The Changing Importance of Quality Aspects in Food Consumption

ABSTRACT - Major consumer trends like health and organic foods figure prominently on the agenda of food businesses and regulatory decision-makers. However, it is not clear from previous research whether rising market shares reflect changes in consumer attitudes, changes in the supply structure, or changes in the pricing of foods. Five scales from the Food-related Lifestyle instrument (FRL) were used in replication surveys in Germany in 1993 and 1996 (N1=1000, N2=1042), France in 1994 and 1998 (N1=1000, N2=1000), and the UK in 1994 and 1998 (N1=1000, N2=1000), measuring the importance of health, price/quality relation, novelty, organic products, and freshness to consumers’ food choices Trends in the importance of these quality aspects were modeled using multi-sample confirmatory factor analysis with structured means. Results indicate that, contrary to widespread expectations, the importance of healthy/ unprocessed foods, organic foods, and fresh foods has been declining in all three countries since the early 1990s. The pattern suggests that the actual consumer trend to organic foods already peaked several years ago, and that the current boom is likely to be a mere short-term consequence of changes in pricing and distribution.



Citation:

Joachim Scholderer, Karen Brunso, and Klaus G. Grunert (2001) ,"The Changing Importance of Quality Aspects in Food Consumption", in E - European Advances in Consumer Research Volume 5, eds. Andrea Groeppel-Klien and Frank-Rudolf Esch, Provo, UT : Association for Consumer Research, Pages: 5-10.

European Advances in Consumer Research Volume 5, 2001      Pages 5-10

THE CHANGING IMPORTANCE OF QUALITY ASPECTS IN FOOD CONSUMPTION

Joachim Scholderer, The Aarhus School of Business, Denmark

Karen Brunso, The Aarhus School of Business, Denmark

Klaus G. Grunert, The Aarhus School of Business, Denmark

Carsten Stig Poulsen, The Aarhus School of Business, Denmark

John Thogersen, The Aarhus School of Business, Denmark

ABSTRACT -

Major consumer trends like health and organic foods figure prominently on the agenda of food businesses and regulatory decision-makers. However, it is not clear from previous research whether rising market shares reflect changes in consumer attitudes, changes in the supply structure, or changes in the pricing of foods. Five scales from the Food-related Lifestyle instrument (FRL) were used in replication surveys in Germany in 1993 and 1996 (N1=1000, N2=1042), France in 1994 and 1998 (N1=1000, N2=1000), and the UK in 1994 and 1998 (N1=1000, N2=1000), measuring the importance of health, price/quality relation, novelty, organic products, and freshness to consumers’ food choices Trends in the importance of these quality aspects were modeled using multi-sample confirmatory factor analysis with structured means. Results indicate that, contrary to widespread expectations, the importance of healthy/ unprocessed foods, organic foods, and fresh foods has been declining in all three countries since the early 1990s. The pattern suggests that the actual consumer trend to organic foods already peaked several years ago, and that the current boom is likely to be a mere short-term consequence of changes in pricing and distribution.

INTRODUCTION

It is customary in food business and consumer policy alike to base strategic decisions on certain "consumer trends" that appear just so obvious that nobody asks if they are really there, after all. A recent example is the organic foods boom. It is estimated that from 1998 to 1999, the volume of the organic foods market has grown by approximately 10% in Germany (Stiftung +kologischer Landbau 2000), 12% in France (USDA Foreign Agricultural Service 1999), and 40% in the UK (The Soil Association 1999).

Judged against this background, it is somewhat surprising that marketing scholars have only paid superficial attention to this trend, usually by means of cross-sectional studies (e.g., Bech-Larsen and Grunert, in press; Kyriakopoulos and Ophuis 1997; Tiilikainen and Huddleston 2000; Th°gersen 1998). However, longitudinal or repeated cross-sectional designs are needed to account for the trend character of the phenomenon.

So far, the authors know of only one study that used an actual panel design. In a study about the influence of personal values on pro-environmental behaviors, Th°gersen (2000) asked 1090 Danish consumers in 1998 and 1999 whether they regularly purchased organic foods. Significant increases in purchase frequency were only found in one out of three product categories (frozen peas), whilst the others did not change significantly.

To make things even more complicated, purely behavioral indicators (like aggregate sales statistics or individual purchase frequencies) are quite unspecific as to the forces that drive a particular trend. Two scenarios may be constructed as to what is actually going on beneath the surface:

$ Scenario 1. The importance of attribute X to consumers’ food choices is in fact increasing. Decision-makers who timely react to this may expect growth of the more sustainable sort, at least when their pricing and distribution decisions are not completely unreasonable.

$ Scenario 2. The importance of attribute X to consumers’ food choices has been rising for some time, but already peaked some time ago. Market shares are still rising, but only because restrictions (such as premium prices and limited supply) have recently been eased. Decision-makers who rely on the sustainability of such a trend are likely to be disappointed soon when short-term growth reaches a ceiling.

The scenarios constitute a classic economic problem: when only quantities supplied and market price are known, it is impossible to determine the shape of the demand curve (Working 1927). Consequently, it is also impossible to decide whether demand has increased, decreased or remained constantBan infinite number of demand curves exist that are consistent with the same equilibria (Figure 1).

The aim of the present study is to look behind a number of alleged consumer trends in European food markets, including healthy/natural foods, organic foods, novelty, freshness, and the price/quality relation. Specifically, we will examine whether these quality aspects have in fact become more important to European consumers, or if their importance is already dwindling again but is still held in high regard by food marketers because the point of inflection went by unnoticed.

The importance of these aspects to consumers’ food choices will be assessed using a number of scales from the Food-Related Lifestyle Instrument (FRL; Grunert, Bruns° and Bisp 1997). Trends in these aspects will be estimated in a structural equation modeling framework, based on data from replication surveys carried out between 1993 and 1998 in Germany, France, and the United Kingdom.

METHOD

Data Collection

The analysis will combine data from six different surveys. All samples were drawn on a household basis with a quota imposed on region:

$ D-93. Data were collected from N=1000 German consumers in 1993. The mean age of the respondents was 45.03 years (SD=16.15), 77 per cent of the respondents were female.

$ D-96. Data were collected from N=1042 German consumers in 1996. The mean age of the respondents was 44.10 years (SD=15.73), 78 per cent of the respondents were female.

$ F-94. Data were collected from N=1000 French consumers in 1994. The mean age of the respondents was 48.16 years (SD=15.43), 91 per cent of the respondents were female.

$ F-98. Data were collected from N=1000 French consumers in 1998. The mean age of the respondents was 48.17 years (SD=15.45), 87 per cent of the respondents were female.

$ UK-94. Data were collected from N=1000 British consumers in 1994. The mean age of the respondents was 44.55 years (SD=15.64), 90 per cent of the respondents were female.

$ UK-98. Data were collected from N=1000 British consumers in 1998. The mean age of the respondents was 44.10 years (SD=14.91), 87 per cent of the respondents were female.

In the D-93, D-96, UK-94 and UK-98 surveys, samples were drawn by means of a random-route procedure. In the F-94 and F-98 surveys, random samples were drawn with an additional quota imposed on age. All interviews were conducted personally at home with the person mainly responsible for the food shopping and cooking in the respective household. First, respondents were asked to answer the 69 items of the Food-Related Lifestyle (FRL) instrument. Then, different modules followed specific to the aims of the study. All surveys concluded with a set of demographic variables.

FIGURE 1

THE IDENTIFICATION PROBLEM ILLUSTRATED: WHEN ONLY QUANTITIES SUPPLIED (Q1, Q2) AND MARKET PRICE (P1, P2) ARE KNOWN, SHAPE AND SHIFT OF THE DEMAND CURVE (D1, D2) CANNOT BE DETERMINED

Measures

The "Importance of quality aspects" domain of the FRL consists of altogether 18 items measuring the six dimensions health, price/quality relation, novelty, organic products, taste, and freshness. Since previous analyses have indicated that the items of the taste dimension have highly undesirable psychometric properties (see Scholderer, Bruns°, Bredahl and Grunert 2000), they will be excluded from the analysis. The remaining items are presented in Table 1. All items were answered on seven-point scales ranging from "completely disagree" (1) to "completely agree" (7).

Model

The aim of cross-cultural surveys in consumer research is to identify differences in response patterns between consumer populations. From a theoretical perspective, revealed differences are a subject of genuine scientific interest. From a methodological perspective, however, they pose difficulties that are all too often impossible to overcome.

The basic problem is the following: suppose we have collected measurements of an observed variable x in two consumer populations " and B and are interested in differences between the expected values of x. A direct test of the hypothesis mx"BmxB=0 would rest on the assumption that, in both populations, x measures an underlying quantity x on a common interval scale f: x=t + l x with invariant location and scale parameters t and l such that differences in x can be meaningfully inferred from differences in x (Krantz, Luce, Suppes and Tversky 1971).

When populations " and B are different cultures or cohorts, and the observed variable x is a questionnaire item, it becomes unreasonable to simply assume a common interval scale for responses on x. Explicit treatment of such biases requires the use of an appropriate psychometric model. Confirmatory factor analysis with structured means (S÷rbom 1974) is the most flexible framework to handle such measurement problems. It represents the observed responses to P items (p=1, 2, ... P) as a linear function of M latent factors (m=1, ... M, M - P), P intercept terms, and P random errors. In multi-sample models, parameters are allowed to differ across groups:

EQUATION (1)

where xg is the Px1 vector of observed variables in group g=1, ... G, tg is the Px1 vector of intercept terms in group g, ?g is the Mx1 vector of latent factors in group g, "g is the PxM matrix of factor loadings in group g, and dg is the Px1 vector of random errors in group g, assumed to be uncorrelated with the latent factors and to have zero expectation. Thus, the expected values of the observed variables are

EQUATION (2)

where ¦g is the Px1 vector of observed means in group g and kg is the Mx1 vector of latent factor means in group g. Finally, the covariance matrix of the observed variables is

EQUATION (3)

where qg is the MxM covariance matrix of latent factors in group g and og is the PxP covariance matrix of random errors in group g.

Across groups, the measurement model can be invariant with respect to each of its five parameter matrices tg, "g, qg, og, and kg. Of particular interest is a measurement model where factor loadings and item intercepts are invariant across groups, defining a congeneric measurement model with group-invariant location and scale parameters. If the constraints hold, the observed variables xg are measured on common interval scales and can be meaningfully compared across groups (Meredith 1993).

TABLE 1

QUALITY ASPECTS ITEMS OF THE FOOD-RELATED LIFESTYLE INSTRUMENT

The present study includes consumer samples from three different cultures and, within the three cultures, samples from two different points in time. Hence, the invariance of parameters will be tested in two steps: (a) In a first step, parameters will be constrained only within cultures, but across time, and (b) in a second step, parameters will be constrained across cultures and across time. After the degree of measurement invariance has been established, change over time will be analyzed using procedures that are appropriate for the particular invariance level.

RESULTS

Normality Check

Since maximum likelihood (ML) estimation of parameters assumes multivariate normality, multivariate skewness and kurtosis statistics were computed for the joint distributions of the fifteen items within each sample. The distributions departed significantly from normality in the D-93 (multivariate skewness=43.964, multivariate kurtosis=337.756, c2=5017.707), D-96 (skewness=22.737, kurtosis=315.425, c2=2309.141), F-94 (skewness=30.853, kurtosis=315.974, c2=3324.497), F-98 (skewness=25.338, kurtosis=305.166, c2=2557.345), UK-94 (skewness=30.884, kurtosis=311.297, c2=3188.863), and UK-98 samples (skewness=27.841, kurtosis=310.163, c2=2803.625; all ps<.001).

Yet because goodness-of-fit tests against multivariate normality are notorious for excessive power in large samples, univariate skewness and kurtosis statistics were inspected to check whether individual items were responsible. None could be identified so that approximate multivariate normality will be assumed.

Measurement Invariance

Five models were specified. Model 0 assumed an invariant factor pattern across the six samples, including five latent factors, and only one non-zero loading for each item (simple structure). The model will serve as the null model in the model comparison sequence. Model 1 assumed that the factor loadings " were invariant over time, but only within cultures. Model 2 assumed them to be invariant over time and across cultures. Model 3 assumed that the item intercepts t were invariant over time, but only within cultures. Model 4 assumed them to be invariant over time and across cultures. All model were estimated by maximum likelihood using LISREL 8.30 (J÷reskog and S÷rbom 1996; J÷reskog, S÷rbom, du Toit and du Toit 1999) and converged without problems.

Goodness-of-fit statistics are shown in Table 2. The RMSEA remained within conventional acceptance limits (RMSEA<.080) for Models 0 through 3, and was only slightly above for Model 4. The CAIC reached its minimum value with Model 1, but did not increase substantially for all models except Model 4. Due to the large samples involved here, the overall c2 and Dc2 values should only be interpreted in relative terms. To improve descriptive accuracy, an incremental TLI (also known as NNFI) was computed for each model comparison, indicating that the only substantial leap occurred between Model 3 and Model 4. Taken together, goodness-of-fit and model comparison statistics suggest acceptable fit for all models up to Model 3, and unacceptable fit for Model 4. As a result, invariance over time and across cultures can be accepted for factor pattern and factor loadings, but not for item intercepts. They are indeed stable over time, but cross-culturally biased so that all subsequent analyses will be conducted within cultures.

TABLE 2

GOODNESS-OF-FIT STATISTICS FOR MEASUREMENT INVARIANCE MODELS

Change over Time

To test for change over time, differences between the latent means of the factors health, price/quality relation, novelty, organic products, and freshness were computed within each culture. Since their estimation at occasions T1 and T2 was based on independent samples, the standard error of any such difference ^k=kT2BkT1 is simply SE(^k)=[SE2(kT2) + SE2(kT1)]1/2 (Goodman 1960), and a t statistic can be computed in the standard fashion t=^k [SE(?k)]B1. The results are shown in Table 3.

The importance of health aspects in the sense of natural, unprocessed foods for consumers’ food choices declined significantly in Germany and the UK, but did not change in France. The price/quality relation of products lost importance in Germany, gained importance in France, but did not change significantly in the UK. Likewise, novelty lost in Germany, gained in France, and remained constant in the UK.

Organic products became less important to consumers in Germany and the UK, but did not change in France. The importance of freshness declined significantly in all countries. Standardized changes for all factors (scaled to zero mean and unit variance) are presented in Figure 2.

DISCUSSION

In food business as well as in consumer policy, we have become used to taking it for granted that, for example, consumers increasingly favor unprocessed and organic foods over highly processed, conventionally grown foods. However, some things change when you look beneath their surfaceBand the rather counterintuitive trends identified in the present paper may be but one example of this.

Contrary to common expectation, the perceived trend towards unprocessed and organic foods seems to be rather one away from unprocessed and organic foods. This may be a bit confusing as the market shares of organic products are still rising at a high pace in most European countries. Yet this is not necessarily a reliable long-term indicatorBthe results of the present study suggest that the importance of eco-related aspects to consumers’ food choices already peaked several years ago (at least in Germany and the United Kingdom, which are the largest national markets for organic foods in the EU).

Largely due to premium pricing and insufficient supply chain management, manifest demand has been low during those years. This has been changing lately, so that the rising market shares may be interpreted as a consequence of lower price and higher availability outweighing the negative trend in consumers’ attitudes. Yet if the trend in consumers’ attitudes does not turn upwards again, market growth may be expected to reach a ceiling all too soon.

The declining importance of freshness appears to complement the trend away from healthy/unprocessed and organic foodsBafter all, a large part of the organic foods business has traditionally dealt in fresh produce. Taken together, both trends may perhaps be interpreted as the by-product of another trend in consumers’ attitudes: convenience foods. Both unprocessed/organic and fresh foods require a certain effort regarding meal preparation, which is obviously at variance with the pre-prepared meals dealt with in the convenience segment.

Whilst the health, organic and freshness trends appear to be hegemonic, the results pertaining to novelty and the price/quality relation of products are somewhat inconclusive. The most obvious feature here is cross-cultural differences. In contrast to France, where the importance of novelty and the price/quality relation has been rising, we observed a strong decline of both aspects in Germany, and no significant changes in the UK. A convenient interpretation would, for example, be one in terms of different degrees of multi-culturalness of the three food cultures. Yet as we have already seen above, common-sensical interpretations like this can be quite misleading. Most importantly, it is not clear whether trends into opposite directions signify divergence or convergence. Although cross-cultural biases in our survey measures prevent us from direct comparisons of their absolute levels, it may also be the case that they approach a common European average, gradually leveling out their different starting points. If that were indeed the case, the apparent divergence could rather be seen as the beginning of a harmonization of European food cultures.

Finally, it should be stressed that our study has its limitations. The design included measurements on two occasions in each country, so that only linear trends could be estimated. It is quite plausible to assume that the true trend has a more complex shape. The 1996 BSE crisis and other food scares, for example, may have triggered temporary increases in consumer attitudes to organic foods that have locally bent the altogether negative trend. A more extensive tracking of consumer attitudes over time is required to account for such phenomena, once again stressing the need for more longitudinal research in marketing.

TABLE 3

CHANGE IN LATENT FACTOR MEANS

FIGURE 2

STANDARDIZED CHANGE IN LATENT FACTOR MEANS

REFERENCES

Bech-Larsen, Tino and Klaus G. Grunert (in press), "Kaufentscheidungen bei Vertrauenseigenschaften: Eine Untersuchung am Beispiel des Kaufes von ÷kologischen Lebensmitteln in Deutschland und DSnemark", Marketing ZFP.

Goodman, Leo A. (1960), "On the Exact Variance of Products," Journal of the American Statistical Association, 55, 708B713.

Grunert, Klaus G., Karen Bruns°, and S°ren Bisp (1997), "Food-related Lifestyle: Development of a Cross-culturally Valid Instrument for Market Surveillance," in Values, Lifestyles, and Psychographics, ed. L. Kahle and C. Chiagouris, Mahwah, NJ: Erlbaum, 337B354.

J÷reskog, Karl G. and Dag S÷rbom (1996), LISREL 8: User’s Reference Guide, Chicago, IL: Scientific Software International.

J÷reskog, Karl G., Dag S÷rbom, Stephen du Toit, and Mathilda du Toit (1999), LISREL 8: New Statistical Features, Chicago, IL: Scientific Software International.

Krantz, David H., R. Duncan Luce, Patrick Suppes, and Amos Tversky (1971), Foundations of Measurement. Volume 1: Additive and Polynomial Representations, New York: Academic Press.

Kyriakopoulos, Kyriakos and Peter A.M. Oude Ophuis (1997), "A Pre-purchase Model of Consumer Choice for Biological Foodstuffs," Journal of International Food & Agribusiness Marketing, 8(4), 37B53.

Meredith, William (1993), "Measurement Invariance, Factor Analysis, and Factorial Invariance," Psychometrika, 58, 525B543.

Scholderer, Joachim, Karen Bruns°, Lone Bredahl, and Kaus G. Grunert (2000), "The Cross-cultural Validity of Food-related Lifestyles", Manuscript submitted for publication.

S÷rbom, Dag (1974), "A General Method for Studying Differences in Factor Means and Factor Structures between Groups," British Journal of Mathematical and Statistical Psychology, 27, 229B239.

Steiger, James H. (1998), "A Note on Multiple Sample Extensions of the RMSEA Fit Index," Structural Equation Modeling, 5, 411B419.

Stiftung +kologischer Landbau (2000), Organic Agriculture Worldwide. Statistics and Future Prospects, Bad Dnrkheim: Stiftung +kologischer Landbau.

The Soil Association (1999), Organic Food and Farming Report 1999, Bristol: The Soil Association.

Th°gersen, John(1998), "The Norm-attitude-behavior Relationship. Theory and Applications in the Environmental Domain," CeSaM Working Paper 1998:14, Aarhus: The Aarhus School of Business.

Th°gersen, John (2000), "Values and Experience in the Development of a Sustainable Consumption Pattern," Paper presented at the Asia Pacific ACR 2000 Conference, March 16-18, 2000, Surfer’s Paradise, Gold Coast, Australia.

Tiilikainen, Aimo and Patricia Huddleston (2000), "Effect of Environmental Soundness on Consumers’ Food Evaluation and Willingness to Buy," Journal of International Food & Agribusiness Marketing, 11(2), 23B43.

Tucker, Ledyard and C. Lewis (1973), "A Reliability Coefficient for Maximum Likelihood Factor Analysis," Psychometrika, 38, 1-10.

USDA Foreign Agricultural Service (1999), France: Organic Food Report 1999, Washington, DC: US Department of Agriculture.

Wandel, Margareta and Annechen Bugge (1996), "Environmental Concern in Consumer Evaluation of Food Quality," Food Quality and Preference, 8, 19B26.

Working, Elmer J. (1927), "What Do Statistical Demand Curves Show?" Quarterly Journal of Economics, 41, 212B235.

----------------------------------------

Authors

Joachim Scholderer, The Aarhus School of Business, Denmark
Karen Brunso, The Aarhus School of Business, Denmark
Klaus G. Grunert, The Aarhus School of Business, Denmark



Volume

E - European Advances in Consumer Research Volume 5 | 2001



Share Proceeding

Featured papers

See More

Featured

Market Structure and Firm Engagement in Divisive Political Issues

Chris Hydock, Georgetown University, USA
Neeru Paharia, Georgetown University, USA
Sean Blair, Georgetown University, USA

Read More

Featured

L8. Recover the Unrecoverable: How Co-Recovery Shifts Consumers ‘Attribution Following a Failed Recovery

Bo Huang, HEC Montreal, Canada
Yany Grégoire, HEC Montreal, Canada
Matthew Philp, HEC Montreal, Canada

Read More

Featured

Liminality, Portals, and Narratives of Transformation

Laetitia Mimoun, HEC Paris, France
Fleura Bardhi, City University of London, UK

Read More

Engage with Us

Becoming an Association for Consumer Research member is simple. Membership in ACR is relatively inexpensive, but brings significant benefits to its members.