On Price Limit Measurement: the Order Effect

ABSTRACT - This study investigated whether the order in which price stimuli is presented to subJects confounds price limit measurement. The results suggest that the order (ascending, descending, random) in which price stimuli is presented for subject evaluation had no statistically significant effect on either the mean lower or upper price limits. However, the results were in the hypothesized direction. Most importantly, the use of a randomized price series exhibited greater variability in both price limits than either the ascending or descending price series. There was no significant difference found in variability between the ascending and descending treatments. As a result, it is possible that the use of randomized price stimuli may increase method variance and mast the true effects of a price limit manipulation.

Citation:

Rustan Kosenko (1989) ,"On Price Limit Measurement: the Order Effect", in NA - Advances in Consumer Research Volume 16, eds. Thomas K. Srull, Provo, UT : Association for Consumer Research, Pages: 248-252.

Advances in Consumer Research Volume 16, 1989      Pages 248-252

ON PRICE LIMIT MEASUREMENT: THE ORDER EFFECT

Rustan Kosenko, Bradley University

ABSTRACT -

This study investigated whether the order in which price stimuli is presented to subJects confounds price limit measurement. The results suggest that the order (ascending, descending, random) in which price stimuli is presented for subject evaluation had no statistically significant effect on either the mean lower or upper price limits. However, the results were in the hypothesized direction. Most importantly, the use of a randomized price series exhibited greater variability in both price limits than either the ascending or descending price series. There was no significant difference found in variability between the ascending and descending treatments. As a result, it is possible that the use of randomized price stimuli may increase method variance and mast the true effects of a price limit manipulation.

INTRODUCTION

Most experimental investigations into consumer evaluation of price is determined to some extent by subject response to some set of price stimuli. Unfortunately, there is some empirical evidence to suggest that the order in which a price stimulus set is presented to experimental subjects may affect subject evaluation of specific prices (Della Bitta and Monroe 1973; Rexeisen 1982). Despite these findings, most pricing research rarely reports whether a random, ascending or descending price stimulus series was used. On the one hand, it may have been intuitively obvious that a randomization procedure was used and did not need to be reported; on the other hand, it may have been that price presentation may not have been expected to have a differential affect on the main effects under investigation (Rexeisen 1982). Unfortunately, randomizing the price series does not eliminate order variability, it only prevents an analysis of the potential order effects (Monroe 1977).

The purpose of this study is to investigate whether the order (ascending, descending or random) that a price series is presented to subjects confounds price limit measurement when the Stoetzel methodology is used.

LITERATURE REVIEW

Assimilation-Contrast (Sherif and Hovland 1961; Sherif, Sherif and Negerall 1965) theory yields the hypothesis that if a series of stimuli is presented for judgment in order of increasing magnitude, it tends to produce higher categories of judgment than if it were presented in order of decreasing magnitude. Essentially then, and in pricing terms, subjects compare cash price stimulus with the other price stimuli in the price series to which they have been exposed. Therefore, if cash successive price stimulus is larger than the prices the subject has seen when the order of the price series is presented in ascending order (e.g, $25, $50, $75. etc.) the subject should judge the price as occupying a "higher' category; if the subsequent stimulus is smaller than the preceding price stimuli (e.g., $75, $50, $25, etc.) then the stimulus should be judged as belonging to a "lower' Category. This latter situation would occur where he price series is presented in descending order (Della Bitta and Monroe 1973; Monroe 1977; Monroe and Petroshius 1981).

In the context of pricing research, Assimilation-Contrast theory suggests that if subjects are initially exposed to relatively high or low prices for a given product their frame of reference, or adaptation level (Helson 1964) is pulled toward the hi,,h or low price stimuli. This frame of reference then serves to anchor subsequent judgments for other prices. Once the high adaptation level or reference point has been established, prices lower than the original anchoring prices would be perceived as "cheaper" or "less expensive" than in the low price anchoring context; once a low adaptation level has been established, prices higher than the original anchoring prices would be perceived as "rr.ore expensive" than in a hi,,h price anchor context (Monroe 1977; Monroe and Petroshius 1981).

Empirical evidence for this effect is rather scant. Only two pricing studies have empirically investigated this potential confound. Della Bitta and Monroe (1973) investigated the order effect within the theoretical context of human psychophysical judgement (Helson 1964; Parducci 1954, 1956; Parducci and Holhe). They presented subjects with a series of prices in either ascending or descending order for eight consumer products (e.g., dress shirt, sports coat, hair spray, blouse), and asked them to rate 11 prices on a seven-point absolute judgment rating scale. They reported that respondents presented an ascending series of prices rated a given price as more expensive than those subjects presented prices in descending order. That is, the mean adaptation level price was higher in the descending than in the ascending condition.

It was almost one decade before someone empirically addressed the methodological concerns espoused by Della Bitta and Monroe (1973). Rexeisen (1982) tested the confounding effect of the price series order effect within the context of the price-quality relationship. His experiment required his subjects to judge the quality of three identical carpet samples. Within each treatment, subjects were first provided the carpet sample without price information, the second carpet sample was provided with a high or low price (57 or 529 per square yard), and the third carpet sample was low priced if the second carpet sample was high priced and vice-a-versa. He reported that the.e was a significant ordering effect in subject judgement of the quality (p <.01), value for the money (p<.05), and perceived worth (p <.10) variates.

These results suggest that the frame of reference of experimental subjects are sensitive to the contextual effect of the order of price presentation (Monroe 1977; Monroe and Petroshius 1981). Furthermore, the order of price presentation could produce systematic changes in the judged value of each price. Price judgments may vary depending on whether a higher or lower price precedes it in the price series and that certain prices would act as anchors for subsequent price judgments (Della Bitta and Monroe 1974, Monroe 1977).

The price order effect is most pronounced in price limit measurement; particularly, where the effect of a experimental manipulation is investigated. Subjects are required to select a price from a price series that represents the lowest and highest price acceptable to them to pay for a specific product under some experimental condition(s). Therefore, a price order effect may increase method variability and could mask the true effects of a manipulation.

Raju (1977), Fouilhe (1970), Cox (1986), and Kosenko et al. (1988) have investigated the effects of various product informational cues on consumer price limits and each has failed to report statistical support for some of their theoretically generated hypotheses. These failures may be due to the masking effect of the stimulus order effect.

HYPOTHESES

The following hypotheses have been generated from the theoretical framework and the empirical evidence presented earlier.

Hypothesis 1:

Subjects asked to evaluate a series of prices in descending order will specify a greater mean lower price limit than subjects asked to evaluate a series of prices in ascending order.

Hypothesis 2:

Subjects asked to evaluate a series of prices in an descending order will specify a higher mean upper price limit than subjects asked to evaluate a series of prices in ascending order.

Hypothesis 3:

There will be greater variability in the lower price limit when subjects evaluate prices presented in random order than in either ascending or descending order.

Hypothesis 4:

There will be greater variability in the upper price limit when subjects evaluate prices presented in random order than in either ascending or descending order.

The rationale for these hypotheses is based on the literature review provided above, and are derived from Assimilation-Contrast and Adaptation-Level theories. Descending prices would set a high adaptation level or reference point and subjects evaluating prices lower than the original anchoring prices would perceive those prices as "cheaper" or "less expensive" than in the low price anchoring context resulting in higher mean price limits. Subjects exposed to ascending prices should judge the subsequent higher prices as "more expensive" due to a low adaptation or reference point. They would t)hen be expected to report lower mean price limits.

The effect of random price presentation is less pronounced. Where subjects are presented prices in random order, the anchoring effect may be more enhanced. Subjects should have difficulty establishing stable anchors from which subsequent price judgments are made. For example, Monroe (1977) indicates that the lowest and highest price of a product line serve as anchors and that those anchors provide informational content for subjects to judge the other prices presented in the price series. Therefore, it is reasonable to assume that the price limits should exhibit greater variability due to the various high/low and low/high price combinations under the random price order condition. Under this condition, subjects should have difficulty establishing singular price anchors and greater variability in the price judgments should occur.

THE EXPERIMENT

Sample. Sixty-seven subjects were selected from two undergraduate classes at a major Northeastern university and were randomly assigned to three treatment groups. Twenty-two subjects were provided with ascending prices, 22 subjects were provided with descending prices, and 23 subjects were provided with the prices in random order.

Product. The product selected for this study was a personal computer. Interviews with local computer retailers confirmed that university students were a major purchasing segment. Retailers also described the most popular configuration purchased by students. That configuration was incorporated in the product description provided to the subjects during the experiment.

Method. The dependent variables were the lower price limit and the upper price limit. These price limits were measured using the Stoetzel (1954) methodology. In this method, subjects were provided with a detailed description of a desktop personal computer and a list of 50 prices ranging from $1500 to $6400. The prices differed by a constant interval of $100 and were presented in either ascending, descending, or random order. The prices used in the experiment included actual market prices.

After reviewing the product description, subjects were provided the price series and asked to respond to two questions:

(1) What is the minimum price that you would be willing to pay for the personal computer (that is, below what price would you seriously doubt the quality of the product?)

(2) What is the maximum price that you would be willing to pay for the personal computer (that is, beyond what price would you feel it would not be worth paying more?).

ANALYSIS

One-way anovas and a series of Hartley's homogeneity of variance F-tests for each of the dependent variables --- lower price limit and upper price limit were used to test the hypotheses. The summary statistics are presented in Table 1 and 2.

Hypothesis 1 was not even weakly supported (p = .276). The order in which the price series was presented to subjects had no statistically signiFicant effect on the mean lower price limit. However, the results where in the hypothesized direction. Subjects i;l the descending treatment reported a greater mean lower price limit ( x = $2186.40) than did subjects provided the price stimuli in ascending order ( x = $1963.60). When the random order treatment was included in the analysis, it was not found to be significantly different from he ascending.descending treatments at p < .30. Nonetheless, in absolute terms the random price treatment produced the highest mean lower price limit (x = $2265.20).

These results may seem to contradict those reported by Della Bitta and Monroe (1973). However, their prediction was not confirmed for the most expensive product of the eight products they tested --Sport Coat. This may suggest that the order effect may be product specific and may have differential effects depending on the price category (e.g., low, medium, high) occupied by the products under investigation.

Hypothesis 2 was not supported (x = 553) There was strong statistical support to suggest that the price series order effect has no effect on the upper price limit. The upper price limit was greater in the descending treatment (x = $3927.30) than in the ascending treatment (x = $3677.30). While these results were not statistically significant, they were in the direction hypothesized by Della Bitta and Monroe (1973). The mean upper limit under the random condition was not statistically significant. However, in absolute terms, the random price stimuli treatment produced the highest mean upper price limit ( x = $4026.10).

Hypothesis 3 was supported (p <.05). Subjects in the random price order condition (S.D. = $884.50) exhibited greater variability than in either the ascending (S.D. = 5401.80; p < .01) or descending (S.D. = $533.00; p <.05) conditions. Where subjects were presented the price series in random order, variability in the lower price limit was almost twice that exhibited in either the ascending or descending treatments. There was no significant difference (p >.10) in variability between the ascending and descending treatments. These findings suggest that the use of random price stimuli increases method variance and may have masked the effects expected by Fouilhe (1970), Raju (1977), Cox (1986), and Kosenko et al. (1988).

Hypothesis 4 was only partially supported (p < .10). There was greater variability in the upper price limit when subjects were provided the price stimuli in random order (S.D. = $1263.20) than when they received the price series in descending order (5931.10). Variability was not found to be different between the random and ascending treatments (S.D. = 51070.80) barely at p > .10.

While the results were mixed, variability was in the hypothesized direction. Unfortunately, these results could not be compared to that reported by Della Bitta and Monroe (1973) study (they did not report variances ) However, the results do suggest that the failure of Raju 1977, Cox 1986, and Kosenko et 21. (1983) to confirm some of their hypotheses may have been due to the additive effect of method variance of the price order effect.

In sum, the results indicate that the use of a random stimulus set increases variability in subject price judgements almost two-fold on both price limits. That increase in variability may be an alternative explanation for the failure to support hypotheses 1 and 2.

CONCLUSIONS

Researchers using random price stimuli may have inadvertently confounded the interpretation of the main effects of their manipulations. This study suggests that the use of random price stimulus may increase variability and thus mask the true effects under investigation. Moreover, the increase in variability may be attributed to method variance.

Equally important is the issue that the use of a large number of prices presented in random order this study provided a price series comprising 50 different prices) may contribute to an additional confound --subject fatigue (Kosenko 1987). Subject fatigue could also add to method variance. The effect of this potential confound should be tested. Such an examination would benefit not only price limits research, but also price-quality research.

The use of an ascending or descending price series relative to a random price series order may be a more palatable alternative despite the reservations of Della Bitta and Monroe (1973), and Rexeisen (1982). There was less variability in the lower and upper price limit scores when the price stimuli were presented in ascending and descending order then when the price stimulus were randomized. Most importantly, there was no significant difference in variability in either the lower or upper price limit between the ascending and descending price order treatments. This reduction in variability attends to the concerns expressed by Cox (1986), and Kosenko and Rah;z (1988) who reported high variability in the upper limit to the detriment of confirming their theoretically generated hypotheses.

Although this study makes use of a limited sample size, the results do generate some concern. This comes at a time when research on price limits seems to be escalating (e.g., Cox 1986; Fouilhe 1970; Raju 1977; Kosenko 1987; Kosenko and Rahtz 1988; Kosenko and Danes 1988). Unfortunately, since all but the first study have used a random price stimulus series they may have inadverately contributed to method variance an(l may have masked that which they had investigated.

The study does suggest that one source of the high variability in the price limits (particularly the upper limit) reported by Cox (1986), and Kosenko and Rahtz (1988) may have Seen due to the price order effect. To enhance external validity, the order effect needs to be investigated across different product categories, as well as among various populations.

TABLE 1

SAMPLE STATISTICS FOR LOWER AND UPPER PRICE LIMITS

TABLE 2

HOMOGENEITY TEST RESULTS

Further, this effect should be manipulated along with other product information cues used by consumers to make purchase decisions.

REFERENCES

Cox. Anthony D. (1986), "New Evidence Concerning Consumer Price Limits," in Advances in Consumer Research, Vol. 13, ed. R. Lutz, Association for Consumer Research. 248-971.

Della Bitta, Albert and Kent B. Monroe (1974), 'The Influence of Adaptation Level on Subjective Price Perceptions," in Advances in Consumer Research, Vol. 1, eds., Scott Ward and Peter Wright, Association for Consumer Research. 359-369.

Fouilhe, Pierre (1970), "The Subject Evaluation of Price," in Pricing Strategy, eds. Bernard Taylor and Gordon Wills, Princeton: Brandon Systems Press, 89-97.

Gabor, Andre and C. W. J. Granger (1964), "Price Sensitivity of the Consumer," Journal of Advertising Research, 4 (December), 4043.

Gabor, Andre and C. W. J. Granger (1966), "Price as an Indicator of Quality: Report on an enquiry," Economica. (February), 13- 70.

Helson, H. (1964), Adaptation-level Theory, New York: Harper and Row.

Monroe, Kent B. (1977), "Objective and Subjective Contextual Influences on Price Perception." Consumer and Industrial Buying Behavior, (eds.) Arch G. Woodside, Jagdish Sheth, and Peter D. Benneth, New York: North Holland, 287-86.

Monroe, Kent B. and M. Venkatesan (1969), "The Concepts of Price Limits and Psychophysical Measurement: A Laboratory Experiment." Proceedings, Fall Conference of American Marketing Association, 345-51.

Monroe, Kent B. and Susan M. Petroshius (1981), "Buyers' Subjective Perceptions of Price: An Update of the Evidence," in Perspectives in Consumer Behavior, 3rd. ed., eds. Tom Robertson and Harold Kassarjian (Glendale, Illinois: Scot, Breman and Co.), 43-55.

Kosenko, Rustan (1987), "A Multi-Method Experimental Investigation of the Effects of Market Price Knowledge on Acceptable Price Range,"Unpublished Dissertation, Virginia Polytechnic Institute and State University. Blacksburg, Virginia 24061.

Kosenko, Rustan and Don Rahtz (1988), "Buyer Market Price Knowledge Influence on Acceptable Price Range and Price Limits," ed. Michael J. Houston, TX, Association for Consumer Research.

Kosenko, Rustan, R. Krishnan and Jeffrey E. Danes (1988), " The Impact of Product-Related Information on Acceptable Price Range and Price Limits," in Proceedings AMA Summer Educators Conference, eds. Gary Frazier et al. (Chicago: American Marketing Association).

Perducci A. (1956), "Direction of shift in the judgement of single stimuli," Journal of Experimental Psychology, 51, 169-70.

Perducci A. (1954), "Learning variables in the judgment of single stimuli." Journal of Experimental Psychology, 70, 272-75.

Perducci A. and R. Hohle (1957). "Restriction of range in the judgement of single stimuli " American Journal of Psychology, 70, 277-75.

Raju, P.S. (1977), "Product Familiarity, Brand Familiarity and Price Influences on Product: Evaluations," in Advances in Consumer Research,, Vol. IV, ed. W. Perreault, Association For Consumer Research, 64-71.

Sherif C. (1963), "Social Categorization as a Function of Latitude of Acceptance and Series Range, Journal of Abnormal and Social Psychology, 2, 158-156

Sherif, Mustafer, and Carl I. Hovland (1961), Social Judgment: Assimilation and Contrast Effects in Communication and Attitude Change, New Haven, CT. Yale University Press.

Sherif, Mustafer, C. Sherif and R. Negergall (1965), Attitude and Attitude Change, Philadelphia: W. B. Saunders Co.

Stoetzel, Jean (1970), "Psychological/Sociological Aspects of Price," in Pricing Research, eds. Bernard Raylor and Gordon Wills, Princeton: Brandon Systems Press, 89-97.

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