1. INTRODUCTION

The theories of persuasive advertising that are still largely accepted today focus on brand differentiation and emphasize change (e.g. “adding values”). Consequently brand advertising is supposed to portray the brand as different and preferably better than its competitors, thus changing or intensifying consumers’ thoughts and feelings about the brand and hence affecting their behavior towards it.

But the content of many advertisements and extensive evidence on the general nature of consumers’ attitudes to brands are not at all consistent with this “persuasive” view of advertising and its effects in the consumer’s mind.

In particular:

1. Many advertisements do not appear to embody persuasive messages and thus are not overtly trying to persuade;
2. Customers’ beliefs about their brands are if anything rather similar from brand to brand. The brands tend to be seen by consumers as competing and more or less substitutable, rather than really differentiated;
3. In the medium term, aggregate brand attitudes are generally stable. When they do change such changes follow rather than precede the associated change in behavior. This casts serious doubt on the supposedly causal powers of consumers’ brand attitudes.

Section 2 looks at the content of advertisements, at what they actually say to people or are generally taken to imply. A simple typology is introduced to classify advertisements into a number of potentially persuasive and non-persuasive message categories. There appears to be much more of the non-persuasive kinds than there would be if advertising practice conformed to the persuasive theory. Yet it is not often claimed that such “non-persuasive” ads are less effective than persuasive ones. And systematic evidence to that effect is yet to be reported.

Next, brand beliefs are examined in Section 3. The evidence here is that, while some brand or especially sub-category differences do show up in the data, brands tend to be seen as rather similar. Customers’ attitudinal belief scores as usually measured (e.g. the percent saying “Tastes Nice”) are predominantly similar from brand to brand, almost irrespective of the particular attribute concerned. There are just too few perceived differences between brands to account for brand choice generally.

And while it is well-known that brand beliefs at the aggregate level are generally stable in the medium term, individual consumers’ expressed beliefs are variable over time. On average, only about 50% of respondents associate the same brand with an attribute when reinterviewed. It is therefore questionable whether such attitudinal responses have the strong causal power often attributed to them. But since each brand’s overall belief scores hardly change between successive interviews, the “variable respondents” are merely associating the given attribute with some other brand. This would be unlikely if they had been persuaded to see the brands as really different.

This perceived similarity of brands also suggests that, insofar as persuasive advertising is intended to differentiate the brands and persuade consumers accordingly, it does not generally succeed in doing so. This is not to say that persuasive differentiation cannot happen, but simply that mostly it doesn’t. “Persuasion” cannot rank as the major or main explanation of why and how advertising works.

Section 4 examines the evidence available to us when attitudes and behavior are in fact changing. It points to attitudes changing after a behavior change, coming into line with behavior rather than directly causing it. This is at odds with the Advertising —> Attitudes —> Behavior sequence underlying the persuasive view of advertising. Again, this is not to say that this persuasive sequence can never happen, but that it is a rare and not a commonplace occurrence.

In marked contrast to the persuasive view, the non-persuasive brand publicity theory of advertising focuses on brand salience, not brand differentiation. Advertising publicizes the brand, keeping it in the public eye. This weak-force theory emphasises reinforcement rather than change. How it differs from the persuasive view is spelt out in “Advertising is Publicity not Persuasion”, Ehrenberg-Bass Institute Report 4 for Corporate Members. It is consistent with and supported by the evidence summarised in this paper.

Amplifications of the main text, technical details and references are given in the Appendix.

2. WHAT ADS SAY OR IMPLY

In this section we consider the overt messages in advertisements. Not what the ads aim to do, nor even what people individually “take away” from them, but what consumers generally agree, when suitably asked, ads more or less explicitly say or seem to say. Do they think any particular ad actually contains a persuasive message, projects an image or identity, or potentially generates an emotional response which could motivate consumers to prefer the brand to its competitors? We think the contents of many ads would not be felt to be like that.

There are, it seems to us, about six different kinds of thing – broadly speaking and keeping it simple – that advertisements for a brand actually seem to say, show, or directly imply. The first four seem at least potentially persuasive:

1. Hard Sell – Some ads shout at us (“Buy Me!”, or even “Buy Me NOW!”. Or “Cut Price!”, “Buy While Stocks Last!”).
2. Give Reasons – Others provide “reasons” for buying that brand, either specific ones (“Has Hazelnuts”) or more general claims (“Better”, “Different” or “New”). Or less functionally perhaps that brand X is for the young or the with-it.
3. Convey Feelings – Some ads are designed to engender an emotional response which could color one’s feelings towards the brand, e.g. that it is warm, caring, and friendly.
4. New News – Ads for a new brand or ads announcing a new feature (“Now has airbags”) are informative. Some at least may seem persuasive, i.e. might change people’s minds. But they may simply be letting them know that brand A has the feature, where they are already predisposed to act once they are aware of it.  In contrast, two other kinds of ads do not really even try to be persuasive:
5. Reminders – Many ads just display the brand name (though often in creative ways). They are simply – or not so simply – just waving to say “Here I am”: e.g. “Nike”, “Always Coca-Cola”, “No 5 Chanel”, etc.
6. A Good Example of the Product – The ad says that X gets clothes clean. But the “benefit” seems to be used as a talking point to publicize X, not as a differentiating selling point for X (i.e. consumers tend to believe that detergents generally get clothes clean).

We are exploring how many ads fall into something like the above broad message types (in a project using full reproductions – Print and TV – “The Form that Advertisements Take” – FAT). Informal results so far suggest that there are more of the “non-persuasive” advertising formats – possibly far more – than is generally acknowledged. We are working to systematize the evidence. But if half or more of all ads turn out “non-persuasive” without their being widely castigated for it, can a persuasive theory still be a candidate for how advertising generally works?

Effects vs Intent

There is however also the major question of effects: whether those ads which appear to carry persuasive messages do actually persuade. (And conversely perhaps, whether apparently persuasion-free ads might in fact “persuade”?). On the face of it strongly persuasive effects seem unlikely. Experienced consumers know pretty well what the products and brands they usually buy are like, and that ads are ads. They see lots of ads without constantly rushing out to the shops, changing their brands, or – it seems – frequently changing what they think or feel about them.

Nonetheless, both the public and many commentators often still think that advertising is persuasive – “Why else would advertisers advertise?”. We think that that is a misuse of persuasive, if the ads neither seriously attempt nor succeed in getting consumers to do or feel things they wouldn’t otherwise do or feel (i.e. convert the indifferent or ill-disposed, or make substitutable brands seem very different).

In any case, as we show next in Section 3, consumers’ brand beliefs are notable for the similarities between brands that they reveal. There does not seem to be much brand differentiation for so-called persuasive advertising to claim the credit for.

3. ATTITUDES TO COMPETING BRANDS

If brands were generally seen as differentiated from one another – either functionally or through their advertising and/or other aspects of the marketing mix – this should show up in consumers’ beliefs about those brands. But as we now describe, the general picture across a variety of product categories is that brands are seen as rather similar (although not of course the same). Some differences between brands are noticed, but they are rarely a dominant factor in consumers’ choice.

The basic finding in Section 3.1 is that buyers or users of competitive brands usually do not differ greatly in how they see the brands they buy. The beliefs of brand A’s customers about Brand A are close to the beliefs of brand B’s customers about Brand B.

We next turn to the exceptions, the differences between brands that are actually perceived. Some attitudinal differences show up for a specific brand or brands (the “Partly Descriptive” deviations discussed in Section 3.2). They often reflect functional variants or sub-categories within the product field.

Other differences between brands are only likely to be picked up after the brand has been tried. These are the “Minor Differences” of Section 3.3 – they are not typically featured in the advertising nor immediately apparent in some other way before the brand is used.

These perceived differences are neither sufficiently commonplace nor important enough to account for brand choice generally.

Finally in Section 3.4 we note the finding that individuals’ brand attitudes are not very consistently expressed (the average “50% repeat-rate”). The implication is that consumers would be unlikely to change their responses in this way if they saw the brands as very different.

3.1 Customers’ Brand Beliefs Are Similar 

Table 1 illustrates that a brand’s customers largely see the brands similarly. Typically, in a survey for UK toothpaste, 55% of the customers of the brand leader, Colgate Dental Cream, thought that it “Promotes Strong Healthy Teeth”. A very similar 57% of customers of the 8th biggest brand, Ultrabrite, thought that about their brand. These percentages have been adjusted to take account of the Double Jeopardy effect – see appendix. The unadjusted scores were 71% for Colgate DC and 51% for Ultrabrite. (The unadjusted scores for all 12 toothpaste attributes are summarized in the first column of figures in Table 4 later, i.e. the free-choice percentages among the regular buyers of each of the brands. The more detailed numbers are also available, e.g. Dall’Olmo Riley 1995).

Similarly, the DJ-adjusted response percentages for each of the twelve different toothpaste attributes covered in the survey (e.g. “Fights Breath Odour”, “Gets Teeth White”, etc.) mostly differ little from brand to brand, as shown by the column of averages in Table 1. This also holds for 4 other UK products and also 4 in the US in the same study – see Table 1a in the Appendix.

Furthermore, non-buyers of a brand do not generally disagree with its customers. Brands still seem largely similar but are less salient to them. Non-buyers (or non-users) are less likely to endorse a positive belief than the brand’s customers (see Table 4 below). But their response levels also do not generally differ from brand to brand. And endorsements of negative versions of an attitude belief are typically low among respondents generally (e.g. Bird, Channon, and Ehrenberg 1970).

Results For The Whole Population

These similarities of belief scores across competitive brands for customers and non-buyers are however not at all apparent in belief scores for the whole population. “All-respondent” beliefs show a marked downward trend from the larger to the smaller brands (e.g. 20 to 30 percentage points for the toothpaste brands as in Table 2). The reasons are that a smaller brand has fewer customers (and more non-buyers) than a larger brand does, and that a brand’s customers are more likely to endorse a positive belief than its non-buyers are (as previously noted).

Many of the attributes measured in Tracking Studies and in Usage & Attitude surveys – and typified in the specific data referred to here – show primarily that a brand’s regular customers tend to say (when asked) that they “value” or “like” the brand (“Tastes Nice”, “Promotes Strong Healthy Teeth” etc.). The measures are therefore mainly “evaluative” in nature.

3.2 The Main Exceptions: “Partly Descriptive” Deviations

Some attribute responses do however stand out for particular brands, or more usually for certain subgroups of brands. Table 3 illustrates such “Partly Descriptive” attributes for the toothpaste brands: Crest is high on “Fluoride” and “For Kids”, and low on “Strong Flavour”; Mentadent P high on “Removes Plaque”; and so on.

Any such perceived difference for a brand is therefore captured as a significant, and in practice highly repeatable, deviation from a brand’s expected score (allowing for the trend with market share and also for the overall response level for that attribute, which is much lower in Table 3 for “Strong Flavour” (16%) than for “Contains Fluoride” (34%).

Such “Descriptive exceptions” occur in most product categories (for a few fairly often, e.g. Breakfast cereals where brands are mostly functionally different variants – Barwise & Ehrenberg 1985). But these deviations are seldom dramatically large (perhaps 10 percentage points or so, as in Table 3). Over the wide range of data analysed they account only for some 10% to 15% of the total observed variation in belief scores.

Many of these “descriptive” deviations relate to functional differences and tend to identify product sub-categories. Little of the variation in belief scores is thus truly brand-specific (and then often related to a long-running advertising campaign or theme for the brand, like “Kind to the Hands” for a particular UK washing powder – Barwise & Ehrenberg 1985). Overall, the extent of perceived brand differentiation seems remarkably modest.

3.3 Further Exceptions: “Minor” Differences 

There are often also “minor” differences between brands which differentiate otherwise near lookalikes. Examples are a different bottle-top or car-door-handle, or for Mueslis say, the degree of sweetness or the proportion of raisins and/or dried banana. Such “minor” features can markedly influence an individual consumer’s on-going brand preference, once they have experienced it (“I like this Muesli better”).

But generally these preferences are not widely shared – the differences are not important enough to be copied by the competition (and hence they remain differences, but “minor”). They are therefore seldom measured in tracking studies. Nor are they usually – if ever – featured in the brand’s advertising. The existence of such real but “minor” brand differentiation does not seem seriously to challenge the general similarity of brands, in the context of advertising.

3.4 Individuals’ Expressed Brand-Beliefs Vary Over Time

Individual consumers’ expressed brand-belief responses vary over time. On average, only about 50% of respondents associate an attribute with the same brand at two separate interviews (Castleberry et al 1994, for intervals of a month up to a year in the UK or 18 months in the US). This varies systematically with the level of the responses (Dall’Olmo Riley et al 1997).

The 50% who do not give the same response as last time associate the attribute with some other brand (since the overall levels of brand-belief responses are steady over time: there is no “leaky bucket” type erosion, see Castleberry et al 1994). The finding holds across both “mainly evaluative” and “partly descriptive” attributes and thus brand-belief associations are not at all uniquely held. The “moveability” of these expressed associations between brands therefore reflects the perceived similarities of the brands. Consumers would be unlikely to respond in this variable way if they saw the brands as generally different. This reinforces the evidence that “brands are brands” as already presented in Section 3.1.

Overall, customers’ attitudes to competitive brands are similar. The clue why brands have very different market shares is, as we stress elsewhere (e.g. Ehrenberg et al 1998), that they have very different numbers of customers, and not that these customers believe very different things about each of their brands.

4. BEHAVIOUR CHANGING FIRST

The persuasive view of advertising emphasises change, through its influence on consumers’ thoughts and feelings about the brand (i.e. Advertising —> Attitudes —> Behavior, basically).

But for much of the time both brand attitudes and behavior are pretty stable. And when change does occur, the evidence we describe here is of Attitudes changing after Behavior rather than preceding it (i.e. Beh —> Att, not Att —> Beh). This casts considerable doubt on the causal presumptions of the persuasive view.

The case for attitudes tending to follow behavior is in three parts. First, intentions-to-buy scores for old, slowly-dying brands are higher than normal, not what one might expect for dying brands. And intentions-to-buy for new, subsequently successful brands are lower than normal, unlike what one might expect for growing brands as we discuss in Section 4.3. The clue to the resolution of this paradox is provided by the way brand attitude scores relate strongly to current and previous buying behavior. This we therefore rehearse first (Sections 4.1 and 4.2).

Secondly, we note the possible influence on brand choice of the “minor differences” between brands (already described in Section 3.3). This can mostly only occur after trial purchase and usage, as minor differences are not usually featured in the brand’s advertising or mentioned on the pack. Indeed, the attitudes towards minor differences are largely derived from past usage experience (Section 4.4). Here also therefore, attitudes follow behavior, rather than preceding it.

Thirdly, the variability over time in individuals’ expressed attitudinal responses shows that they are often neither very firmly nor uniquely linked to just one brand, but tend to be associated with a number of different brands. Lacking a causal focus, it is therefore less surprising that such brand attitudes mostly do not have a direct or decisive influence on brand choice.

There is therefore systematic evidence for Beh —> Att, but little or none the other way round. This leaves Beh —> Att as a much more plausible working-hypothesis than Att —> Beh. Thus advertising will not generally be working by affecting attitudes first, as attitude changes mostly come second, after behavior changes. (This does not mean that attitudes can never influence behavior, but simply that this is rare.)

4.1 Attitudes To A Brand And How Often People Buy It

Holding a positive attribute belief about a brand – or more broadly, “liking it” – is known to vary with whether and how often one is buying or “using” it.

For example, in the same UK toothpaste data as before, 62% of “Frequent” buyers of Colgate Dental Cream said it “Promotes Strong Healthy Teeth”, while only 33% of “Light” buyers of the brand said so, and as few as 18% of “Never” buyers.

Table 4 shows how this steep gradient holds generally for the different toothpaste attributes typically covered in such surveys (“Fights Breath Odour”, “Gets Teeth White”, etc.). It has also been found in a wide range of other cases over many years (especially for the commonly-used free-choice questioning) as summarized in Table 4a in the Appendix (see also Franzen 1994 and references there).

The gradient also holds for overall measures such as consumers’ expressed “Intentions-to-buy”.

This relationship is striking and matters, despite its causal direction being ambivalent. It could be “I like it, therefore I buy it” (Att —> Beh). Or the other way round: “I use it, therefore I like it”, as a rationalization of one’s behavior (Beh —> Att).

4.2 Attitudes To Brands Vary With How Many People Buy Them

A second long-established relationship is that for different brands, more people say “Tastes Nice”, etc., for a big brand than for a small one.

For UK toothpastes again, 50% of housewives said that the brand leader Colgate DC promotes “Strong healthy teeth”. Only 22% said so for the smallest itemized brand, Ultrabrite. Table 5 illustrates this relationship more generally for the average of the 12 beliefs measured for toothpastes. It has also been replicated in many other studies, both published and unpublished (e.g. companies’ U&A surveys, attitudinal tracking studies etc. – see also Table 5a in the Appendix).

But a brand with low attitude scores like Ultrabrite is not really weak, just small. A brand’s frequent buyers have higher attitude scores than less frequent ones (as in Table 4) and small brands have a smaller number of frequent buyers. Hence smaller brands tend to have lower respondent-weighted attitude scores (as in Table 5), even though the scores within buying groups did not differ much by brand size.

This relationship between the belief levels and brand size is again striking and important. But its causal direction is once more ambiguous. The correlation in Table 5 could still be due either to people’s beliefs causing them to choose the brand (Att —> Beh), or to their buying or usage experience causing the brand beliefs to emerge (Beh  —>Att).

4.3 Deviations For New And Slowly-Dying Old Brands

More dynamic situations show up as systematic deviations from the static attitude-behavior relationship in Table 5, with clear cut causal implications.

Figure 1 shows graphically how the number of consumers who express an Intention-to-Buy a brand, I%, varies with the number who “Currently Use It”, U%, similar to the two downward gradients in Table 5. The fit for the general run of established and stable brands in a wide range of product categories is generally close, to within an average deviation of 2 or 3 percentage points (e.g. Bird & Ehrenberg 1966), based on a wide range of BMRB’s API surveys).

But there are occasionally larger deviations of 5+ percentage points. Examples are shown as “I is high” or “I is low”. They occurred when buying behavior for a brand was changing over time, either for:

1. Old and slowly-dying brands, or for
2. New subsequently successful brands.

It might be expected that slowly dying brands, being bought by fewer and fewer people in the future, should have low intentions-to-buy. And that subsequently successful new brands, with more and more people going to buy them, would have the high scores. Particularly if Att —> Beh generally held.

But the evidence of Figure 1 is to the contrary: the above-normal “I is high” cases are for the old slowly-dying brands. And the below-normal “I is low” ones are for the new and subsequently successful brands.

This is because:

• An old slowly-dying brand has a lot more lapsed buyers than a stable brand of similar size, and some of these lapsed buyers still express an intention to buy (as in Table 4). This results in higher than normal scores.
• The opposite occurs for a new brand: it has almost no lapsed buyers since it did not exist more than six or twelve months or so ago. Hence a successful new brand has a short-fall in its scores. It takes quite some time to acquire the “normal” number of lapsed buyers, bringing it into line with other established brands.

In both cases the change in behavior precedes the consumers’ attitudinal response – i.e. their expressed intentions-to-buy. Reported results for both new and fading brands from Y&R’s Brand Asset Evaluator appear in line with this, although they were not initially interpreted in terms of consumers’ past behavior, as here (Agres and Dubitsky 1996, p.24).

Furthermore, in neither case do consumers’ intentions, i.e. an overall attitude which supposedly relates explicitly to the future, actually predict future changes in behavior. Indeed, consumers seldom know months ahead that they will change their brand-choice behavior, or when. Thus their expressed intentions will only be predictive when they go on doing what they already have been doing, i.e. to go on buying the brand.

4.4 Minor Differences

Another clear cut instance of behavior change arising first, before attitude change, is with consumers’ reactions to a brand’s minor or secondary attributes (noted in Section 3.3).

Trial usage of the brand (i.e. behavior) must usually come before the preference or attitude to such a minor attribute can have been formed (e.g. to the degree of sweetness of a Muesli).

This is because:

1. Minor brand attributes are seldom known or believed before the brand has been used. (Even if told about it by a neighbour, one would usually still want to check out oneself that one really does rather like it.)
2. The attributes are seldom mentioned on the pack, or at least not very meaningfully (e.g. just “Sweetened”, or “Contains hazelnuts” for a muesli).
3. Crucially here – such minor attributes are seldom featured, let alone stressed, in any of the advertising to which consumers would have been exposed beforehand.

Minor brand attributes, although at times decisive for some individuals in their longer-term preferences between otherwise close substitutes, are therefore a clear set of cases where brand usage (i.e. experience or “behavior”) virtually always comes before an attitude about the attribute (and brand) is formed.

4.5 Variable Brand-Belief Associations And Brand Choice

We have already noted in Section 3.4 the variability in individuals’ brand-belief associations: that when re-interviewed, on average only about 50% associated an attribute with the same brand as before, while the other 50% then associated that same attribute with some other brand.

This finding suggests that individuals’ expressed brand-beliefs are often neither very firmly nor uniquely linked to the brand. Indeed, for each individual a specific belief is usually associated with a number of brands. It therefore seems much less likely that such generally variable associations could play a causal or decisive part in choosing one repertoire brand rather than another (but mentioning a particular brand could well follow having recently bought it). Having multiple associations is not a sufficient basis for choice.

In contrast, if brand attitudes mostly result from buying and usage experience (i.e. Beh —>  Att), as the evidence of Sections 4.3 and 4.4 suggests, then the fact that consumers mostly have stable repertoires of two or more substitutable brands would lead to the variable or multiple-brand attitudes that are observed.

5. DISCUSSION

In this report we have considered the nature of much brand advertising and how it ties in with the general evidence on brand attitudes. Contrary to the persuasive view of advertising:

1. Many advertisements do not overtly differentiate the brand or present it in a persuasive way. Thus it seems highly unlikely that ads generally can, even in principle, motivate (in terms of reasons or emotions) consumers to prefer one brand over its competitors;
2. The evidence is that brand A’s customers’ attitudes to brand A are pretty similar to brand B’s customers’ attitudes to their brand. Attitudes to brands do not differ enough to support the contention that brands are typically differentiated from one another. There’s not much in practice for so-called persuasive advertising to claim the credit for;

The brand attitude evidence is consistent with attitudes emerging from behavior rather than directly leading to brand choice.

This evidence casts additional doubt on the causal presumptions of the persuasive view.

The alternative view that advertising publicizes the brand, reinforcing and occasionally nudging to how many consumers the brand is salient, is however in line with the content of much brand advertising and the general evidence on brand attitudes. This view is discussed in detail in Ehrenberg-Bass Institute Report 4 for Corporate Members, “Advertising is Publicity Not Persuasion”.

Are We Measuring the Right Things?

It is often alleged that conventional usage and attitude surveys and tracking studies largely do not measure the more emotionally-based responses that really differentiate between brands and motivate their customers (e.g. Gordon & Corr, 1990). Instead, conventional approaches concentrate on what can most easily be quantified: factual or even functional product attributes that mostly do not differentiate between brands (which is in line with what we have argued in Section 4.2).

Some of these critics seem to believe that only qualitative research can be sufficiently sensitive to uncover these intangible but powerful “essences” of a brand. Others, including Gordon, advocate mixed qualitative/quantitative methodologies to overcome the problem (and this belief that emotional aspects can be quantified is crucial). But we are not aware  of any systematic evidence that supports either of these critical views.

Indeed, Brown (1992) doubts that this almost single-minded emphasis of the emotional and disparagement of the functional is warranted, judging by the extensive experience of Millward Brown. But only systematic evidence that directly addresses the issues in contention can hope to resolve such disputes. Anecdotal evidence is not enough.

 

 

APPENDIX

In this Appendix, we briefly summarise how the particular findings reported in the main text of this report generalize to a range of other product categories and attitude measurement techniques and also reflect earlier results reported in the literature.  We also indicate the nature of the attribute beliefs typically measured.

How the Findings Generalize

We give summary tables averaged over a specific set of 9 product categories which  were  measured and analysed in the 80’s:  Ready-to-Eat Breakfast Cereals, Washing Powders, Canned Soups, Toothpastes, and Carbonated Soft Drinks in the UK; and R-T-E Breakfast Cereals, Laundry Detergents, Fast Food Restaurant Chains, and TV News Programmes in the US (including therefore two non-fmcg categories).

These cover some 70 brands and 100-plus attribute beliefs.  The results hold separately for each category as well as on average.

The measurement procedure used was free-choice.  A comparison of free-choice with two forced-choice approaches (scaling and ranking) is reported in Barnard & Ehrenberg (1990).  This showed that free-choice results are replicated by the two forced-choice techniques used there except that the Double Jeopardy trend with brand share among a brand’s regular or frequent buyers is less marked, as expected.  (The theory of Double Jeopardy explains it primarily as a statistical selection phenomenon which occurs when people make free choices among items which are identical except for their market-size – e.g. McPhee 1963, Ehrenberg et al 1990.)

Table 1a shows for the 9 categories that, as in Table 1 just for toothpaste, the average attitudinal response percentages among the regular buyers of each brand vary little from brand to brand, implying that each brand’s customers see the brands as rather similar and not markedly different. The similarity in the brand belief scores is most clearly seen in the DJ-adjusted percentages, but is still notable in the observed percentages with their fairly modest downward trend with market share.

The Double Jeopardy or DJ effect (see above) has been removed from the observed scores in Table 1a by using a simple linear numerical correction procedure to adjust the scores (essentially that the attitude scores vary systematically by about 0.4 percentage points for each buyer percentage point).

Next, Table 4a shows that the likelihood of associating a brand with an attribute belief varies directly with the frequency of buying the brand, as for toothpaste in Table 4 in the main text.  Across each of the nine categories here, a brand’s regular customers are much more likely to endorse a positive statement than its “light”, “lapsed” or “never” buyers are.  The similarity of the scores for the various brands within each buying frequency group (shown here without any adjustment for DJ) again suggests the similarity of the brands.

Finally, Table 5a shows how the respondent-weighted population belief scores vary directly with the brands’ penetrations.  As noted in the main text, the large differences in scores between brands arise because a large brand has more regular buyers and fewer lapsed/never buyers than a smaller brand does. And regular buyers score higher attitudinally than the others, as in Table 4a. It is not that the brand scores differ much within each buyer grouping (as Table 4a showed).

Typically Measured Attribute Beliefs

The attribute beliefs covered in the 9 UK and US product categories analysed are typical of those measured in conventional U&A surveys and tracking studies. Although factual in expression and tending to be somewhat functional, some or even quite a few do have “emotional” or at least “subjective” aspects or undertones, e.g. “Fun for Children to Eat”.

A Selection of Typical Attribute Beliefs

(In alphabetical order; 15 from 108)

• Better for you than other fizzy drinks
• Clothes smell fresh
• Contains fluoride
• Fun for children to eat
• Helps you relax
• Makes the whole wash easier
• Nutritious food
• Pleasing personalities
• Popular with all the family
• Quick service
• Reasonably priced
• Stays crispy in milk
• Suitable for special occasions
• Tastes really fruity
• Value for money

The question is whether more emotional aspects, e.g. caring, friendly or cool, cannot also be adequately summarised in a few well-chosen words to show up any big differences between brands.