Variety - seeking in product choice behavior : theory with applications in the food domain

The primary objectives of the present work are (1) to review the marketing and psychological literature on variety-seeking behavior, (2) to develop a formal model for variety-seeking in product choice behavior and (3) to investigate elements of the proposed model empirically. The present approach specifically focuses on temporal variety-seeking behavior and addresses some of the key issues that have received inadequate attention in previous work on temporal variety-seeking behavior. The main conclusions of the present work will be summarized and discussed in this chapter, and directions for future research in this area will be suggested.Limitations of previous work on variety-seeking behaviorDespite the fact that during the last few years variety-seeking behavior has received considerable research attention in the marketing literature, there are a number of issues that have not been adequately addressed. To a considerable extent this appears to be due to inadequate and inconsistent use of the terminology in this research area (McAlister and Pessemier 1982; Kahn, Kalwani and Morrison 1986). In particular the term variety-seeking behavior has been used rather informally to denote a number of different phenomena. This seriously hampers theoretical progress in this area as it makes it difficult to directly compare results from different studies and to integrate them into a comprehensive theory for varietyseeking behavior. Therefore, in the present work much attention is given to the terminology being adopted. The term variety-seeking behavior is reserved for those instances of consumer switching behavior that are motivated by the utility inherent in variation per se. The present approach thus explicitly distinguishes variety-seeking behavior from derived varied behavior , that is motivated by the more or less delayed consequences of switching behavior rather than the utility inherent in switching behavior per se.Apart from theoretical inconsistencies in the definition of variety-seeking behavior, much of the previous research has also fallen short in the measurement of this type of behavior. Despite the fact that the distinction between true variety-seeking behavior and derived varied behavior has played a prominent role in conceptual analyses of the phenomenon, both in the psychological (e.g. McReynolds 1971a; Deci 1975) and the marketing literature (McAlister and Pessemier 1982; Hoyer and Ridgway 1984; Raju 1984; Mazursky, LaBarbera and Aiello 1987), few studies on variety-seeking behavior that have incorporated it into empirical analysis. Again, the neglect of this distinction in empirical investigations of the phenomenon seriously hinders theoretical progress in the area, as many of the results attributed to variety-seeking behavior may be confounded by elements of derived varied behavior.In the marketing literature, variety-seeking behavior has been studied from two basic approaches. The implicit approach takes observed variation in purchase or consumption histories as a starting point of their attempts to derive insight into variety-seeking behavior. The explicit approach, on the other hand, takes the individual and psychological processes as a starting point of their analyses in an attempt to explain why and when variety-seeking behavior is likely to occur. Both of these approaches have specific strengths and weaknesses. One of the main strengths of the implicit approach is that these studies model variety-seeking behavior from "real-life" consumption data and thereby implicitly consider variety-seeking behavior in the broader context of consumer choice behavior. However, these studies have also specific problems associated with them. One of the most prominent weaknesses concerns the measurement of true variety-seeking behavior. Many of the models suggested within this approach do not allow for a formal distinction between true variety-seeking behavior and derived varied behavior. Therefore, the variety-seeking parameters obtained from these models primarily distinguish repeat purchasing from variation in behavior, without providing insight into the nature of the observed variation in behavior (true variety-seeking behavior versus derived varied behavior). As a result, many of these models are primarily descriptive in nature without providing a detailed insight into the nature of true variety-seeking behavior. The models developed within this approach are becoming increasingly sophisticated, progressing in a direction that in the long-run may allow for a better distinction between variety-seeking behavior and derived varied behavior.Studies within the explicit approach take the underlying psychological processes for variety-seeking behavior as their point of departure. Building on the psychological theories of exploratory behavior, most of these studies have focused on personality characteristics as an explanation for individual differences in variety-seeking behavior intensity. Only recently have these studies begun to consider choice-context related determinants of variety-seeking behavior. Both person-related and context-related determinants have been shown to influence the intensity of variety-seeking in product choice behavior. Thus, these studies are likely to provide a more detailed insight into the psychological processes that may explain why and when variety-seeking behavior will occur. However, studies within the explicit approach also have specific weaknesses associated with them. Many of these studies have used stated behavior rather than actual manifestation of variety-seeking behavior in product choice. Also, the fact than many of these studies have been conducted in controlled experimental settings makes it difficult to capture the phenomenon within the broader context of other consumer choice mechanisms that compete and interact with the desire for variety in determining actual choice behavior.Valid measures for variation in consumption behavior are a prime concern to both approaches to variety-seeking behavior. Although several measures have been proposed, the issue of their validity assessment has largely been ignored. This lack of established validity is problematic as it directly influences the validity of the results obtained, and renders a direct comparison of results from various studies very difficult. Chapter 5 critically reviews measures for variation in consumption that have been proposed in the economics and marketing literatures and provides an empirical investigation into their validity. It is shown that rather than relating to one single underlying construct (variation in consumption), the measures can more accurately be classified into two distinct categories: those that quantify variation at the product level versus those that take into account the attribute composition of the brands switched to and from.Main contributions of the present studyThe variety-seeking modelThe primary aim of the present study is to fill the gaps identified in previous work on temporal variety-seeking behavior. The variety-seeking model developed for this purpose explicitly distinguishes between a static and a dynamic component in consumer evaluation processes of product alternatives. In line with most of the previous research on the phenomenon, variety-seeking behavior is attributed to feedback mechanisms from previous consumption and purchasing, implying that it exerts its influence through the dynamic component of consumers' evaluation processes. The more static component, on the other hand, reflects the consumer's long-term preference for choice alternatives and captures both instrumental and hedonic product attributes.The variety-seeking model allows for a more detailed analysis of the underlying processes of variety- seeking behavior, which relate to changes in perceived hedonic value of choice alternatives under the influence of previous consumption or purchase behavior. Three such underlying psychological processes are identified, Boredom with the choice task is a product-specific decrease in perceived hedonic value of the previously chosen alternative. As a result, the attractiveness of the previously chosen alternative decreases relative to that of all other choice alternatives. In such a situation, another alternative, for which the consumer's unconditional or long-term preference is lower than for the previously chosen alternative, may conditionally become more attractive. Switching behavior in response to this process is a first type of variety-seeking behavior identified in the variety-seeking model (cf. Jeuland 1978). In addition to boredom as an item-specific phenomenon, perceived hedonic value may also decrease under the influence of previous consumption in an attribute-specific manner. Over time, consumers may get satiated not only with repeated consumption of the same item, but also more specifically with certain attributes repeatedly delivered by the product. This process is well documented for sensory attributes, where it is referred to as 'sensory specific satiety' (LeMagnen 1967; Rolls 1986). In such instances, the consumer's evaluative judgment of one or more hedonic attributes changes under the influences of previous consumption. This phenomenon of attribute satiation is identified as the second psychological process underlying variety-seeking behavior (cf. McAlister 1982).Whereas the previous two underlying processes emphasize the reduction in relative attractiveness of the previously chosen alternative vis-à-vis other alternatives, curiosity as an underlying motivation for variety- seeking behavior emphasizes the increase in absolute attractiveness of an alternative not chosen on the previous consumption occasion. Curiosity, the desire to close the information gap between what is known and what one wants to know (Loewenstein 1994), may increase the perceived value of one or more alternatives that have not been chosen on the previous occasion. Switching behavior instigated by the desire to solve product curiosity is identified as the third underlying process for variety-seeking in product choice behavior.The three underlying processes share an important characteristic, namely that they all relate to the stimulation level experienced in choice behavior. Each of these three psychological processes is the result of a discrepancy between the Actual Stimulation Level (ASL) experienced in life and the Stimulation Level that is Optimal (OSL) for the consumer in question. Boredom and attribute satiation reflect sub- optimal levels of stimulation experienced in life (ASL) and variety-seeking behavior in response to these processes is a means of bringing ASL into closer correspondence with OSL. Curiosity on the other hand, is characterized by a mildly supra-optimal level of ASL and variety-seeking behavior to solve curiosity is a means of reducing ASL to bring it into closer correspondence with OSL. Correspondence between OSL and ASL is associated with positive affect. This idea is central to the concept of "Value derived from variety" that summarizes the utility derived from the three types of variety-seeking behavior discussed above. As each of these processes aims at bringing ASL into closer correspondence with OSL, variety- seeking behavior is an inherently pleasurable activity.Variety-seeking behavior as a trade-offThe variety-seeking model states that in actual choice behavior, consumers base their choices on total expected value of consumption of an alternative. This total value assessment comprises a static and a dynamic component, referred to as expected value derived from product-related characteristics (reflected in the hedonic and instrumental long-term value of choice alternatives) and expected variety value (in response to boredom, attribute satiation and curiosity) respectively. In its basic form the variety-seeking model states that at choice occasion t, the consumer's decision to switch from alternative i consumed at t-1 to any other alternative j in the choice set depends on an implicit or explicit comparison of the total expected value of alternatives i and j. If the total expected value associated with consumption of alternative j is higher than that of consuming alternative i again, the consumer is expected to switch. In many instances, the consumer decision to switch or not will depend on a tradeoff between the two sources of total expected value. One of the contributions of the varietyseeking model is that it makes this trade-off explicit, thereby putting variety-seeking behavior into the broader context of consumer choice behavior rather than treating it in isolation.Variety-seeking behavior is conceived of as one of the consumer choice mechanisms that competes and interacts with other relevant choice mechanisms (summarized in value derived from product-related characteristics) in determining choice behavior. Only when the varietyvalue inherent in switching behavior is the decisive motivator for variation in behavior, is the behavior referred to as true variety-seeking behavior. When the value derived from productrelated characteristics is decisive, variation in behavior is referred to as derived or extrinsically motivated varied behavior.By considering variety-seeking behavior in the broader context of consumer choice behavior, the formulation of the variety-seeking model not only allows for a formal classification of observed variation in behavior as either variety-seeking behavior or derived varied behavior, it also provides an explicit framework to structure determinants in choice behavior that may either stimulate or reduce the occurrence of variety-seeking in actual product choice behavior.Determinants of variety-seeking behaviorIn terms of determinants of variety-seeking behavior, the present work extends previous work by considering product-related determinants and their interaction with the person-related determinants are also considered. Central to the hypothesized interaction between person- and product-related determinants is the notion that product-related determinants operate as controlling factors on consumer choice behavior. Building on cognitive evaluation theory (e.g. Deci and Ryan 1985), the variety-seeking model suggests that these product-related determinants put extrinsic pressure on consumer choice behavior, thereby pressuring choice in a certain direction and reducing the consumer's perceived freedom in choice. As a consequence, the controlling factors are hypothesized to limit the expression of the intrinsic desire for variety in consumers who otherwise would be quite likely to engage in varietyseeking behavior. The present approach suggests that consumers with a high intrinsic need for variety will be particularly sensitive to controlling factors in choice behavior, and in addition to main effects for person-related and product-related determinants, hypothesizes that the product-related determinants will interact with consumers' variety- seeking tendency.Person-related determinants of variety-seeking behaviorTwo hypotheses regarding the main effect of person-related determinants of variety-seeking behavior were empirically tested. The first hypothesis states that consumers with a higher variety-seeking tendency are more likely to engage in variety-seeking behavior than those with a lower variety-seeking tendency. A domain-specific scale, VARSEEK, was developed that specifically taps variety-seeking tendency with respect to foods. The construct validity of the measurement instrument was investigated extensively and confirmed. The nomological validity of the VARSEEK-scale was also confirmed, both in terms of more general personality scales to which it is hypothesized to relate (general OSL and OSL in the consumer context) and in terms of manifestations of variety-seeking behavior (both self-report measures and actual behavior). VARSEEK's relationship with variety-seeking behavior confirms that variety-seeking tendency is an important determinant of variety-seeking behavior. Confidence in VARSEEK's predictive validity was further enhanced in a large scale study on consumer panel data that explicitly allowed for the distinction between true variety-seeking behavior and derived varied behavior. On these "real-life" choice data, the role of VARSEEK as a determinant of variety-seeking behavior was confirmed, both relative to repeat purchases and derived brand switches.Our second hypothesis with respect to person-related determinants of variety-seeking behavior states that VARSEEK, as a measure specifically tapping consumers' variety tendency with respect to foods, should have higher predictive validity than both (a) general measures for OSL and (b) consumer specific measures for OSL, when the purpose is to predict variety-seeking in food consumption. The rationale behind this hypothesis is the "principle of measurement correspondence" (Ajzen 1987), which states that higher predictive validity will be achieved when the predictor concept (e.g. personality variables) is measured at the same level of specificity as the behavior purportedly being predicted. Consistent support was found for VARSEEK's predictive superiority vis-à-vis a general personality measure for OSL (CSI; Steenkamp and Baumgartner 1995). However, only weak support was found for VARSEEK's predictive validity vis-à-vis a measure for OSL in the consumer context (EBBT; Baumgartner and Steenkamp 1994), in particular with respect to EBBT's subscale for Exploratory Acquisition of Products (EAP). In its definition, "a consumer's tendency to seek sensory stimulation in product purchase through risky and innovative product choices and varied and changing purchase experiences" (Baumgartner and Steenkamp 1994: 6), the EAP-subscale bears high similarity with our concept of variety-seeking tendency. Empirical results reveal that the two constructs are closely related, as is evidenced by their bivariate correlation of 0.670 (p < .001). Although VARSEEK has a slight predictive advantage when the purpose is to predict variation in food choice behavior, the size of this advantage is not likely to compensate for the food-specific nature of the VARSEEK-scale. Although future confirmation is required, it seems that the more general nature of EAP would make this subscale a promising alternative for the VARSEEK-scale when the purpose is to predict variety-seeking in product choice behavior outside the domain of foods.Product-related determinants of variety-seeking behaviorContext factors as a determinant of variety-seeking behavior intensity have only recently begun to attract attention in the marketing literature. Examples include purchase strategy (Simonson 1990), display format (Simonson and Winer 1992), consumers' mood during decision making (Kahn and Isen 1993) and context variation (Menon and Kahn 1994). The present study extends this stream of research by deriving and testing specific hypotheses for what we refer to as "product-related" determinants of variety-seeking behavior. These hypotheses follow from the variety-seeking model's assumption that whether or not varietyseeking behavior will occur depends of the magnitude of the variety value inherent in switching behavior relative to the magnitude of the difference in value derived from productrelated characteristics (hedonic and instrumental value) associated with the alternatives switched from and to. Product-related determinants of variety-seeking behavior may exert their influence through both of these value-components. Consequently, the variety-seeking model suggests three classes of product-related determinants:1. those that affect value derived from variety2. those that affect the difference in value derived from product-related characteristics3. those that simultaneously affect value derived from variety and difference in value derived from product-related characteristicsThe present work further extends previous work in this area in that it not only considers the main effect of these product related determinants of variety-seeking behavior, but in addition hypothesizes that these product-related determinants will interact with the person-related determinants. Consumers low in variety- seeking tendency are not likely to derive value from variety and thus are not likely to engage in variety- seeking behavior, irrespective of the product-related characteristics. Consumers high in variety-seeking tendency, on the other hand, are quite likely to express their intrinsic desire into actual variety-seeking behavior unless product-related determinants exert a controlling effect on the choice task.Several hypotheses with respect to product-related determinants of variety-seeking behavior and their interaction with variety-seeking tendency were empirically tested in a large-scale consumer panel. In addition to recording brand choice behavior over time, the data collection procedure also identified underlying motivations for brand switching, thus allowing for a distinction between true variety-seeking behavior and derived varied behavior. The data collection procedure adopted thus permitted a test of the hypothesized determinants of variety-seeking behavior relative to both repeat purchasing and derived varied behavior. Empirical support was found for the hypotheses with respect to low product-category involvement, small perceived differences among the choice alternatives, low brand loyalty and high hedonic features as product-related determinants that stimulate variety-seeking behavior vis-à-vis repeat purchasing and derived varied behavior. All of these characteristics distinguished variety-seeking behavior in a statistically significant sense, with the exception of small perceived differences relative to derived switching behavior. Mixed support was found for the hypothesized interactive effects. Consistent support was found for the hypothesized interactive effect between variety-seeking tendency and low involvement. In addition, relative to repeat purchasing, support was found for the hypothesized interactive effect between variety-seeking tendency and high hedonic features of the product category as a determinant of variety-seeking behavior. The direction of the interactive effect between variety-seeking tendency and small perceived differences was opposite to the hypothesized direction. Overall, the results of the present study provide considerable evidence for the variety-seeking model's central notion that variety-seeking behavior is under the joint influence of person-related determinants, product-related determinants and their interaction.Concluding remarksMeaningful investigation of variety-seeking behavior intensity requires an a priori specification of a set of choice alternatives among which the phenomenon is being investigated. This researcher-based demarcation of the relevant choice set is to some extent arbitrary and can be made at different levels of abstraction of product definitions. For example, variety-seeking behavior may be investigated at the level of different product types within a particular product category, such as different types of vegetables, fruits, drinks (coffee, tea, beer, soft drinks etc), or desserts (e.g. ice cream, flavored yogurt etc). However, the phenomenon can also meaningfully be explored at the level of different items within a product type (e.g. different flavors of yogurt, different brands of beer etc.). The level of abstraction at which variety-seeking behavior is investigated will influence the relative importance of the different underlying psychological processes for variety-seeking behavior and consequently the variety-seeking behavior intensity observed. The varietyseeking model attempts to account for these differences in variety-seeking behavior intensity through the incorporation of product-related determinants that may generalize across product levels. Perceived differentiation among the alternatives in the choice set to a large extent accounts for these differences. For example, at the level of product types within a particular product category, alternatives in the consumer's choice set are likely to have a considerable degree of perceptual variation, while all being capable of satisfying the identified need adequately (i.e. low preferential differentiation). At this level, all three underlying psychological processes for variety-seeking are likely to contribute to the stimulation of variety-seeking behavior. On the other hand, at the level of brands within a narrowly defined product type (e.g. coffee or cigarettes), the perceived perceptual differentiation among alternatives is likely to be considerably smaller. In such instances, attribute satiation is less likely to be an important underlying motivator for variety-seeking behavior, simply because the items in the (researcher-defined) choice set only have a very limited capacity of relieving this attribute satiation. Consequently, at this level of product definition, variety-seeking behavior will mainly result from boredom and curiosity and variety-seeking intensity may be lower. On the other hand, for some products, perceptual differentiation within the product type can be considerable even when preferential differentiation is small. For example, this would be the case for different flavors of a product type, such as flavored yogurt. In those instances when perceptual differentiation will be higher, all three underlying processes for variety-seeking behavior may contribute and variety-seeking behavior intensity would be expected to be higher.Suggestions for future researchThe present study filled in a number of gaps in the theoretical account for variety-seeking behavior. The variety-seeking model and the empirical work evolving from it suggest a number of promising avenues for future research in the area of variety-seeking behavior.As most other contributions to the variety-seeking literature, our variety seeking model only takes into account the influence of the most recent consumption experience on present choice behavior (i.e. a first-order process). Within the implicit approach to variety-seeking behavior, higher-order models have been proposed (e.g. Jeuland 1978; McAlister 1982; Kahn, Kalwani and Morrison 1986; Lattin 1987; Bawa 1990), but not in combination with the strict separation between true variety-seeking behavior and derived varied behavior. Future investigations might focus on the extension of the present model to higher-order feedback effects, as well as on the adequacy of the present first-order formulation Vis-à-vis extended model formulations. The strict distinction between true variety-seeking and derived varied behavior is of crucial importance to the power of this comparative test as marketing-mix influences on variation in behavior tend to drive observed purchase behavior toward lowerorder (Kahn, Kalwani and Morrison 1986).Three underlying processes for variety-seeking behavior were identified. Future research might explore each of these underlying processes in more detail as well as their relative impact on variety-seeking behavior in different choice contexts. Boredom with the choice task and attribute satiation are psychological processes that relate to sub-optimal levels of stimulation. More insight is needed into these processes, in particular with respect to the type of products and product attributes that are more likely to stimulate boredom and attribute satiation than others, and the managerial implications in terms of product-line development.The underlying process of curiosity in consumption behavior might be another fruitful avenue for future work. This motivation has direct marketing implications in terms of new product development and product communication. Curiosity may be an important consumer motivation that can be appealed to in an attempt to attract customers for a new product introduction. in this respect, curiosity poses an interesting paradox: it may stimulate product trial, but it also is relatively easily satisfied. After an initial product trial, it will be hard to retain curiosity-motivated customers for the new product. Relevant issues that willrequire further investigation are how "new" or different a new product introduction should be to stimulate variety-seeking behavior without casting doubt on the product's extrinsic value, and which characteristics the new product should have in order to extend curiosity-motivated product trial into a basis for repeat purchasing or brand loyalty. As curiosity results from an information gap between what is known and what one wants to know (Loewenstein 1994), this type of research may build on insights on consumer knowledge development and knowledge structuring.We suggested several product-related determinants of variety-seeking behavior that have not yet been subjected to empirical investigation. Future research will be needed to empirically test product-related determinants developed in the present work and to suggest additional determinants that may contribute to the explanation of the circumstances under which variety-seeking behavior is more or less likely to be a determinant factor in consumer choice behavior. The product-related determinants that affect variety-seeking behavior intensity simultaneously through variety value and through the difference in value derived from product-related characteristics would be particularly interesting research topics. This category of product-related determinants includes the related concepts of product-category involvement, perceived differences among alternatives and perceived risk. Depending on the type of attributes involves, perceived differences among choice alternatives are hypothesized to simultaneously stimulate and reduce variety-seeking behavior intensity, where the balance between these two effects depends on product-related characteristics such as product-category involvement and perceived risk. More research is needed to provide a more detailed insight into this joint effect of perceived differentiation among choice alternatives.The variety-seeking model is a conceptual model that allows for the integration of previous research findings on determinants of variety-seeking intensity by expressing them in terms of their effect on two value components: value derived from product related characteristics and value derived from variety. From our variety-seeking model specific hypotheses were developed for product-related determinants. Although not specifically addressed in the present work, hypotheses concerning marketing-mix influences on varietyseeking intensity may similarly be expressed in terms of the two value-components of the variety-seeking model. For example, the variety-seeking model suggests that controlling factors may limit the expression of variety-seeking behavior among consumers who are high in variety-seeking tendency. Thus, marketing efforts that aim at eliminating these controlling factors would be expected to increase variety seeking behavior, whereas those that impose controlling factors would be expected to stimulate variety seeking behavior. Consider, for example, a major brand. For such a brand, the variety-seeking model suggests that varietyseeking behavior among current users might be discouraged by marketing efforts (e.g. through product positioning or advertising) that emphasize the switching costs in terms of the differential in value derived from product-related characteristics. Minor brands in the same market may stimulate consumers to switch away from the major brand by marketing efforts that reduce the switching costs in terms of the differential in value derived from productrelated characteristics (e.g. price discounts) and/or by appealing to the intrinsic desire for variety (e.g. emphasizing the novelty and change). Future research will be needed into the effect of marketing-mix variables on the intensity and direction of variety-seeking behavior. As specific hypotheses directly follow from the variety-seeking model, confirmation of these hypotheses can provide further support for the validity of the variety-seeking model. Again, empirical assessment of hypothesized effects of marketing-mix variables will critically depend on the separation between true variety-seeking behavior and derived varied behavior.We explicitly considered variety-seeking behavior as one of the many choice mechanisms relevant for consumer choice behavior and developed specific hypotheses with respect to why and when consumers' variety-seeking tendency is likely to be a determinant consideration in choice behavior Vis-à-vis repeat purchasing and derived varied behavior. A fruitful avenue for future research would be to consider the situations under which different forms of derived varied behavior, such as situational and normative considerations, price- and product-related considerations, and habit reversion, are more or less likely to be determinant factors in consumer choice behavior. Together with the results of the present work, this line of research would provide a more detailed picture of consumers' trade-offs in the complex phenomenon of consumers' product choice behavior.