Effect of Odd Pricing on Choice of Items From a Menu

ABSTRACT - Subjects who were asked to make selections for a five-course meal from a menu were found to have a greater likelihood of choosing a particular item when that item was odd-priced than when it was even-priced. Although the number of items on the menu did not affect the size of this odd pricing effect, the study does suggest some situational variables which may determine when and to what degree odd pricing will affect consumer choice.

Citation:

Robert M. Schindler and Lori S. Warren (1988) ,"Effect of Odd Pricing on Choice of Items From a Menu", in NA - Advances in Consumer Research Volume 15, eds. Micheal J. Houston, Provo, UT : Association for Consumer Research, Pages: 348-353.

Advances in Consumer Research Volume 15, 1988      Pages 348-353

EFFECT OF ODD PRICING ON CHOICE OF ITEMS FROM A MENU

Robert M. Schindler, University of Chicago

Lori S. Warren, University of Chicago

ABSTRACT -

Subjects who were asked to make selections for a five-course meal from a menu were found to have a greater likelihood of choosing a particular item when that item was odd-priced than when it was even-priced. Although the number of items on the menu did not affect the size of this odd pricing effect, the study does suggest some situational variables which may determine when and to what degree odd pricing will affect consumer choice.

INTRODUCTION

Odd pricing refers to the practice of expressing a price so that its ending (i.e., its rightmost digits) causes it to fall just below a round number. For example, a price setter might choose to use the "odd-ending" price of $4.95 rather than the round number (or "even-ending") price of $5.00. The widespread and persistent use of odd pricing by retailers (Friedman 1967; Georgoff 1972; Kreul 1982; Rudolph 1954; Twedt 1965) suggests that this technique may have effects on the consumer which are beneficial to the retailer.

However, research on the effectiveness of odd pricing has failed to provide evidence for a consistent effect. Three field studies (Dalrymple & Haines 1970; Georgoff 1972; Ginzberg 1936) have reported odd pricing effects for some conditions or products, but not others. Laboratory studies have yielded similarly inconsistent findings. Dodds and Monroe (1985) and Georgoff (1972) failed to find an effect. Lambert (1975) and Alpert, McGrath, and Alpert (1984) reported an effect for some products or groups but not others. Schindler (1984) and Schindler and Wiman (1986) have found evidence for an effect of odd pricing on long-term memory, but have not found evidence for a more immediate effect of this pricing technique.

The Amount of Attention Used to Process a Price

Such inconsistent results strongly suggest that there are one or more aspects of the situation where a consumer responds to price which determine whether or not the consumer will respond more favorably to an odd-ending price than to the equivalent even-ending price. This study was designed to test the role of one such situational variable, the amount of attention consumers use when perceiving a price.

This attention variable was chosen from consideration of a commonly proposed mechanism of the odd-pricing effect (e.g., Barker & Anderson 1935, p. 166; Bush & Houston 1985, p. 599; Lambert 1975). According to this mechanism, an odd-pricing effect occurs because consumers tend to ignore the rightmost nines and fill their places with "default" values (e.g. zeros) or with words which could be used for any rightmost digits and therefore connote numbers lower than nine. Thus, a consumer may think of $5.98 as "around five dollars," or may consider $395 as "three hundred and some dollars."

Although this mechanism seems plausible, it is hard to overlook that consumers will typically deny that they just ignore the rightmost nines. Some will claim that they round odd prices up to the even price. Others will claim that they consider all of the digits in a price, not just the leftmost ones. And, in fact, it is hard to believe that consumers are so foolish as to ignore numbers which could represent a significant amount of money.

A possible solution to this problem is suggested by Brenner and Brenner's (1982) observation that rounding an odd price upwards or processing all of the digits of a price involves considerably more effort than processing only the leftmost digits (see also Nagle 1987, p. 248). Consumers may be fully aware that they should round odd prices upward or process all of the digits, and perhaps they often do so when there are no other tasks making demands on the consumer's limited processing capacities (see Jacoby & Olson 1977). But, in situations where there are other stimuli competing for their limited cognitive resources, consumers may fail to take the extra time or make the extra effort to refocus their attention from something else to processing these less important digits, the rightmost nines. Instead, they may end up simply ignoring these digits. In other words, while it is hard to imagine that consumers are foolish enough to ignore the rightmost nines, it is not at all hard to imagine that they are lazy enough to do so.

How well does this attention variable account for the inconsistencies in the past research on odd pricing effects? For the field studies, it is impossible to reconstruct how much attention the consumers may have been expending in the various purchase situations involved. But, in the laboratory studies, it could be argued that the subjects usually applied a great deal of attention to each price. In most of the experiments (Dodds & Monroe 1985; Georgoff 1972; Schindler 1984; Schindler & Wiman 1986), the subjects were presented with the prices one at a time in a situation where they were not pressed for time. Thus, these subjects certainly had the time to give their full attention to each price. The subjects in the Lambert (1975) and Alpert et al. (1984) studies were required to deal with four prices within one minute, and thus may have been a bit pressed for time. However, these experiments involved the subjects in a game, and this may have increased their alertness and motivation and, as a result, increased their likelihood of fully attending to each price.

Thus, these laboratory studies may all have created situations which led the subjects to allocate enough attention to each price so that they rounded the odd prices upward or fully processed the rightmost digits. It could be for this reason, then, that these studies failed to show a pronounced immediate effect of odd pricing on judgment or choice.

Design of the Experiment

This study was designed to test whether the amount of attention consumers use to process a price plays a role in determining the size of the odd pricing effect which is likely to occur. We chose to investigate this attentional variable in a situation where the subjects had to choose food items from a simulated restaurant menu. It was felt that this task, being a relatively familiar one to the subjects of this experiment, would have a good chance of evoking realistic habits of attentional allocation. Further, both casual and more systematic observation (Kreul 1982; Marketing News 1986) indicates that restaurants often use odd-pricing policies.

In order to create one situation where the subjects would be likely to expend a large amount of attention to each price and one situation where they would be likely to expend only a small amount of attention to each price, we varied the number of alternative items on the menu. In the short form of each menu, the subjects made choices from among only 19 alternatives. In the long form of each menu, the subjects were faced with a total of 58 alternatives. Since the subjects who received the long menus had to make choices in the same amount of time as was given to the subjects who received the short menus, it was predicted that the long-menu subjects would pay less attention to each price than would the short-menu subjects and therefore would be less likely to consider the rightmost digits or round the odd prices up to the even numbers. Thus, they should show a larger odd pricing effect than would the short-menu subjects.

Formally, our hypotheses can be stated as follows:

H1: Subjects will show a greater likelihood of choosing a menu item when it is odd-priced than when it is even-priced.

H2: Subjects in the long-menu condition will show a greater tendency to choose odd-priced menu items over even-priced menu items than will the subjects in the short-menu condition.

METHOD

Design of the Menus

Each menu was printed on a single 8.5" by 11" sheet of paper. All of the menus were printed in the same typeface and displayed the names of alternative food items grouped under the following six headings: Appetizers, Soups, Salads, Entrees, Desserts, and Beverages. In all of the menus the headings occupied approximately the same positions on the page.

Each subject received one menu and was instructed to select one of the alternatives under each of the six headings. In other words, they were asked to order a five-course meal. For each of the five courses (i.e., all of the headings except Beverages), the alternatives consisted of at least one even-priced item, at least one odd-priced item, and at least one filler item. Each even price ended with the digits 00 (e.g., $5.00), each odd price ended with the digits 95 (e.g., $6.95), and the price of each filler item ended with either 25, 50, or 75 (e.g., $5.25, $6.50, $2.75).

Price ending was varied by constructing two versions of each menu. If an item was odd-priced in the A version of a menu, then that item would be even-priced in the B version of that menu. Each item which was even-priced in the A version of a menu was odd-priced in the B version. For example, the Baked Flounder was priced at $8.95 in the A version of each menu. In the B version of each menu, it was priced at $9.00. The Broiled Pork Chops were priced at $6.00 in the A version of each menu, and were priced at $5.95 in the B version of each menu. Half of the subjects in the experiment received the A version of a menu and the other half received the B version. By this means, price ending was varied while keeping constant the name of the item and its approximate position on the page.

The filler items on the menus served two purposes. The first concerned realism. Since restaurant menus typically do not consist entirely of items priced at or just below whole dollar amounts, it was felt that exclusive use of such prices in this study might lead some subjects to suspect that odd pricing was a critical variable. Note that the filler items were chosen from the same pool of possible items as were the odd- and even-priced items and, except for price ending, were indistinguishable from the odd- and even-priced items.

The second purpose of the filler items was that they served as a means of measuring the price and quality sensitivity of the subjects' choices and thus made it possible to confirm that the subjects were taking both price and quality into account. The price sensitivity of the subjects' choices was tested by varying the price of each filler item by exactly one dollar (note that this left unchanged the price endings of the filler items). The quality sensitivity of the subjects' choices was tested by systematic placement of those filler items which pretesting showed were relatively well-liked and those filler items which pretesting indicated were somewhat less well-liked. These two variables, price and quality of the fillers, were crossed. For example, a particular position under Entrees which was reserved for a filler item would contain Roast Breast of Turkey (well-liked) for $7.50 in 1/4 of the menus, Roast Breast of Turkey for $8.50 in 1/4 of the menus, Roast Pocket of Veal (less well-liked) for $7.50 in 114 of the menus, and Roast Pocket of Veal for $8.50 in 1/4 of the menus. Each of the resulting four menus had both an A version and a B version (which differed as to the endings of the nonfiller items), thus defining eight different menus.

For each of these eight menus there was both a long form and a short form. Each short menu was constructed from the corresponding long menu by simply deleting items until a minimum of three per course (one odd-ended, one even-ended, and one filler) was reached (the Entree category was allowed five rather than three items in order to maintain realism). Each heading in the short and long forms of the menus had the following number of alternative items:

TABLE

Thus, the long forms of the menus contained over three times the number of items as the short forms of the menus. Since each of the eight menus previously described had both a long form and short form, the experiment contained a total of 16 different menus.

The Booklets

Each subject in the experiment received a four-page booklet. The first page of the booklet contained a description of the dining scenario and the instructions for the experiment (see Appendix) [A slightly different, earlier version of this scenario was used for the first 86 subjects.]. The description of the dining situation was designed to encourage consideration of price. It was felt that, without such encouragement, the lack of involvement of real money in the experimental task could lead subjects to respond on the basis of their food preferences alone.

On the second page of the booklet was one of the 16 different menus. On the third page of the booklet was a series of post-choice questions which asked the subjects about (1) the difficulty of the task, (2) the plausibility of the scenario, (3) the importance of price in their choices, and (4) what they believed the experiment was about.

On the fourth page of the booklet was a menu which was an exact copy of the menu which appeared on Page 2 of the booklet except that all prices were omitted. The purpose of this page was to test the subjects' short-term recall of the prices of the menu items that they had chosen on Page 2. This fourth page was stapled to Page 3 so the subjects would be unaware that they were going to be asked to recall prices. It was expected that if the long menu succeeded in taxing the subjects' attention, then the price recall for the chosen items in the long menu would be lower than for the chosen items in the short menu. Thus, this surprise recall task was included as a manipulation check, albeit an imperfect one.

Subjects

Two hundred and two university graduate and undergraduate students served as subjects. Data from 17 subjects were discarded because they failed to make a choice for all of the five courses of the dinner. Thus, the final sample size was 185.

The subjects were not told anything about the critical variables in this experiment until after the data collection was complete. The subjects' answers to the post-choice question about the purposes of the experiment failed to reveal any subjects who had correctly guessed the critical variables.

Procedure

The experiment was carried out during regularly scheduled class meetings. The booklets were handed out so that the 16 booklet types were distributed randomly to the members of each class. After each subject had received a booklet, the instructions and description of the dining situation were read aloud and any procedural questions were answered. Then, the subjects were given one minute to make their menu choices. Next, they were asked to complete the post-choice questions. Finally, they were given the instructions to complete the surprise recall task. The first three-fourths of the subjects were instructed not to spoil the recall measure by looking back to the second page of their booklets. For the last fourth of the subjects, the pages were coded so that the first three pages could be handed in before the recall task was begun. Analysis of the data indicated that these two procedures produced quite similar recall results.

RESULTS

The results of this study can be divided into two categories: (1) those concerning the success of the technique, and (2) those concerning the substantive hypotheses of the study. The results concerning the technique will be reviewed first.

Results Concerning the Technique

The technique used in this study was designed to lead the subjects to take into account both price and quality when they make their choices. To test whether they in fact did so, the price and quality of the filler items were systematically varied. A two-way analysis of variance was run on the number of filler items chosen in each of the four price-quality conditions. The results indicate that more filler items were chosen when they were given the lower, rather than the higher, prices (1.10 filler items/subject vs. 0.67 filler items/subject; F[1,181 = 10.96, p < .002). Also, more filler items were chosen when they were of higher quality (i.e., well-liked), rather than lower quality (i.e., less well-liked) (1.20 filler items/subject vs. 0.55 filler items/subject; F[1,181] = 26.15, p c .001). There was no statistically significant interaction between these two variables (F[1,181] < 1).

These results also make possible an estimation of the price elasticity of demand for the menu items. The prices-of the filler items were varied by exactly one dollar, and the average price of the filler items (i.e., the midpoint between the higher and lower prices) was $4.34. Use of this midpoint for the computation of percent change (Samuelson 1980, p. 361) indicates that there was a 23% change in price. A similar calculation indicates that a 49% change in demand was associated with this price difference. This yields a price elasticity estimate of -2.13, which is quite typical of the price elasticities found in the food industry (e.g., Ghosh, Neslin, & Shoemaker 1983).

The surprise recall test was designed to check whether or not the manipulation of menu length was effective in determining the amount of attention the subjects would pay to each menu item; if less attention was paid to the items in the long menus, then recall should be poorer. Measuring recall as the percent of prices exactly correct, the mean for the short-menu subjects was 53.3% and the mean for the long-menu subjects was 52.7%. The difference between these means was not statistically significant (t[173] < 1). Further, the menu length manipulation was expected to affect the likelihood of the subjects rounding an odd price up to the even price. If the percent of odd prices which were recalled as the equivalent even prices (i.e., the price five cents higher) is used as a measure of the degree to which odd prices were rounded upwards, the menu length manipulation again appears to have been ineffective. In fact, the long-menu subjects were more likely to have recalled an odd price as the equivalent even price than were the short-menu subjects (15.4% vs. 6.2%, t[171] = 2.29,. p < .025).

On the other hand, while the recall results showed the menu length manipulation to be ineffective, a further analysis shows that the long-menu subjects did not, in effect, shorten their menus by considering only the first few alternatives under a heading. The mean item-order of the chosen items was considerably higher in the long-menu condition than in the short-menu condition (5.0 vs. 2.0; t[183] = 15.17, p < .001). Furthermore, post-choice questioning also indicated that the long-menu subjects did not use some strategy to make their choices as easy as those of the short-menu subjects. On a 4-point scale with 1 being "easy' and 4 being "difficult,' the long-menu subjects rated the task as being more difficult than did the short-menu subjects (1.8 vs. 1.4, t[183] = 3.55, p < .001).

Effects of Price Ending on Choice

To determine the effects of price ending on choice, the number of odd-priced items chosen and the number of even-priced items chosen was computed for each subject. Each of these numbers could be between 0 and 5 (since the experiment concerned the choices in each of five courses of a meal). However, the number of odd-priced items and the number of even-priced items chosen need not add up to five, because the subjects sometimes chose filler items [In fact, the total need not add up to five even if the chosen fillers are added in. This is so because a minor design error made it necessary that some of the filler items be excluded from the analyses.].

It should be emphasized that no subject saw the same item with both an odd and an even price ending. However, each item that was odd-priced in the A version of a menu was even-priced in the B version of that menu (and vice versa). Since half of the subjects received A versions of a menu and the other half received B versions, a greater overall tendency to choose odd-priced items could not be due to the particular items involved. Rather, it would mean that a particular menu item would be more likely to be chosen when it was seen with an odd price than when it was seen with the equivalent even price.

The mean number of odd- and even-priced items chosen from the long and the short menus can be seen in the Table. In both menu conditions, items were more likely to be chosen when they were odd-priced than when they were even-priced. However, this odd pricing effect was not larger for the long menus than for the short menus. A two-way analysis of variance was run on these data. The number of choices made by each subject was the dependent variable, the price ending of the chosen items was a within-subjects independent variable, and the menu length was a between-subjects independent variable. This analysis indicated that the main effect of price ending was significant (F[1,1831 = 7.56, p < .01), the main effect of menu length was not significant (F[1,183] = 1.06, p < .4), and the price-ending by menu-length interaction was not significant (F[1,183] < 1). Thus, H1 was supported, but H2 was not.

TABLE

MEAN NUMBER OF ODD- AND EVEN-PRICED ITEMS CHOSEN FROM THE SHORT AND LONQ MENUS

It should be pointed out that this greater likelihood of an item being chosen when it is odd-priced is an odd-pricing effect, as opposed to resulting simply from the fact that the odd prices were five cents lower than the equivalent even prices. Considering that the mean price of a nonfiller menu item was $5.50, the average increase in demand due to odd pricing yields an estimated price elasticity of -23.0. This is an order of magnitude more extreme than the price elasticity computed from the one-dollar price changes in the filler items and indicates a greater price elasticity than is usually seen in consumer products. Such a greater-than-expected price elasticity around prices which are just below a round number is in fact a technical definition of the odd pricing effect (e.g., Georgoff 1972, pp. 6-7).

If one uses the short-term recall results of this study as a measure of the level the subjects perceived the prices to be, it could be determined whether or not this greater-than-expected price elasticity is due to the subjects perceiving this five cents difference as a much larger difference. These results indicated that the average odd price was recalled as being 8.1 cents less than it actually was, and the average even price was recalled as being 0.4 cents more than it actually was (these recalled-actual price differences were not significantly different, t[184] = 1.47, p < .2). Thus, there is not evidence from the recall data in this study that prices which are only slightly below a round number are perceived as being substantially below the round number.

One additional analysis was conducted in order to further explore the lack of the hypothesized interaction between the odd pricing effect and menu length. This analysis tested the possibility that the results were distorted by those subjects who found the scenario less than fully plausible. A perceived plausibility scale created from the post-choice questions on the topic was added to the two-way analysis of variance as a third independent variable. While there was a tendency for the subjects who found the scenario more plausible to show a larger odd pricing effect in the long menus than in the short menus, this price-ending by menu-length by perceived-plausibility interaction was not significant (F[1,1793 = 1.27, p < .3).

DISCUSSION

This study provides clear evidence that pricing an item just below a round number can increase its likelihood of being chosen to a greater extent than would be expected on the basis of the few pennies involved. This finding may not be surprising, because many have long believed that odd pricing has just such an effect. However, this finding is important, because of the shortage of published evidence that this belief in odd pricing is justified.

Although this study was successful in producing an odd pricing effect, it was not successful in yielding information about the role attention plays in mediating the occurrence or the size of the effect. The failure of the menu-length manipulation to affect the size of the odd pricing effect is inconclusive, since the recall results raised questions about whether this manipulation was effective in producing differences in attention. It has been pointed out (e.g., Thaler 1986) that subjects who are paid for accuracy in experimental tasks typically do not perform any better than subjects who are not paid for accuracy. This suggests that perhaps the mere agreement to participate in an experiment leads subjects to, when necessary, expend considerable effort in the task. Such conscientiousness may well have elicited in the subjects of this experiment enough processing capacity to attend to all the digits of the prices of even 58 menu items over the course of a minute. Thus, the role of attention load in the odd pricing effect remains a topic for future research.

Although this study was not successful in providing evidence for the role of one situational variable, the amount of attention allocated, the study was successful in demonstrating a pronounced odd pricing effect. Why was a consistent odd pricing effect found in this study and not in several previous studies? Consideration of the ways in which this experimental situation differed from that of previous studies suggests some other situational variables which may mediate odd pricing effects.

In this study, the subjects were asked to choose among alternatives, some of which were odd-priced, some even-priced, and some neither. ID previous studies (e.g., Dodds & Monroe 1985; Georgoff 1972; Schindler 1984; and Schindler & Wiman 1986), subjects were asked to rate products with either even or odd prices rather than choose among them. Considering that important differences have been found between rating and choice situations, including preference reversals (e.g., Lichtenstein & Slovic 1971; Goldstein & Einhorn 1987), the role of this variable in odd pricing effects is worthy of further research.

Another way in which the present study differs from previous odd pricing research concerns the presence of a social context. The studies cited above do not report presenting a social context in the experimental situation. The scenario used in the present study provided the subjects with a clear social context: a dinner partner who cared about how much was being spent. While this scenario was included to encourage subjects to take price into account, it may have also help produce the odd pricing effect. For example, a consumer who considers all of the digits of odd prices or rounds them upwards when evaluating alternatives may assume that other people do not do so. Thus, when trying to create an impression of being thrifty, the consumer might feel it would be helpful to choose items which are odd-priced. This possibility is in fact consistent with the present finding of an odd pricing effect on choice in the absence of recall evidence that the subjects themselves perceived the odd prices as being substantially lower than the equivalent even prices.

Thus, this study provides some evidence for the existence of an odd pricing effect and serves as a starting point for research on the situational variables which play a role in affecting when such an odd pricing effect will occur and how strong the effect is likely to be.

APPENDIX

REFERENCES

Alpert, Mark I., John E. McGrath, and Judy 1. Alpert (1984), "Magic Prices: An Extension," Proceedings and Abstracts of the Thirteenth Annual Meeting of the Western Regional Conference, American Institute for Decision Sciences.

Barker, Clare W. and Ira D. Anderson (1935), Principles of Retailing, New York: McGraw-Hill. Brenner, Gabrielle A. and Reuven

Brenner (1982), "Memory and Markets, or Why Are You Paying $2.99 for a Widget?" Journal of Business, 55(1), 147-158.

Bush, Paul S. and Michael J. Houston (1985), Marketing: Strategic Foundations, Homewood, IL: Richard D. Irwin.

Dalrymple, Douglas J. and George H. Haines, Jr. (1970), "A Study of the Predictive Ability of Market Period Demand-Supply Relations for a Finn Selling Fashion Products," Applied Economics, 1(4), 277-285.

Dodds, William B. and Kent B. Monroe (1985), "The Effect of Brand and Price Information on Subjective Product Evaluation," in Elizabeth C. Hirschman and Morris B. Holbrook (eds.), Advances in Consumer Research, Vol. 12, Provo, UT: Association for Consumer Research, 85-90.

Friedman, Lawrence (1967), "Psychological Pricing in the Food Industry," in Almarin Phillips and Oliver E. Williamson (eds.), Prices: Issues in Theory, Practice, and Public Policy Philadelphia: University of Pennsylvania Press, 187-201.

Georgoff, David M. (1972), Odd-Even Price Endings, East Lansing, MI: Michigan State University.

Ghosh, Avijit, Scott A. Neslin, and Robert W. Shoemaker (1983), "Are There Associations Between Price Elasticities and Brand Characteristics," Patrick E. Murphy et al. (eds.), 1983 AMA Educators' Proceedings, Chicago: American Marketing Association, 226-230.

Ginzberg, Eli (1936), "Customary Prices," American Economic Review, 26 (June), 296.

Goldstein, William M. and Hillel J. Einhorn (1987), "Expression Theory and Preference Reversal Phenomena," Psychological Review, 94(2), 236-254.

Jacoby, Jacob, and Jerry C. Olson (1977), "Consumer Response to Price: An Attitudinal Information Processing Prospective," in Moving Ahead in Attitude Research, eds. Yoram Wind and Marshall Greenberg, Chicago: American Marketing Association, 73-86.

Kreul, Lee M. (1982), "Magic Numbers: Psychological Aspects of Menu Pricing," The Cornell Hotel and Restaurant Administration Quarterly, 23 (August), 70-75.

Lambert, Zarrel V. (1975), "Perceived Prices as Related to Odd and Even Price Endings," Journal of Retailing, 51 (Fall), 13-78.

Lichtenstein, Sarah and Paul Slovic (1971), "Reversal of Preferences Between Bids and Choices in Gambling Decisions," Journal of Experimental Psychology, 89, 46-55.

Marketing News (1986), "Penny Pricing Facing Pinch from Nickel Discounts: Restaurant Study," (April 11), 14.

Nagle, Thomas T. (1987), The Strategy and Tactics of Pricing: A Guide to Prof table Decision Making, Englewood Cliffs, NJ: Prentice-Hall.

Rudolph, Harold J. (1954), "Pricing for Today's Market," Printers' Ink, 247 (May 28), 22-24.

Samuelson, Paul A. (1980), Economics, 11th edition, New York: McGraw-Hill.

Schindler, Robert M. (1984), "Consumer Recognition of Increases in Odd and Even Prices," in Thomas C. Kinnear (ed.), Advances in Consumer Research, Vol. 11, Provo UT: Association for Consumer Research, 459-462.

Schindler, Robert M. and Alan R. Wiman (1986), "Consumer Recall of Odd and Even Prices," Working paper, University of Chicago.

Thaler, Richard N. (1986), "The Psychology and Economics Conference Handbook: Comments on Simon, on Einhorn and Hogarth, and on Tversky and Kahneman," Journal of Business, 59(4, Part 2), S279-S284.

Twedt, Dik W. (1965), "Does the '9 Fixation' in Retail Pricing Really Promote Sales?", Journal of Marketing, 29 (October), 54-55.

----------------------------------------