The behavioral economics of consumer brand choice: patterns of reinforcement and utility...

Behavioural Processes 66 (2004) 235–260

The behavioral economics of consumer brand choice: patterns ofreinforcement and utility maximization

Gordon R. Foxalla,∗, Jorge M. Oliveira-Castrob, Teresa C. Schrezenmaiera

a Cardiff Business School, Cardiff University, Cardiff, UKb Instituto de Psicologia, Universidade de Bras´ılia, Brası́lia, DF, Brazil

Abstract

Purchasers of fast-moving consumer goods generally exhibit multi-brand choice, selecting apparently randomly among asmall subset or “repertoire” of tried and trusted brands. Their behavior shows both matching and maximization, though it isnot clear just what the majority of buyers are maximizing. Each brand attracts, however, a small percentage of consumers whoare 100%-loyal to it during the period of observation. Some of these are exclusively buyers of premium-priced brands whoare presumably maximizing informational reinforcement because their demand for the brand is relatively price-insensitive orinelastic. Others buy exclusively the cheapest brands available and can be assumed to maximize utilitarian reinforcement sincetheir behavior is particularly price-sensitive or elastic. Between them are the majority of consumers whose multi-brand buyingtakes the form of selecting a mixture of economy- and premium-priced brands. Based on the analysis of buying patterns of 80consumers for 9 product categories, the paper examines the continuum of consumers so defined and seeks to relate their buyingbehavior to the question of how and what consumers maximize.© 2004 Elsevier B.V. All rights reserved.

Keywords:Consumer behavior; Elasticity of demand; Brand choice; Behavioral Perspective Model; Fast moving consumer goods

1. Introduction

Within marketing science, the analysis of brandchoices for fast-moving consumer goods, based onaggregate data, shows that most individuals tend topurchase a variety of brands within a product cate-gory. More specifically, such results indicate that, insteady-state markets: (a) only a small portion of con-sumers buy just one brand on consecutive shoppingoccasions, that is, few consumers remain 100% loyalto one brand; (b) each brand attracts a small group of100%-loyal consumers; (c) the majority of consumersbuy several different brands, selected apparently ran-domly from a subset of existing brands; (d) existing

∗ Corresponding author. Tel.:+44-2920-874-275.E-mail address:[email protected] (G.R. Foxall).

brands usually differ widely with respect to penetra-tion level and not so much in terms of average buyingfrequency (i.e., how often consumers buy it during theanalysis period); and (e) brands with smaller penetra-tion levels (or market shares) also tend to show smalleraverage buying frequency and smaller percentagesof 100%-loyal consumers (i.e., “double jeopardy”).These results have been replicated for some 30 foodand drink products (from cookies to beer), 20 cleaningand personal care products (from cosmetics to heavycleaning liquids), gasoline, aviation fuel, automobiles,some medicines and pharmaceutical prescriptions,television channels and shows, shopping trips, storechains, individual stores, and attitudes toward brands(cf. Dall’Olmo Riley et al., 1997; Ehrenberg, 1972;Ehrenberg et al., 1990; Ehrenberg and Scriven, 1999;Goodhardt et al., 1984; Uncles et al., 1995).

0376-6357/$ – see front matter © 2004 Elsevier B.V. All rights reserved.doi:10.1016/j.beproc.2004.03.007

236 G.R. Foxall et al. / Behavioural Processes 66 (2004) 235–260

So sure are the relationships involved that a math-ematical model has also been developed to describesuch regularities, the Dirichlet Model (e.g.,Ehrenberget al., 1990), which has been used to predict the mar-ket insertion of new products (Ehrenberg, 1993), toanalyze the effects of promotions (Ehrenberg, 1986;Ehrenberg et al., 1994), and to evaluate patterns ofstore loyalty (Ehrenberg and England, 1990; Kengand Ehrenberg, 1984; Sharp and Sharp, 1997;Uncles and Ehrenberg, 1990). Nonetheless, despitethe wide replication of such patterns, which havebeen raised by some authors to the status of “empiri-cal generalizations” in marketing (e.g.,Uncles et al.,1995), little is known about the variables and theunderlying behavioral mechanisms that influence andexplain consumers’ brand choices. The marketing lit-erature is not forthcoming, for instance, about the fac-tors responsible for shaping the subset of the brandsthat compose a product category among which con-sumers choose in practice (their “consideration sets”)and what Ehrenberg calls the “repertoire” of suchbrands actually purchased (their “purchase sets”).

It is a basic axiom of modern marketing thoughtthat sales are produced not simply by price actingalone, any more than by product attributes, or adver-tising and other promotional means, or distributioneffectiveness acting singly, but by a combination ofall four of these influences on demand that constitutethe “marketing mix.” As marketing science has devel-oped as a separate discipline, it has de-emphasized theinfluence of price on demand (the principal focus ofthe economist’s purview) and stressed the non-priceelements of the marketing mix, notably the promo-tional activity involved in brand differentiation (DeChernatony and McDonald, 2003; Jary and Wileman,1998; Watkins, 1986). Behavioral economics, partlybecause of the stress it has placed on the economicsof animal responding in experimental situations,where the sole reliable analogue of the influences onconsumer demand ruling in the market place relatesto price, has necessarily followed the reasoning andmethodology of the economist rather than the mar-keting scientist. The non-price marketing mix has,therefore, not featured in the research program ofbehavioral economics.

The assumption that consumers maximize utility insome way or other—a preoccupation of the economicsapproach—is, nevertheless, common in the marketing

literature.Krishnamurti and Raj (1988), for example,state that “the consumer chooses that alternative whichmaximizes his (or her) utility,” although they recog-nize that this is a latent or unobservable utility which isassumed rather than tested (cf.Rachlin, 1980). Basedon this maximization assumption, one could expectconsumers to choose the cheapest brands that offer theattributes and characteristics that they are looking for.Although the price of different brands is certainly onevariable that is expected to influence brand choice, asexemplified by the literature on the effects of promo-tions (e.g.,Ehrenberg, 1986; Ehrenberg et al., 1994;Bell et al., 1999), empirical evidence showing that con-sumers tend to maximize when choosing across brandswas not available before recent research on the behav-ioral economics of brand choice (Foxall and James,2001, 2003; Foxall and Schrezenmaier, 2003). In thispaper, we extend this research from the analysis ofsingle cases to that of panel data for some 80 con-sumers purchasing 9 product categories, examining indetail the relationship between price and quantity de-manded in relation to the functional and symbolic at-tributes of brands which influence the composition ofconsumers’ consideration and purchase sets.

1.1. Previous research

Foxall (1999a), Foxall and James (2001, 2003),and Foxall and Schrezenmaier (2003)adopted tech-niques refined in choice experiments in behavioraleconomics and behavior analysis to investigate brandchoice. Three types of analysis were used: matching,relative demand, and maximization.

1.1.1. Matching analysisThe results of choice experiments with nonhuman

animals in behavior analysis gave support for the de-velopment of the matching law, which in its simplestform asserts that organisms in choice situations matchthe relative distribution of responses to the relativedistribution of the reinforcers they obtain (Herrnstein,1961, 1970). In its more general form, the generalizedmatching law (Baum, 1974, 1979) states that the ra-tio of responses between two alternatives is a powerfunction of the ratio of reinforcers, that is,

B1

B2= b

(R1

R2

)s

(1)

G.R. Foxall et al. / Behavioural Processes 66 (2004) 235–260 237

where B represents responses,R represents rein-forcers, and the subscripts 1 and 2, choice alternatives.The parameterb, obtained from the intercept of thelinear log–log formulation of the law, is a measure ofbiased responding between the alternatives, usuallyrelated to asymmetrical experimental factors such asdifferences in response cost between the alternatives.The parameters, the slope of the linear log–log for-mulation, is interpreted as a measure of sensitivity inresponse distribution with changes in reinforcer dis-tribution, which indicates that the individual favors,more than predicted by precise matching, the richer(s > 1) or poorer (s < 1) schedule of reinforcement.In behavioral economics, the parameters can also beused as an estimate of the level of substitutability ofthe reinforcers in the situation, in which case there isevidence suggesting that it should be equal or closeto 1 for substitutable commodities, and negative forcomplementary commodities (cf.Baum and Nevin,1981; Foxall, 1999a; Kagel et al., 1995).

Foxall and James (2001, 2003)applied this typeof analysis to data obtained from consumers’ brandchoice. Consumer choice was analyzed for brands thatwere substitutes, non-substitutes and independent, for1-, 3-, and 5-week periods. Matching and maximiza-tion analyses were based on relative measures of pricepaid and amount bought, which considered the relationbetween the amount paid for (or amount bought of) thepreferred brand and the amount paid for (or amountbought of) the other brands in the consumer reper-toire. As predicted, substitute brands showed matchingwhereas independent brands showed some evidenceof anti-matching. Their results also showed some ev-idence that consumers tend to maximize the amountthey pay in relation to the amount they buy within theirbrand repertoire by purchasing the cheapest brand (al-though they sometimes also bought some more expen-sive brand). Similar results have also been reportedby more recent research (cf.Foxall and James, 2003;Foxall and Schrezenmaier, 2003).

1.1.2. Relative demand analysisWhereas matching analysis relates the actual

amount of a reinforcer obtained to the actual amountof behavior expended in obtaining it, an understand-ing of consumer decision making in the face of com-peting sources or reinforcement offered at a varietyof programmed behavioral costs or prices requires

a different kind of analysis. Matching analysis plotsthe quantity obtained of a commodity as a positivelyaccelerating function of the amount paid for it. Bycontrast, the sensitivity of the quantity demanded ofa commodity to its ruling market price is expressedby economists in terms of the demand curve. Oneof the assumptions underlying the demand curve isthat as the unit price of a commodity increases, itsconsumption will decrease (Madden et al., 2000).This is demonstrated when demand curves plottedon logarithmic coordinates show consumption to bea positively decelerating function of unit price. Thesensitivity of quantity demanded to price is expressedin economic terms as “price elasticity of demand”which at its simplest relates the percentage change inamount consumed to the percentage change in price(Houston and McFarland, 1980; see alsoHursh, 1980;Hursh and Bauman, 1987).

In an attempt to incorporate some of the featuresof naturalistic marketing settings involving consumerchoices among competing brands whose relative pricesmight influence selection decisions, Foxall and James(following Kagel et al., 1980) employed relative de-mand analysis which presents the relative amounts ofbrands A and B as a function of their relative prices.Their results, albeit for a restricted sample of individ-ual consumers and covering a small number of productcategories, found downward-sloping demand curveswhich indicated a degree of price sensitivity on the partof the buyers investigated (Foxall and James, 2001,2003).

1.1.3. Maximization analysisAnalyses to reveal whether the observed consumer

behavior was maximizing returns on price expendedwere undertaken following procedures developedby Herrnstein and Loveland (1975), Herrnstein andVaughan (1980). On concratio schedules,1 there is afixed probability of reinforcement for each response,

1 A ratio schedule is one in which a specified number of re-sponses has to be performed before reinforcement becomes avail-able. Fixed ratio schedules keep the number of required responsesequal from reinforcer to reinforcer; variable ratio schedules allowthe required number of responses to change from one reinforcer tothe next. Concurrent variable ratio schedules, usually abbreviatedto conc VR VR, allow simultaneous choice to be investigated. Itis this arrangement that most clearly resembles the purchases ofbrand within a product class.


which can be expressed as the reciprocal of theschedule parameter. Concurrent VR30 VR60 refers toresponse alternatives which have respective reinforce-ment probabilities of 1/30 and 1/60. On ratio sched-ules, the probability of reinforcement is independentof response rate (something not true of VI scheduleswhere the probability of reinforcement is inverselyproportional to rate of responding). Although mostresearch on matching and maximization has beenundertaken in laboratory settings which incorporateVI schedules, VR schedules are more probable innaturalistic settings (Herrnstein, 1982; Herrnstein andLoveland, 1975; Herrnstein and Prelec, 1991;Herrnstein and Vaughan, 1980; Vaughan andHerrnstein, 1987).

Faced with conc VR40 VR80 schedules, theindividual’s maximal probability of reinforcementis obtained by responding exclusively on the VR40schedule. Matching theory makes the same predic-tion for conc VR VR schedules, claiming that maxi-mization is under these circumstances a special caseof matching (cf.Rachlin, 1980). Previous research,subject to the limitations of scope noted above, con-firmed that consumers tend to maximize by generallypurchasing the least expensive brand available withintheir consideration set (Foxall and James, 2001,2003).

1.2. Research issues

Taken together, these results indicate that, withintheir repertoire of brands, consumers show price sen-sitivity, maximizing (most of the time), and matching(which refers to the relation between the amount theyspend and the amount they buy). Based on such find-ings, one can predict that consumers will buy, moreoften than not, the cheapest brand among those thatthey usually buy, although one still does not know whythey usually buy a certain set of brands and not others.The fact that consumers tend to buy the cheapest brandwithin a restricted set of brands rather than the cheap-est of all brands available in the product category indi-cates that not all brands are perfect substitutes for theothers. Even though they may be functionally equiva-lent for the consumer, the brands are not entirely equiv-alent, that is, consumer preferences reflect more thanfunctional utility. This additional source of utility isusually rationalized in the marketing literature as stem-

ming from rather nebulous “branding” considerations.Branding is not, however, a quantifiable construct andan important objective of the research reported herewas to clarify its basis as an extra-functional sourceof reinforcement.

Although research to date is indicative that the prin-ciples and methods of behavioral economics can beusefully applied to consumer brand purchasing, thereis clearly need for a more extensive investigation ofa larger, systematically-selected sample of consumerspurchasing a wider range of products in order to as-certain how far previously reported results are gener-alizable. It is necessary to take into greater considera-tion the differences between the typical consumptionpatterns of laboratory subjects which can be shownto be sensitive to price (or its analogue) and those ofconsumers in supermarkets who are subject to a muchwider spectrum of choice under the influence of theentire array of marketing mix variables available toretailers. For example, an expectation of demand anal-ysis as it is employed in the behavioral economics lit-erature is that when consumers choose between qual-itatively identical reinforcers which vary in terms ofthe unit prices that must be paid for them, the brandwith the lower or lowest unit price will be exclusivelychosen (Madden et al., 2000). This is the predictionof both matching and maximization theories with re-gard to choice on conc VR VR schedules. However,research in these theoretical traditions typically takesplace within laboratory settings that restrict choice totwo alternatives, one or other of which must be se-lected at any choice point. Consumer brand choice ismore complicated than this in that numerous choicesare usually available to the consumer within a givenproduct category, more than one of which may beselected on a single shopping occasion (Foxall andSchrezenmaier, 2003).

A source of difference among brands, related tothis and other aspects of consumer choice, stems fromthe distinction between utilitarian and informationalbenefits offered by different brands, as proposedby the Behavioral Perspective Model (Foxall, 1990,1994, 1996, 1997, 1998). According to this proposal,the behavior of the consumer can be explained bythe events that occur before and after the consumersituation, which influence directly the shaping andmaintenance of consumer behavior in specific envi-ronments. The consumer situation, in turn, is defined


as the intersection between the consumer behaviorsetting and the consumer learning history. The con-sumer behavior setting—a supermarket, a bookstore,or a rock concert—includes the stimuli that form thesocial, physical and temporal consumer environments.As purchase and consumption are followed by differ-ent consequences in different settings, the events inthe setting become predictive of such consequences,building a learning history that relates elements of thesetting to different consequences. According to theproposal, antecedent events present in the consumerbehavior setting signal the possibility of three typesof consequences: utilitarian reinforcement, informa-tional reinforcement, and aversive events. One majorcharacteristic of economic behavior is that it involvesboth aversive and reinforcing consequences, for onehas to give away money or rights (i.e., loss of gener-alized reinforcers) in order to get products or services(i.e., reinforcing events).

Utilitarian reinforcement consists in the practicaloutcomes of purchase and consumption, that is, func-tional benefits derived directly (rather than mediatedby other people) from possession and application of aproduct or service. It is reinforcement mediated by theproduct or service and refers to consequences associ-ated with increases in the utility (i.e., use value) forthe individual (“pleasant”) obtained from the productor service. The utilitarian, most obvious, consequenceof owning a car, for example, is to be able to go fromone place to the other, door to door, not depending onother people’s time schedules and avoiding being ex-posed to weather conditions, as usually happens whenone uses public transportation.

Informational reinforcement, on the other hand,would be symbolic, usually but not exclusively medi-ated by the actions and reactions of other persons, andwould be more closely related to the exchange valueof a product or service.2 It does not consist in infor-

2 Following Wearden (1988), we use “informationalreinforcement” to refer to performance feedback. The term“informational” carries excess baggage for many behavior analystssince it may appear to make cognitive inferences. Given the ex-amples we provide in the text, it may appear that “social” wouldbe a more acceptable and accurate alternative. However, “social”does not entirely capture what we mean by “informational” whichincludes rewards for adhering to social mores,and physicalsources of feedback such as lines on the road that convey animpression of speed, or the fullness of one’s shopping trolley. A

mation per se but in feedback about the individual’sperformance, indicating the level of adequacy and ac-curacy of the consumer’s behavior. Whereas utilitarianreinforcement is associated with the functional andeconomic consequences of purchasing and consuminggoods or services, informational reinforcement is de-rived from the level of social status and prestige thata consumer obtains when purchasing or using certaingoods. According to Foxall, informational and utili-tarian reinforcements would be orthogonal, and mostproducts and services would involve, in different lev-els or proportions, both types of reinforcement. Then,according to this analysis, the person who drives aJaguar© or Bentley© gets, in addition to door-to-doortransportation (utilitarian), social status and approvalfrom friends and acquaintances who see that car as aprestigious product, and from the general public thatsees him or her driving around in a socially desirablecar. The social status and prestige received are theinformational, symbolic, consequences that the con-sumer obtains, which are usually related to brandingor the level of brand differentiation of the product (cf.Foxall, 1999a).

The specific combination of utilitarian and infor-mational reinforcement made available by purchaseor consumption of a particular product is known asthe “pattern of reinforcement” controlling these re-sponses.Foxall and James (2001, 2003)argued thatpattern of reinforcement influences consumers’ brandchoices and that it is a key to understanding whatconsumers maximize. Different consumers might, forexample, select brands belonging to different levelsof informational reinforcement, some buying mostlyhighly differentiated whereas others buy relatively un-differentiated brands. The differences in patterns ofbrand choice, including the set of brands that consti-tute each consumer’s brand repertoire, may be a con-sequence of individual differences in responsivenessto different types of benefits. This idea gains even

concomitant consideration arises in the functional definition ofrules as “plys,” which involve the mediation of other people andwhich are therefore social, or as “tracks,” which depend on therule-follower’s “reading” the physical environment, e.g., in theprocess of following directions to get to a supermarket (Zettleand Hayes, 1982). Informational reinforcement thus remains ourdesignation of choice for this phenomenon since it includes bothpersonally-mediated and nonpersonally-mediated performancefeedback.


more force when we consider that branding is usu-ally related to price, higher-differentiated brands beingmore expensive than less differentiated ones, and thatconsumers have different income levels. Then, indi-vidual buying patterns may be predominantly related,for example, to minimizing costs, maximizing utili-tarian reinforcement, maximizing informational rein-forcement, or to particular combinations of these. Ifthis is so, consumers may differ with respect to priceresponsiveness related to informational and utilitarianbenefits.

The research reported here tested predictions arisingfrom these considerations using data from a consumerpanel. Panel data are especially valuable for longitu-dinal studies because changes in purchasing behaviorcan be monitored very accurately by continuous mea-surements (Crouch and Housden, 2003). Furthermore,diary panel data are considered to be very precise andless susceptible to errors than those obtained throughconsumers’ reporting their past behavior in surveys(Churchill, 1999). Hence, they are particularly valu-able when collecting multifarious information on vari-ables such as price, shopping occasion, brand name,and so on. The special significance of this researchtechnique for the present research lies in the fact thatthe data were obtained non-experimentally, by elec-tronically tracking real consumers spending their realdiscretionary income.

The two main purposes of the investigation were asfollows. First, in order to ascertain the generalizabil-ity of earlier research findings to consumer behaviorin marketing-dominated contexts, three analyses wereundertaken in order to determine whether the brands inquestion were in fact close substitutes (matching anal-ysis), whether brand choice was sensitive to price dif-ferentials (relative demand analysis), and whether con-sumers could be said to maximize returns (maximiza-tion analysis). Second, in order to gauge consumers’responsiveness to price and non-price marketing mixelements, the brands of 9 food product categories wereranked according to their informational and utilitarianlevels. The proportion of purchases made by eachconsumer at each brand level was computed, whichserved as basis for grouping consumers accordingto the level of brands they bought most. To test fordifferences in price responsiveness, price elasticitiesfor consumer groups and individual consumers werecompared.

2. Method

2.1. Sample and procedure

The market research company, Taylor NelsonSofres, provided consumer panel data for 80 Britishconsumers and their total weekly purchases in 9fast-moving consumer goods categories over 16weeks. Taylor Nelson Sofres is one of the largestand best-known companies in its field and clustersconsumer purchasing data on its so-called TNS Su-perpanel on a range of consumer goods from 15,000randomly selected British households. Data collec-tion is operationalized as follows: after each shop-ping trip, members of the panel scan their purchaseditems into a sophisticated handheld barcode readerby passing the scanner across the barcodes, whichnowadays are printed on all packaged supermarketproducts. The data are then automatically sent toTaylor Nelson Sofres for central processing withoutany further voluntary contribution from the panelparticipants. The retail outlets at which purchaseswere made was also identified for each shopping oc-casion, and included major UK supermarkets suchas Asda (a subsidiary of Wal-Mart), Tesco, andSainsbury.

The 9 product categories that served as basis for thisresearch were: baked beans, cookies, cereals, butter,cheese, fruit juice, instant coffee, margarine, and tea.In more detail, the following information was recordedon each shopping occasion for each consumer: brandspecification (i.e., different versions of the same prod-uct category were classified as different brands, e.g.,Corn Flakes and Rice Krispies© by Kelloggs), pack-age size, name of the supermarket/shop, date, num-ber of units, and total amount spent. As the analysisof brand choice requires information concerning ac-tual purchase across several buying opportunities, datafrom consumers who bought, within each product cat-egory, fewer than four times during the 16-week pe-riod were disregarded.

2.2. Measures and analyses

2.2.1. MatchingIn consumer research, the matching law becomes

the proposition that the ratio of amount of moneyspent for a brand to the amount spent on other brands


within the product category will match the ratioof reinforcers earned (i.e., purchases made as a re-sult of that spending) of that brand to the amountbought of other brands within the product category.The first of these, theamount paid ratio, was oper-ationalized as the ratio of money spent on “BrandA,” defined as the most frequently purchased brand,to money spent on “Brand B,” i.e., the amountspent on the remaining brands purchased within therequisite product category:Amount paid for BrandA/Amount paid for the remaining brands in the prod-uct category (B). The amount bought ratiowas cal-culated, in terms of the physical quantity acquired,as: Amount bought of Brand A/Amount bought ofBrand B (the remaining brands of the product cate-gory). Logarithmic transformations were used for theanalyses.

2.2.2. Relative demandIn order to devise relative demand curves for the

product categories, a demand analysis expressed theratio of amount bought of the dominant brand (A) tothe amount bought of the remaining brands in that cat-egory (B) as a function of the ratio of the relative aver-age prices of the dominant brand to the average priceof other brands purchased from the appropriate prod-uct category (the relative price ratio). In operationalterms, the relative price ratio= mean price of BrandA/Mean price of other brands in the repertoire (B).The amount bought ratio was calculated as in the caseof the matching analysis. Again, log transformationswere used for the analyses.

2.2.3. MaximizationTo ascertain whether maximization is occurring, fol-

lowing Herrnstein and Loveland (1975), Herrnsteinand Vaughan (1980), we plotted the amount boughtratio against probability of reinforcement. The lat-ter is operationalized as the reciprocal of the priceof brand A over the reciprocal of the price of brandA plus the reciprocal of the mean of the prices ofthe other brands in the consumer’s consideration set(“Brand B”): 1/PA/(1/PA + 1/PB). If the step func-tion described by the data points falls to the rightof the 0.5 line on the abscissa then the purchaser ismaximizing by selecting the favorite brand (A) whichis also the least expensive (Herrnstein and Loveland,1975).

2.2.4. Schedule analogiesTo ascertain how consumers make decisions, it is

necessary to have some idea of how they integrateprice data and brand choice responses over time,notably from shopping trip to shopping trip. In thelaboratory this can be achieved without undue diffi-culty by the imposition of a schedule of reinforcementwhich programs the relationships between dependentand independent variables. Researchers who are con-cerned with the behavioral analysis and explanationof non-experimental behavior face the difficulty ofascertaining with precision whether brand choice innaturalistic settings occurs, by analogy, on a seriesof fixed ratio schedules (represented by the prices ofeach brand obtaining on each purchase occasion) or,aggregated over several such occasions, on variableratio schedules. The question we are seeking to an-swer is whether consumers take into considerationonly the prices of the brands in their considerationset that are in force on each discrete shopping trip,or whether their behavior (brand choice) reflects theprice–quantity relationships for competing brandsthat are in force over the extended period representedby a series of shopping trips. This led us to under-take two analyses for each product category studied.The first treated the schedules as a sequence of fixedratio relationships by expressing measures of amountbought as a function of measures of prices for (a)weekly periods, representing (albeit by analogy ratherthan programming) the situation in which experimen-tal subjects face a sequence of FR schedules, and (b)periods of 3 weeks, for which the data were aver-aged, similarly representing an experimental situationgoverned by VR schedules.

2.3. Utilitarian and informational reinforcement

To investigate possible effects of informational andutilitarian reinforcement values on brand choice, anattempt was made to identify different levels or mag-nitudes of informational and utilitarian reinforcementoffered by the brands available (i.e., bought by con-sumers in the sample) in each product category. The setof alternative brands and product characteristics avail-able in a supermarket within each product categorycan be interpreted as a set of programmed contingen-cies of reinforcement, which specify what responses(e.g., how much one has to pay) are followed by what


consequences (e.g., product characteristics). A majorpart of marketing activities, according to this interpre-tation, is to plan and establish contingencies for thebehavior of consumers (Foxall, 1999b). Not all pro-grammed contingencies, however, have the desired orplanned influences on behavior, and that is why an im-portant issue for marketing managers and academics isto identify the actual effects of different contingencies(i.e., the effects of the price and non-price elements ofthe marketing mix on consumer choice). The analysesof informational and utilitarian reinforcement levelspresented below follow the same logic, that is, theyrefer to programmed levels of informational and util-itarian reinforcements, which may or may not influ-ence particular instances of consumer choice. In otherwords, in the case of marketing activities, an event thatwas planned to have high reinforcement magnitude,vis-à-vis its aversive components (costs), can in facthave low reinforcing value for consumers (e.g., inno-vations that do not attract people or are too expensive).

Considering that there are no general units to mea-sure utilitarian and informational reinforcement lev-els, these were identified based on a forced rankingsystem in which three informational and two utilitar-ian levels were ascribed to each product category. Thisclassification was chosen due to our interest in makingcomparisons across product categories and was in partinfluenced by our sample (not all brands and brandtypes were bought by our sample during the period).Levels of informational and utilitarian benefit cannotbe defined absolutely: they depend ultimately on theinterests of researchers. More levels of utilitarian rein-forcement, for example, could be identified for someproduct categories, such as cookies and cheese, but anequal number of levels across products was thought tobe desirable for the present analysis.

In the case of supermarket food products like thoseinvestigated here, increases in utilitarian level can beidentified by the addition of (supposedly) desirable at-tributes. Such attributes usually add value to the prod-uct or its consumption, are mentioned on the packageor product name, and justify increases in price. More-over, in most cases, several general brands offer prod-ucts with and without these attributes. For the productcategories in question, utilitarian levels were identi-fied based on additional attributes (e.g., plain bakedbeans versus baked beans with sausage) and/or differ-entiated types of products (e.g., plain cookies versus

chocolate chip cookies). In the case of differentiatedproduct types, several general brands usually offer thedifferent product types, charging differentiated pricesfor them (e.g., plain cookies are cheaper than moreelaborate cookies for all brands examined).

Informational reinforcement, by contrast, isstrongly associated with brand differentiation in thatthe most promoted and best known brands are usuallyassociated with higher levels of prestige, social status,and trustworthiness. In the case of the supermarketproducts investigated here, informational reinforce-ment level is closely associated with brand differen-tiation, which in turn is usually also related to pricedifferentiation. If one compares the level of branddifferentiation of, say, Asda Smart Price© and Heinz©

plain baked beans, Heinz is clearly the better known,more differentiated and consequently more expensivebrand, offering a higher level of informational rein-forcement. This kind of difference among brands hasbeen interpreted in the present work as differences ininformational reinforcement level. It should be notedthat informational reinforcement level as specifiedhere does not exclude the possibility of there alsobeing differences in utilitarian reinforcement betweentwo informational magnitudes. Corporate representa-tives of any differentiated brand would argue stronglythat their products differ from those of other compa-nies in terms of their “utilitarian” attributes, such asquality of raw materials and ingredients, productionprocedures, health control, and such like. Similarly,consumers of differentiated brands may also assertthese brands’ functional superiority, e.g., that theytaste much better than other cheaper brands, whichwould imply differences in utilitarian reinforcementlevel. The classification adopted does not excludesuch possibilities, since most consumer behavior gen-erates both types of consequences. Nevertheless, theranking of informational reinforcement is based onthe predominant difference that one can find betweenproducts, offered by different brands, that usuallyhave almost identical formulations (cf.Ehrenberg,1972; Foxall, 1999a) and may not even be distin-guished by consumers on the basis of their physicalcharacteristics (e.g., in blind tests).

The ranking of informational reinforcement levelwas based on the following general criteria: (1) in-creases in prices across brands for the same producttype (e.g., plain baked beans, plain cookies or plain


Table 1Levels of informational reinforcement

The cheapest own (retailer) brands (Asda Smart Price©, Tesco Value©, Sainsbury Economy©) Level 1Own (retailer) brands that do not mention good value for money or economy

(Asda, Tesco, Sainsbury) and cheapest specialized brandsLevel 2

Specialized brands (Heinz©, McVities©, Kelloggs©, Lurpak©) with higher prices Level 3

cornflakes) were considered to be indicative of differ-ences in informational levels; (2) the cheapest storebrands (e.g., Asda Smart Price©, Tesco Value©, Sains-bury Economy©) were considered to represent thelowest informational level (Level 1); (3) store brandsthat do not mention good value for money or econ-omy (e.g., Asda, Tesco, Sainsbury) and cheapest spe-cialized brands were usually considered to representthe medium informational level (Level 2); and (4) spe-cialized brands (e.g., Heinz©, McVities©, Kelloggs©,Lurpak©), with higher prices, were considered to rep-resent the highest informational level (Level 3). Theclassification is shown inTable 1.

3. Results and discussion

General statistics for the sample are shown inTable 2.

3.1. Preliminary analyses

3.1.1. MatchingAs has been noted, when the matching law is ex-

pressed logarithmically as a power function, unity ofthe exponents is frequently interpreted as indicating

Table 2Number of consumers, total and average (per consumer) number of purchases, total and average (per consumer) amount spent, average(per quantity) price, average unit price, total and average number of brands purchased, and percentage brand loyalty, calculated for eachproduct category

Product Number ofconsumers

Totalpurchases

Averagepurchases

Totalspent

Averagespent

Averageprice

Unitprice

Totalbrands

Averagebrands

Brandloyalty (%)

Baked beans 39 265 6.79 4.52 0.62 0.07 0.51 32 2.18 74.59Cookies 59 1125 19.07 14.02 0.74 0.30 0.63 230 8.93 28.99Butter 21 174 8.29 9.84 1.17 0.28 0.76 21 2.24 76.91Cereals 56 691 12.34 20.09 1.56 0.27 1.46 125 5.64 42.24Cheese 45 447 9.93 13.38 1.38 1.91 2.76 95 5.24 43.66Coffee 19 144 7.58 18.32 2.51 2.09 2.09 31 2.95 65.29Fruit juice 34 336 9.88 13.99 1.52 0.72 1.05 43 2.91 65.77Margarine 50 401 8.02 8.75 1.12 0.19 1.01 55 2.70 67.77Tea 32 199 6.22 11.67 2.02 0.61 1.66 30 1.94 77.93

the perfect substitutability of the alternative rein-forcers. Measures ofs that deviated only in the small-est degree from unity were generally found (Figs. 1and 2, Table 3). In Tables 3–5, we reportsvalues as theoutput from the regression equations, and theb (bias)values as the intercept from those equations. This gen-eral pattern of substitutability (Foxall and Schrezen-maier, 2003) is consistent with the findings of earlieranalyses in which anti-matching and substantial under-matching (Kagel et al., 1995) were demonstrable onlyfor gross complements belonging to separate productcategories (Foxall and James, 2001, 2003). However,as Table 3 indicates, there are two exceptions. Un-dermatching is apparent in the case of baked beanswhether data are integrated over a series of 1-week pe-riods (the so-called “FR” schedule) or over 3-week pe-riods (“VR”). Consumers in this case selected Heinz©

baked beans to an extent disproportionate with thatpredicted by strict matching: There were in total 265purchases of baked beans (including baked beans withsausages and other complements), 52% of which werefor Heinz plain baked beans alone. Their “favorite”brand was the most expensive. In the case of fruit juice,undermatching is apparent from the “FR” schedule.This anomaly arises from the single outlying valueshown inFig. 1. When this value is excluded from the

244G

.R.

Foxa

lle

ta

l./Be

ha

viou

ral

Pro

cesse

s6

6(2

00

4)

23

5–

26

0

-0.4 -0.2 0.0 0.2 0.4

Amount Bought Ratio (logarithms)

-0.2

0.0

0.2

0.4

0.6

-1.3 -1.1 -0.9 -0.7

-1.5

-1.3

-1.1

-0.9

-1.6 -1.4 -1.2 -1.0 -0.8 -0.6 -0.4 -0.2

-2.0

-1.6

-1.2

-0.8

-1.6 -1.4 -1.2 -1.0 -0.8 -0.6

-0.5

-0.7

-0.9

-1.1

-1.3

-1.5

-1.4 -1.2 -1.0 -0.8 -0.6 -0.4 -0.2

-0.2

-0.4

-0.6

-0.8

-1.0

-1.2

Am

ou

nt

Paid

Rati

o (

log

ari

thm

s)


Am

ou

nt

Paid

Rati

o (

log

ari

thm

s)

Baked Beans

-2.0 -1.8 -1.6 -1.4 -1.2 -1.0 -0.8 -0.6

-1.8 -1.6 -1.4 -1.2 -1.0 -0.8 -0.6 -0.4

Amount Bought Ratio (logarithms) Amount Bought Ratio (logarithms)

-2.2

-1.7

-1.2

-0.7

Am

ou

nt

Pa

id R

ati

o (

log

ari

thm

s)


Am

ou

nt

Pa

id R

ati

o (

log

ari

thm

s)


Am

ou

nt

Pa

id R

ati

o (

log

ari

thm

s)


Am

ou

nt

Paid

Rati

o (

log

ari

thm

s)


Am

ou

nt

Pa

id R

ati

o (

log

ari

thm

s)

Amount Bought Ratio (logarithms)A

mo

un

t P

aid

Ra

tio

(lo

ga

rith

ms

)

-2.0

-1.5

-1.0

-0.5

-0.9 -0.7 -0.5 -0.3 -0.1 0.1

-1.0

-0.4

-0.2

-0.8

-0.6

0.0

-1.35 -1.10 -0.85 -0.60 -0.35 0.10 0.51

-1.6

-0.4

-1.2

-0.8

0.0

Am

ou

nt

Pa

id R

ati

o (

log

ari

thm

s)Cookies Butter

Cereals Cheese Coffee

Fruit Juice Margarine Tea

Fig. 1. Matching analysis: “FR” schedules. Amount paid ratio as a function of amount bought ratio for all 9 product types (all data are aggregated across purchasers withina product category).

G.R

.Fo

xall

et

al./B

eh

avio

ura

lP

roce

sses

66

(20

04

)2

35

–2

60

245

0.15 -0.10 -0.05 0.00 0.05 0.10


0.0

0.1

0.2

0.3

0.4

Am

ou

nt

Paid

Rati

o (

log

ari

thm

s) Baked Beans

-1.5 -1.4 -1.3 -1.2 -1.1 -1.0 -0.9


-1.6

-1.5

-1.4

-1.3

-1.2

-1.1

Am

ou

nt

Paid

Rati

o (

log

ari

thm

s)


Am

ou

nt

Paid

Rati

o (

log

ari

thm

s)


Am

ou

nt

Paid

Rati

o (

log

ari

thm

s)


Am

ou

nt

Paid

Rati

o (

log

ari

thm

s)


Am

ou

nt

Paid

Rati

o (

log

ari

thm

s)

Amount Bought Ratio (logarithms)A

mo

un

t P

aid

Rati

o (

log

ari

thm

s)


Am

ou

nt

Paid

Rati

o (

log

ari

thm

s)


Am

ou

nt

Paid

Rati

o (

log

ari

thm

s)Cookies

-0.9 -0.7 -0.5 -0.3 -0.1

-1.2

-1.0

-0.8

-0.6

-0.4

-0.2

Butter

-1.000-0.975-0.950-0.925-0.900 -0.875 -0.850 -0.825

-1.15

-1.10

-1.05

-1.00

-0.95

Cereals

-1.2 -1.1 -1.0 -0.9 -0.8-1.1

-1.0

-0.9

-0.8

Cheese

-0.60 -0.55 -0.50 -0.45 -0.40 -0.35 -0.30 -0.25

-0.6

-0.5

-0.4

-0.3

Coffee

-0.9 -0.8 -0.7 -0.6 -0.5 -0.4

-1.2

-1.1

-1.0

-0.9

-0.8

-0.7

Fruit Juice

-1.10 -1.05 -1.00 -0.95 -0.90 -0.85

-1.00

-0.95

-0.90

-0.85

-0.80

-0.75

Margarine

-0.76 -0.71 -0.66 -0.61 -0.56 0.51

-0.65

-0.60

-0.55

-0.50

-0.45

Tea

Fig. 2. Matching analysis: “VR” schedules. Amount paid ratio as a function of amount bought ratio for all 9 product types (all data are aggregated across purchasers withina product category).


Table 3Matching analysis

FR schedule VR schedule

R2 Beta Intercept R2 Beta Intercept

Baked beans 0.637∗∗ 0.813∗∗ 0.152∗∗ 0.242 0.657 0.161Cookies 0.941∗∗ 0.972∗∗ 0.112 0.992∗∗ 0.997∗∗ −0.300∗∗Butter 0.991∗∗ 0.996∗∗ −0.183∗∗ 0.994∗∗ 0.998∗∗ −0.172∗∗Cereals 0.928∗∗ 0.966∗∗ −0.182∗ 0.982∗∗ 0.993∗∗ 0.137Cheese 0.893∗∗ 0.949∗∗ 0.035 0.986∗∗ 0.995∗∗ −0.252∗∗Coffee 0.934∗∗ 0.969∗∗ −0.079∗ 0.970∗∗ 0.989∗∗ −0.069Fruit juice 0.541∗∗ 0.756∗∗ −0.415∗ 0.990∗∗ 0.996∗∗ −0.321∗∗Margarine 0.952∗∗ 0.977∗∗ 0.030 0.089∗ 0.925∗ 0.023Tea 0.978∗∗ 0.990∗∗ −0.017 0.956∗∗ 0.983 −0.108∗∗

∗ P < 0.05.∗∗ P < 0.01.

Table 4Relative demand analysis



Baked beans 0.024 −0.211 −0.060 0.027 −0.520 −0.099Cookies 0.113 −0.415 −2.22∗∗ 0.422 −0.753 −2.996∗Butter 0.333 −0.013 −0.834 0.077 0.009 −0.293Cereals 0.057 0.114 −0.871∗∗ 0.675 0.870 −0.559∗Cheese 0.085 0.382 −1.072∗∗ 0.510 0.795 −1.047∗∗Coffee 0.007 0.254 −0.243 0.142 0.379 −0.247Fruit juice 0.319∗∗ −0.604∗ −3.382∗∗ 0.612 −0.842 −3.219∗Margarine 0.040 −0.185 −0.928∗∗ 0.297 −0.165 −0.924∗∗Tea 0.634∗∗ −0.811∗∗ −1.184∗∗ 0.136 −0.384 −0.975

∗ P < 0.05.∗∗ P < 0.01.

Table 5Maximization analysis



Baked beans 0.015 0.284 −2.102 −0.020 0.485 −1.54Cookies 0.462∗∗ 0.706∗∗ −1.26∗∗ 0.477 0.780 −1.355Butter −0.056 −0.138 10.75 −0.312 −0.128 15.67Cereals −0.034 −0.188 0.378 0.633 −0.851 0.869∗Cheese 0.029 −0.307 0.315 0.595 −0.834 0.809Coffee −0.019 −0.233 0.147 −0.080 −0.436 2.189Fruit juice 0.126 0.429 −3.092 0.426 0.754 −4.911Margarine −0.073 0.057 0.067 −0.310 0.132 0.062Tea 0.678∗∗ 0.836∗∗ −5.885∗∗ −0.188 0.330 −3.345

∗ P < 0.05.∗∗ P < 0.01.


analysis, the value ofR2 is 0.991, and that for betais 0.996. The results for fruit juice indicate the meritof integrating the data over 3- as well as 1-weekperiods: Even a price-conscious consumer would beexpected to “deviate” from matching from time totime in order to secure variety (or because he or shewas buying on behalf of another household mem-ber). This is consonant with the finding reported byFoxall and James (2003)in the case of single butterpurchaser who deviated from her usual highly-pricesensitive pattern of choice on occasion simply in orderto obtain the flavor advantages of a premium-pricedbrand.

3.1.2. Relative demandThe expectation that logarithmically plotted de-

mand curves would show consumption to be a pos-itively decelerating function of unit price (Maddenet al., 2000) was generally though not universallysubstantiated. Relative demand curves for 6 of the9 product categories are, as expected, downwardsloping, though that for butter is approximately hor-izontal; three were upward sloping (Figs. 3 and4, Table 4). Moreover, while the curve for bakedbeans (which were identified as anomalous withrespect to matching) is positively decelerating, theR2 and beta parameters indicate a weak relation-ship between relative price and relative quantity de-manded, which is in keeping with the interpretationof consumers’ brand perceptions advanced above.Although the results are generally in line with the ex-pected price–quantity relationship, the wide dispersalof data points reflected in the many low values ofR2

and beta suggest that more precise methods be soughtfor the demonstration of price–demand associa-tions.

3.1.3. MaximizationThe maximization analysis indicates that most

consumers consistently chose the cheapest brand oneach shopping occasion regardless of product cat-egory (Figs. 5 and 6, Table 5). This is consistentwith the behavioral economics approach pioneeredby Herrnstein and Loveland (1975), Herrnstein andVaughan (1980), but close examination of the resultsreveals a more complicated pattern of choice thanis apparent in the studies of nonhumans undertakenby those authors. First, whilst consumers generally

selected the cheapest brand within their considera-tion set, these “repertoires” in many cases comprisedonly premium, highly differentiated brands. Manyconsumers did not maximize in any “absolute” sense.In each product category, own-label, or store brand,and economy versions existed which were consider-ably cheaper than those actually purchased. Second,whilst research with nonhuman subjects is typicallylimited to only one choice on each occasion, con-sumers are able to purchase more than one brandeven on a single shopping trip. As is apparent fromFigs. 5 and 6, for seven products, consumers inthe aggregate maximized by purchasing the favorite(cheapest) brand (Brand A); for two products, how-ever (cheese and margarine), this pattern was notfound. This same overall pattern was found for boththe “VR” and the “FR” schedules. However, evenfor the seven product categories where consumersmaximized by purchasing Brand A, there is a com-plication which arises from the nature of consumerchoice in the marketplace and which is not encoun-tered in laboratory research with either human ornonhuman animals in which choice is constrained.Although most consumers maximized in the sensethat they purchased the cheapest brand within theirconsideration set, many also purchased a secondbrand priced substantially higher on the same occa-sion. The maximization analyses undertaken basedon the behavioral economics literature was thus in-capable of indicating comprehensively the patternof consumer brand choices in relation to a simplevalue-for-money criterion. (The diagonal lines inFigs. 5 and 6indicate the distribution of data pointsthat would have occurred if consumers had exhibitedprobability matching. By contrast, the step functionindicated by the vertical distribution of data pointsindicates a consistent preference for one or otheralternative.)

These patterns of choice are consistent with find-ings reported in the consumer research and marketingliteratures on branding which portray consumers’consideration sets as a function of the level of qual-ity required for a variety of consumption settings.It was these broader considerations that led to thesuggestion that consumers maximize some combi-nation of utilitarian and informational reinforcementand which resulted in the following more detailedanalyses.

248G

.R.

Foxa

lle

ta

l./Be

ha

viou

ral

Pro

cesse

s6

6(2

00

4)

23

5–

26

0

-0.155-0.130-0.105-0.080-0.055-0.030-0.005 0.020

Average Price Ratio(logarithms)

-0.5

-0.3

-0.1

0.1

0.3A

mo

un

t B

ou

gh

t R

atio

(lo

gar

ith

ms)

Average Price Ratio(logarithms)Am

ou

nt

Bo

ug

ht

Rat

io (

log

arit

hm

s)


ou

nt

Bo

ug

ht

Rat

io (

log

arit

hm

s)


ou

nt

Bo

ug

ht

Rat

io (

log

arit

hm

s)


ou

nt

Bo

ug

ht

Rat

io (

log

arit

hm

s)


ou

nt

Bo

ug

ht

Rat

io (

log

arit

hm

s)


ou

nt

Bo

ug

ht

Rat

io (

log

arit

hm

s)


ou

nt

Bo

ug

ht

Rat

io (

log

arit

hm

s)Average Price Ratio(logarithms)A

mo

un

t B

ou

gh

t R

atio

(lo

gar

ith

ms)

Baked Beans

-0.385 -0.360-0.335-0.310-0.285-0.260-0.235-0.210

-2.0

-1.6

-1.2

-0.8

Cookies

-0.226 -0.221 -0.216 -0.211-1.5

-1.1

-0.7

-0.3

0.1

Butter

-0.255 -0.230 -0.205 -0.180 -0.155 -0.130 -0.105-1.4

-1.2

-1.0

-0.8

-0.6

Cereals

-0.3 -0.2 -0.1 0.0 0.1-1.8

-1.4

-1.0

-0.6

Cheese

-0.185 -0.160 -0.135 -0.110 -0.085 -0.060 -0.035

-1.0

-0.8

-0.6

-0.4

-0.2

Coffee

-0.39 -0.37 -0.35 -0.33 -0.31

-1.6

-1.2

-0.8

-0.4

Fruit Juice

-0.05 0.00 0.05 0.10 0.15

-1.6

-1.4

-1.2

-1.0

-0.8

-0.6

Margarine

-0.09 -0.07 -0.05 -0.03 -0.01 0.01-1.5

-1.1

-0.7

-0.3

Tea

Fig. 3. Relative demand analysis: “FR” schedules. Amount bought ratio as a function of average price ratio for all 9 product types (all data are aggregated across purchaserswithin a product category).

G.R

.Fo

xall

et

al./B

eh

avio

ura

lP

roce

sses

66

(20

04

)2

35

–2

60

249

-0.13 0.11 -0.09 -0.07 -0.05 -0.03 -0.01Average Price Ratio (logarithms)

-0.15

-0.10

-0.05

0.00

0.05

Am

ou

nt

Bo

ug

ht

Rat

io (

log

arit

hm

s)

Average Price Ratio (logarithms)Am

ou

nt

Bo

ug

ht

Rat

io (

log

arit

hm

s)


ou

nt

Bo

ug

ht

Rat

io (

log

arit

hm

s)


ou

nt

Bo

ug

ht

Rat

io (

log

arit

hm

s)


ou

nt

Bo

ug

ht

Rat

io (

log

arit

hm

s)


ou

nt

Bo

ug

ht

Rat

io (

log

arit

hm

s)


ou

nt

Bo

ug

ht

Rat

io (

log

arit

hm

s)


ou

nt

Bo

ug

ht

Rat

io (

log

arit

hm

s)Average Price Ratio (logarithms)A

mo

un

t B

ou

gh

t R

atio

(lo

gar

ith

ms)

Baked Beans

-0.29 -0.28 -0.27 -0.26 -0.25 -0.24

-1.5

-1.4

-1.3

-1.2

-1.1

-1.0

-0.9

Cookies

-0.222 -0.221 -0.220 -0.219 -0.218 -0.217 -0.216

-1.0

-0.8

-0.6

-0.4

-0.2

Butter

-0.18 -0.17-0.16-0.15-0.14-0.13-0.12

-1.00

-0.95

-0.90

-0.85

Cereals

-0.03 -0.01 0.01 0.03 0.05 0.07

-1.2

-1.1

-1.0

-0.9

-0.8

Cheese

-0.12 -0.10 -0.08 -0.06 -0.04

-0.6

-0.5

-0.4

-0.3

Coffee

-0.40 -0.39-0.38-0.37-0.36-0.35-0.34

-0.9

-0.8

-0.7

-0.6

-0.5

-0.4

Fruit Juice

-0.015 0.010 0.035 0.060 0.085 0.110 0.135 0.160

-1.10

-1.05

-1.00

-0.95

-0.90

-0.85

Margarine

-0.048 -0.046 -0.044 -0.042 -0.040 -0.038 -0.036-0.80

-0.75

-0.70

-0.65

-0.60

-0.55

-0.50

Tea

Fig. 4. Relative demand analysis: “VR” schedules. Amount bought ratio as a function of average price ratio for all 9 product types (all data are aggregated across purchaserswithin a product category).

250G

.R.

Foxa

lle

ta

l./Be

ha

viou

ral

Pro

cesse

s6

6(2

00

4)

23

5–

26

0

0.0 0.2 0.4 0.6 0.8 1.0

Relative probability of reinforcement

0.0

0.5

1.0

1.5

2.0

2.5

Am

ou

nt

Bo

ug

ht

Rati

o


Am

ou

nt

Bo

ug

ht

Rati

o


Am

ou

nt

Bo

ug

ht

Rati

o


Am

ou

nt

Bo

ug

ht

Rati

o


Am

ou

nt

Bo

ug

ht

Rati

o


Am

ou

nt

Bo

ug

ht

Rati

o


Am

ou

nt

Bo

ug

ht

Rati

o


Am

ou

nt

Bo

ug

ht

Rati

o


Am

ou

nt

Bo

ug

ht

Rati

o

Baked Beans

0.0 0.2 0.4 0.6 0.8 1.0

0.00

0.05

0.10

0.15

0.20

Cookies

0.0 0.2 0.4 0.6 0.8 1.0

0.0

0.2

0.4

0.6

0.8

1.0

Butter

0.0 0.2 0.4 0.6 0.8 1.0

0.06

0.11

0.16

0.21

Cereals

0.0 0.2 0.4 0.6 0.8 1.0

0.0

0.1

0.2

0.3

Cheese

0.0 0.2 0.4 0.6 0.8 1.0 0.0

0.2

0.4

0.6

0.8

1.0

1.2

Coffee

0.0 0.2 0.4 0.6 0.8 1.0

0.0

0.1

0.2

0.3

0.4

0.5

0.6

Fruit Juice

0.0 0.2 0.4 0.6 0.8 1.00.00

0.05

0.10

0.15

0.20

0.25

Margarine

0.0 0.2 0.4 0.6 0.8 1.0

0.0

0.2

0.4

0.6

Tea

Fig. 5. Maximization analysis: “FR” schedules. Amount bought ratio as a function of the relative probability of reinforcement for all 9 product types(all data are aggregatedacross purchasers within a product category).

G.R

.Fo

xall

et

al./B

eh

avio

ura

lP

roce

sses

66

(20

04

)2

35

–2

60

251

0.0 0.2 0.4 0.6 0.8 1.0


0.6

0.8

1.0

1.2

Am

ou

nt

Bo

ug

ht

Ra

tio


Am

ou

nt

Bo

ug

ht

Ra

tio


Am

ou

nt

Bo

ug

ht

Ra

tio


Am

ou

nt

Bo

ug

ht

Ra

tio


Am

ou

nt

Bo

ug

ht

Ra

tio


Am

ou

nt

Bo

ug

ht

Ra

tio


Am

ou

nt

Bo

ug

ht

Ra

tio


Am

ou

nt

Bo

ug

ht

Ra

tio


Am

ou

nt

Bo

ug

ht

Ra

tio

Baked Beans

0.0 0.2 0.4 0.6 0.8 1.0

0.04

0.06

0.08

0.10

0.12

Cookies

0.0 0.2 0.4 0.6 0.8 1.00.0

0.2

0.4

0.6

0.8

Butter

0.0 0.2 0.4 0.6 0.8 1.0

0.10

0.11

0.12

0.13

0.14

0.15

Cereals

0.0 0.2 0.4 0.6 0.8 1.0

0.05

0.07

0.09

0.11

0.13

0.15

Cheese

0.0 0.2 0.4 0.6 0.8 1.00.21

0.31

0.41

0.51

Coffee

0.0 0.2 0.4 0.6 0.8 1.0

0.10

0.15

0.20

0.25

0.30

0.35

Fruit Juice

0.0 0.2 0.4 0.6 0.8 1.0

0.08

0.10

0.12

0.14

Margarine

0.0 0.2 0.4 0.6 0.8 1.00.15

0.20

0.25

0.30

Tea

Fig. 6. Maximization analysis: “VR” schedules. Amount bought ratio as a function of the relative probability of reinforcement for all 9 product types(all data are aggregatedacross purchasers within a product category).


3.2. Individual patterns of choice acrossinformational reinforcement levels

To demonstrate how individuals choose across dif-ferent informational reinforcement levels,Fig. 7showsthe percentage of the total quantity of goods boughtof brands at each informational level by each con-sumer for each product category. In the figure, the

Fig. 7. Percentage of quantity purchased of brands at each informational level (Level 1: black bars; Level 2: empty bars; Level 3: stripedbars) by each consumer of each product category as a function of average price paid per consumer.

black, empty, and striped bars represent the percentagebought of brands classified at informational Levels 1,2 and 3, respectively. Each vertical bar in the figurerepresents data for one consumer. Data for individ-ual consumers were plotted as a function of the aver-age price (total amount spent divided by total quantitybought) paid by each consumer during the 16-weekperiod. Wider or narrower bars in the figure indicate


larger and smaller numbers of consumers included inthe analysis of different product categories.

In general, increases in average price paid were as-sociated with decreases in the percentage of brandsbought at Level 1 of informational reinforcement andincreases in the percentage of brands bought at Level3. Considering that the average brand price was one ofthe criteria to classify brands at different informationallevels, this may seem a trivial finding: clearly, by def-inition, the more the consumers buy Level-3-brandsthe higher should be the average price they paid. How-ever, when one considers that the figure shows data forindividual consumers, some non-trivial findings canbe noted. First, it becomes clear that most consumersbought mostly brands at one particular informationallevel, rather than across all levels. The percentage ofconsumers who bought 70% or more of goods at oneparticular informational level is: for baked beans 92%,tea 91%, coffee 84%, margarine 84%, butter 81%, ce-reals 68%, fruit juice 68%, cheese 64%, and cookies58%. This indicates that the majority of consumersmake 70% or more of their purchases within one par-ticular informational level.

A second non-trivial aspect of the data is the factthat, when buying across informational levels, con-sumers tend to buy more brands at adjacent informa-tional levels than at more distant levels (e.g., buyingLevels 1 and 2 more than Levels 1 and 3). A thirdrelevant tendency shown in the figure is the wide dif-ference in the average price paid across consumers,with some consumers buying mostly the cheapestbrands while others bought the most expensive ones.This finding could be deduced from the patterns ofbuying mostly brands at the same informational level,just described above, but it is not a trivial one, forit suggests that consumers’ brand-repertoires may beinfluenced by economic variables such as consumer’sbudget. This has not been reported in the literaturethat describes consumers’ multi-brand buying pat-terns. Similar analyses also indicate that, for 8 of 9product categories, most consumers also made thelarge majority of their purchases within the same levelof utilitarian reinforcement. The percentage of con-sumers that bought 70% or more of brands belongingto the same utilitarian level is: for butter 91%, bakedbeans 85%, coffee 84%, tea 84%, cheese 82%, fruitjuice 77%, margarine 74%, cereals 66%, and cookies,42%.

3.3. Consumer groups

These findings invite comparison of the buying pat-terns of consumers grouped by their predominant pur-chasing of brands having specific patterns of informa-tional and utilitarian reinforcement. Hence, consumerswere classified in one of six groups, derived from thecombination of the three levels of informational andthe two levels of utilitarian reinforcement, on the basisof the informational-utilitarian level of the brands theybought more frequently. The six groups were named asfollows: Group 1—Informational Level 1 and Utilitar-ian Level 1; Group 2—Informational Level 1 and Util-itarian Level 2; Group 3—Informational Level 2 andUtilitarian Level 1; Group 4—Informational Level 2and Utilitarian Level 2; Group 5—Informational Level3 and Utilitarian Level 1; and Group 6—InformationalLevel 3 and Utilitarian Level 2.

Groups’ buying patterns were compared in terms ofelasticity of demand, using the equation:

log Quantity= a − b(log Price) (2)

as suggested byKagel et al. (1995). Some modifica-tions of the measures of quantity and price were nec-essary for the following reasons. First, price variationthroughout the 16-week period was not very wide andcan be expected to be even less so within each con-sumer group, since the classification of individuals insuch groups was dependent upon the informationallevel of the brands they bought most frequently,which in turn were classified in part based on theiraverage price. Therefore, each consumer group canbe expected to have a different price average withina relatively restricted range of prices. Second, theanalysis of purchases of brands by a particular con-sumer group for each product category would reducedramatically the number of data points available tocalculate price elasticities. For example, in the case ofbaked beans, there was no consumer classified in con-sumer Group 2, which would restrict the analysis forthe product category. One possible solution for thisproblem would be to aggregate all the data obtainedfrom all the products and then calculate price elastic-ities for each consumer group. This solution wouldpose another type of measurement problem. Consid-ering that the measurement scales (and even units)of quantity and price varied greatly among product


categories, it would be difficult to calculate one singleregression line using data from different products.

One way of overcoming all such problems wouldbe to use measures of quantity and price relative tothe average of each consumer group (e.g.,Bell et al.,1999). These relative measures can be calculated by di-viding the quantities bought (and prices paid) on eachpurchase by the average quantity bought (and aver-age price paid) of each product within each consumergroup. The resulting data would provide an estimate ofprice elasticity relative to the consumer group mean,that is, it would provide an estimate of changes inquantities as a function of changes in prices above andbelow the mean of each consumer group. Data fromeach product would be “standardized” to the prod-uct mean for each group, yielding unitless ratio val-ues above and below 1.0. Data from all products andgroups would become comparable in terms of respon-siveness around the mean.

This procedure was adopted in the analyses de-scribed next. Each quantity data point for the regres-sion was calculated by dividing the quantity boughton a shopping occasion by the average quantity forthat specific consumer group for that specific product.Analogously, each price data point for the regressionconsisted of the price paid on a given shopping oc-casion divided by the average price paid by that spe-cific group when buying that specific product. Then,for example, the quantity bought of Tesco Value© in-stant coffee by a specific consumer on a given shop-ping trip was divided by the average quantity of in-stant coffee bought by all consumers in Group 1 (In-formational and Utilitarian Level 1). This same proce-dure was used to calculate the correspondent measuresof price. A regression analysis was then conductedwith all data points obtained for all consumers classi-fied in Group 1, including data points from all prod-

Table 6Parameters of equation (log Quantity= a − b(log Price)), calculated for each consumer group, the significance level of the regression (P),and the standard error of the estimate ofb

Consumer group R2 a b S.E. P

Group 1 0.22 −0.13 −0.50 0.038 <0.000Group 2 0.42 −0.15 −0.86 0.076 <0.000Group 3 0.46 −0.17 −0.73 0.027 <0.000Group 4 0.35 −0.13 −0.59 0.030 <0.000Group 5 0.25 −0.12 −0.66 0.041 <0.000Group 6 0.22 −0.06 −0.41 0.033 <0.000

uct categories. The same was done with the data forthe other five consumer groups (the number of paireddata points,N, for the six groups ranged from 179 to897).

The results are shown inTable 6. All regressions,using the above equation (calculated with relativemeasures of quantity and price), were statisticallysignificant (i.e.,P < 0.000 for all groups). The valuesof R2 were not very large and ranged from 0.22 to0.46, indicating that other variables that did not enterthe equation also influenced the quantities consumersbought. The values of standard error were all 10 ormore times smaller than the corresponding coeffi-cients of price elasticity,b, suggesting accurate esti-mations of the latter. All price elasticity coefficientswere negative indicating that the quantity consumersbought tended to decrease with increases in price.Moreover, all coefficient values were between 0 and−1.0, indicating that demand was inelastic for allconsumer groups. Despite these similarities, the abso-lute values of elasticity coefficients, shown inFig. 8,were lower for the extreme groups, Groups 1 and 6,than for the other groups, suggesting that consumersthat buy predominantly intermediate-level brandsshowed higher price responsiveness than those buyingpredominantly the least- and highest-differentiatedbrands (split-sample reliability analyses confirm thistrend).

3.4. Intra- and inter-brand elasticities

The observed decreases in the quantity bought withincreases in prices, indicated by negative elasticitycoefficients, may, however, have been associated withdifferent response patterns by different groups. Thetendency to buy larger quantities when prices arelower may be related to one or more of the follow-


Informational/Utilitarian Consumer Group

6543210

Price E

lasticity C

oeffic

ients

.9

.8

.7

.6

.5

.4

.3

.2

.1

0.0

Fig. 8. Price elasticity coefficients calculated for each group of consumers classified according to the informational and utilitarian level ofthe brands they predominantly purchased.

ing three patterns: (1) buying larger quantities of aproduct when its price was below its usual, average,price rather than when its price was above its aver-age price (i.e., intra-brand or absolute elasticity); (2)buying larger quantities when buying brands belong-ing to cheaper, lower informational levels than whenbuying brands belonging to more expensive, higherinformational levels (i.e., informational inter-brandor relative elasticity); and (3) buying larger quanti-ties when buying brands belonging to cheaper, lowerutilitarian levels than when buying brands belongingto more expensive, higher utilitarian levels (i.e., util-itarian inter-brand or relative elasticity). One way ofmeasuring such patterns is to decompose the globalprice elasticity coefficient into three different coeffi-cients, namely, intra-brand, informational inter-brand,and utilitarian inter-brand coefficients. This analysiswould yield an equation in which the quantity boughtwould be a function of intra-brand changes in price,informational reinforcement levels of the purchasedbrands, and the utilitarian reinforcement levels of the

purchased brands, that is,

log Quantity= a − b1(log Intra-Brand Price)

− b2(log Informational Level)

− b3(log Utilitarian Level) (3)

Intra-brand price was obtained by dividing theprice paid for the brand by the average price for thatsame brand in the sample. Relative values of quantity,intra-brand price, informational level and utilitarianlevel with respect to their respective consumer groupaverages, analogous to those used to obtain globalelasticity coefficients, were used. Regression coeffi-cients were obtained for each consumer group.

The results are summarized inTable 7. All regres-sions were statistically significant (i.e.,P < 0.000 forb1 for all groups). The values ofR2 were not very largeand ranged from 0.06 to 0.36, indicating that othervariables that did not enter the equation also influencedthe quantities consumers bought. Only three, out of 18,values of standard error were 10 or more times smaller


Table 7Parameters of log Quantity= a − b1(log Intra-Brand Price) − b2(log Informational Level) − b3(log Utilitarian Level), calculated for eachconsumer group, the significance level of the regression (P), and the standard error of the estimates ofb1, b2, andb3

Consumer group R2 a b1 S.E. P b2 S.E. P b3 S.E. P

Group 1 0.17 −0.11 −0.53 0.084 <0.000 −0.33 0.061 <0.000 −0.61 0.082 <0.000Group 2 0.36 −0.13 −1.51 0.252 <0.000 −0.84 0.113 <0.000 −0.32 0.141 0.024Group 3 0.30 −0.13 −0.92 0.067 <0.000 −0.52 0.078 <0.000 −0.72 0.063 <0.000Group 4 0.15 −0.10 −0.74 0.078 <0.000 −0.39 0.072 <0.000 −0.24 0.070 0.001Group 5 0.21 −0.12 −0.70 0.063 <0.000 −0.29 0.080 <0.000 −0.69 0.074 <0.000Group 6 0.06 −0.05 −0.58 0.117 <0.000 −0.09 0.083 0.291 −0.23 0.074 0.002

than the corresponding coefficients of price elastic-ity, b, suggesting that coefficient estimations were notvery accurate (although split-sample reliability anal-yses corroborated the observed patterns). Collinearityanalyses yielded values of tolerance and variance in-flation factor close to 1.00, suggesting that there wasno significant covariance among variables included inthe equation (Hair et al., 1995). All price elasticitycoefficients were negative indicating that the quantityconsumers bought tended to decrease with increases inintra-brand price variations, informational level, and

Informational/Utilitarian Consumer Group

6543210

Price E

lasticity C

oeffic

ients

1.6

1.4

1.2

1.0

.8

.6

.4

.2

0.0

intra-brand

informational

utilitarian

Fig. 9. Intra-brand, informational inter-brand and utilitarian inter-brand price elasticity coefficients calculated for each group of consumers,classified on the basis of the informational/utilitarian level of brands they predominantly purchased.

utilitarian level. Moreover, with the exception of theintra-brand coefficient for Group 2 (−1.51), all coeffi-cient values were between 0 and−1.0, indicating thatall three types of demand tended to be inelastic for allconsumer groups. Despite such similarities, the abso-lute values of intra-brand, informational inter-brand,and utilitarian inter-brand elasticity coefficients dif-fered across consumer groups, as shown inFig. 9.

Intra-brand elasticity coefficients were lower forGroups 1 and 6 than for the intermediate groups, show-ing a decreasing trend from Group 2 to Group 6.


This suggests that consumers buying predominantlythe cheapest, least-differentiated brands (i.e., Group 1)do not change much the quantity they buy as a func-tion of changes in brand price relative to their usual(average) price. This result suggests a tendency to-ward buying the cheapest brands, irrespective of other,slightly more expensive, brands. If this interpretationis correct, the observed pattern for intra-brand elas-ticity, which was largest for Group 2 and decreasedsystematically as group classification increased up toGroup 6, can be interpreted as suggesting that respon-siveness to intra-brand changes in price decreases asgroup classification increases. In other words, if thelow intra-brand elasticity observed for Group 1 is aconsequence of buying the cheapest brands most ofthe time, these findings point to the conclusion thatas the level of differentiation of the purchased brandsincreases (i.e., as the price of purchased brands in-creases), the responsiveness of consumers to changesin prices decreases.

Informational inter-brand elasticities were smallerthan intra-brand elasticities for all six groups and fol-lowed a similar pattern, with Group 1 showing a lowcoefficient, Group 2 showing the largest one whichdecreases systematically with increases in groupclassification up to Group 6. This suggests that con-sumers buying mostly the least-differentiated, cheap-est brands do not change much the quantities they buyas a function of informational brand level, whereasthe responsiveness to informational reinforcement ofthose buying intermediate-level brands decreases sys-tematically with increases in the informational levelof the predominantly purchased brands. This valueis close to 0 for Group 6, suggesting that consumersthat already usually buy the highest informational andutilitarian level brands are not sensitive to changesin informational level (similar to a “satiation” effect,since satiated animals are not expected to be respon-sive to food, i.e., to do things to get food).

Utilitarian inter-brand elasticity, indicated by thefilled circles, was higher for the three groups thatbought predominantly low utilitarian-level brands (i.e.,Groups 1, 3 and 5) than for the other three that boughthigh utilitarian-level brands. This finding indicatesthat consumers who buy predominantly brands withlow utilitarian levels tend also to buy smaller quan-tities of higher utilitarian brands, whereas those thatbuy predominantly brands with high utilitarian levels

do not seem to vary much the quantities they buy as afunction of utilitarian brand level. Hence, the utilitar-ian inter-brand elasticities followed a slightly differentpattern from the informational inter-brand elastici-ties, though like them they were mostly smaller thanintra-brand elasticities. Group 1 is the only exceptionwith a coefficient larger than that of the intra-brandelasticities, if only marginally. Whereas the other twocurves follow a similar pattern, the shape of the util-itarian curve is different in that it follows a zigzagcourse with Group 2 showing a lower coefficient thanGroups 1 and 3, and similarly Groups 4 and 6 dis-playing a lower coefficient than their neighbor groups.The implications of this pattern are more complicatedbecause it suggests that consumers buying mostly atutilitarian Level 1, i.e., Groups 1, 3 and 5, are moresensitive to changes in utilitarian level than consumerswith a preference for utilitarian Level 2, independentof the informational level of the brand. For example,consumers who mostly buy the least-differentiated,cheapest brands (i.e., Group 1) are more likely to buylarger quantities than consumers who buy at but ahigher utilitarian but at the same informational level(i.e., Group 2). Consumers of Group 3 however, witha lower utilitarian level than Groups 2 and 4 but ahigher informational level than Group 2 and the sameinformational level as Group 4, is in turn more respon-sive to utilitarian reinforcement than both Groups 2and 4.

4. General discussion

As predicted by both matching theory and maxi-mization theory, we have confirmed that choice onconc VR VR schedules exhibits both matching andmaximizing. However, the examination of consumerchoice in naturalistic environments raises a numberof complications for behavior analysis and behavioraleconomics that are not evident from the experimentalanalysis of choice. While the realities of consumer be-havior in affluent, marketing-oriented economies haveimplications for behavioral economics, the techniquesof analysis which behavioral economics makes avail-able to the marketing researcher also elucidate the na-ture of brand choice in the market place.

A common assumption in aggregate studies of con-sumer choice conducted by marketing scientists is that


brands within a product category are functional al-ternatives and that consumers will include a brandwithin their repertoire or purchase set only if it em-bodies the physical and functional benefits that arecommon to all members of that category (Ehrenberg,1972, 1993). This proposition is seldom supported byempirical evidence. Although the discovery of match-ing on conc VR VR schedules is both expected andperhaps in some respects trivial, it is important forthe sort of analysis we have undertaken in that it con-firms that the alternative brands considered are indeedsubstitutes in the assumed sense. The very-nearly per-fect matching that we have found is a characteristic ofchoices that are near-perfect substitutes (Kagel et al.,1995).

Another common assumption in the marketing lit-erature is that price plays a relatively small part inthe determination of consumer choice: brands thatare highly differentiated by advertising command apremium but the consumer is generally portrayed asrelatively insensitive to such differentials. Non-priceelements of the marketing mix (i.e., promotional tac-tics, brand attributes, and distribution strategies) arethought to be more influential than price factors for af-fluent consumers operating within marketing-orientedeconomies (Foxall, 1999b). The relative demandand maximization analyses, which were intended toshed light on the sensitivity of consumer demand toprice differentials among competing brands, presentan equivocal impression of the relationship betweenmarket prices and quantity demanded. WhileFigs. 3and 4indicate the expected relationship, denoted bydownward-sloping relative demand curves, the evi-dence for the remaining product categories is mixed.The maximization analysis suggests that consumersare in some respects sensitive to price levels whenmaking decisions about how much of a brand tobuy relative to other brands in the consideration set.However, the interpretation of the data summarizedin these figures and inFigs. 5 and 6must includethe phenomenon of single shopping trip multi-brandpurchasing. Although a consumer may exhibit eco-nomically rational price sensitivity by purchasing thecheapest brand in her consideration set, her generalsensitivity to price may be confounded by her pur-chasing a premium-priced alternative at the sametime. Hence, our results are equivocal on the questionwhether consumer brand choice is sensitive to price.

The consequent need to attend to non-price elementsof the marketing mix led us to the analyses of priceelasticity of demand which take into considerationboth the utilitarian (functional) and informational(symbolic) benefits gained by consumers from thebrands they purchase and use.

The evidence is that consumers choose their reper-toire of brands on the basis of the informational andutilitarian level of reinforcement programmed by thebrands. This is likely to be related, among other things,to their budgets, which we were not able to take intoconsideration. However, it is also of marketing sig-nificance in that it provides opportunities for the par-titioning (segmentation) of markets. There do seemto be clearly definable segments based on combina-tions of the utilitarian and symbolic benefits of pur-chase and consumption and the cost minimization.These factors encourage consumers to choose brandswithin a given range defined in terms of these vari-ables. Most purchasing takes place within a fairlynarrowly defined range and consumers who switchout of that range generally move only to an adjacentrange.

Consumer groups, classified on the basis of theinformational/utilitarian level of the brands they buymostly, show different responsiveness to changes inprices, with extreme groups showing the lowest lev-els of responsiveness (possibly for different reasons).Price elasticities can be decomposed into intra-brandand at least two types of inter-brand elasticities, in-formational and utilitarian, according to the type ofreinforcing events that influence consumer choice.Intra-brand elasticity can be interpreted as a measureof responsiveness to the aversive consequences ofgiving up money (Alhadeff, 1982). Therefore, choicepatterns can be interpreted as being determined bydifferent combinations of the tendencies to avoidaversive consequences, maximize informational re-inforcement and maximize utilitarian reinforcement.A pattern that minimizes financial loss, showingminimum responsiveness to informational attributesand some to utilitarian ones, seems to characterizechoices of consumers in Group 1. The responsivenessto informational and utilitarian attributes related tochanges in price seems to be an inverse function ofhow much of these the consumer obtains regularly.So, the results showed increasing responsiveness toinformational reinforcement from Group 6 (who ob-


tain higher levels of it) to Group 2 (who obtain lowerlevels of it). The same was observed for utilitarianattributes, for those groups buying lower levels of util-itarian attributes (Groups 1, 3, and 5) showed higherresponsiveness to this aspect of the brands than thosethat buy higher levels of utilitarian attributes moreregularly (Groups 2, 4, and 6).

Elasticity coefficients can be interpreted as mea-sures of consumer “satiation” level, since the lessfrequently consumers purchase a given reinforcingdimension the higher their responsiveness to thatdimension. In the case of intra-brand elasticity thistendency is probably related to available budget. Theonly exceptions were obtained for consumers that buythe least differentiated brands most of the time, forwhom elasticity coefficients seem to reflect a patternof buying the cheapest products in the category.

From the point of view of theBehavioral Per-spective Model, the analysis has demonstrated thatrelatively high and low utilitarian and informationalreinforcement can be used to classify consumer be-havior even within the narrow range represented byfast-moving consumer goods. In previous analyses,these variables, along with the relative openness ofthe consumer behavior setting, have been employedto categorize broader patterns of consumer behav-ior. Within that categorization, the purchase of foodproducts is classified in terms of low utilitarian andlow informational reinforcement in a relatively opensetting. That categorization is meaningful when buy-ing fast-moving consumer goods is compared withbuying and using other kinds of product and service(Foxall and Yani-de-Soriano, 2004), but the demon-stration of this paper is that these structural elementsof the consumer situation also provide means of clas-sifying consumer behavior within those broader cate-gories.

The results for the elasticities of demand, es-pecially those for intra-brand, inter-utilitarian andinter-informational elasticities, suggest that the ex-planatory variables investigated are far from the onlyinfluences on brand choice. Nevertheless, along withthe results for the inter-group elasticities of demandwhich provide somewhat stronger evidence of a link,they indicate that utilitarian and informational rein-forcement have distinct effects on brand choice andthat they may form the basis of the partitioning ofmarkets and strategies of market segmentation.

Acknowledgements

We gratefully acknowledge grants (SGS/LB/0431/A& SGS/00493/G/S1) from The Nuffield Foundation,London, to Foxall in support of this research, and toCAPES and CNPq, Brazil, for financial support toOliveira-Castro.

References

Alhadeff, D.A., 1982. Microeconomics and Human Behavior:Toward a New Synthesis of Economics and Psychology.University of California Press, Berkeley, CA.

Baum, W.M., 1974. On two types of deviation from the matchinglaw. J. Exp. Anal. Behav. 22, 231–242.

Baum, W.M., 1979. Matching, undermatching, and overmatchingin studies of choice. J. Exp. Anal. Behav. 32, 269–281.

Baum, W.M., Nevin, J.A., 1981. Maximization theory: someempirical problems. Behav. Brain Sci. 4, 389–390.

Bell, D.R., Chiang, J., Padmanabhan, V., 1999. The decompositionof promotional response: an empirical generalization. Market.Sci. 18, 504–526.

Churchill Jr., G.A., 1999. Marketing Research: MethodologicalFoundations. The Dryden Press, Fort Worth, TX.

Crouch, S., Housden, M., 2003. Marketing Research for Managers.Butterworth-Heinemann, Oxford.

Dall’Olmo Riley, F., Ehrenberg, A.S.C., Castleberry, C.B.,Barwise, T.P., Barnard, N., 1997. The variability of attitudinalrepeat-rates. Inter. J. Res. Market. 14, 437–450.

De Chernatony, L., McDonald, M., 2003. Creating PowerfulBrands. Elsevier/Butterworth-Heinemann, Oxford.

Ehrenberg, A.S.C., 1972. Repeat-Buying: Theory and Applications.North-Holland Publishing Company, Amsterdam.

Ehrenberg, A.S.C., 1986. Pricing and brand differentiation.Singapore Market. Rev. 1, 5–13.

Ehrenberg, A.S.C., 1993. New brands and the existing market. J.Market Res. Soc. 33, 285–299.

Ehrenberg, A.S.C., England, L.R., 1990. Generalising a pricingeffect. J. Ind. Econ. 39, 47–68.

Ehrenberg, A.S.C., Scriven, J., 1999. Brand loyalty. In: Earl, P.E.,Kemp, S. (Eds.), The Elgar Companion to Consumer Researchand Economic Psychology. Edward Elgar, Cheltenham,Gloucestershire, pp. 53–63.

Ehrenberg, A.S.C., Goodhardt, G.J., Barwise, P., 1990. Doublejeopardy revisited. J. Market. 54, 82–91.

Ehrenberg, A.S.C., Hammond, K., Goodhardt, G.J., 1994. Theafter-effects of price-related consumer promotions. J. Advert.Res. July/August, 11–21.

Foxall, G.R., 1990. Consumer Psychology in BehaviouralPerspective. Routledge, New York.

Foxall, G.R., 1994. Environment-impacting consumer behaviour:a framework for social marketing and demarketing. In: Baker,M.J. (Ed.), Perspectives on Marketing Management. Wiley,Chichester, pp. 27–56.


Foxall, G.R., 1996. Consumers in Context: The BPM ResearchProgram. Routledge, New York.

Foxall, G.R., 1997. Marketing Psychology: The Paradigm in theWings. Macmillan, London.

Foxall, G.R., 1998. Radical behaviorist interpretation: generatingand evaluating an account of consumer behavior. Behav. Anal.21, 321–354.

Foxall, G.R., 1999a. The substitutability of brands. Manage. Decis.Econ. 20, 241–258.

Foxall, G.R., 1999b. The marketing firm. J. Econ. Psych. 20, 207–234.

Foxall, G.R., James, V.K., 2001. The behavioral ecology of brandchoice: a preliminary analysis. Eur. J. Behav. Anal. 2, 209–220.

Foxall, G.R., James, V.K., 2003. The behavioral ecology of brandchoice: how and what do consumers maximize? Psych. Market.20, 811–836.

Foxall, G.R., Schrezenmaier, T.C., 2003. The behavioral economicsof consumer brand choice: establishing a methodology. J. Econ.Psych. 24, 675–695.

Foxall, G.R., Yani-de-Soriano, M.Y., 2004. Situational influenceson consumers’ attitudes and behavior. J. Bus. Res., in press.

Goodhardt, G.J., Ehrenberg, A.S.C., Chatfield, C., 1984. TheDirichlet: a comprehensive model of buying behaviour. J. RoyalStat. Soc. A147, 621–643.

Hair, J.F., Anderson, R.E., Tatham, R.L., Black, W.C., 1995.Multivariate Data Analysis. Prentice-Hall, Englewood Cliffs,NJ.

Herrnstein, R.J., 1961. Relative and absolute strength of responseas a function of the frequency of reinforcement. J. Exp. Anal.Behav. 4, 267–272.

Herrnstein, R.J., 1970. On the law of effect. J. Exp. Anal. Behav.13, 243–266.

Herrnstein, R.J., 1982. Melioration as behavioral dynamism.In: Commons, M.L., Herrnstein, R.J., Rachlin, H. (Eds.),Quantitative Analyses of Behavior. Vol. II: Matchingand Maximizing Accounts. Ballinger, Cambridge, MA,pp. 433–458.

Herrnstein, R.J., Loveland, D.H., 1975. Maximizing and matchingon concurrent ratio schedules. J. Exp. Anal. Behav. 24, 107–116.

Herrnstein, R.J., Prelec, D., 1991. Melioration: a theory ofdistributed choice. J. Econ. Perspect. 5, 137–156.

Herrnstein, R.J., Vaughan, W., 1980. Melioration and behavioralallocation. In: Staddon, J.E.R. (Ed.), Limits to Action: TheAllocation of Individual Behavior. Academic Press, New York,pp. 143–176.

Houston, A.I., McFarland, D.J., 1980. Behavioral resilience and itsrelation to demand functions. In: Staddon, J.E.R. (Ed.), Limits

to Action: The Allocation of Individual Behavior. AcademicPress, New York, pp. 177–204.

Hursh, S.R., 1980. Economic concepts for the analysis of behavior.J. Exp. Anal. Behav. 34, 219–238.

Hursh, S.R., Bauman, R.A., 1987. The behavioral analysisof demand. In: Green, L., Kagel, J.H. (Eds.), Advancesin Behavioral Economics, vol. 1. Ablex, Norwood, NJ,pp. 117–165.

Jary, M., Wileman, A., 1998. Branding: The New Wealth Creators.Palgrave Macmillan, Basingstoke, Hants.

Kagel, J.H., Battalio, R.C., Green, L., 1995. Economic ChoiceTheory: An Experimental Analysis of Animal Behavior.Cambridge University Press, Cambridge.

Kagel, J.H., Battalio, R.C., Green, L., Rachlin, H., 1980. Consumerdemand theory applied to choice behavior of rats. In: Staddon,J.E.R. (Ed.), Limits to Action: The Allocation of IndividualBehavior. Academic Press, New York, pp. 237–267.

Keng, K.A., Ehrenberg, A.S.C., 1984. Patterns of store choice. J.Market. Res. 21, 399–409.

Krishnamurti, L., Raj, S.P., 1988. A model of brand choice andpurchase quantity price sensitivities. Market. Sci. 7, 1–20.

Madden, G.W., Bickel, W.K., Jacobs, E.A., 2000. Three predictionsof the economic concept of unit price in a choice context. J.Exp. Anal. Behav. 73, 45–64.

Rachlin, H., 1980. Economics and behavioral psychology. In:Staddon, J.E.R. (Ed.), Limits to Action: The Allocation ofIndividual Behavior. Academic Press, New York, pp. 205–236.

Sharp, B., Sharp, A., 1997. Loyalty programs and their impact onrepeat-purchase loyalty patterns. Int. J. Res. Market. 14, 473–486.

Uncles, M., Ehrenberg, A.S.C., 1990. The buying of packagedgoods at US retail chains. J. Retail. 66, 278–293.

Uncles, M., Ehrenberg, A.S.C., Hammond, K., 1995. Patterns ofbuyer behavior: regularities, models, and extensions. Market.Sci. 14, G71–G78.

Vaughan, W., Herrnstein, R.J., 1987. Stability, melioration andnatural selection. In: Green, L., Kagel, J.H. (Eds.), Advancesin Behavioral Economics, vol. 1. Ablex, Norwood, NJ,pp. 185–215.

Watkins, T., 1986. The Economics of the Brand. McGraw-Hill,London.

Wearden, J.H., 1988. Some neglected problems in the analysis ofhuman operant behavior. In: Davey, G.C.L., Cullen, C. (Eds.),Human Operant Conditioning and Behavior Modification.Wiley, Chichester, pp. 197–224.

Zettle, R.D., Hayes, S.C., 1982. Rule-governed behavior: apotential framework for cognitive-behavioral therapy. In:Kendall, P.C. (Ed.), Advances in Cognitive-Behavioral Researchand Therapy. Academic Press, New York, pp. 73–117.

The behavioral economics of consumer brand choice: patterns of reinforcement and utility...

Documents

Transcript of The behavioral economics of consumer brand choice: patterns of reinforcement and utility...