A behavioural theory of brand loyalty is derived in this dissertation from an examination of the literature of consumer behaviour, repetitive behaviour, and related psychological phenomena. The theory proposes that there are three classes of motivation of consumer brand loyalty -learnt brand loyalty, where shoppers are characterised by strength and independence, especially in decision-making, and high perceived risk; conditioned brand loyalty, found amongst shoppers with low self-esteem, a fear

more »

... of change, and subject to certain brand choice demands from reference group members; and economic brand loyalty, where it is money-saving, rather than psychological or social influence factors, that initiates (negative) brand loyalty. The theory has been formulated to predict brand loyalty generally for all low-cost, regularly purchased household products. However, in testing theoretical propositions, we have found differences between products in the variances of the dependent variable, and conclude that these are a major source of the differences between products in the fit of the data to the theoretical propositions. To break down this variability, the product samples were sub-divided into four groups, according to degree of value of the dependent variable. Although there continued to be differences between product sub-groups in the DV variance, more evidence was found in support of the general theory, with a major exception being that conditioned brand loyalty seems to occur only occasionally. Moreover, we have some evidence to suggest that another source of difference in brand loyalty motivation between products lies in market structure and activity variations, e.g., variations in amount of cents-off advertising at point of sale or in the press, in width of product distribution. These differences, we infer, are linked to cognitive (learnt) differences between products and, as well, to the absolute level of brand loyalty for a product. On reviewing our findings, we conclude that there is support for the three-factor theory of brand loyalty, provided that (a) between-product-market differences in structure and activity are regarded as second-order determinants of brand loyalty, and (b) conditioned brand loyalty is hypothesised as occurring only when there is a genuine reference group conformity effect on brand choice. ** This is probably the most common definition employed in marketing, but has been criticised by Day, and especially Jacoby, as noted later in this chapter. Probably the earliest study in this area was the work of Cunningham ( 26 ). Using panel purchase data for seven regularly purchased household products, he found no relationships between brand loyalty and (a) any socio-economic variables, (b) size of purchase. However, he found that purchasing on deals tended to be concentrated amongst those with low brand loyalty. In 1964, Farley (38) reported research that tested hypotheses derived from Stigler's theory of the economics of information, which argued that certain socio-economic household characteristics were associated with brand loyalty. Under this theory, he proposed, inter alia: 1. "If ... brands are generally considered good substitutes for one another, then households which put considerable effort into gathering market information should appear less brand loyal toward all products than those families which search relatively little ...".(p. 370). 2. "If brand preferences are not strong, heavy buyers should also appear less brand loyal than light buyers", (p. 371). 3. "... the higher a household's income, the more a family member values his or her time. (Hence) if brands are considered good substitutes, high income families should be more brand loyal than low income families", (p. 371). Farley found, contrary to expectations, that "high income families tend to be brand disloyal", and concluded that "the hypotheses arising from combining the economics of information with the assumption that consumers consider brands as good substitutes were generally not supported". The only hypothesis supported -on the assumption that heavy buyers search more for "market (price?) information" -was that heavy buyers tend to be disloyal. It should be noted that Farley's assumption that brands were "good substitutes for one another" was not tested, which, if not met, he notes, makes the economics of information model inappropriate. Suggesting avenues for further research, he speculated that store loyalty, promotional activity and point-of-purchase factors may need to be investigated. Kuehn and Frank had earlier made similar suggestions. Abandoning the economics of information approach, Farley (37) re-approached the same data with an ad hoc set of hypotheses, finding that "Consumers as expected tended to be less loyal ... towards products with many brands"available, where number of purchases and dollar expenditures per buyer are high, where prices are relatively active, and where consumers might be expected to simultaneously use a number of brands of the product. As expected, consumers are brand loyal in markets where brands tend to be widely distributed and where market share is concentrated heavily in the leading brand", (p. 13). He was thus able to show that market environmental characteristics such as dealing activity, brand leader dominance, width of distribution, explained a good deal of the variance in loyalty between products. While Farley's researches are particularly important in their contribution to scientific knowledge, it is essential to remember, using summated differences between consumers as his independent variables, that he sought for explanations of the differences between products in respect of overall loyalty to each product. Frank, Douglas and Polli (45) found some stimulation in Farley's explicit theory of the determinants of brand loyalty, which influenced their study of brand loyalty, based on Chicago Tribune panel data for 1961. Using a set of socio-economic, shopping, purchase and consumption variables as predictors, they undertook multiple regression analyses of brand loyalty for 44 products. 23 of the 44 multiple R's were statistically significant, but only 2 six were above 0.20. Examining the proportion of partial r 's for each variable that were significant across all products, they found that the variables most generally associated with brand loyalty were: 1. Total household consumption -positively associated with brand loyalty, and opposite in direction to their hypothesis. 2. Ase of youngest child -negatively associated with brand loyalty. They suggested this might mean that the housewife had more time to search for alternatives. Proportion of purchases in National food stores -negatively associated. The writers could suggest no rationale here, though one might surmise that such stores have a wide brand array and more price specialling, factors which should tend to induce brand switching. 4. Average price per unit -positively associated. They suggest that this may indicate consumers who concentrate their purchases on premium brands, where there are fewer alternatives •k and hence a lower probability of brand switching. This suggestion, they point out, is dependent upon brands not being completely substitutable. 5. Proportion of purchases devoted to small package sizesnegatively associated. They suggest that small package purchasers may be more frequent shoppers, and therefore better informed of merchandising and price variations. (However, this hypothesis may compound inter-purchase time with package size). Although Frank, et al, actually utilised little of Farley's formal theory to generate hypotheses -there was no theoretical justification for most of their variables -they provided in their discussion a number of suggestions (as above) for further research.