How much was your shopping basket? Working memory processes in total basket price estimation

Although much attention has been paid to recall of a single price, research is lacking in understanding the process of how consumers estimate the total price of a shopping basket. Drawing on research on numeric cognition, memory processes, and mental accounting, we show in five studies that the accuracy of total price estimation as well as the timing of such estimation is systematically influenced by several factors. We find that the length (in syllables) of the prices in the basket and the attention that consumers pay to the prices influence the accuracy of the calculation of the total basket price. Furthermore, our studies also show that the timing of the calculation is influenced by the nature of the items in the basket (i.e., unrelated vs. complementary items). © 2009 Society for Consumer Psychology. Published by Elsevier Inc. All rights reserved. A shopper enters a clothing store and buys a pair of jeans for $108 and a shirt that would go perfectly with them for $49. Another shopper enters a discount store and purchases a fishing pole for $108 and a writable DVD pack for $49. Later, their friends ask them how much they spent on their shopping trip and both shoppers attempt to estimate the total price. When will they calculate the total price of their basket? At the time of the purchase, or when they are asked later? Will the timing of the calculation influence price estimate accuracy? What factors might influence their estimates' accuracy? Price estimation can be a complex process. Prices, like all other marketplace cues, are not processed in isolation; they are processed in a context—for instance, in conjunction with the prices of other items (e.g., buying a shirt and a pair of jeans that fit well together vs. buying a fishing pole and a writable DVD pack). Price recall has been a topic of concern in marketing, but little attention has been paid to the factors influencing estimates of the total price of a basket of items or to other potential contextual influences on price estimation and recall (Dickson & Sawyer, 1990; Estelami, 2003; Estelami&Lehmann, 2001;Monroe&Lee, 1999;Vanhuele & Drèze, 2002; Vanhuele, Laurent, & Drèze, 2006). ⁎ Corresponding author. E-mail addresses: [email protected] (D. Luna), [email protected] (H.-M.(C.) Kim). 1 Both authors contributed equally to this research. 1057-7408/$ see front matter © 2009 Society for Consumer Psychology. Publish doi:10.1016/j.jcps.2009.03.003 A systematic understanding of the factors surrounding total price estimation is important because most consumers buy multiple items in a single shopping trip (Manchanda, Ansari, & Gupta, 1999) and thus being aware of the total amount spent influences their future spending decisions more than remembering the price of only one item, especially because most consumers do have a budget (Blackorby, Lady, Nissen, & Russell, 1970). Therefore, we focus on the estimation of the total price of multiple items in this research. We argue that linguistic processing is important in total price calculations. In particular, we investigate the role of phonological encoding and show that prices with longer number names (e.g., “seven” vs. “two”) lead to worse total price estimate accuracy compared to prices with short number names. Thus, our findings extend recent research that investigates single price recall (Vanhuele et al., 2006). We provide a theoretical framework for the cognitive processing of multiple prices. Our theoretical framework revolves around the notions of (a) stimulus length limits on the capacity of workingmemory (Baddeley, Thomson, &Buchanan, 1975), and (b) information chunking in working memory (Miller, 1956). The calculation of multiple prices can be seen as an instance of chunking of several pieces of information into a higher-level piece. The memory literature suggests that two factors can influence chunking: the length of the items to be processed and non-length factors (e.g., semantic relatedness of items in basket, ed by Elsevier Inc. All rights reserved. 347 D. Luna, H.-M.(C.) Kim / Journal of Consumer Psychology 19 (2009) 346–355 attention to stimuli). Based on the working memory literature (Baddeley, 1986), we suggest that length and non-length factors influence total basket price estimation, but via different components of working memory (i.e., the phonological loop and the central executive, respectively).We argue that both types of factors should be considered jointly to understand the process of estimating the total price of shopping baskets. Number length and total price calculation After a stimulus is perceived and attention is drawn to it, it enters working memory. Working memory consists of three components: the central executive, a component that performs a supervisory attentional role; the visuo-spatial scratch pad, which is specialized for visual/spatial tasks, and the phonological loop, which holds information in a phonological form (Baddeley & Hitch, 1974). The phonological loop is the key component in our research, although as we will see later, the central executive also has a role in calculation when it comes to factors unrelated to number length that influence calculation accuracy (Fürst & Hitch, 2000; Logie, Gilhooly, & Wynn, 1994). In particular, we focus on the phonological loop's function as a passive phonological store directly concerned with speech perception. The phonological loop captures language that is visually or aurally presented and holds it for further processing. The capacity of this store is limited to the information that can be phonologically rehearsed approximately in two seconds (Baddeley, 1986). One strategy to increase the amount of information that can be held in working memory, therefore, would be to shorten the length of words so more of them can fit into the phonological loop (Baddeley et al., 1975). Recent consumer research has investigated the role of number name length (henceforth number length) on single price recall (Vanhuele et al., 2006). The key process we examine is the calculation that takes place when consumers encounter multiple prices. The phonological coding of multiple prices The triple-code model developed by Dehaene (1992) implies that consumers may process numbers (i.e., price information) in three types of codes. Consumers may vocally or subvocally read a price (e.g., “79” as “seventy-nine”) and process it using the phonological, or sound-based, code. Alternatively, consumers could process a price as a visual symbol (e.g., “79” as “79”), thus using the visual code, or they could rely on the analogue magnitude code, processing only the relative magnitude of a price (e.g., “$79” as “expensive”). As suggested by Vanhuele and Drèze (2002), the encoding of a price depends on the task at hand. For example, in a price recognition task, consumers tend to rely on the visual code, and when comparing a price with a reference price, consumers tend to use the analogue magnitude code. In contrast, when asked to recall prices, consumers rely more on the phonological code. Therefore, estimates of the total price of multiple items should be influenced more by phonological encoding than by visual or analogical processes (see also Wyer, Hung, & Jiang, 2008). Research on numeric cognition andmental arithmetic provides further support for the idea that multiple prices are phonologically encoded. Consider shoppers who see an ad for a camera phone, listed at $327, and the charger for the phone, a necessary item, priced at an additional $17. How do they calculate the total price? The phonological loop of working memory plays an important role in mental calculations of that type (for a review, see DeStefano& LeFevre, 2004). In particular, the phonological loop tends to be more involved in mental calculations when the calculation is not presented in a form that facilitates graphic solution of the problem (Trbovich & LeFevre, 2003). It follows from this discussion that when stimuli (e.g., prices) are encoded phonologically in a more efficient manner (i.e., in shorter words), the amount of information that can fit into working memory can increase. Thus, if we shorten the names of objects or numbers, we can hold more of them in working memory. In other words, memory span will increase inversely proportionally to the time it takes to rehearse the information being encoded. In particular, if price X has a shorter number name than price Y, individuals' working memory span for numbers should be longer when they process X than Y. The difference in length between X and Y can arise if number names in one language are generally shorter than their equivalents in a different language (e.g., English and Korean, respectively), or it can happen within a single language (e.g., “seven” vs. “two”). In this research, we investigate both cases. Confirming cross-language differences, Ellis and Hennelly (1980) found that working memory span was shorter in Welsh than in English and attributed this effect to the longer number names in Welsh. Similar differences have been found in other languages (Vanhuele et al., 2006). So far, we have argued that the phonological loop is involved in the processing of multiple prices and, therefore, individuals can hold more numerical information in their working memories when the prices are short. We now focus on the calculation of the total price, suggesting that such estimation can be considered an instance of chunking. Chunking in working memory Chunking refers to the grouping of several single stimuli into a compound element. For instance, the digits 1-9-9-5 could be grouped into one compound element, 1995, which would have a unique meaning to the individual (i.e., a particular calendar year). The notion of chunking dates back to Miller (1956), who argued that approximately seven chunks could be held in memory at any one time, although more recent research puts that limit at about four chunks (Cowan, 2000). We suggest that calculating the total price of a multiple-item basket is similar to chunking diverse elements into one. In both processes various items are combined into one; both result in more efficient information processing, expanding the capacity of working memory, and both are influenced by similar factors. One of those factors is stimulus (e.g., number) length. The phonological loop can be overtaxed by long numbers, so the ability to chunk them—that is, to add them up, would be limited by number length. Contextual and long-term memory factors 348 D. Luna, H.-M.(C.) Kim / Journal of Consumer Psychology 19 (2009) 346–355 also play a role on the successful chunking of incoming information items (Bousfield, 1953; Tulving & Patkau, 1962), and we suggest they do so via the central executive component of working memory, which is in charge of interacting with long term memory and directing resources to specific tasks (Baddeley, 1986). Contextual factors include the external punctuation of some items, which may serve to create groupings of those stimulus items (McLean & Gregg, 1967). Long-term memory influences chunking operations. For example, if individuals do not know that IBM is a company, they would not be able to chunk the letter I-B-M together (Eysenck & Keane, 1995). Long-term memory factors that can influence whether individual items are chunked together include item cooccurrence (Deese, 1959; Hulme, Stuart, Brown, & Morin, 2003; Stuart & Hulme, 2000). The results of that research suggest that items that typically co-occur in individuals' everyday experiences are more readily chunked together than non-co-occurring items. In summary, mental calculations of multiple prices seem to involve both the phonological loop and the central executive. The pilot study and studies 1–2 address the phonological loop and its limitations. This component of working memory has severe capacity constraints, which can make it difficult to process prices with long number names. Therefore, the ability to mentally perform price calculations (i.e., chunk multiple prices) will depend on the number name length of the individual prices (Ellis, 1992). Studies 3–5 center around the role of the central executive in the chunking process.