Marketing Letters 5:1, (1994): 91-100 © 1994 Kluwer Academic Publishers, Manufactured Jn the Netherlands.

"Out of Sight, Out of Mind": Pantry Stockpiling and Brand-Usage Frequency BRIAN WANSINK Assistant Pr(~fessor of Business Administration, Tuck School, Dartmouth College, Hanover. NH 03755, (603) 646-3981 ROH1T DESHPANDE* E. B. Osborn Professor of Marketing, Tuck School, Dartmouth College, Hanover, N i l 03755, (603) 646-1336

Key words: stockpiling, usage frequency, salience, versatility

Abstract Both researchers and brand managers have suggested that price promotion-induced stockpiling can increase a household's usage frequency of a product. Empirical findings, however, contradict this relationship. In reconciling this inconsistency, laboratory results reported in this paper suggest that stockpiling may have the greatest effect on a product's usage frequency when usage-related thoughts about the product are highly salient. These results also suggest that when stockpiling stimulates usage frequency, it can do so by increasing perceptions of a product's versatility. These findings have implications for the advertising versus promotion debate. They suggest that consumer promotions and advertising might play a j o i n t and complementary role in increasing product usage: promotions by encouraging stockpiling, and advertising by building the asage-related salience needed to deplete the stockpiled inventory.

Brand managers are witnessing an unprecedented emphasis on consumer sales promotions and on the pantry stockpiling such promotions encourage (Helsen and Schmittlein, 1992). The assumption by some of these managers is that pantry stockpiling, in turn, increases a household's usage frequency of a product. Nevertheless, there is little empirical evidence of any such increase (Btattberg and Neslin, 1990, p. 134). The few studies in this area show either no support (Moore and Winer, 1978) or limited, inferential support (Ward and Davis, 1978). To reconcile this inconsistency, we propose a behavioral model of this relationship and describe the results from a laboratory experiment that is designed to demonstrate the viability of this model. Finally we discuss theoretical contributions along with the implications for the advertising versus sales promotion debare.

*The authors are grateful to the Marketing Science Institute and to the Amos Tuck School of Business Administration at Dartmouth College for their financial support of this project. They are additionally grateful to Scott Neslin, Jim Lattin, and Jeff Inman for helpful comments on earlier drafts.

92

BRIAN WANSINK AND ROHIT DESHPANDÉ

1. Background: stockpiling and usage 1.1. Prior theoretical research Past research on the relationship between stockpiling and usage has implied a cost-based model of inventory depletion that focuses on purchase costs and on carrying costs (Assuncao and Meyer, 1993). For instance, it has been analytically suggested that price-promoted products tend to be stockpiled and used more frequently than products that are perceived as more expensb¢e because they were not bought on promotion (Meyer and Assuncao, 1990). In addition, the results of a household panel study suggested that carrying costs (such as inventory space constraints) may discourage extreme levels of inventory (Blattberg, Peacock, and Sen, 1976), leaving consumers to frequently use a stockpiled product until its inventory level returns to an acceptable level. A concern with this cost-based model is that it assumes that consumers are attentive to the prices they pay for products and that they are attentive to their household inventory levels. Such assumptions may not be valid. First, it is unclear whether consumers can accurately estimate the price of stockpiled products, as they frequently cannot eren recall a product's price immediately after its purchase (Dickson and Sawyer, 1990). Furthermore, recent evidence suggests that many people tend to underestimate their inventory level of a product (Wansink and Neslin, 1994). These considerations raise the question of whether we should expect pantry stockpiling to have any noteable effect on usage frequency.

1.2. Prior empirical research EmpiricaI evidence for a positive relationship between stockpiling and usage is not conclusive. Ward and Davis (1978) observed that increases in coupon use also increased the volume of orange juice a family purchased. The inference was made that increases in purchase frequency necessarily imply increases in usage frequency. Yet we know only that orange juice was purchased, not that it was actually used. As Blattberg and Neslin (1990, p. 130) state, "It might be relatively easy to ger consumers to stockpile tuna fish or soup or a cake mix, hut getting the consumer actually to use these products is a different problem." Blattberg and Neslin's assertion is supported by the results of a field experiment conducted by Moore and Winer (1978), who gave consumers various-sized bottles of soff drinks and various-sized boxes of spaghetti to use at home. They hypothesized that the availability of a product should produce more frequent use. This, however, did not occur.

2. Hypotheses: stockpiling and usage salience Ward and Davis (1978) contend that when orange juice is stockpiled, it is more freely used in place of water, coffee, tea, milk, or soft drinks. In effect, orange

PANTRY STOCKPILINGAND BRAND-USAGE FREQUENCY

93

juice is used more frequently because it is seen as a more versatile, substitutable product. Orange juice, however, might not be the most appropriate product on which to base conclusions about stockpiling and usage. When in a liquid form, orange juice is perishable, needs refrigeration, and usually assumes a visually prominent position in most refrigerators. In short, orange juice might be a frequently substituted beverage because a person's thoughts about drinking it are frequently made salient since (1) they know it is perishable and taust be used and (2) they see the container every time they open the refrigerator. This orange juice example, hence, illustrates a key point: Before usage can occur, usage-related thoughts about a product must be salient, Such usage-related thoughts are situation-specific and will affect choice sets (Shocker, Ben-Akiva, Boccara, and Nedungadi, 1991). In short, when usage-related thoughts are generated about a stockpiled product, the product may be used more frequently if it is perceived as versatile and substitutable. This usage-salience model can exist independently of the cost model of inventory depletion that was described earlier. Hence, we would expect that When a product is both stockpiled and when usage-related thoughts of it are made salient, a person will use the product more frequently than if it is stockpiled but not salient or if it is salient but not stockpiled. Support of this usage-salience model would also indicate why past empirical studies, such as the work of Moore and Winer (1978), have failed to show significant results. If usage-related thougbt s about the product were not highly salient, we would not expect significant increases in usage. That is, if the product is out of sight, it is out of mind. This hypothesis also complements Assuncao and Meyer's (1993) argument that stockpiling influences usage primarily through its effect on perceived price. Perceived price will not be an issue unless the product has been stimulated to salience and is being considered for use.

3. Methodology 3.1. Subjects and design Subjects were recruited through eight Parent Teacher Associations (PTAs). A $6 donation was given to each PTA for each member who participated in the study. Of the 219 people who participated, the only individuals selected for the study were those who had eaten (or served) canned soup at least once in the past year. Of the 191 subjects who had done so, 87 percent were between the ages of 30 and 45, all were primary meal planners (92 percent were female), and 32 percent were employed outside the home. Their educational background was heterogeneous. The study used a 2 × 2 between-subjects design where product-usage salience (low-usage-related salience and high-usage-related salience conditions) was crossed with the stockpiling and nonstockpiling conditions. Subjects were randomly assigned to the experimental conditions.

94

BRIAN WANS1NKAND ROHIT DESHPANDÉ

3.2. Procedure and treatments Subjects gathered in groups of between 11 and 35 at the schools where their respective PTAs met. They were asked to take alternate seats in classrooms, and they were told that they would be asked a variety of questions dealing with issues ranging from home economics to leisure time pursuits. Each subject was then given a booklet with instructions that manipulated their usage salience of canned soup. Usage-related salience was manipulated by asking subjects in the highusage-related salience condition to describe the last time they ate or served canned soup and the thoughts they had when doing so (cf. Fazio and Zanna, 1981). Subjects in the low-usage-related salience condition were instead asked two questions that involved describing their last experience with an unrelated topic (picnicking in a public park). After finishing this booklet, subjects in the nonstockpiling condition were simply told, "We would now like to ask you some questions about canned soup." Subjects in the stockpiling condition were told, "We would now iike to ask you some questions about canned food. As you answer these questions, imagine that in the back of your pantry you have numerous cans of food that you bought long ago, such as canned vegetables, canned soup, canned tuna." To make their inventory levels vivid, subjects in both stockpiling and the nonstockpiling conditions were asked to write down the number of cans of soup that they visualized having in inventory. (As described in the following section, the manipulation of usage-related salience and of stockpiling both had their intended effects on the relevant manipulation-check questions). Subjects were next asked their usage intentions for the upcoming month and asked questions about their attitudes (ninepoint Likert scales), their beliefs (nine-point semantic differential scales), and their past usage frequency of canned soup.

4. Results

4.1. Manipulation checks Both manipulations were successful. The stockpiling manipulation was successful in that the subjects in the stockpiling condition were more likely to say they imagined their kitchen to be "stocked up with canned soup" when compared with those subjects in the nonstockpiling condition (F1.187 = 8.3; p < .01). In particular, they imagined themselves having an average of 7.2 cans in their pantry as compared to the 6.1 cans that subjects in the nonstockpiling condition had imagined (F1.187 = 4.2; p < .01). The usage-related salience manipulation was successful because subjects in the high-usage-related salience condition were more likely to have thought of "a specific instance in which they served canned soup" when compared with those subjects in the low salience condition (Fl.187 = 14.3; p < .01), and they were also

PANTRY S T O C K P I L I N G A N D B R A N D - U S A G E F R E Q U E N C Y

95

more likely to have thought of "specific uses for soup" (ff'l,J87 = 8.6; p < .01). As we would expect, the stockpiling manipulation had no effect on either of these indicators of usage-related salience.

4.2. Stockpiling and usage ß'equency We hypothesized that when a person's usage-related thoughts about a stockpiled product are made salient, usage intentions will be higher than if the product were neither stockpiled not salient. As can be seen in Table 1, this hypothesis was supported. Subjects who visualized a stockpiled pantry and who had high usagerelated salience estimated they would use approximately twice as many cans each month (X = 8.0 c__ans) when compared with subjects with low levels of usagerelated salience (X = 3.7 cans)_)or when compared with subjects who had nonstockpiled levels of inventory (X = 4.7 cans). This interaction was significant for both the number of cans they expected to use in the upcoming month (F~,~s7 = 6.3; p < .01), as weil as for the number of occasions in which they estimated they would eat soup (FL~~7 = 6.0; p < .0l). These results remained significant when family size and prior usage frequency were included as covariates. If we focus only on the two columns in Table I where usage-related salience is high (columns 2 and 4), we see that subjects in the stockpiling condition more strongly believed (6.4 versus 5.2) that "canned soup goes well w/th õther foods" (Fj,94 = 5.6; p < .05) and more strongly believed (7.5 versus 6.5) that "canned soup can be eaten at any time" (FL94 = 5 . 7 ; p < .05). It is important to understand that stockpiling only increased these perceptions of canned soup's versatility when usage-related thoughts were highly salient. When usage-related thoughts about soup were not highly salient, stockpiling had no statistically significant impact on these perceptions of versatility (6.3 versus 5.8; 7.2 versus 6.8). These results suggest that perceptions about a product's versatility might mediate usage. This was examined using tests of mediation, which regressed the dependent variable - the number of cans u s e d , against the two measures of versatility and against indicator variables that accounted for the four (2 x 2) experimental conditions. The adjusted R 2 from this model was then compared with the adjusted R 2 of a model that did not include these measures of versatility. In the full model, the R 2 increased from .28 to .37. A partial F test showed this increase to be statistically significant (p < .01), indicating that perceptions of versatility can mediate the number of cans these subjects intended to use. This effect of versatility on one's usage appears to exist independently of one's perceived price of the product. When the two measures of price perception were included in the previously specified regression model, neither made a statistically significant contribution. Given the results in Table 1, this should not be surprising. Contrary to the assumption that stockpiling decreases one's perceptions o f a product's price, perceptions of price were unaffected by stockpiling. Last, although perceptions of versatility are enhanced by stockpiling, brand

96

BRIAN W A N S I N K A N D R O H I T D E S H P A N D É

~

6)

..= e-~

eq

eq

e-

¢)

.8 e-

¢) e~

.o .--õ~

e-

o~

z~

~"

eq

eß

e-ù

E.

Z

Z

eß

PANTRY S T O C K P I L I N G A N D B R A N D - U S A G E F R E Q U E N C Y

~.

"2.

--.

,w

t"-

e-,

2 o z~ ~--Jl

=o ~i ~~

a,,

~D

"6 I~ Ir v" v" Z

9"7

98

BRIAN WANSINKAND ROH1TDESHPANDÉ

attitude appears to be instead influenced by usage-related salience ( E l , i s 7 = 5 . 1 ; p < .05). Such a result is consistent with research in social psychology that shows that salience can influence one's general affect for an object by strengthening beliefs about attributes (Tesser and Conlee, 1975).

5. Discussion

To examine this stockpiling-usage relationship we used a laboratory design to control the internal validity of the experiment rather than the external validity of it. This enabled us to complement previous research (e.g., Moore and Winer, 1978). However, before we can draw any conclusions from this study, we need to note how the limitations of the research design constrain the generalizability of these findings. Because both of the key manipulations (stockpiling and salience) were operationalized in a laboratory environment, their effect in a more rigorous inhome usage context still needs to be confirmed. Furthermore, we can only speculate about the effect of price on usage since price was not directly manipulated in this experiment. Last, the results of this study pertain only to products that are easily stockpiled and have elastic utilities (i.e., products that can be substituted for others with little difficulty). Because of these limitations, out results are suggestive rather than conclusive. They suggest that usage-related salience has an important influence (along with stockpiling) on the usage frequency of a product. To summarize the key findings of this study: (1) stockpiling appears to have the greatest influence on usage ffequency when usage-related thoughts about a product are concurrently salient; (2) stockpiling can influence usage by increasing a favorable user's perception of the product's versatility; and (3) such effects may occur independently of any effect that price perceptions may have on usage frequency.

5.1. Implications for consumer prornotion and advertising The results of this paper underscore the role that usage-related salience might play in helping clear a stockpiled product out of household inventory. Since usagerelated salience can be stimulated by usage-oriented advertising (Wansink and Ray, 1992), this model has direct relevance for the advertising and promotion debate. Advertising campaigns might use a message strategy and a media plan to stimulate usage-related thoughts, but these thoughts will not increase a customer's usage frequency if the product is not available in inventory. To the degree that sales promotions encourage the stockpiling of products, they create the opportunity for such increased usage. In summary, both advertising and consumer promotions appear to play a strong role in stimulating usage. When sales promotions encourage stockpiling, they increase one's opportunity to use a product. When advertising encourages usage-

PANTRY STOCKPILINGAND BRAND-USAGEFREQUENCY

99

related thoughts about the product, it increases the likelihood that one will consider using the product in a particular situation.

5.2. Pt~)positions for future research The findings of this exploratory study highlight how usage-related salience may drive the usage of a stockpiled product. Future research should examine usagerelated salience in greater detail. We might expect that it can be internally stimulated by how frequently the product is used and by how recently it was last used. Usage-related salience might also be externally stimulated by marketing communication efforts or by the exposure to the product that occurs as one "forages" in the kirchen. Hence, we can suggest two propositions:

Proposition 1: Stockpiling a product will increase its usage frequency if a. the product is one that is frequently consumed; b. the product has been recently consumed; c. the household has been frequently exposed to marketing communication regarding the product; d. the household has been recently exposed to marketing comunication regarding the product; and e. the product is stored in a salient location (such as on a table or counter or in the front portion of a cupboard or refrigerator). The effect that stockpiling has on usage is product-specific. The product-related factors that mediate the usage of stockpiled products also need to be examined in more detail. For instance, if perishable products tend to be stored in a salient locatiõn (see Proposition Re) and consumed with greater frequency and recency, their usage should be more dramatically influenced by stockpiling than nonperishable products. Similarly, we expect that the usage of highly substitutable products will be more influenced by stockpiling than products that have a narrow function, such as disposable diapers. This perceived substitutability of a product should be even further enhanced if a product's purchase price is perceived as relatively low and if the storage costs associated with it are perceived as relatively high. Hence, we expect

Proposition 2: Stockpiling a product will increase its usage frequency if the product is a. b. c. d.

perceived as widely substitutable, highly perishable, perceived as having been bought on promotion, and perceived as having high storage costs.

100

BRIAN WANSINK AND ROHIT DESHPANDÉ

As Helsen and Schmittlein (1992) noted, it is critical to understand what factors influence the relationship between a household's stockpiling of a product and their usage frequency of it. T/bis study indicates that orte of these factors - usage-related salience - needs to be considered in the specification of future stockpiling models to avoid having the behavioral relevance of these models be "out of sight, out of mind."

References Assuncao, Joao, and Robert J. Meyer. (1993). "The Rational Effect of Price Expectations on SalesPrice Relationships," Marketing Science, forthcoming. Blattberg, Robert C., and Scott A. Neslin. (1990). Sales Promotion: Concepts, Methods, and Strategies. Englewood Cliffs, NJ: Prentice-Hall. Blattberg, Robert C., Peter Peacock, and Subrata K. Sen. (1976). "Identifying the Deal Prone Segment," Journal of Marketing Research 15 (December), 143-154. Dickson, Peter R., and Alan G. Sawyer. (1990). "The Price Knowledge and Search of Supermarket Shoppers," Journal of Marketing 54 (July), 42-53. Fazio, Russell H., and Mark P. Zanna. (1981). "Direct Experience and Attitude-Behavior Consistency," In L. Berkowitz (ed.), Advances in Experimental Social Psychology (vol. 14). San Diego, CA: Academic Press. Helsen, Kristiaan, and David C. Schmittlein. (1992). "Some Characterizations of Stockpiling Behavior Under Uncertainty," Marketing Letters 3(1), 5-17. Meyer, Robert J., and Joao Assuncao. (1990). "The Optimality of Consumer Stockpiling Strategies," Marketing Scienee 9 (Winter), 18--41. Moore, William L., and Russell S. Winer. (1978). "An Experiment to Determine the Effects of Package Size on Consumption." Working paper, Graduate School of Business, Columbia University, New York. Shocker, Allan D., Moshe Ben-Akiva, Bruno Boccara, Prakash Nedungadi. (1991). "Consideration Set Influences on Consumer Decision-Making and Choice: Issues, Models, and Suggestions," Marketing Letters 2(3), 181-197. Tesser, Abraham, and M.C. Conlee. (1975). "Some Effects of Time and Thought on Attitude Polarization," Journal of Personality and Soeial Psychology 31 (June), 262-270. Wansink, Brian and Scott A. Neslin. (1994). "Risk Aversion and the Underestimation of Household Inventory." Working paper, Amos Tuck School of Business Administration, Dartmouth College, Hanover, NH. Wansink, Brian, and Michael L. Ray. (1992). "Estimating an Ad's Impact on One's Consumption of a Brand," Journal ofAdvertising Research 32 (May-June), 9-16. Ward, Ronald W., and James E. Davis. (1978). "A Pooled Cross-Section Time Series Model of Coupon Promotions," American Journal ofAgricultural Eeonomics 60 (November), 393-401.