What Healthy-Living Means: Trialing a New Qualitative Research Tool

Published in International Journal of Market Research (June 2002: 44.2. pp 223-234)

Summary: The shared meaning of ‘Healthy-Living’ for a set of consumers is unpacked into its component memes (genes of meaning) using a new associative meaning mapping technique. This simple qualitative technique extends the method of associative group analysis and uses a Darwinian ‘Survival of the Fittest’ rationale to build up a group-level summary map of meaning from individuals’ chains of associations around Healthy-Living. The utility, validity and reliability of the technique are discussed. As a practical application of grounded theory, it is suggested that the resulting ‘meme maps’ may be useful in providing insight and creative stimulus for marketing initiatives that ‘connect’ with consumers.

Introduction: Making Connections
One of the principal tasks of qualitative market research is to inform marketing initiatives in such a way that they make meaningful connections with target consumers. Unless such initiatives ‘connect’ with consumers they cannot be compelling, or even meaningful. For example, the message ‘hatsumei’ is utterly meaningless until it is associated, that is, connected in the mind with the notions ‘innovation’ and ‘Japanese’: Hatsumei quite simply means innovation in Japanese. Similarly, a brand name such as Volvo is meaningless when stripped of the associations that consumers imbibe it with: The associations ‘safety’ and ‘car’ are the connections that make Volvo meaningful and distinctive to many people.

Of course, that meaning is made out of associations in the mind, and that this meaning can be unpacked by associative research techniques such as word association, is an established insight not only in market research but one that runs through philosophy, psychology and linguistics (Galton 1880, Warren 1916, Freud 1924, Noble 1952, Deese 1965, Anderson and Boyer 1973, Law and Lodge 1984, Fiske and Taylor 1991, Sowa 1991, Chisnall 1992, Gordon and Langmaid 1995, Zerubavel 1997, Greenbaum 1998, Gordon 1999, Buzan and Buzan 2000). Part of a battery of ‘projective techniques’, so called because they work by getting respondents to make sense of stimuli (such as a word, ink blot, partial sentence, third person or cartoon story) by projecting their own subjective meanings onto them, associative techniques can produce associative meaning maps that unpack the meaning of things into a chains of associations, that can be thought of as ‘genes of meaning’ (Zipf 1965).

Figure 1: ‘Genes of Meaning’ of Healthy-Living

For example, a research participant may be invited to free associate around Volvo, and may indeed respond by projecting subjective meaning onto the stimulus with the associations ‘car’ and ‘safe’. By extending the capture of associations into chains of associations, say by producing the association chain ‘Volvo – safe – children’, researchers can gain useful insight into what safety means to the consumer in the context of Volvo, and crucially, how safety may be communicated in a way that makes compelling and meaningful connections with consumers.

Two Cheers for Associations
As well as tapping into the associative structure of the human mind, and in addition to providing succinct insight – capturing in several words what many hours of alternative investigation and lengthy analysis may take to uncover – word association exercises are methodologically sound insofar as they neither pose leading questions nor impose answers: The combination of structured questioning and free responses in word association tasks allows data capture to proceed in a relatively unbiased fashion. For these reasons, word association exercises are an established, respected and oft’ used technique in the qualitative researcher’s toolbox that are employed in the unpacking of meanings and the informing and development of meaningful marketing initiatives.

However, like many qualitative techniques that have their origins in clinical psychology, word association exercises suffer from the ‘problem’ of generating insights about particular individuals, not groups or whole target populations. Consumer research, on the other hand, is typically more concerned with generating insight at the level of a target group or population for the simple reason that marketing initiatives tend to be directed as groups or target populations rather than specific individuals. This renders word association exercises somewhat maladapted to the needs of consumer research, producing as it does associative meaning maps that unpack what something means to somebody, rather than unpacking shared meaning within a group.

A ‘Survival of the Fittest’ Solution
One simple solution to this problem has been to first capture word associations from a sample of individuals, and then code the associations into similar themes and plot only the most representative, by frequency of occurrence, on a summary map. This way, weak idiosyncratic associations are left out of the group map, whilst the ‘fit’, dominant associations are mapped using what is essentially a simple ‘Survival of the Fittest’ rationale. For example, Szalay and Deese applied this rationale in what they called 'the method of associative group analysis' (Szalay and Deese 1978). Essentially, the technique involves capturing single associations around a concept, and then coding them into meta-associations by thematic similarity. By rating these codes in terms of the degree to which they evoke positive or negative feelings, the most frequently made association themes can be plotted by frequency and value-direction on a bar chart. The resulting ‘semantograph’ provides a simple group-level map that represents only the fittest, that is, the most representative ‘genes of meanings’ of a concept for a group.

Building a Meme Machine
However, as originally specified, this method of associative group analysis is limited to mapping only immediate associations around a concept, rather than providing a richer unpacking of meaning in chains of associations: There is little new news in the finding that Volvo means safety, but there may well be new news in unpacking what safety means to people in the context of Volvo. Now, there is no reason in principle why the method of associative group analysis could not be extended to do this, but such a move would present a coding nightmare as soon as more than a handful of discrete and discrepant association chains had to be integrated into some meaningful whole.

Fortunately, the daunting prospect of having to manually sift through hundreds of discrepant and idiosyncratic association chains made by individual consumers can be simply obviated by using simple but smart features of user-friendly databases such as Microsoft Access. To exploit the database capacity to structure and link qualitative data, we extended the method of associative group analysis in order to unpack and map two levels of associations around the concept of ‘Healthy-Living’ for 150 people. Specifically, we set up a simple macro function (automated procedure) in Microsoft Access to capture, sort and link chains of associations by applying the same Darwinian survival of the fittest rationale as originally specified: As association chains were entered, frequently made associations were reinforced, whilst idiosyncratic and unrepresentative associations would fall out of the analysis (Figure 2).

Figure 2: Specification for mapping a collective mindset

Methodologically speaking, this is simply an automated application of ‘grounded theory’; the qualitative research approach that generates models of meaning through iterative loops of data integration and analysis (Glaser and Strauss 1967, Becker 1993, Pidgeon 1996, Pidgeon and Henwood 1996). By feeding individual chains of associations made by consumers into a database pre-programmed to automatically create or reinforce links between associations, the database ‘learns’ and builds a blueprint, iteratively and ground up, of the dominant and representative associative chains made by that population. In this way, a single integrated associative meaning map evolves through the progressive addition, interlinking, and resorting by link strength, of associations. In essence, this involves no more than the straightforward process of progressively building up a coding frame of the data that links categories by association, and then ranks them by link strength. By setting a limit to the number of associative links any association can have, the coding frame is kept manageable and free from impractically over-long lists associations. Specifically, by allowing new associations to replace the weakest linked associations when this limit is reached, idiosyncratic and unrepresentative associations are removed in a conceptual Darwinian survival of the fittest contest. Using the feature of databases such as Microsoft Access to automatically output data in a simple graphical interface, it is possible to generate intuitive group-level associative meaning maps that we call a ‘meme maps’, following the Darwinian neologism ‘meme’ for associative units of memory that describe ‘genes of meaning’ in the mind[i] (Semon 1921, Blum 1963, Dawkins 1976, 1982, Young 1978, Minsky 1986, Durham 1991, Bollen 1996, Bollen and Heylighen 1998, Marsden and Bollen 1999, Marsden 2000, 2000a).

This extended method of associative group analysis was trialled in July and August 2000 at the University of Sussex in the context of an NPD healthcare project with the marketing consultancy Brand Genetics. UK and US adult consumers were invited by chain email to play a simple online word association game (Figure 3) around the concept of ‘Healthy-Living’, and after two weeks 142 people had completed the game, with each participant having made three association chains. By linking the online word association game directly to the database, the meme map, i.e. the associative meaning map, automatically evolved through some 1278 iterations, as respondents either reinforced or created new associations to produce the output shown in Figure 4.

Figure 3: Data capture with an online word association game

Figure 4: Unpacking the genes of meaning of ‘Healthy-Living’

A visual inspection of the output shows the ‘core’ central concept (Healthy-Living) lying at the centre of the meme map, linked to a hierarchy of associative chains that progressively unpack the meaning of the concept into its component memes. In this way, each node unpacks into a ranked list (1-5) of its dominant defining memes (most commonly made associations). Additionally, each meme is followed by a polarity rating in brackets, which is simply the modal rating respondents gave that association in terms of the degree to which it evoked, for them, positive or negative feelings (+3 to -3). Conceptually, the meme map can be thought of as a blueprint of the semantic DNA that progressively unpacks the dominant ‘genes of meaning’ of ‘Healthy-Living’ in the mindset of participating consumers. For example, the meme map shows that the dominant meme in ‘Healthy-Living’ is ‘natural’ whose own genes of meaning can be unpacked, in the context of ‘Healthy-Living’, into ‘fresh’, ‘pure’ and ‘organic’. Overall, the meaning of ‘Healthy-Living’, was described by the memes of ‘natural’, ‘balanced diet’, ‘being well’, ‘keeping fit’ and ‘eating well’; concepts that the meme map then unpacks into their own defining memes.
 

Validity and Reliability
In terms of commercial utility, the genius of qualitative research may lie in its capacity to generate insight and provide a creative stimulus for informing successful marketing initiatives (cf. Gordon 1999). . Within such a ‘generative’ context of understanding and interpretation, applying the empiricist standards of quantitative research of validity and reliability to judge qualitative research such as this may be somewhat disingenuous (Woolgar 1996). Nevertheless, unless the findings appear to have validity and reliability in terms of being ‘meaning-adequate’ (retaining the meanings of those researched) and consistent, where expected, with the findings of previous or subsequent similar analyses, the credibility of the technique may be undermined (c.f. Weber 1949).

To assess this, what could be called an interpretative validity and reliability of the findings, the research was repeated with 120 new respondents, and a second meme map was evolved (Figure 5). Having completed the online word association game, respondents were taken to a web page depicting the first meme map and asked whether or not they thought it captured the important associations around ‘Healthy-Living’. Of the 76 participants who replied, 59 (78%) replied ‘yes’, indicating a high degree of validity.

Figure 5: A replication of the ‘Healthy-Living’ meme map with different consumers

Assessing reliability was more problematic since it required a qualitative assessment of the conceptual overlap of the two maps. Of course, there was a theoretical possibility that the replication could have produced an identical map in terms of both structure and labels given to the associated concepts. However, outside the most naïve empiricism, this would not be expected, especially given the grounded approach through which the map evolves. Accordingly, the second map was indeed different from the first, but crucially there was a good degree of conceptual continuity between the two; ‘good diet’ (Map 2) – balanced diet (Map 1); ‘active life’ (Map 2) - ‘keeping fit’ (Map 1); ‘health foods’ (Map 2) - ‘eating well’ (Map 1); feeling good (Map 2) – ‘being well’ (Map 1); and with less overlap ‘good for you’ (Map 2) – ‘natural’ (Map 1). Thus, although the specific words inputted as associations differed, as would be expected, the dominant themes emerging were similar. Against this, it should be noted that the hierarchical structure of the associations differed substantially between the two maps, with consistency between the maps limited to painting an overall qualitative picture of meaning, as opposed to mirroring any quantitative ranking of associations. In sum, the maps certainly fall short of what would be expected, but not necessarily achieved, in a quantitative perceptual mapping study, but they do appear to be consistent and potentially insightful in the provision of an understanding of what healthy-living means to a target population.

Discussion: Applications and Limitations
Simple database technology allows the qualitative method of associative group analysis with its survival of the fittest rationale to produce associative meaning maps, dubbed here ‘meme maps’, that unpack the dominant meanings of a concept within a target population. In doing so, meme maps build on a research insight that dates back to the Aristotelian insight developed by the British Empiricists that ideas are stored in memory and are only meaningful by association. This idea has travelled through the centuries well, and underpins contemporary understanding of memory and meaning, in which associative networks of memory are understood to imbue objects of experience with meaning.

However, it is important to note that neither associative techniques nor meme maps are held up here as some kind of panacea for the qualitative investigation of meaning, for they are not. Associative techniques are poor devices for capturing the rational, linear, cognitive thought and thought processes that are also important in consumer understanding, and they tend to ignore the discursive and situational aspects of meaning – that ideas have meaning within the context of a particular discourse and background circumstance (Szalay and Deese 1978).

Bearing these caveats in mind, meme maps are perhaps best seen as insight maps that capture a population-level qualitative picture of meaning using associative techniques. In this way, the principal utility of meme mapping may be in their capacity to generate creative stimulus for marketing initiatives, rather than tracking or measurement tools (Figure 6).
Figure 6: Creative stimuli for meaningful ‘Healthy-Living’ marketing initiatives

For example, by combining positive associations with the central concept, in the form of ‘Healthy-Living that makes you feel good’ or ‘Natural Healthy-Living’, the maps generate attractive new consumer promises with the capacity to connect with consumers and that can be used to stimulate ideas for marketing initiatives (cf. Boden 1992, Buzan and Buzan 2000, Clegg and Birch 1999, Marsden and Bollen 1999, Simonton 1999). Likewise, negative associations could be used to direct creativity away from negative associations such as ‘health foods’ and ‘dieting’. In this way, and in the context of a ‘creative turn’ in consumer research, meme maps may provide the raw material, the ‘genes of meaning’, for engineering compelling next-generation ideas with the potential to resonate and make meaningful connections with consumers.
 

Dr Paul Marsden is a research psychologist at the London School of Economics

Back to Published Work
 

References
Anderson, J. R. & Bower, G. H., (1973). Human Associative Memory. Washington DC: Winston & Sons.

Becker, P. H., (1993). ‘Common pitfalls in published grounded theory research’. Qualitative Health Research, 3, pp254-260.

Blum, H.F., (1963). On the origin and evolution of human culture. American Scientist, 51, pp32-47

Boden, M., (1992). The Creative Mind, London: Abacus.

Bollen, J. & Heylighen, F., (1998). ‘Modelling the Microstructure of Memes’. Proceedings 15th International Congress on Cybernetics (International Association of Cybernetics), Namur.

Bollen, J., (1996). ‘Self-structuring Hypertext Networks that learn to reflect the common semantics of their users: applications to the WWW’. Proceedings of the Annual Meeting of the Belgian Psychological Society, Brussels.

Buzan, T.,and Buzan, B. (2000). The Mind Map Book. Millennium edition. London: BBC Worldwide.

Chisnall, P.M., (1992). Market Research (4th edition) Maidenhead: McGraw-Hill Dawkins, R., (1982). The Extended Phenotype. Oxford, OUP.

Clegg, B., and Birch, P., (1999). Instant Creativity. London: Kogan Page.

Dawkins, R., (1976). The Selfish Gene. Oxford: OUP.

Deese, J., (1965). The Structure of Associations in Language and Thought. Baltimore: John Hopkins Press

Durham, W.H., (1991) Coevolution: Genes, Culture and Human Diversity Stanford: Stanford University Press.

Fiske, S. T. & Taylor, S. E., (1991). Social Cognition. New York: McGraw-Hill.Baltimore: John Hopkins Press.

Freud, S., (1924). Collected Papers. London: Hogarth.

Galton, F., (1880). ‘Psychometric experiments’. Brain, 2, pp149-162.

Gatherer, D.  (1998). ‘Why the thought contagion metaphor is retarding the progress of memetics’. Journal of Memetics - Evolutionary Models of Information Transmission,2. http://www.cpm.mmu.ac.uk/jom-emit/1998/vol2/gatherer_d.html

Glaser, B. & Strauss, A., (1967). The Discovery of Grounded Theory. Chicago: Aldine.

Gordon, W. and Langmaid, R., (1995). Qualitative Market Research: A Practitioner’s and Buyer’s Guide. Cambridge: Cambridge University Press

Gordon, W., (1999). Good Thinking: A Guide to Qualitative Research. Oxford: Admap Publications.

Greenbaum, T.L., (1998). The Handbook for Focus Group Research (2nd edition) Thousand Oaks: Sage.

Law, J. & Lodge, P., (1984). Science for Social Scientists. London: Macmillan.

Marsden, P, (2000). The Werther Effect, Doctoral thesis: University of Sussex

Marsden, P. & Bollen, J., (1999). ‘Help advertising evolve: Clone consumer thought patterns’. Admap, 393, pp37-39.

Marsden, P., (2000). ‘Brand selection, naturally: A case study’. Proceedings of the Market Research Society Conference March 2000.

Marsden, P., (2000a). The Werther Effect. Doctoral Thesis: University of Sussex.

Minsky, M. (1986). The Society of Mind. New York: Simon and Schuster.

Noble, C. E., (1952). ‘An analysis of meaning’. Psychological Review, 59, pp421-430.

Pidgeon, N. & Henwood, K., (1996). ‘Grounded theory: Practical implementation’. In J. T. E. Richardson, (ed.), Handbook of Qualitative Research Methods, pp86-101, Leicester: BPS Books.

Pidgeon, N., (1996). ‘Grounded theory: Theoretical background’. In J. T. E. Richardson, (ed.), Handbook of Qualitative Research Methods, pp75-86, Leicester: BPS Books.

Simon, R. (1921) The Mneme. New York: Macmillan.

Simonton, D.K., (1999). Origins of Genius. London: OUP.

Sowa, J., (ed) (1991) Principles of Semantic Networks: Explorations in the Representation of Knowledge. San Francisco: Morgan Kaufmann.

Szalay, L., & Deese, J. (1978). Subjective Meaning and Culture: An Assessment through Word Associations. Hillsdale, NJ: Lawrence Erlbaum Associates.

Warren, H. C., (1916). ‘Mental Association from Plato to Hume’. Psychological Review, 23, pp208-230.

Weber, M., (1949) The Methodology of the Social Sciences. Translated and Edited by Edward A. Shils and Henry A. Finch. New York: The Free Press.

Woolgar, S., (1996) Psychology, qualitative methods and the ideas of science.’ In J. T. E. Richardson, (ed.), Handbook of Qualitative Research Methods, pp11-25, Leicester: BPS Books.

Young, J. Z. (1978). Programs of the Brain. Oxford: OUP.

Zerubavel, E., (1997). Social Mindscapes. Cambridge, Mass: Harvard University Press.

Zipf G.K., (1965) [1935] Psychobiology of Languages. Mass. MIT Press.

[i] The concept of a meme, here used to describe an internal semantic node in the associative structure of memory, has also been used to denote received ideas and practices as they exist in their objectified form – as cultural behaviours and artefacts. For a discussion of meme definitions, see Gatherer (1998) and associated commentaries. The use of the term here follows from Dawkins (1982) revised conception of the term he coined that is consistent with Semon’s (1921) ‘mneme’ to describe received idea stored in memory, Blum’s (1963) ‘mnemotype’ as internally stored information, and Young (1978) and Minsky’s (1986) notions of ‘mnemon’ and ‘neme’ as basic units of associative memory.