A. S. C. Ehrenberg

The Design of Replicated Studies

Abstract Replication is little discussed in the statistical literature nor practiced widely by statistically minded researchers. It is needed not merely to validate ones findings, but more importantly, to establish the increasing range of radically different conditions under which the findings hold, and the predictable exceptions. This article describes how to design highly differentiated replications. The irrelevance and/or impossibility of identical replications are also discussed. Practical illustrations of the success and failure of replicated studies are given.

Repetitive Advertising and the Consumer

ABSTRACT This paper is one of 18 selected by the Editorial Review Board of The Journal of Advertising Research to be a ‘classic’ - an article that has withstood the test of time. First published in 1974, Ehrenberg examines the role of advertising by looking at advertising and consumption in general, then discussing competition among brands and the factors affecting brand choice, particularly for established brands of frequently bought goods. He concludes the advertisings main role is to reinforce feelings of satisfaction with brands already bought.

Rudiments of Numeracy

SUMMARY Many tables of data are badly presented. It is as if their producers either did not know what the data were saying or were not letting on. Some precepts for improved data presentation are discussed.

Brand Advertising As Creative Publicity

ABSTRACT Our view of brand advertising is that it mostly serves to publicize the advertised brand. Advertising seldom seems to persuade. Advertising in a competitive market needs to maintain the brands broad salience-being a brand the consumer buys or considers buying. This turns on brand awareness, but together with memory associations, familiarity, and brand assurance. Publicity can also help to develop such salience. This publicity view of advertising should affect both the briefs that are given to agencies (e.g., that cut-through is more important than having a persuasive selling proposition) and how we then evaluate the results. But since few advertisements seem actively to seek to persuade, how much do the advertisements themselves have to change, rather than just how we think and talk about them?

Market segmentation for competitive brands

Although market segmentation is widely described as a major marketing tool, questions whether brands which are broadly similar and competitive are bought by identifiably different consumer segments. Notes that few, if any, examples of marked brand segmentation are cited in the literature. Reports on a new international study of the characteristics of brand purchasers in over 20 grocery product categories using consumer panel data, which reveals that there is little brand segmentation. Finds that the consumer profiles of competitive brands differ little in terms of the commonly‐used classification measures such as socio‐demographic characteristics, and that brands in the same product category tend to be bought by similar kinds of people.

Predictability and prediction

A result can be regarded as routinely predictable when it has recurred consistently under a known range of different conditions. This depends on the previous analysis of many sets of data, drawn from different populations. There is no such basis of extensive experience when a prediction is derived from the analysis of only a single set of data. Yet that is what is mainly discussed in our statistical texts. The paper discusses the design and analysis of studies aimed at achieving routinely predictable results. It uses two running case history examples

A replication study of two brand-loyalty measures

Abstract In a systematic check across 34 US product categories, two standard measures of brand-loyalty are found to be closely predictable from the Dirichlet model of buyer behaviour in most cases. This means that market share is generally the dominant factor, but that there are also certain submarkets and isolated deviations. The general role of replication studies is also briefly considered.

The Problem of Numeracy

Abstract Lack of numeracy is due mainly to the way data are presented. Most tables of data can be improved by following a few simple rules, such as drastic rounding, ordering the rows of a table by size, and giving a brief verbal summary of the data.

The Elements of Lawlike Relationships

SUMMARY Much of what we know in science is as yet confused by uncertainty. But implicit in our observed phenomena are many lawlike relationships saying that if this occurs, so does that. The systematic uncovering of such relationships and their subsequent applications to practical problems are how science and how technology progress. The meaning of a lawlike relationship lies in the variety of empirical conditions under which it is known to hold. Such generalized or lawlike relationships are of course familiar and widely used in practice, but the statistical literature does not give much explicit help in deriving them. The present paper tries to set out some elementary aspects by means of a simple case-history. 1.1. Generalized Relationships A LAWLIKE relationship between two observed variables x and y describes how variable x varies with variable y. Examples are the way the pressure of a gas tends to vary with its volume, how the distance travelled by a falling body increases with the time taken, or how childrens height is related to their weight. Lawlike relationships of this kind are simple and elementary things. They are used to describe and understand and predict relationships between observable phenomena. The essence of a lawlike relationship is that the equation in question-say that y = ax+ b-describes approximately how the variable y has been found to vary with variable x under different specified conditions of observation. A case-history illustration is given in Section 2 in terms of the relationship log w = 08h + 04 between childrens height h and weight w. This relationship is known to hold for children of different ages from 5 years to 13 years, for boys and for girls, for children in higher, middle and lower socio-economic classes, for children in Birmingham, in Canada and in France, and pre-war, during the war and post-war. Fig. 1 illustrates some of the different kinds of data for which this equation holds. The basic problem which needs to be considered in this paper is therefore the fitting of an equation to different sets of data in the same variables. Three or four other examples of simple quantitative laws may help to clarify what is at issue. First, a typical case from physics:

Theory or Well-Based Results: Which Comes First?

I have been asked by the editors to contrast two current research traditions in marketing, what I call the theoretical-in-isolation or TiI and the empirical-then-theoretical or EtT approaches. First I will define the two approaches, give examples especially of the less well-documented EtT, describe a more advanced research tradition (ETET...), and comment on my rather black-and-white evaluation of TiI (no good) and EtT (good). I will then set out the main characteristics of the two traditions as I see them, illustrate differences between TiI and EtT in terms of research into buying behavior, and note how EtT, unlike TiI, seems to have led to routinely predictable results, to grounded theory, and to a range of practical applications.