RIPPLE
.
Consumers’ attitudes towards regional food products.
A comparison between five different european countries1
Laurent TROGNON*, Jean-Paul BOUSSET**,
Joanne BRANNIGAN*** and Louis LAGRANGE*
*ENITA de Clermont-Ferrand, France, **Cemagref de Clermont-Ferrand France,
***Scottish Agricultural College, Scotland
NB: on trouvera les bases mathématiques de la méthode d'analyse dans https://educator6.webnode.fr/probablement/ et d'autres résultats dans le volet Applications de la page sus-citée.
Abstract
This paper reports the results of a comparative analysis of the main explanatory factors of consumer behaviour relating to regional quality products in five EU countries. Bayesian Network Technology was used to process data collected in the consumer task of the RIPPLE programme.
Consumer's country of origin was shown to be the main factor of difference between the study countries. But the distinction between consumers from the North and those from the South of Europe does not appear so evident, and rather than a marked typology, the analysis suggests a set of indistinct groups partly overlapping.
This analysis also shows that consumers’ knowledge, perception and attitude are as important in explainingbehaviour patterns as the socio-demographic factors. Moreover, these four groups of factors strongly interact to influence consumer behaviour.
Keywords : consumer attitude and behaviour, regional quality food products, Bayesian Belief Network
Introduction
The growing importance of "region" in the marketing of products has become apparent in acompetitive and saturated market-place. With particular reference to food products, it has been suggested that regional labels have been used to differentiate and add value to products (Ritson and Kuznesof, 1996). This potentially enhances the competitive advantage of regional food producers, and helps to maintain the development of the local economies (Tregear et al., 1997). Bouquery (1994) estimated that the market value of "typical" products (most of them are identified with a territorial identity) was 7.5% of the EEC food market (45 billion Euro) and estimated an increase of 1 to 2 billion Euros per year. In some EU countries, consumer interest for these products has been rising since the mid 1980’s. However, this should not be perceived as a mere fashion phenomenon, but as a deep trend, as has been observed in France. Indeed, the fact that particular qualities are linked to traditional know-how and to particular regions, means that these products meet the French consumer's need for confidence and roots in food purchasing behaviour (Rochefort, 1995).
Given this development in the regionalisation of markets, the European Union research programme RIPPLE (Regional Images and the Promotion of Quality Products and Services in the Lagging Regions of the European Union) was established. The main objective of the research was to investigate the producer, consumer and institutional aspects of quality products and services (QPS) within particular European lagging regions. The aim of the consumer task was to investigate consumer awareness and perception of regional images with regard to QPS, and the influence they might have on consumers’ purchase behaviour.
In this paper we report the results of a comparative analysis of the main explanatory factors of some consumer attitudes and behaviour relating to QPS, carried out on data collected in the RIPPLE consumer task. The first part recalls the research method and the data analysis procedure. The second part presents the results of the comparison (main explanatory factors of consumer behaviour and similarities and differences between countries). Finally, we conclude and discuss in the context of using Bayesian networks, the limits of such an analysis, and the perspectives for further research.
1. Materials and Method
1.1. The "Consumer Objective" and the Concept of Quality
The general objective of the RIPPLE project was "to help public and private institutions develop strategies, policies and structures to aid the successful marketing and promotion of quality products and services in the lagging regions of the European Union". The general project objective comprised of four interrelated sub-objectives including the "consumer objective". It aimed "to explore the consumer’s perceptions, wants and needs in relation with the purchase of quality products and services from specific lagging regions ; to examine consumer’s perceptions of the links between location, quality image and actual product and service characteristics ; and to identify the social, psychological and economic factors influencing consumer’s behaviour as regards to the products and services of lagging regions".
One of the prior tasks of the project was to define a common concept of QPS. Indeed, many definitions of quality exist within the social sciences field (Sylvander et al., 1994). Quality is a subjective product attribute, and difficult to define and quantify for the purposes of research. Quality, with regard to products and services, may be determined or signified via many different criteria, for example, the use of official quality marks ; the use of specified products (inputs) or processes ; or associations with tradition, territory or craftmanship. Perceptions of quality may vary between individuals, different social groups, regions or countries. Irrespective of these variations, QPS are normally able to command premium prices in the market place. Thus, one of the essential characteristics of quality is that it is a positional attribute, describing a product that is above the minimum market standard (Parrott et al., 1997).
With regard to QPS, there would appear to be an increase in awareness and consumption levels, thus suggesting that their differentiated nature is attractive to particular consumers in the market-place. However, from the marketing perspective, there is still much to learn about these consumers’ perceptions of quality and their association of quality with place of origin. From the institutional perspective, assistance may be required for producers to establish complete marketing approaches for effective development and increased competitive-ness for these lagging regions of the European Union. Research has indicated that consumers are aware of the regionality of products, and food products in particular (Tregear et al, 1997 ; DGAL, 1994 ; INAO, 1995). However, the consumers’ evaluation depends significantly on the perceived quality and authenticity of the products.
The perception of quality is a subjective and individual assessment, and is influenced by existing beliefs, attitudes and general environmental conditions (Chisnall, 1985). However, for the purposes of our research, it was essential to work to an agreed definition, therefore the RIPPLE team adopted the following statement :
a quality product is one "which is differentiated in a positive manner by way of one or more factors from the standard product, is recognised as such by the consumer and therefore commands a market benefit if it is effectively marketed" (Scottish Food Strategy Group, 1993 quoted in Jenkins et al. 1997).
1.2. The Consumer Survey
1.2.1. Questionnaire contents
The consumer questionnaire contained both closed questions, which provided comparative factual informa-tion about the consumers, and open-ended questions, which provided an understanding of the consumer’s perceptions, attitude, and behaviour. It was composed of four main parts :
*
Part 1 dealt with consumer perceptions of quality ;
*
Part 2 examined consumer perceptions of regional QPS in general ;
*
Part 3 examined consumer perceptions of regional imagery in relation to specific QPS in the study regions ;
*
Part 4 collected consumer classification data.
Taking into account the consumer objective and the wide definition of quality products, 5 kinds of questions were asked of each interviewee. These were related to :
*
Socio-demographic profile
(age ; gender ; occupation ; income levels ; educatio-nal level ; household status ; residence location related to the study region ; main household decision maker regarding QPS) ;
*
Perceptions
(the perceived characteristics of a QPS ; perceived difficulties in obtaining QPS in the market-place) ;
*
Knowledge
(the study region ; QPS from the study region ; QPS from the study country ; QPS from the other partner study regions ; official quality marks used in the study country ; information channels used to find QPS) ;
*
Attitudes
(level of confidence in official quality marks ; price premium for QPS ; importance of the region of origin in the purchase decision ; support for rural small and medium sized enterprises (SMEs) when buying QPS ; regulation for QPS ; image of QPS from rural SMEs) ;
*
Behaviour
(purchase location ; purchase frequency).
Figure 1 : Variable groups describing the behaviour of QPS consumers
1.2.2. Main assumptions about the explanatory factors of consumers’ behaviour
The first four groups formed the explanatory variables to describe the consumer's behaviour in relation to QPS awareness and purchase ; Figure 1 shows the make-up of the groups according to questions from the survey. These groups could be further categorised into personal consumer characteristics (socio-demographic, psycholo-gical, and psychographical) and consumer environment characteristics (culture, social class, family, and situational factors) which are the main explanatory factors of the consumer's behaviour (Filser, 1994)
The Socio-demographic characteristics were considered as the primordial explanatory factors that could influence any other variable. The questions related to Knowledge, which can be associated with the con-sumer's environment characteristics, and the questions related to Perception and Attitude were considered as intermediary factors to explain consumer behaviour.
Therefore, the main assumption was that the socio-demographic, perception, knowledge, and attitude factors interact in order to influence in fine consumer behaviour. As Figure 2 shows, there are many possible channels of interaction amongst the variable groups
Figure 2 : Relationship between the explanatory factors of attitudes and behaviour of QPS consumers
1.2.3. Sampling method
Six partner countries were involved in the research (Finland, France, Greece, Ireland, Spain, and the United Kingdom (UK)), each completing approximately 400 consumer surveys using the quota method (gender, age and occupation criteria).
For the specific analysis presented in this paper, consumers who were unaware of QPS or who no longer purchased QPS were eliminated. Data were used from2 Finland, France, Ireland, Spain and the UK. Therefore, the sample used for this analysis consisted of 1500 QPS consumers.
1.3. Survey Analysis Methodology
The data collected attempted to identify the differences between and the similarities amongst EU countries regarding the behaviour of QPS consumers.
1.3.1. Modelling Stage
Using the socio-demographic, knowledge, perception and attitude factors, the behaviour of QPS consumers was modelled and analysed using the Bayesian belief networks technology [Bousset et al, 1999]. The Bayesian belief network technology refers to Bayesian Probability Theory and to the Graphical Modelling concept.
The Bayes’ Theorem, generalised by Laplace [Laplace, 1951], is the basic starting point for inference problems using probability theory as logic. Bayes' theorem gives the rule for updating belief in a Hypothesis H (i.e. the probability of H) given additional evidence E, and background information (context) I :
p(H|E,I) = p(H|I)*p(E|H,I)/p(E|I) [Bayes Theorem]
where :
*
p(H|E,I), the posterior probability, gives the probability of the hypothesis H after considering the effect of evidence E in context I.
*
p(H|I) is the prior probability of H given I alone ; that is, the belief in H before the evidence E is considered.
*
p(E|H,I), the likelihood, gives the probability of the evidence assuming the hypothesis H and background information I is true.
*
1/p(E|I), is independent of H, and can be regarded as a normalising or scaling constant. The information I is a conjunction of (at least) all of the other relevant statements to determine p(H|I) and p(E|I).
Graphical Modelling consists of using Directed Acyclic Graph (DAG) to represent conditional probabilities leading the values of a set of variables. Such a DAG is called a Bayesian belief network. In other words, a Bayesian belief network is a DAG with a probability table for each variable (also called a node).
The Bayesian belief network representing the probabilistic relationships existing between the socio-demographic characteristics, the knowledge, the perceptions, the attitudes and the behaviour of QPS consumers, was built using the Belief Network Power Constructor [Cheng 1997].
1.3.2. Analysis Stage
Measurement of the similarities and differences among the partner countries was analysed by using three methodologies : Structural Analysis ; Sensitivity Analysis ; and Inductive Inferences.
Structural Analysis pointed out the direct and total motivity/dependence index of each variable/node of the network. The direct/total motivity index of a given node i is the percentage of nodes directly and indirectly influenced by node i. The direct/total dependence index of a given node i is the percentage of nodes that directly and indirectly influence the node i.
Sensitivity Analysis pointed out how much the value of certain dependent variables/nodes Q (e.g., attitudes, behaviours) could be influenced by a single finding at each of the other nodes F : the minimum and maximum belief that each state q of dependent node Q can take due to a finding f at a given motive node F, and the standard deviation (RMS change) of probability P(q|f) about P(q) due to a finding at F, with the finding at F distributed by P(f). There was also measurement of the mutual information of each pair of QF adjacent nodes [Spiegelhalter 1993].
Inductive Inferences pointed out, for a given state q of certain dependent nodes Q (e.g., attitudes, behaviours), the most probable states f for the other nodes F, by searching for the states f of nodes F that give the state q to the node Q with the maximum [a posteriori] probabilities. The set of node states F(f) having a maximum a posteriori probability = 1 is called “most probable explanation” of the state q of Q, or “most probable configuration” explaining the state q of Q ; e.g. : most probable configuration of the French buyer of quality products ; most probable configuration of a buyer of QPS in a supermarket ; etc. The lower a posteriori probabilities indicate the relative probabilities of the other states f given that typical features are found in the most probable configuration. Two or more states f of the same node F may have a maximum a posteriori probability = 1. This indicates that there is more than one “most probable explanation” of the state q of Q, with an equal probability.
2. Results
The Results section first presents the findings about the general behaviour of the European QPS consumer : the main explanatory factors of behaviour ; and the characterisation of the European QPS consumer. Then it compares the behaviour of QPS consumers among the study countries : the main explanatory factors of behaviour according to the country ; and the similarities and differences of the consumer's profile.
2.1. General Behaviour Patterns of the European QPS Consumer
2.1.1. Explanatory Factors for European QPS Consumers' Behaviour
The structural analysis of the Bayesian belief network built from the collected data (Figure 3) validated the assumed relationships between the socio-demographic, perception, knowledge, attitudes, and behaviour variables ; these factors strongly interact. Moreover, calculating the motivity and dependence indices, it quantifies the relative importance and the role of each factor in the behavioural model (table 1). Two main results appeared.
Figure 3 : Explanatory Factors for QPS Consumers’ Behaviour – a Bayesian Belief Network (country effects included)
The country of residence, with a score of 100, had the most influence on other factors of consumer behaviour, accompanied by the socio-demographic characteristics of the consumer.
Considering the motivity index (country effect included), the consumer knowledge (about QPS, regions, and official quality marks), perceptions (about quality), and attitude have together similar weightings of importance to the socio-demographic factors.
Further, the structural analysis also gave insight into each of the factors ; the observations are recorded below.
Table 1 : Total Motivity and Dependence Indeces of Criteria
|
Criteria |
Motivity index |
Dependence index |
||
|
Country effects |
Country effects |
|||
|
included |
excluded |
included |
excluded |
|
|
Country |
100 |
|
|
|
|
Social Profile (S) |
24 |
51 |
|
|
|
Perceptions (P) |
9 |
21 |
41 |
15 |
|
Knowledge (K) |
9 |
16 |
34 |
28 |
|
Attitudes (A) |
7 |
30 |
28 |
23 |
The analysis for the probabilistic relationships existing between socio-demographic, perception, knowledge and behaviour showed that the general behaviour of the QPS consumer varied according to the country of residence. From a marketing perspective, such a finding is to be expected, as the behaviour of consumers is influenced by environmental issues such as culture, tradition and heritage.
On examination of the more specific activities of consumer behaviour, it was found that certain ones were not significantly directly influenced by the country of residence. For example, gaining access to QPS in the market-place, and the level of awareness and knowledge of quality marks were found to be not directly influenced by the country of residence. However, there was indirect influence through the relay variables (see further).
Looking beyond the impact and influence of country of residence on consumer behaviour, there are other factors which were found to have a strong influence. The age of the consumer was particularly influential when purchasing QPS, mainly on the criteria for perceiving quality and differentiation factors. Other socio-demographic factors to be considered included the source of the main income into the household, and the educational level of the respondent. Some influence did come from the gender of the respondent, and their place of residence relative to the study region.
The level of knowledge and awareness of QPS were also very important in determining consumer behaviour. The research found that the source used to obtain QPS information had particular impact on consumers’ behaviour towards these products. The source type would determine the level of detail of information, the mode of delivery and its perceived reliability.
The nature of consumer behaviour comes as a result of consumer attitudes and perceptions, which are themselves created through information and experience in the market-place. Attitudes are formed over time, have the potential to change, and are influenced by many forces. The research found that the attitudinal factor of confidence in official quality marks was extremely influential on consumer behaviour, with the expectation to pay more for QPS having little influence. With the perception factors, differentiation of the QPS from other similar products was influential. It would therefore appear that QPS consumers prefer the visual confirmation of quality through the official certification, which in turn influences the perception of product attributes and the comparison with other products in the market-place.
Consumer behaviour, as generally observed in the five EU countries, would appear to be influenced greatly by all factors of socio-demographic characteristics, perception, knowledge and attitudes. In particular, factors such as country of residence, information source used, and product differentiation all play a key role in the development and variation of consumer behaviour.
The removal of the country of residence from the analysis allows to better evalute the relative weight of the socio-demographic characteristics, perception, knowledge and attitude factors. So that, if the socio-demographic characteristics factors are the main explanatory factors for consumer behaviour, the attitudes factors are in second place. Perception and knowledge factors follow on from that.
Each one of these three factor groups must also be considered from the dependence index perspective. The total dependence index for a variable expresses the amount of influences received from other variables. When a variable is both motive and dependent, it is called a relay variable. It appears then, that these factors have a rather high dependence. Therefore they are important relay variables. So, among the attitude factors, confidence in quality labels was directly influenced by the socio-demographic characteristics, perception, and knowledge factors (respectively through the number of QPS known, the source of main income, and a quality feature) and had direct impact on the purchase frequency of QPS. Among the knowledge variables, the preferred information source was an important relay variable. Upstream, it was influenced by the country of residence and the age of the interviewee, and downstream it influenced the degree of confidence in official quality marks and the purchase frequency of QPS.
2.1.2. Characterisation of European Consumer Behaviour
The structural snalysis indicated the variables which were influenced by and/or had an influencing effect on other variables. To complement those findings, the sensitivity analysis pointed out precisely how the Behaviours, Attitudes and Perceptions were influenced. The analysis was carried out on frequency of purchase of QPS, the purchase location, and the source used to obtain information about QPS. We would like to draw the reader's attention to the interpretation of this part. This is an ascendant analysis of the consumer beha-viour, therefore, the consumer's country of residence is used as an explanatory variable. The consumer profile for each country is presented later, in section 2.2.
*
Purchase frequency
The behaviour of QPS consumers was observed in examining the frequency with which QPS were purchased. The consumers were asked to give three examples of QPS which they purchased and to give an indication of how often the QPS were purchased on average. These answers were processed to build three variables : purchase frequency of the most frequently purchased QPS ; purchase frequency of the second most frequently purchased QPS ; and purchase frequency of the less frequently purchased QPS.
The sensitivity analysis indicated that the purchase frequency of the most frequently QPS purchases depended significantly on : the country of residence (table 2), the number of QPS known, and the purchase location (table 3). At the EU sampling level, the educational level and the occupation of consumers had very little impact. However, when the country of residence variable was excluded, the percentage of labels known and the location of purchase became more important, while the influence of socio-demographic characteristics was insignificant.
Table 2 : Sensitivity analysis : Frequency of purchase of the most frequently purchased QPS (by Country)
|
Maximum frequency of purchases |
observed in |
|
For festive occasion |
Finland |
|
Less than once a month |
Ireland |
|
Less than once a week |
France |
|
More than once a week |
Spain |
There was a positive link between the number of QPS quoted and the purchase frequency : the more the consumers can quote numerous QPS, more frequently they purchase QPS.
Table 3 : Sensitivity analysis : Frequency of purchase of the most frequently purchased QPS (by Purchase location)
|
Maximum frequency of purchases of the “first” QP |
was observed when |
|
For festive occasion |
From producers |
|
Less than once a month |
In supermarket |
|
less than once a week |
In small shops in urban area |
For the second QPS identified, the sensitivity analysis indicated that frequency of purchase depended signifi-cantly on : the country of residence, the percentage of the studied regional QPS known, the percentage of quality labels known, and the information source used. The age of consumers had very little influence.
The maximum frequency of "for festive/special occasion" items for the “second” most frequently purchased QPS was observed in Finland. The maximum frequency of "once a week", "less than once a week" and "more than once a week" were observed in France. The maximum frequency of purchases for "less than once a month" was observed in UK.
This purchase frequency seems to be linked to the information source typically used by the QPS consumer (table 4).
Table 4 : Sensitivity analysis : Frequency of purchase of the most frequently purchased QPS (by Source of Information)
|
Maximum frequency of purchases of the “second” QP |
was observed when the information vector is |
|
For festive occasion |
From producers |
|
Less than once a month |
From neighbours |
|
Once a week |
From market organisations |
The purchase location for QPS depended significantly upon the country of residence only (table 5).
Table 5 : Sensitivity analysis : Purchase location (by Country)
|
Maximum frequency of purchases of QP |
was observed in |
|
Directly from the producer |
Spain |
|
In supermarket |
France |
|
In small shops in urban area |
Finland |
In the observations made for the purchase of the second QPS, the information source used appeared to be of significance in the decision-making process. The sensitivity analysis showed that the variable for information source depended on the country of residence (table 6), and the age of the consumers.
Table 6 : Sensitivity analysis : Source of Information (by Country)
|
Moste Frequent users |
was observed in |
|
The media |
Ireland |
|
Their neighbours |
France |
|
Individual producers |
Finland |
Those consumers most likely to use information circulated by consumer and producer organisations were found to be older than 55 years. Users of information circulated by market organisations and producers were mostly aged between 35 and 44 years. Information circulated by media and neighbours was most likely utilised by consumers aged 25 to 34 years.
The most frequent users of information circulated by the media and by producer organisations were found in Ireland. The most frequent users of information circulated by their neighbours were found in France. The most frequent users of information circulated by the individual producers were found in Finland.
* Sensitivity of Perceptions of Consumers about the Quality Factors of a Given Product
The QPS consumers were asked to rank the three most important factors which they thought characterised a QPS in the market-place. These factors are shown in Table 7, and their relationship with other variables was examined.
Table 7 : Characteristics of a QPS
* QPS are differentiated from other similar products and services in some way
* QPS has an official quality mark
* QPS is associated with a particular area/region
* Consumer can trace the product to the original producer
* QPS is more expensive than average
* QPS is made with special production methods
* QPS is well presented
Consumer perceptions as to the importance of quality marks in defining the quality of a given QPS depended significantly on : consumer perception about the importance of differentiation of QPS products from other similar products, the country of residence, and the information source used. The age of consumers had little influence.
The consumers that considered a QPS as mainly a product with an official quality mark were most frequently found in Finland (around half of them quoted this characteristic in first or second place). Consumers ranking this feature in third place were most frequently found in Ireland. Consumers ranking it in a lower position or not at all were most frequently found in UK.
Association with a specific area has been used as a traceability mechanism for QPS sourcing, and in some ways, authentication. The sensitivity analysis found that area association depended significantly on : the consumers’ perception as to the importance of quality marks in quality definition, their perception as to the importance of product differentiation, the country of residence, and the information source used. Again, the age of consumers had only a slight influence.
The consumers that coupled quality with a geographical area were most frequently found in Spain (ranking the feature in first or third place) and France (ranked in second place). The consumers that ranked it in a lower position or not at all were found most frequently in Finland.
Specific production methods can sometimes be used as an indicator of the authenticity and quality of a QPS. The sensitivity analysis showed that perceptions on production methods as an indicator of quality significantly depended on : consumers’ perception as to the importance of presentation, the country of residence, consumers’ perception of the importance of product differentiation, and the information sources used. The age of the consumer had very little influence
The consumers that considered presentation as an indicator of quality were most frequently found in Ireland (around half ranked this features in first or second place). Consumers who ranked it in third place were most frequently found in the UK. Consumers who ranked it in a lower position or not at all were most frequently found in Finland.
2.2. Similarities and differences in Behaviour of QPS Consumer among Countries
After the general analysis of the behaviour of European QPS consumers, it is now appropriate to examine the similarities and differences of QPS consumer behaviour amongst the partner countries. The general approach of the main explanatory factors using structural analysis is presented first, followed by a comparison of the consumer's profiles.
2.2.1. Explanatory Factors for Behaviour of QPS Consumers Regarding its Country of Origin
This approach of the main explanatory factors among countries stems from a structural analysis of the Bayesian network carried out for each country. Table 8 summerises the synthetic data used for the comparison.
* Socio-Demographic (S-factors)
The S-factors were more motive in France and the UK than in the other countries. These differences arose from the number of highly motive S-factors in each country : 5 in France and in the UK ; 4 in Spain ; 3 in Ireland and in Finland. In other words, not every S-factor used in the model had an influence on QPS consumer behaviour.
Moreover, depending on the country, the most motive factors were not the same. For example, location of residence relative to the study region was one of the most motive factors in France, the UK and Finland, but it had very little influence in Ireland and Spain. The gender factor had very little influence in France, the UK and Finland, whereas it was important in Spain and in Ireland. The level of education of the respondent and the main source of income into the household were also important according to the country. Both were very motive variables in France and the UK ; only the level of education was important in Ireland and Finland, and the source of household income was very motive in Spain.
Table 8 : Motivity and Dependence Indices of Criteria in Each Country
|
|
France |
UK |
Ireland |
FinIand |
Spain |
|||||
|
|
Motivity |
Depend. |
Motivity |
Depend. |
Motivity |
Depend. |
Motivity |
Depend. |
Motivity |
Depend. |
|
Socio-Demo |
78 |
|
68 |
|
48 |
|
49 |
|
56 |
|
|
Perceptions |
19 |
21 |
21 |
40 |
6 |
26 |
19 |
73 |
17 |
30 |
|
Knowledge |
9 |
35 |
11 |
50 |
9 |
39 |
9 |
41 |
18 |
40 |
|
Attitudes |
14 |
42 |
4 |
53 |
11 |
42 |
8 |
26 |
6 |
33 |
* Perception (P-factors)
The P-factors were more motive in France, the UK, and Finland than in the other countries. The heterogeneity of motivity for the variables differed according to the country, although some patterns did emerge. In France, the UK and Spain, the perception that QPS would be more expensive was the most motive P-factor, whereas it was very little in Spain and Ireland. Conversely, in these countries, an official quality mark as a quality feature was the most motive P-factor, while it was of little importance in France, the UK and Spain.
* Knowledge (K-factors)
The K-factors were more motive in Spain and the UK than in the other countries. Whatever the country, the information source used to find QPS was one of the most motive factors : first place in France, Ireland and Spain ; second place in UK ; and third place in Finland.
2.2.2. Comparative Analysis of Profiles of QPS Consumers
To draw the consumer profile the inductive inferences method was used. The method identifies which is/are the most probable configuration(s) of the profile. For each of perception, knowledge, attitude and behaviour factors, in each country, the characteristics of the typical consumer (TC) are described below.
a) Similarities
The typical consumer (TC) of each country :
1. seldom perceives QPS as being more expensive than average ;
2. knows the study regions in the country, but does not know the other study regions ;
3. thinks the origin of the product is important in the purchasing process ;
4. thinks QPS supports local or regional businesses (only some French TC seemed to have some doubts about this point) ;
5. expects to pay extra for QPS and thinks it is worth the extra cost ;
6. is unlikely to meet any difficulties in locating QPS in the market-place ;
7. claims to have knowledge about official quality marks.
b) Differences
* Perceptions
The French, UK and Spanish TC perceived association with a particular area or region as an important quality feature. The Finnish TC considered having an official quality mark as the most important quality characteristic. The French, UK and Spanish TC did not perceive official quality marks as important quality characteristics. However, with a lower maximum a posteriori probability, it was ranked in second or third position by the Spanish and UK TC. The Irish TC considered official quality marks as important features. Differentiation of QPS from other similar products was considered by the French TC as being most important. In the other countries, the TC did not consider differentiation as important. However, with a lower maximum a posteriori probability, the UK and Spanish tended to also rank it in first place.
* Knowledge (K-factors)
Knowledge of the regions.
The Finnish and Spanish TCs were not particularly familiar with the other study regions. The French TC seemed to know only the Greek regions of Achaia/Korithia. However, the Irish and UK TC showed a wider knowledge of European regions. The Irish TC knew the regions in Wales (UK), Valencia (Spain) and Normandy (France). With a lower maximum a posteriori probability, there was also some knowledge of Grampian (UK). The UK TC had knowledge of Cork (Ireland), Valencia (Spain) and Normandy (France). With a lower maximum a posteriori probability, there was also knowledge of Auvergne (France) and Sligo/Leitrim/Roscommon (Ireland).
Knowledge of the products
The French and UK TC had the wider set of quality food products (QFP), quoting between 1 and 4 QFP. The Irish TC quoted 1 or 2 QFP ; the Finnish and Spanish TC tended not to quote a specific number of QFP.
The French, UK and Finnish TC knew 80 to 100% of the QPS from their respective regions. The Irish and Spanish TC knew 20 to 40% of the QPS from their regions. Does this lack of knowledge express a restrictive choice of products made by the researcher, a small set of products due to the local consumption habits, a low sensitivity to this product type, or a lack of information to aid awareness ?
When asked about the full list of QPS from the other study regions, only the French TC quoted up to 40% of the products. The other nationalities recognised on average approximately 20%.
Knowledge of official quality marks
Of course, every TC claims to have knowledge of the official quality marks. However, the set of official quality marks quoted for each study region varied by country. The UK TC had strong knowledge of the marks recognising 80 to 100% of the list. The Finnish and the French TC recognised 60 to 80%, but with a high probability they recognised 40 to 60% of the list. The Irish and Spanish TC tended to recognise 20 to 40% of the list. Such a result shows that, if the list of official quality marks was considered comparable between each country, in most of the countries, consumers were not well informed about or aware of official quality marks. Nevertheless, a high level of recognition does not mean the consumer really understands the meaning of the marks (Trognon et al., 1999).
Sources of Information to find QPS
Only the Irish TC used media sources of information to find QPS. The other nationalities (Spanish TC with a lower probability) used and preferred neighbours as an information source (friends, relatives, etc.).
* Attitudes (A-factors)
The French, Irish and Spanish TC had a high degree of confidence in the official quality marks. The UK TC had a more moderate confidence in them.
* Behaviours
Purchase frequency. Regarding the purchase frequency of QPS, the main behaviour of the Finnish TC was to buy QPS infrequently for festive or special occasions. The Spanish TC had the most probable behaviour : for festive or a special occasion, or more than once a week. The French, UK and Irish TC purchased on a frequent (weekly) basis. However, they had several likely purchase behaviours.
Purchase location. For the Finnish and Spanish consumer, the preferred purchase location for QPS was directly from the producer. The French consumer preferred to purchase from a supermarket ; this was also the case for the Irish consumer, but to a lesser extent. The UK consumer purchased from small shops within the area.
3. Conclusion and Discussion
3.1. Bayesian network Technology
By providing graphical means for representing and manipulating probabilistic knowledge, Bayesian networks overcome many of the conceptual and computational difficulties of rule-based systems [Pearl, 1988]. Their basic properties and capabilities can be summarised as follows :
1. Graphical methods make it easy to maintain consis-tency and completeness in probabilistic knowledge bases. They also prescribe modular procedures of knowledge acquisition which significantly reduce the number of assessments required.
2. Independent relationships can be dealt with explicitly. They can be articulated by an expert, encoded graphically, read off the network, and reasoned about, yet they remain forever robust to numerical imprecision.
3. Graphical representations uncover opportunities for efficient computation. Distributed updating is feasible in knowledge structures that are rich enough to exhibit inter-causal interactions (e.g., "explaining away"). And, when extended by clustering or conditioning, tree-propagation algorithms are capable of updating networks of arbitrary topology [Pearl 1988].
4. The combination of predictive and abductive inferences resolves many problems encountered by first-generation expert systems and renders belief networks a viable model for cognitive functions requiring both top-down and bottom-up inferences.
5. The causal information encoded in Bayesian networks facilitates the analysis of action sequences, their consequences, their interaction with observations, and their expected utilities, and hence the synthesis of plans and strategies under uncertainty [Pearl, 1994a].
6. The isomorphism between the topology of Bayesian networks and the stable mechanisms that operate in the environment facilitates modular reconfiguration of the network in response to changing conditions, and permits deliberative reasoning about novel situations.
On the introduction of new elements for the analysis and the modelling of complex events, the use of the Bayesian network Technology in consumer behaviour research should develop.
3.2. Results
The analysis presented some limits, and these are linked to the methodology background :
*
There is an inherent difficulty in building a question-naire common to six countries, five languages, twelve labs, and a multi-diciplinary team (geographers, economists, marketers, rural developers).
*
There were most likely a number of semantic difficulties. The question must be asked - do the concepts of QPS and of regionality have the same significance in the mind of each interviewee in the various regions ? The collected information shows the concepts are rather unclear, even within a region.
*
As in most marketing research, what was described as behaviour was studied through the interviewee's declarations. Therefore, it was the consumers’ attitudes which were examined rather than observing their real behaviour patterns.
*
Consumer behaviour is dynamic, however this research was static in nature.
*
Distortion of the survey : it is likely that regional and country differences may partly result from the distortions linked to the researchers who made the interviews. The individual techniques employed create a natural bias to the research.
*
The variables used do not cover all the likely consumer behaviour explanatory factors.
Despite its limitations, the research strives to identify the differences and similarities in consumer behaviour relating to regional QPS across a number of European states. There was no clear distinction in behaviour patterns between consumers from Northern and Southern European states, and rather than presenting a marked typology, the results suggested a set of indistinct groups partly overlapping.
The analysis shows and quantifies explanatory factors of consumer behaviour other than the current socio-demographic characteristics typically found in consumer profile construction. Knowledge, perception and attitude factors, when combined, are seen to be as motive as the socio-demographic factors. Moreover, these four groups of factors strongly interact to influence in fine the consumer behaviour.
NOTES
(1) A contribution based on the RIPPLE Programme (RIPPLE FAIR3 CT96 1827 : Regional Images and the Promotion of Quality Products and Services in the Lagging Regions of the European Union).
(2) Results from Greece were not available for the paper.
Bibliography
BOUQUERY J.M. (1994). Caractéristiques de qualité particulières et chaîne de qualités patrimoniales : définition et implications pour le marketing et le développement. In Multon, JL (coord), pp. 103-116.
BOUSSET J.P. et al. (1999). Integrating case and model-based reasoning for thinking about the future : Results from France. RIPPLE Project (FAIR 3 CT 96 1827), Working Paper 12. Cemagref, Unité de Recherche ET, Clermont-Ferrand.
CHENG J. et al. (1997). An algorithm for Bayesian belief network construction from data. Proceedings of AI & STAT'97, Fort Lauderdale, Florida.
CHENG J. et al. (1997). Learning belief networks from data : an information theory based approach. Proceedings of the Sixth ACM International Conference on Information and Knowledge Management.
CHISNALL, P. (1985). Marketing : A Behavioural Analysis. Second Edition. McGraw-Hill.
CHOW C.K. et al. (1968). Approximating discrete proba-bility distributions with dependence trees. IEEE Tran-sactions on Information Theory, Vol. 14, pp. 462-467.
DIRECTION GÉNÉRALE DE L'ALIMENTATION, (1994). Perception et image des produits alimentaires portant les reconnaissances officielles de qualité. Paris : DGAL, June p. 18.
ESPERCIA J. et ali. (1999). Consumer Survey Results and Analysis : Spain. Working Paper 9, RIPPLE (FAIR3 CT96 1827).
FILSER M. (1994). Le comportement du consommateur. Dalloz, p. 426.
INAO (1995). Attitudes des français vis-à-vis des produits de terroir et des produits sous appellation d’origine contrôlée. Note de synthèse de l’enquête Goût, Gastronomie et Gourmandise, Paris : INAO, p. 6.
JENKINS T., PARROTT N. (1997). Marketing in the Context of Promotion of Quality Products and Services in the Lagging Regions of the European Union. Working Paper 4, RIPPLE FAIR3-CT96-1827, p. 10.
LAPLACE P.S. (1951). A Philosophical Essay on Probabilities, unabridged and unaltered reprint of Truscott and Emory translation. Dover Publications, Inc., New York, original publication date 1814.
MARSAT J.B., et al. (1999). Synthèse France : Consumer Survey Results and Analysis : France. RIPPLE (FAIR3 CT96 1827).
McINTYRE B., HENCHION M. (1999). Consumer Survey Results and Analysis : Ireland. Working paper 9, RIPPLE (FAIR3 CT96 1827).
PARROTT N., JENKINS T., LAMPARD J., LEAT P., WILLIAMS F. BRANNIGAN J. (1999). Consumer Survey Results and Analysis : United Kingdom. Working Paper 9, RIPPLE (FAIR3 CT96 1827).
PARROTT N., KNEAFSEY M., JENKINS T., ILBERY B. (1997). Guidelines for Questionnaire Surveys : Theoretical Perspectives on Quality Products and Services, Synopsis from Geography and Marketing, RIPPLE FAIR3-CT96-1827, p. 2.
PEARL J. (1988). Probabilistic reasoning in intelligent systems : networks of plausible inference. Morgan Kaufmann
RIPPLE (1996) “Technical Annex of Project” FAIR3 CT96 1827, p. 2.
RITSON C., KUZNESOF S. (1996). “The Marketing of Rural Food Products and its Role”. In P. Allanson and M. Whitby (eds), Sustaining the Rural Economy : The Rural Economy and the British Countryside. London Earthscan.
ROCHEFORT R. (1995) La société des consommateurs. Editions Odile Jacob, p. 267.
SÖDERLUND A. (1999). Consumer Survey Results and Analysis : Finland”, Working Paper 9, RIPPLE (FAIR3 CT96 1827).
SPIEGELHALTER D.J., PHILIP DAWID A., STEFFEN L., LAURITZEN G., ROBERT G. COWEL (1993). Bayesian Analysis in Expert Systems. In Statistical Science, Vol. 8, No. 3, pp. 219-283.
SYLVANDER B., LASSAUT B. (1994). L'évolution contemporaine du secteur agro-alimentaire et l'incertitude. In Multon, JL (coord). La qualité des produits alimentaires, politique, incitations, gestion et contrôle, Paris : Lavoisier, Tec&Doc, 2ème édition, pp. 28-59.
TREGEAR A., MOXEY A., KUZNESOF S. (1997). Marketing of Regional Foods : A Policy Perspective Paper presented to Agricultural Economics Society Annual Conference, Edinburgh, March 1997.
TROGNON L., LAGRANGE L., MARSAT J.B. (1999). Perception des produits alimentaires régionaux de qualité par le consommateur. In Actes du colloque SFER INRA ENITA, Les signes officiels de qualité - Développement agricole : Aspects techniques et économiques, April 14-15,1999, Paris : Ed. Tec & Doc Lavoisier, pp. 265-271.
TROGNON L., LAGRANGE L., JANIN S. (1999). Consumer Attitudes Towards Regional Food Products. A Case-Study for Auvergne, in Food for the Consumer - Measurements of Consumer Attitudes, AIR-CAT meeting reports, Consumer Attitudes Towards Typical Foods, The European Food Consumer 5, 1, AIR-CAT Matsforsk ed., pp. 30-39.
TROGNON L., (1998). The Influences of Territorial Identity on the Consumer Preferences, in Food for the Consumer - Measurements of Consumer Attitudes, AIR-CAT meeting reports ; workshop : Consumer Preferences for Products of the Own Region/Country and Consequences for the Food Marketing, 4, 3, AIR-CAT Matforsk ed., pp19-27