I'. .00.:m wcm mm.mm pm mmofipa Q0» pficx coupoo mo mcoHmeoLmQ postopa rcm .mcoHQMOHUcH zpfimcmaopq mcfizzn .mmpmefiumm msfim>ul.om mqm<9 213 .Hm>mH mo. pm pcmoflmficmflmm mHH. mmo.m :ma.m mm :mo.m wwo.m mo mmmcopsa HmSpfinm: Immmzopsa Umpmazoamo :03. omm.a mmm.m om mmo.m mmm.m a» N5m: an cmmmuzww an cmmm mmmm.mn m::.m Hmm.m om mma.m oafi.m mu weapLOQEH pmoE COficHQo czonmucwfiph mp UmccmEEoomm mmm. mmm. HH:.H om wmm. mmm.a an :mfiazpmsHmOHpompm me.Hn wom.m omm.: om mwm.m mm:.: mm pcmsmomfiqmp pcmzvmmeEmpfi manmpso H:H.Hu mm~.H :sm.m om mum.fi m:o.m aw pmmmnopsa m>HmH3QEHImmmcoLSQ nmccmam mmo. mmfi.m mm©.: om Hmo.m mmm.: Hm aocm comfipmaeoo umpquoo on awn p.:mmov pH Ham. moo. mo:.a mm Hem. w::.a :5 manmpsmmmamlasmmms moa.a mom.m wmm.: om mzfl.m mmm.: mm pmmsn mOfipmupomwno 0: moapm mow. wom.a mow.: mm H©>.H mwm.: mm mmmcopzm pcmwpbummmzopsm mammmmomCCD mmflmom Hmfipcmpmwufia ofiucmsmm 0mm. cam.H 5mm.m mo omm.a m5~.m NF zpfimcmaopm wcfizsm 22m. mm~.m mow.w mm :mm.m mam.w pm mmeHumm mafim> .>mQ .>mQ ofipmm N n.pm saw: 2 U.pm cme z cofipmmsa cm>m UUO It‘ll 'I'I'I, .oo.Hm ucm mm.a pm umofiga Ufizah mcficmmfio go mcofipawopma pozcopu 6cm .mcofipmofivcfi mufimcmaopa mcHzSQ .mmumefipmm mzfim>nn.am mqm<9 21“ In L“‘_ ‘ Hmm. moo.m :mfi.m mm :mo.m 5mm.m mm mmmzopsa Hmspfipmz Immmsopsa Uwpmazofiwo mza.fi omm.a mmm.m om mmo.m mam.m mm N3m: mp cmmmnzmm an cmmm mmo.fiu m::.m Hmm.m mm omm.m mom.m Hm pcmupoaefi umoE coficfiqo czolmocmfipm an chCmEEoomm 0mm. mmm. HH:.H om owH.H :mm.a Ho cmfiazmeHwOHpompm m:H.H- mcm.m mmm.z om 30H.m 503.: cm pamEmomHQmp ucmsvmmeEmpH mfinmpzo mwm. mmw.fi 33m.m om :mo.m 0mm.m Ho pmmmcogsa m>HmH5QEHImmmcopSQ Umccmam woo. mwfi.m mmm.z om :mm.m Nom.: om aonm cemfimeEoo ImmeEoo 0p zma p.:mmoc pH :mo. mom. mo:.H mm mHo.H mmm.a Hm mHQmLSmmmHmnasmmm: ems. oom.m mmm.: om @©H.m OOH.: om pmmzp mofipmapomwpo o: moapm :mH. mom.a mom.: mm mmm.a omw.: ow mmmcopsm pcmwgznmmmcopsm mpmmmmomcc: mmamom Hwfipcmpmgmfio oapcwEmm mmm. ofim.fl “mo.m mm m:m.a mom.m mo zpfimcmaopm mcfimsm mm». mma.m ocw.m mm Hmm.m mmm.© mm mmeHpmm mafim> .>mQ .>ma ofiumm N U.pm cmmz z v.pm 2mm: 2 soapmmza cm>m @wc .oo.am new mm.e pm omOHpa saga“ mcficmmfio mo mcofiuamogma pozcoaa 6cm nmcofimeHUCH mpfimcmaopq wcH>39 .mmmeHpmm mSHm>II.mm mqm.H 5mm.m QHH mmmcopsa HmSpfipmm Immmzopsa Umpmazoamo H:O.H- Hmm.fi ozfi.m 50H mom.H omm.m :HH mcmz >9 :mmmnzmg an cmmm 0mm. mmm.a mom.m 50H wow.H omw.m oHH pcmppomefi uon coacfiao czolmncmfipm mp UmccmEEoomm pom. omo.m wmo.: moa mmH.m Hmm.: NHH :mfiazmeHmOHpowpm mam. u omw.a wma.m on waw.a moo.m mHH pcmEmomHQmp pcmsmemIEmuH mapmpza mom. I oom.a mam.m coa :oo.m zom.m mHH pmmmcopsa m>HmH3QEHImmm£opza nonsmHm wow. I mmm.a :ow.m moa nww.a :m:.m mHH Q0£m COmHLwQEoo umpmano on man u.cmmov pH mmm. :oo.m wmm.m moa swo.m omm.m mHH mHQMLSmmmHmlasmmmD mflm. mow.a omm.: moa :mm.a mow.z maa Lm>59 moapmluomnoo oc mOfipm wmm.a mom.H me.m Boa mm:.H Haw.m HHH mmmnopsm pcmwuanmwmnopzm zpmmmmomcca mmamom Hmfiucmpmmuflo oapcwemm mmfi. ch.fl m:a.m OHH mo:.a mwa.m omH apfimcmgopm mcfihsm asp. . mma.m mm©.m HHH ”mo.m m~:.m omfi mpmefipmm mzam> .>mo .>wo ofipmm N n.pm 2mm: 2 n.pm cam: : cofipmmza cm>m was ll. .oo.mm new mm.~m pm cmOfipQ ppficm aficx m.cme no mcoflpamopma posvopa Ucm .mcofiumOHUcfi muwmcmqoua wcfizzn .mmpmeflpmm msfim>ln.mm mqmm Hm>m H No. pm chOHNNCNHmQ H mo. pm NCNONNHCNHNN m:o.a mom.H mmo.m Fm 2mm.H wwm.m mma mmmcopsm Hmzpflpwz Immm50p: UmpmHSOHmo pomm.mn mmm.H mom.m mm mmw.a mom.m :ma mew: an :mmmnzwm K3 :mmm mom. . mmo.m omN.: ooa 0H0.N om©.: QNH pcmppoaefi pmoe coflcfiao czoumwcmfipm an UmvcmEEOOmm HHO.H OH:.H omo.m OOH :Nm.a 32:.m mma cmHH>pmnHm0Hpompm memo.a mHH.H :mN.H Fm mme.H H©H.m :ma pamEmomaqmp pcmzmemIEmpH mapmpso mmm. om>.H oom.m mm mom.a mmm.m mma pmmmcopsa m>fimH5QEHammmcopsa omccmam Nam. Hm©.H mmw.m mm 0mm.H NQN.m mmfi aosm cemfipmaeoo ImLmQEoo 0p zma p.2mmoo pH new. :mm.a mmm.m HQH mmm.H mH:.m mma manmLSNNmHmnasmmmD 0mm. ONm.H NNH.m mm mmN.H HHN.m mNH Lamas mafiamuuomnno 0: mofipm can. mwm.a moH.m moa :m©.H owm.m mmH mmmzopsm pcmwpbummmcopsm zgmmmmomccb mmamom Hmfipcmpmmmfio QNpcmEmm mmN.H Nflm.fi mo:.m moH :mm.a :NN.m NNH Npfimcmaopm wcfizsm mmfi. I moo.m Nwm.m moa wom.a Nmm.m wmfi mpmefiumm msHm> .>ma .>mo ofipmm N n.pm saw: 2 ©.pm :mm: : cofiummza cm>m Ugo HEX? q 0 . La pmaaopp :cpse 07:.m :1. m m :N N mcofipamo Lma posuopg new 217 Hmm. Hom.fi mwn.m mm mm:.H :mm.m mm mmmsopzm Hmzpfinmz Immmnopsa UmpmH20Hmo :mo.fiu Nm©.H mo:.m Nm mmN.H mNm.: 0: New: Np cmmmusz Np cmmm ONm. HmN.H mNm.m Nm mmq.fi NON.m H: pcmppogefl pmoe coficfiao czolmncmapm mp UmUCmEEoomm mac. NH@.H :mm.m mm mmm.a Nam.m Hz nmfiazmeHNOHuompm Han. mmm.a Nmfi.m om mom.H Nam.m H: pszmomHQmp pCmSUmLmIEmpH mfiompso :Nm.H- HmN.N wmo.z om mom.a mNm.m o: pmmwnopza m>HmH5QEHImmm£oL5Q omccmam Hmm. mm:.a mmm.m om H@N.H mwm.m o: mocm comfipmasoo smLmQEoo 0p mma p.2mmov pH own. mwm.fi Nmm.m mm New.fi ooo.m H: mHQNLSmmmHmnasmme zoo. NHH.N mzm.: mm mmn.a mwm.: o: pm>39 moapmnuomhno 0: mOfipm mm:.a oo:.H mo:.m Nm mww.fi 5mm.m H: mmmcopsm pcmwppummmcopzm mpwmmmomccs mmamom Hmfipcmpmpgflm oapcwEmm HmN. :mm.a mam.m mm OHM.H mom.m N: Npflmcmaopm Ncfizsm mum. mmm.a mam.z mm :mo.a OHH.m m: mpmefipmm msam> .>mQ .>mQ :Ofiummsa N Emmi 7 0 mm N n.pm :mmz 7 n.pm Hp 11‘ cm>m use it!" .\\i\!‘zfi‘\l\u 1fiwnfiunnuv HH‘ Q o no Il.m mqm m .OO.Nm TCQ 00 am HUQOfiUCfi %UH 218 .Hm>mH No. pm NCNOHNflcwfimm mCOHme 3mm. mm:.a m:m.m mm mam.a :mm.m mm mmmzogsg Hmzpfipmm ummmzonsa UmpmHSono Nmm.fin moH.H mom.m mm mom.H oom.m mm New: Np cmmmuzmm Np :mmm HNm. . Nmm.H :Nz.m mm moo.N mom.m :m pamppoqefi pmos coficfiao czolmucmfipm an omncmEEoomm :m:. I HNm.H mom.m mm mom.H wmo.m mm cmflaxpmnHNONNUNNN mam. omN.H wmo.m mm mmm.a m:H.N mm pcmEmomHamp pcmzvmmeEmpH manmpsa mwz. OOH.N NH:.m mm NH:.N www.m mm pmmmgopsa m>HmH5QEHImmmcoL5Q @mccmam moo. . NNo.H omm.m mm Nzo.m mHN.m mm rom comfimeEOQ ImmeEoo on zmq p.2mmoc pH mom. :mfi.m Ham.m Nm CNN.N Nmm.m mm manmpsmmmamuasmmm: mam. ONH.N :mN.: Nm moo.m mNN.m mm amass mofipmnpomfino 0: mafipm Nam. u NH©.H mom.m em mNN.H mam.m mm mmmcopsm ucmmpblmmmcopzm mummmmomcc: mmamom HmfipcwpmMMHQ Ufipcmemm oHH. ; NNm.H NNm.m mm mNm.H mmm.m mm Npfimcmaogm wcfizsm mmmm.m Hmm.H oa:.z mm moo.N mm:.m mm mpmefipmm msam> .>mo .>mo prmm N U.pm saw: 2 n.pm cmmz z COHummSG cm>m woo “““H“H“““H“H““Hl‘ii!‘\io\l -1 &Q pmm QHU UCM Qfifio MW QEHumw ®3H®>II.WN Mdm<9 . m mm.mw um Umofi % mcmaoua mcfi>zp mm» .00 mm am new .mcoHNNOanfi pH opmu pustop 219 mwm.fl mm:.H mmm.m @m :o:.a mmm.m mm mmmcopsm Hmspflpm: Immwnops UmpmHSOHwo mNo.HI mmm.a omm.m om NHN.H mmw.: mm New: Np cmmmIsz Np cmmm mmm. I mom.H me.m mm mm:.H HNm.m mm pamppoqefi pmoe coficfiao m m m m m m m m czolmucmfipm hm cmccmEEoomm : . I ow.H NN . w mNN.H HNm. m :mfifi meHmoHpompm czfi. I mm:.a mmm.m cm mam.fl mmN.N mm a newsmomaamp pcm: mmeEmpfi manmpso mmm. I wNm.H oom.m mm mmm.H Nmm.m 2m pmm¢gopsa m>fimH3QEHImmmcopsa Umccmam 0mm. NNN.H omm.m om Hmm.fi mzm.m mm a aocm momflpmasoo Impm E00 0» m p.:wmov pH Ham. mmm.a ooo.m om mmm.H mzfi.m mm mflnmpsmmmHmIfismmma :NH. I N:®.H mmo.m mm Nmm.H mmo.m 3m pmzsn mOHmepomnpo 0: mafipm Nmm.H mmm.H oom.m mm mmm.a HNm.N mm amazopsm pcmwpalmwmcopsm mpmmmwomccz mmamom Hmfiucmpmmufim oapcmEmm Hmo. I m:m.a mNN.m mm omm.H omN.m mm Npfimcmaopm Ncflzzm can. mmm.H mmo.m mm mmm.m omm.m mm mumsfipmm msfim> .>ma .>mm a . . . COHpmm: o m. N. P5 :32 7 .niLm :8: r E m I cm; 30 \II‘ Eilh‘ CG QHLO .Hnau |\.N.N mqmdwrb nW|\n“Hv\\ . Ucm mm.mm pm chHmw mmmwmwmomq mcfixzn mmumEHumm mSHQ> . . A. H CO @eQLQ UCN amCOHme.© 220 moo.fi mm:.H mmm.m mm How.a wNN.N mm mmwcopsm Hmspfipmm Immmcops omumHSOHmo mos. mmm.a omm.m mm Nmo.fi mo:.m Nm New: mp cmmmIsz Np cmmm :mm.HI moN.H me.m wm HmN.H mNm.m Nm pcmppoaefi wmoe coficfiao czonmocmfipm n UmucmEEoomm mom.HI mom.H NNm.m mm NHN.H :Nm.m Nm nmfiazmeHmofiuompm Non. I mm:.H mmm.m mm mmm.H Nmfi.m wm w acmEmomHng pcm: mLmIEpr manmpso mmo.a mNm.H oom.m mm HmN.N mmo.: mm pmmmnopsa m>HmH3QeHlmmmnop3Q owccmam wa. NNN.H omm.m om NN:.H mmm.m mm a aocm momfipmqeoo Imam 500 cu m u.:mmoc pH HNm. mmm.a ooo.m mm mNm.H Nmm.m mm mapmpsmmmHmIflsmmm: mom. I N:@.H @mo.m mm NHH.N mzm.z Nm pmzsn mONNNIpomnno 0: mafipm mmm. I mNm.H oom.m cm om:.a mo:.m Nm mmmcogsm NemwpDImmwcopsm mammmmomccz mmamom Hmfipcmpmmufim oapcmEmm . . Q m >. mwo. wzm.H mmN.m wm :mm.a wfiw.m mm NUHMMMEMWMm mmfimm HNQ. I mmw.fi MWN06 om mmm H mam : IIII \.\I§|\\HIII.IIIII||II\\“||I!.|!I“I|I (HO III I mapr nh“““““““““““““““nnuflwm OO. UM U®OHLQ MMMWWWMOWMQWWWMMD ammumEfiUmw ®3HN> mm m4 . o .. CH 00 OVOLQ ECG amCOHpMOHU mcofi pflmemu DDSU 221 mmz. HO:.H mmm.m mm mm:.H HN©.N mm mmmzopsa HmspHnmm Immmcopsa Umumazoamo mam. NHm.H mmm.: mm mmN.H m~m.: on News as :mmmIsz Np cmmm mom. mm:.H HNm.m mm mm:.H NON.m H: pamppoaeH pmoe coHcho czonmwcmfihm >9 UmUQmEEoomm mom. @NN.H HNm.m mm mom.H NHm.m H: :mHHHmeHNOHpomHm mac. mam.H omm.m mm mom.H NHm.N H: pcmsmomHamp quSUmLmIEmpH manmpso mHo. mwm.H Nmm.m Hm oom.H mNm.m o: pmmmcopsa m>HmH5QEHImmmnopsq Umccmam me. Hmm.H mam.m mm H@N.H mNN.m o: aocm cememaeoo Impmaeoo ou mma p.:mmoc pH NNm. mmm.H mHH.m mm Nmm.H ooo.m H: mempsmmmHmIHzmmm: NNH. Nmm.H mmo.m :m NwN.H mNm.: o: pmzsn mOHmepomnno 0: mOHpm NHH. wmm.H HNm.N mm MNm.H Nmm.m H: mmmcopsm pcmwpblmmmcopsm zpmmmmomCCD mmHmom Hmflpcwpmumfle oaucmEmm NHm. omm.H omN.m mm QHm.H mom.m N: HpHmcmaopm wcHNsm mom. @Nm.m omm.m om 3mm.H oHH.m m: mpmermm msHm> .>wo .>mo oHHmm n.pm cum: 2 n.pm cmmz z coHummsa cm>m vac .mm.mm mam mm.mm pm umOHpQ mcofipqmopma poznona rem .mcoHpmoHUCH ll Il'lll'li. “I“ pmm QHU 62m QHLO no xufimcmqopq mcflzsn ammpmefipmm msHm>II.mm mqmHmH3QEHImwmcoLSQ Umccme moo. NNN.H omm.m mm HoN.H mNN.m o: aonm cemHgmaeoo ImeQEoo on man p.£wmou pH III mmm.H ooo.m mm Nm©.H ooo.m H: mHDNHSNNmHmIHzmmmD Hmm. I m:m.H wmo.m mm mm:.H mmo.: o: gmzzn mOHHmIpommno 0: mOHLm mmm.H mmm.H oom.m mm mNm.H 5mm.m H: mmmconsm pcmwpDImmmzopsm mummmmomcca mmHmom HmecmmemHQ QHHCNEmm Hmm. I m:m.H mNN.m om QHm.H mom.m N: HpHmcmgopm NcHN:m mom. mmw.H mmo.m mm :mo.H mHH.m m: mmeHpmm msHm> .>mo .>mo oprm N @.pm 2mm: 2 ©.pm 2mm: 2 coHpmmsa cm>m wvc I'l'll. . l .oo.mw new mm.©m pm UmOHLQ pmm QHU Gem QHmo mo . I mcoHuqmoLmQ posvopa wcm amcoHumoncH mpHmchoua mcH>zn .mmmeHpmm msHm>II om mqmHHEmm mpmHnmEEH CH mmHm mcomEom own. I mom.m H:m.o H: HHm.m mm:.: mm mHHEmu mpHucm mo mm: cH:.H NmH.m m:m.¢ :m Nwo.H m:m.m mm mm: :30 :HsHm wchmmHo m:o. Nmm.m Hmm.: o: mmo.m OHm.: N: :NHN m m: tmaH.m mmC.H cmu.m mm oHH.m H:N.: m mHHEmu mpvawEEH :H mme mcomEom mumwpgcpaaw pcr :HHEwm mLHpcm no mm: v::. :H:.m 1H:.: NC :m:.H mmm.: mm mm: :30 memzo ucmHm cepuoo mHm. mmN.H cHo.m mm :No.a mmm.m H: umHm m m: N:m. mrH.m mmm.m cm mm.m mHC.m om :HHENQ mpwHUmEEH :H mmHm mcomEom : g L mHHEmh mpHucm no mm: mmm. I NN®.H :::.u :: mzH.x mcc.m mm mm: C30 mo: :Hc: ceppoo mc:. mcH.a chp.\ a: cH:.m H:m.w mm “NHN m m: wwm. rm:.m a::.u c mom.H mmH.w HH :HHEmE mpmvaEEH CH mmHm mcomEom ccr.H :mm.m mam.v aw HuC.H mac.\ mm mHHEmu mLHucm no mm: Hy~.H mHH.a :Hm.u um mmu.H :cw.h um mm: :30 LmHHopm mHLHNLw>om m m. I MHm.m c(:.m mm m_c.H :o:.: NH :NHN a m: omxm.mI mN:.H Ho:.p Nu NNQ.H co:.: mm sHHgma mHmHUoEEH :H mmHm mcowEow cum. I mN:.H Hm:.u an :Hm.n cou.u mm :HHEmu mLHpcm 90 mm: :&:.H :NH.m wam.L Nu oxH.n :w\.m :u mm: :3: ppmo :chsmg .Hm. I Nna.m :;..: um N.o.r :ao.: :m pNHN m m: oxr.HI mmH.r cmr.h mm :cc.H NC:.: ow :HHETE awmwuoEEH.cH mmHm. mcomEom L H 2 :HHENQ mLHpcm go mm: CMH. wmm.m Nrm.u :5 :uc.r c:m.m mm mm: :30 mppocm m.:mE©3 on. I m:c.m c:m.: ma :Cm.a ::m.: Hm ppr m m: Cam. I uac.m :H:.: on mmH.m NHm.: 0H >.HHEHC mpmHUmEEH :H mmHm mcomEom 2 g L >HHEmu mLHpcm no mm: mum.HI :m:.m ::m.m m: o:c.u Nmm.: om mm: czo mumpa Hm3mwo mam. mu .N NNH.: H: m:c.m Cum.“ N: HNHN m m: mwm.H mmb.m 0mm.w mm mmc.m mmH.N :m HHHENN mpmHUmEEH :H mme mcowEom m m m mHHEmu mLHpcw mo mms HHH. I mmH.N Hno.N um Nmo.m mNH.N mo wm: :39 as HHngpmo quwm .>$O .>®D N go : :.:N :NIE r .um cmm: : mHom NcHasm Um>Hwome .mpowmnsm IIIIII-III I. cw>m five .conzHHH mOHLq ccm mHOL meOpmso Um>Hmopr .mpoanSN no mHmmHmcmH Ho. pm :cmOHuchHmm .Hm>mH me. p: acmOHNHcmHm: .Ho>¢H mo. pm pcmoHCchHmo .Umpgoqwg mco: 2235 0'“ a .w.anm;m Lou HccHoHLu3u2H mmmcoammp no meezzm HZWH.J. \w Nx.C.n.. Fags...“ NIH Quad m m r m >Hwirg mpmwmeEH :H mnHm mcomEom >::.c CH Nm\.m mH..: CH :HHENL mLHpcm ac om: r w m: CBC cc.cm ncm mm.mm um umm QHQ ocw QHzo ._ .. :NHN m m: r\_ L) I (V f‘ crc.: nu NHr.H an w m >HHEra mpmvaEEH CH mmHm wcomEom c.n.u gH pmg.H acx.: mH >HHEmg oLHucm Lo mmH m m was :3: o¢.mm sum m©.mm pm pom qH: :cm quo .x_.: :; Cni.a :rw.; cm _ uuHm m m< I M nHHqu mumHUeEEH CH ome mcancm I r.C IH amn.H :::.g :H mHHEmu mLHIcm no mu: m w was C30 oo.ww 1cm mc.mm um me QHn cc: QHQU _H:.; H: Hc:.H Nov.; mm ppHm m a: r N usfir; quHUeEEH CH cme mcochw .2n._ n_ :H4.H mg.» L; aHHEmu oLHpcm Lo mm: :I... .a on.H moH.u :: mm: :30 . LmHHoLm :cgsk or: m...: my .Ic.z mmm.u m: pmHm m m: arr.» Hg mHa.H ch.m mu AHHErg meHUmEEH CH mem mcomEom ,H :H zHHEwL mLHpcm we mm: ~:H.m Tm m:H.m ooo.m c: mm: :30 uLHcm pch n.C®E "t' ‘ .u 226 . . . I . . . HHH E: 2: HH :2 :HH HH 3202022320: N@:. I MNN.H HNm.m m: 0mm.H mNm.m Nm mmmzopzq 0» cmHm mmo.H Hmm.H N:c.m mm :HH.N Nmm.: mN mmmcopsa :0: :Ha No:.HI 0:0.H me.m am Hmw.H m:m.: mm posnoga cmmmnopsm mo: :Hc: coupoo ocNH.N mmH.N m:m.m :w NN®.H NHm.: mm mmmgopsa 0» cmHa poc oo mom. I ONN.N N:@.: :m Nwm.m omm.: mm mmmcopsa 0: cmHN N:m.H mNo.N mm:.m mm mmm.H mH:.: ON mmmnopsa poc :Ho m:m. :N:.N mHm.: :m :mm.H omN.: mm posnopa :mmmzopsm Lmuwdo ocmHm coupoo :mm.H QNN.N NmH.w NHH mam.H ::m.w No mmmgopsa 0: cqu won on p ,H p mmmnohza o» :mHm o:m.H HHm.N Nmm.w :HH mmm.H omo.w om mmmnousa :0: :Ho p L p podUOLQ ommmsopzm chHHOLm mflnwmeTwN/mm mom. I m:o.m CNH.m HHH 0:0.N omo.m HQH mmmgopza 0: cmHa :0: oo m m m mmmgopsa o» :mHm NQH. NQH.N :Hm.m :HH :mH.N mmm.m moH wmmgopza No: wHo m m m nonwoaa commcopzm meo zpbcsmq N::. I mHm.N CNI.: NN :o:.H Hmm.: :m ommgogzq 0: :mHQ :o: oo m:m. I Hmm.N CNN.m m: :No.N :mo.m :: mmmzopsa 0: cmHN Hmo.HI mmH.N omc.m mm Nco.H mm .: NN wmmnopsa we: :HQ moo. Hmo.m NN:.m om NNH.N oom.m mm poscopu commcopzm mppocm .wcmEo3 HmN. I GNH.N w::.: wm :HH.N mmH.: HN mmmzopsa 0: :qu :0: oo wa. I m:m.m m:m.m mm NNH.N m:m.m Hm mmmcopza 0: :mHN m5H.HI 5mm.m 055.: mu mmH.m m:m.: $5 mmmcopza no: GHQ Nam. I NC:.N :NN.m mm o:m.m HHm.m m: posvopa :mmmcopzm mmmgo stmwo Nmm.H NNH.N N:N.@ No ooo.m N:N.w mm mmmgopsa ow cmHa :0: oo mwo.H mom.H mom.N NN 0mm.H :mo.m NH mmmcopsa 0: cmHN m:m.H :mm.m m5:.u moH moo.m mmw.w om mmmcopsa no: GHQ mmm. omo.H ::m.: mH mam.m oom.N cm posnoga nmmmgopsm mmm HHmwuLmo oHpmm .>mo .>wQ N Lo p U.um cam: : v.pm cam: 7 mcoHpmpoqum mpzpsm Ucm mocmHLmme ummm I :¢>: cue I . .mcoHumuomaxm mmmzopza mp3psw USN mocprmaxm mcH>sp pmma :cm NNHHHpHpomgmzm consHHH .mpomesm mg: :0 mHmNHmcmH No. pm NCNOHNHcNHmo .Hm>mH mo. pm :CMUHHHcmHmo .mHmszcm pom pcmHoHpuzmcH mmmcoammp no meESZQ .nmppoamp mcozm awe. Nwo.H omo.: mm NNm.N omm.m Hm mmmcouza 0: cqu :0: on n L n mmmcopsa 0p :mHm NoN. owm.H :c:.: mm :mN.N HmN.m mm mmmgopsa no: UH: Q L ; NQSUOLQ Ummmzopsm oo.mw as: wo.mw - pm :mm oHo wcm QHLN om:o.m CNo.H N:N.: mm HNN.H omN.m N mmmcopzq 0: cqu :0: oo 2 L . 2 mmmcopsa 0p cmHm H.HNHN oHo.H mc:.: Nm wHo.N :Nm.m :m mmmzopza :0: :Ho 2 g L podUOLQ Ummmcopzm oo.m: ncm mm.N: pm pmm QHQ ::w cho :mo.H aco.H NN:.: cm :mc.H N:o.: Nm mmmcogzq 0: :mHQ :0: oo L L L mmmzopsa 0p :mHm :TN. :mc.N mm:.: Cm mwc.H coa.m o: mmmsopzq :02 :Ha .E L I: nonwopa vmmmgopsm oo.NH :cm mo.mm pm pmm QHQ ucm QHLD mmH. I :HH.N :NQ.C Nc Noc.H NON.N omH mmmcopza 0: :mHa :oc 0: L L L mmwgopza Op cmHm mNH. I Noc.m N::.m HcH mmo.H c:m.m cHH mmmgopza :0: :Ho 2 L L pQSGOLQ UmmmLULSm LmHHOLm :CLSB 02: mm:. m:N.N mcH.m Nm :Nc.H Ccm.m m: mmmzopza 0: cmHa :o: oo :cH.H :::.H :ca.: :u mcH.N NNN.m m mmmcopsa 0: cmHN NHN. I ocm.m mNN.u mm mHo.N oom.u N: mmmcogzq poc oHo :00. I H:c.N :Hc.: :: CV0.N NMK.m Nm posuopq ommmgopsN . . :pHcm :Hc: m.:mz Ho:.HI cmc.m mcH.w m xmc.m 5Hn.m 5: wmmcopzq o» :wHa pee on c::. :MH.N :ca.N :m Nmm.m oo:.N om mmmgopsa 0: :mHN :Nc.HI c:c.m N:u.w N HON.N ocH.C mm mwmcogzq :0: :HQ :uo. mm:.m omn.o om mm5.m :mm.w 0H uosnopq @mmmcopsm oo.H: Us: mm.: :m :HsHm NchmmHo -o"" Ir .11 .II IqIndlnt III,.II. It... . u b 228 03:8 25 com. I NmH.N omN.o :m Hoo.m mo:.@ m: MHH. moo.N N:N.© Nm SON.N :wm.o mm H0059” :me.:p;EII oo.H: cc: mm.: pm :HsHm mchmmHo mmH. I omo.N Nm:.m Hm NMH.N mm:.m No mNmHHoo Naom HNN. I Hmm.H oHo.m Hm moo.N mNm.: N: Hoogom sz: sun: go: pHc: coupoo Nom.H H:H.N mom.m :0 mHm.H N::.: H: mNmHHoo msom NHH. mmN.N mN:.: mm Hmo.N HNm.: w: Hoosom zNH: can: memsa UCmHm couuoo :mN:.N m::.N HmN.© mo :NN.H mNH.N om mmmHHoo msow NmH.HI omo.N mmN.w mm Nmm.H mmm.m m: Hoonom :NH: 3pc: .IHmHHOLm mflflfimLmN/mm :om. :oo.N ©:m.m mo NHH.N me.m No mNmHHoo «Sow mmN. I NHN.N Hmm.m Nm mom.N NHN.m m: Hoocom :NH: sage ammo 5pvcsmq meo.NI HN:.N No:.m HN NNo.N Nmm.: mm mNmHHoo NEON m:w.H mNm.N :NN.: Hm omm.H Hw:.m Nm Hoocom :NH: sane mphocm m.cmEo3 mmo.HI NmN.N omH.m mm ©0N.N HmN.: No mNmHHOO msom HmN. I :Nm.N N:o.m N: Nom.H Nom.: :m Hoogom :NH: :Nce mmmpo Hmzmmo New. mNm.N w::.w No wNH.N NHN.N ow mNmHHoo Naom oom.H Nmo.N Hm©.© mm MNo.N :NN.N m: Hoogom :NH: ups: me HHmmppmo oHpmm .>mo .>mo N No : n.pm CNN: 2 :.:m cum: 2 Hm>mq choproznm cm>m Ugo .mHm>mH HmcoHpmoswm new mpHHHpHmeQNSN COHmsHHHII.mm mgmOH NO. OO OOOOHOHONHNO N:H.H HNO.N OON.: ON HON.N OO:.O ON ONOHHOO OeOm NNN. I NO:.H NNH.O OH O:N.N OOO.O OH HOOOOO :NH: ONO: OO.NO OOO NN.NO OO OOO OHQ OOO OHOO momm.m mmo.H :Ow.m mm w5w.H mmm.: 5H mmeHoo mEom ONN.H NON.H NHH.O NH :NN.H OOO.O NH HOOOOO :NH: 2pc: oo.NN Ocm NN.5N OO OON OHO Ocm OHOO NHN.H HNN.H NON.: :N HON.H ON:.O HN ONOHHOO OEON N::.H NON.H NNH.: OH OOO.H OON.: ON HOOOOO ONHc spa: OO.NO OOO NN.OO Om OON OHQ OOO OHOO NON. I NNO.H NNN.O NO NON.H ONN.O NO ONOHHOO OEON HmN. mmc.m ::O.m m: H50.H OOo.m mm Hoocom cch sane pmfifi Ohm :CLSB O7: ONN. I N:H.N NO0.0 OO N:H.N OON.O NO ONOHHOO OEOO NOO. I OOH.N HOO.O m: NNN.H OOO.O OO HOOOOO :NH: sane ONHON OHON O.cOz 230 EH8 OOHN . . NH ONO. ONN.N NNN.O OH mom m 5:: m H:N.HI NNO.N NNH.N O: ::O.N NO0.0 0m pwaaoo mpflpg OO.HN OOO NN.N OO OHOHN N:HOOOHO OH0.H m00.H mm0.: 0m NmH.m 0mm.m Hm LNHHoo msHm Om:.HI 0mm.m m:m.m N: 0H0.m 5Hm.: N: pmHHoo much mo: OHON OOOOOO 0:0.HI N50.m mm:.: mm 000.m 000.: mm pmHHoo msHm mmmm.m mOm.m H5:.m Hm 500.H mO:.: m: pmHHoo mpch pmpmza ncmHm coupoo mO5. 550.H 00:.O 0m 505.H m00.0 Hm LmHHoo msHm mmO. Nmm.m 0ON.O 0m 00H.m NOm.O :: pmHHoo mpHc3 LOHHopm mHnHmpm>mm 00m. me.m :mH.m Om m0m.m 0mm.m mm LmHHoo msHm 50m. Hm0.m HHm.m mm mOH.m H:m.m :: pmHHoo mpch ppmo ancsmq m5m.H 0mm.w 0m5.: mm mNN.H 05m.m 0m LmHHoo msHm NMH.HI :H5.m :Om.m mm 005.H mm5.: H: pmHHoo mpan Oppocm m.:m503 OHm. I 5NO.N mm0.: :m 505.H H5:.: 5m pmHHoO msHm w:00.mI 5OH.m 0mm.m O: 0:0.H 0m:.: 0: pmHHoo mpch mmmpm HNSNNO 000.H 0H0.H 00m.O mm 55m.m Nm0.5 Om LmHHoo msHm Now. N0:.m Hmm.O 0: mm0.m 0m0.0 5: pmHHoo mpch wmm HHNNLLNQ OHNNm .>m0 .>m0 N no u O.pm 2mm: 2 ©.um saw: 2 :oHNNOHNHmmmHo HNCOHpm05000 cm>m 000 .Ummc UHonmmsog no COHONOHuHmmmHo HNCOprasooo 02m NpHHHpHuQmONSO COHmzHHHII.:m mqmmH Ho. .Hm>mH mo. OO OOOOHOHONHNO OO OOOOHOHONHNO wo:.H 5mm.H mH5. mwm. :00. 0mm. 0:0. mmm. m5m.m NOO.: oom.m m:m.m 00:.: 000.m :ON.O N:N.O :H5.m omm.m HH NH OH 0H CH mH mm mm Hm mm MKS MCD :r CON Ln OO:.O OO:.O m5m.: :mm.: 0mm.m :5:.m mm5.m 0Hm.m 0m 5m 0m m: pmHHoo msz NOHHOO OOHO3 OO.NO OcO NO.NN OO OON OHO OOO OHOO LmHHOO msHm NOHHOO OOHO3 OO.NO OOO NO.NO OO OON OHO OOO OHOO pmHHoo msHm NOHHOO OOsz OO.NO OOO NN.OO OO NON OHO OOO OHOO pmHHoo msHm OOHHOO OOch NOHHONN zaps: Oz= LNHHoo msHm NOHHOO OOHON OOHON OHON m.cOz 232 Hfiw mxpog aw“? 50H.HI HHO.m 000.0 00 0H0.H :0:.0 0m 0EHP a NNN. I NOH.N NNN.O NO NOO.N HNm.O N: ::o: OOOO 00H: 5:m.H :05.H 005.: 00 000.0 000.0 00 xpoz no: mmon m0H3 mo: :HON OOOOOO 20:5.m 00m.m 005.0 :m 005.H 005.: 0m mEHp HHsm mxpoz mMH3 :0m.H 0H:.m m00.m 00 :05.H H00.: 0: xpoz mmov mmHz 0H0. 550.m 000.: 00 :mH.m 000.: 00 xpoz poc mmow mmHz memsa UCmHm coupou :0m.H 000.0 500.0 00 0:0.m 000.5 :m mEHp HHsm mxpoz OMHB 505.H 00:.m 000.0 H0 000.m 000.0 5: xpoz mmou 00H: 000. I m00.m 00:.0 50 005.H 00:.0 00 xpoz po: 0000 MO0H3 LmHHOLm mHnHmpm>mm 0H0. 000.0 0H0.0 m 0mH.m 000.0 00 mEHp HHsm mxpoz 00H: mm:. 00H.m :0m.0 H0 000.0 00:.0 5: xpoz mmow 00H: 0mm. I 00H.m 00H.0 0- 000.0 0H0.0 :0 xpoz no: mmon mmHE NLNQINLUCSNH 500.HI 000.0 00:.0 mm :55.H 005.: 00 mEHp HH30 mxpoz 00H: N00.HI 000.0 0::.0 00 0HH.m 000.: 0: x003 mwov mmHz 00H.H mom.m 050.: 00 ::o.H :0:.0 m0 xpoz po: 0000 00H: muposm m.cmE03 H00. I 00H.m m:0.0 :m 00H.m 000.: 00 mEHp HHsm mxpoz 00H3 mm0. I 00:.m m5m.0 m: mm0.m 000.: 00 xLo3 mmow mNHB 000. I H0m.m 000.: 50 00m.m 000.: 00 XL03 p0: mmoc mmHz mmmga Hmzmmo 0mH. 0H0.H 000.0 mm 000.H 000.5 00 mEHp HHsm mxpoz m0H3 000. 00H.m 000.0 m0 0:H.m 0:0.0 0: xpoz mmoc mmHz mm::.m 00m.m m5H.0 :0 00H.m 05H.5 00 xpoz po: mmoc OMHB wmm HHmzLLmo .>mQ .>m0 N no u 0.00 :00: z 0.00 :00: z mapmum xgoz 0.00H3 cm>0 000 .UHozmmzoz 00 000: meEmm mcHxLoz 0cm szHHanQmUNSN COHmsHHHII.0m mqm<9 233 .H0>0H HO. 00 pcOOHchNHNO .HO>OH NO. ON OOOOH0H00H0O .HO>OH OO. 00 OOOOHOHCNHNO G F- r—4F‘4r—4 L?“ (“O CI.“ 0: \N‘ C‘J O\ m [\_ ,_I m C rm (‘0 (‘0 (‘O '00 IO .'r—4 (\JCNJCO :H .CI) Our—4 mH mm D 0". C“. (\J r-4 0 MCI: Lf‘ LI" (‘1 Lf‘I C‘d LO» II (\J ‘3' O 0 CL (\J C\ C; Li\ m :r m [\ O C’\ LI" KL LP HH NH NH :H 0H 0H 0H 0 H: OEHp HHON Oxgoz 00H: XLOZ mmOU QMHB xcHOZ ”NOE m®OU ®MH3 OO.ON OOO NO.NO ON 000 OHQ OOO OHOO OEHO HHsN Oxpoz 00H: 0L03 0000 00H3 0003 00: 0000 00H3 OO.NN OCO N0.NO 0O O00 OH0 OOO OHgo OEHp HHO0 Oxpoz OOH: xuoz 0000 00H3 0L03 00: 0000 00H3 OO.NN OCO N0.ON OO OO0 OHO OOO OHOO OEHO HHON ONLOz OOH: 0L03 0000 00H3 xLo3 000 0000 00H3 LOHHOLN :OLON OZ: OEHO HHOO Oxpoz OOH: 0003 0000 00H3 0003 00: 0000 00H3 00HO0 0H0: O.OOz OEH: HHOO 0:0O3 00H: x003 0000 00H3 0003 00: 0000 00H3 OO.HN 0cm N0.0 OO OHOHN N:HCOOHO I 0. WNW.“ 23M EEO: :ON. O:H.N NHN.O OO OO:.N HO:.O NO 33% OOO. ONH.N OON.O OH NNO.H HNO.O :H OO.HO OOO NO.N NO OHOHN N:HOOOHO HON. OOO.H ONO.O OO HHO.N NOH.O :N OOHNNOz OON.H NH:.H ON0.0 NN :ON.N OO0.0 ON OHNOHm woe 0H2: OOONOO OOO. ONH.N HHO.: NO :NO.H OON.: HN OOHNNOz OON.H ONH.N NOO.: OH NON.H NNH.O NN OHNch chpmSQ UCmHm coupoo NNN. HNN.N NON.O NO NNN.H ONO.O ON OOngmz NNN.H NOH.N OOH.O HN OON.H NOO.O HN OHNOHN chHHOch mHDHmth/mm OH:. HOH.N N:H.O OO ONN.N HON.O ON OOHpNOz ONN. OON.H ONN.O HN ONH.N ON:.O HN OHNOHm ppmoINp0csmq ::N. OH:.N NNO.O NO NHO.N O:O.: ON OOHNNOz NNO. OOO.N OOO.O NH NNO.N o:0.0 ON OHNOHm mppocm 0.C0§03 ONN. :NN.N NOO.: ON ONH.N OOO.: NO OOHNNOS ONNO.NI HON.N OO0.0 ON NNN.N NNN.: HN OHNch 00000 Hmsmmo ONOO.N OOH.N ONO.O OO NNH.N NOH.N ON OOHNNOz Hmm. NO:.N NNN.O NH OHN.H ONH.O :N OHNOHm 000 HHmzppwo oHpmm .>0Q .>0Q N NO O 0.00 COO: z 0.00 :00: 7 050000 HOOHNO: c0>m E00 .OOOOOO HOOHNOE OOO NOHHHOHOOOOOOO COHOOHHHII.ON 000:0 .H0>0H HO. 00 0:00H0H:0H0: .Hm>mH mo. um acmOOmflcmem 235 OOO. ONO.H . NOO.: ON OHm.N NON.O ON 00H::0z OON. NOO.H OO:.O O NOO.N OOO.O O 0H0:H0 O0.00 0:0 O0.00 :0 :00 OH: 0:0 :H:O OOO.m OOO.N OOO.: Om NOH.N OOO.O ON 00H::0z OOO.H NOO.H mm0.m O OON. OOO.: O 0H0:H0 O0.00 0:0 OO.NO :0 :00 OH: 0:0 OH:O MOO. OHO.H OON.O Om OON.H OOO.: Hm 00H::0z OOO.H OH:.N NOO.: N NNO.N ONN.O O 0H0:H0 OO.NO 0:0 O0.00 00 :00 OHO 0:0 OH:O ONO. HNO.N OOO.O HN OOO.H OmN.O NOH 00H::0z ONH. NNN.H mmm.O HN OOO.H HOm.O HN 0H0:H0 LQHHOch :CcHSH. OZ: NOO. . ONH.N OOO.O ON ONH.N OOm.O OO 00H::0z NNO. NOH.N ONN.O NN NHN.H OOO O ON 0H0:H0 uhfiim pacx m.CmZ 236 CO 0070 ONO.H OOH.N ONO.O ON OOO.H HNO.O ON OHH.H- OOH.N HON.O N OOO.N NOO.O ON mOsmm va OOH. . OOO.H OOO.O OO ONO.N OOO.O OO OOION 000 mme 00:0 :00000 OOH.H- NOO.N OOO.O N NON.H HOO.O NN OO 0000 0ONO.H OOO.H OO0.0 OO OHO.N OOO.O OO OOuOO 000 OON. HOO.N OOO.O NO OOO.H NOH.O NO OOION 000 memzm ccmam £00000 ONO. NOO.N OOO.O ON OOO.H OOO.O ON OO 00>O NOO. I OON.N ONN.O OO ONO.H OON.O NN OOIOO 000 NOO. NOO.N ONO.O HO NOO.H ON0.0 OO :1 OOION 000 pmflfiogm 0H2O000>0m OOO.H- OON.H OOO.O ON OOO.N NOO.O ON OO 00>O HOO.H OON.N OOO.O OO OOO.N OOO.O ON OOIOO 000 OOO. OOO.N OOO.O NO ONH.N OOO.O OO NO-ON 000 . ppmoxwuncsmq OOH. OOO.N OOO.O OO OOH.N HOO.O NN OO 00>O OOO. I ONO.N OOH.O ON OHO.H ONN.O HO OOIOO 000 ONO. u ONO.N HOH.O NO HOO.H ONO.O OO OOION 000 mchOSm. m 0 29:03 OOO.H- OOO.H OOO.O ON NHO.N OOO.O ON OO 00>O OOO. NOO.N OON.O NN HON.N OOO.O OO OOuOO 000 0HOO.H- NOO N OOH.O OO NHO.N OON.O OO OOuON 000 0000a H030wo OON. OHO.N OON.O OO OHO.H OOO N ON + OO 000 ONH.H OOH.N NOO O OO OOH.N ONH.N ON OOIOO 000 NON.H NOO.N OO0.0 OO OON.N HOO O OO OOION 000 wmm OHmzppmo .>0Q .>0Q 0 :o 0 0.00 :00: z 0.00 :00: z 0w< cm>m UGO .000 0:0 N00H0000O00000 :0000HHH--.NO 00000 237 .H0>0H O0. 00 0:00H0H:0H00 OHO.H- OOO.H OOO.O OH OOO.N OOO.O O OO 00>0 OOH. OOO.N ONO.O O NNN.N OON.O O OOIOO 000 NOO.H OOO.H ONO.O OH OOH.N OO0.0 0N OOuON 000 00.3 0:0 OOOO 00 000 0H0 0:0 00:0 OON. HHH.H ONO.O O ONN.H NOO.O N 0O 0000 OOO.H NON.N OOO.O NH ONO.N OOO.O O OmuOO 0mO 0ON0.N NOO.H ONO.O 0H OOO.H HOO.O NN O ION 000 OO.OO 0:0 OO.NO 00 000 000 0:0 OH:O HOO. OOO.H OHN.O N OOO.H OH0.0 O OO 0000 OON. OO .H OOO.O OH OOO.H ONO.O OH OOIOO 000 OHH.H OOO.H NO0.O 0H HOO.N ONO.O OH OOION 000 00.NO 0:0 O0.00 00 000 000 0:0 OH:0 OOO. . OOO.H NOO.O ON NOH.N OOO.O ON 0O :0>0 ONN. . OOO.N OOO.O ON NOO.H OOO.O NO OOnOO 000 NN. OOO.N OOO.O NO OOO.H ONO.O O0 OOION 000 hmflflOLm :CLSE O7: 0NOO.N- NNN.N OOH.O NN OOO.N OOH.O 0O 0O 00>0 ONO. OOO.H OO0.0 ON OOO.H OOO.O OO OOuOO 000 NHO. . ONH.N OO0.0 N. ONH.N OO0.0 NO OOuON 000 000:0 00:0 0.:02 0OOO.N- OOO.N OOO.O ON OOO.N OHO.O OH OO 00>0 0NOH.N HOO.N OOO.O ON ONN.N OOO.N NH OOnOO 000 HOO.H- OOH.N HOO.N NO OOH.N OOO.O OO OOION 000 OO.HO 0:0 OO.O 00 000HO 0:H:00H0 238 OOO. u OOO.N OOO.O OO OOO.N OOH.O OO OOO.OHO 0000 OOO. OOO.N OON.O OO OOH.N NOO.O HO OO0.00 00 OOO.OO HOO. OON.H NOO.O ON OOO.H HHH.O NN OO0.00 00 000 0H:O :00000 OON.H HOH.N NOO O O OOO.H NO .O OO OOO.OHO 0000 OOO. u ONN.N ONO.O 0O OOO.N OO0.0 NO OO0.00 00 OOO.OO OOO. O N.N NON.O ON HOO.N NO0.0 HO OO0.00 00 mem3Q Ccmam coppoo HOO.H NNO.N OON.O 0O OOO.N OON.O 0O OO0.0HO 0000 OON. . OOO.N OOO.O 0O OON.H ONO.O OO OO0.00 00 OOO.OO OOO.H OOH.N NOO.O ON OHO.H OOO.O NN OO0.00 00 LMHHOch mflflwmth/mm ONO. NON.N NOH.O OO NNN.N OOO.O 0O OO0.0HO 0000 ONN. OOO.N NON.O NO OON.N OO0.0 OO OO0.00 00 OOO.OO NOO. . OON.H OOO.O N OOO.N ONO.O NO OO0.00 00 0000 0000304 :OOO.N- OOO.N ONN.O NO ONO.H ONN.O OO OO0.0HO 0000 NOO.H O N.N ONN.O OO HHO.H OH0.0 OO OO0.00 00 OO0.00 OOO. ONO.N HON.O N OON.N HOO.O ON OO0.00 00 mpLOLW m0C®E03 OHN. . NOO.N OOO.O 0O OOH.N OOO.O 0O OOO.OHO 0000 OOO. . ON0.H OOH.O OO OOH.N OON.O OO OO0.00 00 OO0.00 NON.H- OOO.N OO0.0 ON OOO.N HOO.O HO OO0.00 00 00000 O03000 00NO.H NOO.N OOO.O OO HOH.N NHO.N HO OOO.OHO 0000 HON. OOO.H NON.O HO OON.N HNO.O OO OO0.00 00 OO0.00 OOH. I NON.N ONN.O 0N OHO.H OOO.O NN OO0.00 00 Mdm OO000000 oflpmm .000 .>0Q N 00 0 0.00 :00: 0 0.00 :00: : mEOQCH :00: 000 .mEoocO 0:0 zpflfiwpflpamomzm COszaaHll.mm mqmma l'".nl|'| Ilell' mpospomm mmomo pcm mommSUm mo Esm cofipmfipm> mo condom 'l’l’n' 1' .Axv mamm mom maomflfim>m omfipcwzopoe CH wCOHumfipm> Loo wcfiumshom .Amv moamm mo mocmfipm>oo mo mammfimc¢ ”doe pacx coupoo a.mm503nu.mm mqmmmmm :H Lommm a.mo a.mHm H.mHm H mpcmspmope 2 Wm. Rm: a.mama mémmaa m 383 a.mme 5.5mmHH m.wHome m monoum w.mmmm :.Hmme 0.5HHmmm mm Hmpoe onmp mumsom mo mopmsqm m cam: mo Esm mm“ xx“ mxw ma coHpmHmm> wo momzom :onmomwom mpozpopm mmopo Eomm :oHpmH>mQ Ucm mmpmsqm mo Esm U. III I‘ll it!!! .Axv mHmm mom mHomHHm>m omecmnome CH wCOHpmHhm> pom mcHumSWUm .nmv monm mo mocmHmw>oo mo mHmmec< ”annm pch m.cmzll.m: mqm<9 TABLE ““.--Laundry Cart: 263 Analysis of variance of sales. Source of Sum of Df Mean E Variation Squares Square ratio Total 2“.0 23 Stores 8.0 5 Weeks 1.0 3 Treatments .2 1 .2 Error 1“.8 1“ 1.1 .189 26“ Ill .. . ‘ . fl..- . ...._...51, n- mmH. H. H H. mcmmE psoEpmomp ompmSmp< :H a.» m.mH H.Hm N.N» mH Loppo msHQ pcmEummpB w. mH m.w @.mH m.mH m.:> :H Loppm m. m.H m.m H mucoEpwmpB w.m @.: m.mm m mxmmz m.~H v.0: w.mmH m mmLOpm m.mm m.©© m.mHm mm Hmpoe onmL opmzom ma mopmsvm a :mmz mo Esm mm“ xx“ mxw ma conmomem Eomm coprH>oo mpospopm mmOLo pcm wmpmsqm mo Esm coprHpm> mo mopsom .Hxv mem pom oHomHHm>m mmecmcopoE CH mcoHpmHLm> pom wchmsmpm .sz monm mo mocmHmm>oo mo mHmszc< “aaaaoam =case oz=uu.m: mamae 265 TABLE “6.-—Reversib1e Broiler: Analysis of variance of sales. Source of Sum of Df Mean 3 Variance Squares Square ratio Total 37.0 23 Stores 17.8 5 Weeks 10.2 3 Treatments .“ 1 .“ Error 8.6 1“ .6 .667 “1 I_-.-:3—q'£ ”at Jun... 1.. .~0_ u L... “a w ‘1...» 266 conmomem Eomm :oHpmH>mQ muozoomd mmopo pom mopmsam mo 53m m.wm memoE pCmEpmmpp UoprnU< H.0HH m.MHH m.wm a.mmm mH moppo msHQ pcmEpmmLB m.mw H.mm HH.wHV m.mmH :H moppm H.mm m.m: o.wm H mpcmEpmmLB m.mm m.mm H.2mmH m mxom3 m.HmH o.:mm w.m:m m mmpoum m.mmm m.:m: m.sHmm mm Hmpoe mpmzdm mm moLmSUm mo Ezm mmw zxw mxw mo coHumem> no momsom ‘1‘"‘ll III. Il| .Axv oHMm pom mHomHHm>m mmecmzoLoE CH mCOHumHmm> mom wcHuwszm .sz mmHmm mo mocmem>oo mo mHmmHmc< ”Loumsa pcon coupooln.sa mqmqs 267 .. . .. 71.1 '1. .I will. - .I J J r... ::N. m.om H m.wm mCmoE pCmEpmon poumzmp< :H H.mooH o.Hwom m.zmmH m.:wmm mH COCCm msHQ pCoEpmoCB w.wHH mH m.mHmH a.mme m.mmoH m.~m>~ :H COCCm H.:om o.mHm 0.5»: H mpCoEpmmCB m.mH Am.mmHv a.mHmH m mammz m.HmOH N.Nwwm a.mwmwm m moCOCm m.HMHm m.MHmn ©.Hw:w: mm HmpOB onwC mCmsom mo moCmsqm m Cmoz no Esm mzw mxw mxw mo ConmoCmmm EOCm coHumH>mQ muosp0Cm mMOCo pCm mopmsvm mo Esm CoHpmHCm> no moCsom I'll. .Axv mHmm Com mHCmHHm>m mepCMCoCmE CH mCoHpmHCm> Com mCHumshpm .Hmv monm Co monHCm>00 mo mHmmHmC< ”mam HHmaCCmouu.m: mqm