- Email: [email protected]
Do pig farmers preferences bias consumer choice for pork? Response to critique of the pork preference studies
Meat Science 85 (2010) 788–791
Contents lists available at ScienceDirect
Meat Science journal homepage: www.elsevier.com/locate/meatsci
Do pig farmers preferences bias consumer choice for pork? Response to critique of the pork preference studies T.M. Ngapo a,*, J. Fortin a, J.-F. Martin b a b
Agriculture and Agri-Food Canada, Food Research and Development Centre, 3600 Bd Casavant Ouest, Saint-Hyacinthe, Québec, Canada J2S 8E3 Unité de Maladies Métaboliques et Micronutriments, INRA, Theix, 63122 St-Genès-Champanelle, France
a r t i c l e
i n f o
Article history: Received 30 October 2009 Received in revised form 8 March 2010 Accepted 11 March 2010
Keywords: Consumers Preferences Pork Choice Colour Fat Marbling Drip
a b s t r a c t Québec consumers and pig farmers selected their preferred chop from 16 images that had been modified to give 16 treatments: two levels each of fat cover, colour, marbling and drip. The selection process was repeated eight times from different groups of chops. Fat cover (47% preferred lean) and colour (44%, light red) were the most frequently chosen characteristics. No significant differences were observed between farmers and consumers preferences (v2 test, P < 0.05). Two preference-based clusters were found; 41% preferring dark red, lean meat and 59%, light red, lean meat, without marbling or drip. Choice-based clusters showed no significant links with either individual socio-demographic items, including pig farmer as occupation, or the three socio-demographic-based clusters observed (v2 test, P < 0.05). No evidence was found to suggest that the choices of pig farmers differed from those of consumers and, therefore, inclusion of pig farmers in consumer panels would not bias consumer choice for pork. Crown Copyright Ó 2010 Published by Elsevier Ltd. All rights reserved.
1. Introduction Using photographic images varying systematically in appearance, a programme was started in France to identify the most important characteristics of fresh pork which determine consumer choice and to show how consumer segmentation in choice relates to socio-demographic differences (Ngapo, Martin, & Dransfield, 2004). This study was undertaken in 23 countries (Chen, Guo, Tseng, Roan, & Ngapo, 2010; Cho et al., 2007; Cipolli, Silveira, Ngapo, & Dransfield, 2003; Fortomaris et al., 2006; Ngapo, Martin, & Dransfield, 2007a, 2007b; Verbeke et al., 2005) and data collection sites in many countries included agricultural shows. These shows provided a cross-section of the general public with great variation in socio-demographic attributes including consumer occupation. Regardless of this demonstrated variability, when the results of this study have been presented, comments have invariably been raised about sample bias due to high participation of pig farmers. While of no consequence to the international study, these comments evoke the question ‘‘do pig farmers really have a different preference for pork than the general public?” After all, pig farmers are merely consumers of pork, who happen to raise pigs.
* Corresponding author. Tel.: +1 450 768 3300; fax: +1 450 773 8461. E-mail address: [email protected] (T.M. Ngapo).
The entire Canadian panel in the international study comprised 248 Quebecers and 802 Albertans, from which a representative sample of 200 consumers was selected (Ngapo et al., 2007a, 2007b). An additional 50 Québec pig farmers were surveyed as a part of the Québec panel, but due to concerns of potential sample bias these responses were not used. These responses provide an opportunity to determine if pig farmers’ preferences differ from those of consumers in Québec, the aim of the present study.
2. Materials and methods 2.1. Pork chops The method and chop characteristics are described in detail by Ngapo et al. (2004). Briefly, photographs of 16 commercial pork chops were computer-modified to give two levels of each of the characteristics: fat cover, colour, marbling and drip. The resulting 256 (2 2 2 2 16) images have been published as a book (Dransfield, Martin, Miramont, & Ngapo, 2001) which can be used as a tool for analysing the importance of those factors in consumer choice. The book is comprised of six series of which series 1 + 2, 3 + 4, and 5 + 6 each contain all 256 images. A series constitutes 16 (A4) pages or eight double-pages. Every double-page contains the 16 different chop shapes and each chop represents one of the combinations of the four characteristics studied. Therefore, every
0309-1740/$ - see front matter Crown Copyright Ó 2010 Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.meatsci.2010.03.022
789
T.M. Ngapo et al. / Meat Science 85 (2010) 788–791
double-page contains a complete set of all 16 combinations of the two levels of each of the four characteristics. Both the order of representation of the characteristics with respect to the chop shape and the position of the chops in a double-page are randomised. It is important to note that the chop shape was not a factor studied. 2.2. Consumers Consumers were pork eaters older than 15 years of age and chosen at random. Consumers were surveyed in July and August of 2002 in Québec at agricultural shows and workplaces. In total 298 consumers from Québec completed the survey, including 50 pig farmers. 2.3. Selection Each consumer selected their preferred chop from each doublepage showing the 16 combinations of two options of each of the characteristics of colour, fat cover, marbling and drip. The selection was repeated from eight different double pages which showed the same 16 appearance characteristics and chops, but in different combinations. The consumers also completed a short questionnaire asking basic socio-demographic and purchase- and eating-behaviour information (Table 1). The questionnaire was an abridged version of that used in the international survey (Ngapo et al., 2007b). Each new consumer was given a series in the order 1–6 so that all series were used approximately equally throughout a survey period.
consistently selected. The results were quantified by the definition that if P6 of 8 choices for one consumer are the same for a given characteristic, the choice is a ‘real’ choice (P < 0.14). If <6 choices are the same, selection for the given characteristic is considered to be ‘inconsistent’. This test assumes a binomial distribution of the results (P = 0.5). Significant differences in the number of choices were observed using the v2 test (P < 0.05). The number of characteristics used to form consistent choices was then obtained by calculating the percentage of consumers using 0, 1, 2, 3 or all 4 characteristics. For each consumer and using all choices, the sum of the number of times each combination of characteristics was selected is calculated. Using the sums for each combination, a contingency table is constructed comprising the consumers and their choices. Using this contingency table, a correspondence analysis using all 15 dimensions was undertaken accounting for 100% of the variability. The coordinates of each consumer obtained in the 15 dimensions of the correspondence analysis were used as the basis for cluster analyses. Firstly, a hierarchical cluster analysis using Ward distance was undertaken using the SAS CLUSTER procedure (SAS, 1999). The number of clusters to be retained was selected by considering the ‘distance’ between clusters and the profile of the resulting graph. A disjoint cluster analysis was then carried out using the SAS FASTCLUS procedure (SAS, 1999) forcing the consumers into the different clusters. Links between the consumer choice-based clusters and questionnaire items were determined using v2 test. Similarly to the choice data, correspondence and cluster analyses were undertaken to define clusters based on the questionnaire items.
2.4. Data analyses and statistical methods 3. Results and discussion 2.4.1. Analysis of choices The choices consumers made were divided into three categories for each characteristic; in the first two categories the consumer actually chose one of the two levels of the given characteristic, whereas for the third category the given characteristic was not
3.1. Consumers The socio-economic questionnaire composition and responses are given in Table 1. There were proportionately more consumers
Table 1 Questionnaire composition with responses attained for the entire consumer sample, Québec consumers and pig farmers. Question
What is your age (years)? Gender? Marital status? How many people live in your household?
How often do you eat meat?
Response options
<35 years P35 years Female Male Single Married 1 2 3 4 5+ Every meal, everyday Once/day, everyday Several times/week Once/week or less
Consumers (number) Quebecers
Farmers
All
Quebecers
Farmers
84 213 189 107 87 198 27 132 45 46 34 50 116 116 14
66 181 173 73 81 154 25 116 37 33 23 38 88 107 13
18 32 16 34 6 44 2 16 8 13 11 12 28 9 1
28 71 63 36 29 66 9 44 15 15 11 17 39 38 5
27 73 70 29 32 63 10 47 15 13 9 15 35 43 5
36 64 32 68 12 88 4 32 16 26 22 24 56 18 2
6 168 72 151 93 50 20 37 55 105 190 60
3 44 3 37 13 11 2 17 19 25 39 14
3 70 25 63 36 57 21 18 25 43 76 25
2 68 29 61 38 54 22 15 22 42 77 24
6 88 6 74 26 85 15 34 38 50 78 28
The following make reference to fresh pork and not to sausages, ham or other processed pork products How often do you eat pork? Everyday 9 >Once/week 212
Consumers (%)
All
In the percentage columns, given as a percent of those who claim to have changed their frequency of consumption of pork.
790
T.M. Ngapo et al. / Meat Science 85 (2010) 788–791
Table 2 Selection of visual chop characteristics by Quebec consumers and pig farmers according to consistency of choice (P6 out of 8 choices deemed a selection for a given level of characteristic; <6 deemed inconsistent). Selection (% consumers)a
Selection (number of consumers)
a
All
Quebecers
Farmers
All
Quebecers
Farmers
Colour
Dark Light Inconsistent
64 130 104
57 103 88
7 27 16
21 44 35
23 42 35
14 54 32
Fat cover
Fat Lean Inconsistent
20 141 137
17 116 115
3 25 22
7 47 46
7 47 46
6 50 44
Marbling
Marbled Not marbled Inconsistent
31 74 193
24 64 160
7 10 33
10 25 65
10 26 65
14 20 66
Drip
Drip No drip Inconsistent
29 96 173
27 80 141
2 16 32
10 32 58
11 32 57
4 32 64
No significant differences in selection were observed between farmers and Quebecers for any given level of a characteristic by v2 test (P < 0.05).
less than 35 years of age in the group of farmers (36%) than the group of Quebecers (27%). More than two thirds of the Quebec consumers were female compared to less than a third of the farmers. A third (32%) of the Quebecers were single, whereas almost all (88%) of the farmers were married. Almost a half of the Quebec consumers lived in two-person households and about 12% in each of households of 1, 3, 4 or 5+ persons, whereas a third of the farmers lived in two-person households, about a quarter in households of 4 or 5+ persons and 16% in three-person households. Few lived in one-person households reflecting the small proportion of single farmers. Almost all respondents eat meat at least several times a week of which 80% of the farmers eat meat at least once a day compared to 50% of the Quebec consumers. Frequency of pork consumption was also higher in the farmers, with 94% eating pork more than once a week compared to 70% of the Quebecers. A quarter of the farmers claimed to have changed their pork consumption recently of which most (85%) have increased their consumption. In comparison, 38% of the Quebecers have changed their pork consumption recently, of which about half have increased their consumption. With responses from more than three quarters of both farmers and Quebec consumers, the most cited reason for liking pork is its taste. 3.2. Consumer choices The number of choices observed for all four characteristics are presented in Table 2. The more selected choices are the same for Quebecers and farmers with averages being 44% light red colour, 47% lean fat cover, 25% no marbling and 32% no drip. This analysis also shows that most of the choices for marbling (65%) and drip (58%) can be considered random based on the inconsistent selection for a given level of characteristic. While a greater proportion of the producers (54%) than consumers (42%) appeared to prefer the light red pork, none of the choices of consumers and producers were significantly different by v2 test (P < 0.05). Of the characteristics studied, the number that the consumers and producers used to make their selection, based on the ‘real’ and ‘inconsistent’ choices above, was similar for farmers and consumers with, on average, 19% using one characteristic, 41% using two and 23% using three. Only 7% used all four characteristics, and another 7% of consumers did not use any. It is notable that twice as many producers (14%) as consumers (7%) did not use any of the characteristics studied to make their choices. A 15-dimensional correspondence analysis projecting the consumers and producers depending on their choices of the 16 different combinations showed no consensus for any particular
Table 3 Percent of consumers selection within chop characteristics for each cluster with significant differences (shown in bold) as determined by v2 (P < 0.05) compared to results of the entire panel. Clusters (% consumers) 1
All (% consumers) 2
Colour
Dark red Light red Inconsistent
53 0 47
0 73 27
21 44 35
Fat cover
Fat Lean Inconsistent
11 36 53
4 55 41
7 47 46
Marbling
Marbled Not marbled Inconsistent
19 16 65
5 31 64
10 25 65
Drip
Drip No drip Inconsistent
12 29 59
8 35 58
10 32 58
41 121
59 177
Consumers in cluster (%) Consumers in cluster (number)
combination of the four characteristics studied. The first dimension, accounting for 16.7% of the total variability, corresponded to colour, the second, accounting for 10.5% of the total variability, to fat cover and the third accounting for 8.4% of variability, to drip. 3.3. Choice-based clusters Cluster analyses were undertaken to group respondents with similar strategies for pork chop selection (Table 3). The clusters can be defined as follows: Cluster 1 of 121 consumers (41%): preferred dark red, lean meat. Cluster 2 of 177 consumers (59%): preferred light red, lean meat, without marbling and without drip. The strongest preferences differentiating the two clusters were those of colour with 53% of respondents in cluster one preferring dark red meat and 73% in cluster two preferring light red meat. The distribution of the individual questionnaire items over the two choice-based clusters of farmers and consumers combined was tested by v2 (P < 0.05) and no differences in distributions between the clusters were observed.
T.M. Ngapo et al. / Meat Science 85 (2010) 788–791
Correspondence and cluster analyses were undertaken to derive clusters based on the socio-demographic and purchase- and eating-behaviour items, items from the questionnaire. The hierarchical cluster analysis showed three different clusters. These clusters would be of interest if a significant link was observed with the choice-based clusters. However, no significant link was observed between the choice-based and questionnaire-based clusters when assessed by v2 test (P < 0.05) and therefore the questionnaire-based clusters were not further investigated. 3.4. Occupation impact Studies demonstrating a significant effect of consumer occupation on food preferences are few. One might question as to how a consumer’s occupation could influence food preferences, yet on many occasions when presenting consumer preference data obtained at agricultural shows, comments from the audience have been raised about sample bias due to high participation of pig farmers. Questioning these scientists, economists and marketers as to the reasoning behind the supposition that pig farmers might have different preferences for pork chops than other consumers, generally evoked responses of the belief of an acquired superior knowledge of meat quality as a result of producing the animals from which the meat is attained. While this logic is questionable, it is not unreasonable to think that pig farmers might consume more pork than the average consumer, which, through greater experience with the product, might in turn influence their preferences. Many models exist to predict consumer behaviour of which experience plays an important role. For example, Steenkamp (1989) and Anderson (1994) presented frameworks analysing the quality perception process for food products based on the assumption that qualities sought in a food product by a consumer are always experience and credence qualities. Grunert (1997) expanded this model to include the determinants of experience as opposed to expected quality and applied this model to meat. While there are numerous such models, few studies on the impact of experience on consumer choice of foods are reported, but rather studies tend to focus on the ‘‘before purchase” part of the models excluding experience, or describe the impact of experience at consumption. This is no surprise considering the complexity of the relationship between experience and expectation (Frederickson & Kahneman, 1993; Mela, 2001; Varey & Kahneman, 1992) and the difficulty in measuring the former which is a consequence of intensity and duration of experience, and is subject to distortions in memory (Rozin & Tuorila, 1993). One study that demonstrates the influence of experience with a product on the perception of that product is that of Turner and Collison (1988) who undertook 285 hedonic acceptance evaluations of whole meals and their components. Results showed that customers who were either catering staff or catering students, and therefore directly involved in the food industry, scored lower than other customers. In this study, the occupation of the consumers directly impacted on experience with the food product studied which in turn impacted on preference. In the present study, 94% of pig farmers claimed to eat pork more than once a week demonstrating that this occupational group has greater experience with the product than the Quebec consumers (70% claimed the same consumption frequency). However, no differences in visual preferences of pork chops were observed. Of the four characteristics studied, the most important in both Quebec consumers and farmers choices of pork chops were fat cover (with 47% of the entire panel preferring lean fat cover) and colour (44% preferring light red meat).
791
4. Conclusions Two clusters of preferences for pork chops were identified from a survey of Quebec consumers that included 50 farmers. These clusters showed no significant links with either individual sociodemographic items of the consumers, including farmer as occupation, or the three socio-demographic-based clusters identified on the whole sample. The most important characteristics in Quebec consumers and farmers choice of pork chops were fat cover and colour. Using this novel choice system of within-consumer preference replication and image manipulation, no evidence was found to suggest that pig farmer’s choices of pork differ from that of consumers in Quebec and inclusion of pig farmers in the consumer panel would not bias consumer choice for pork. Acknowledgements The authors would like to acknowledge Égide, France and the Ministère de Relations Internationale (MRI), Québec, Canada for the travel grant accorded to Mr. J.-F. Martin. The authors also thank Dr. E. Dransfield for his invaluable role in this project and without whom the study would not exist. References Anderson, E. S. (1994). The evolution of credence goods: A transaction approach to product specification and quality control. MAPP working paper no. 21. The Aarhus school of business, Aarhus [Grunert, K.G. (1997). What’s in a steak? A crosscultural study on the quality perception of beef. Food Quality and Preference, 8, 157–174]. Chen, M. T., Guo, H. L., Tseng, T. F., Roan, S. W., & Ngapo, T. M. (2010). Consumer Choice of Pork Chops in Taiwan. Meat Science, 85(3), 555–559. Cho, S., Park, B., Ngapo, T., Kim, J., Dransfield, E., Hwang, I., et al. (2007). Effect of meat appearance on South Korean consumers’ choice of pork chops determined by image methodology. Journal of Sensory Studies, 22, 99–114. Cipolli, K. M. V. A., Silveira, E. T. F., Ngapo, T. M., & Dransfield, E. (2003). Consumer preferences of pork chops – The Brazilian survey. In Proceedings 5th Pangborn sensory science symposium: A sensory revolution (p. 187), 20–24 July 2003, Boston, USA. Dransfield, E., Martin, J.-F., Miramont, J., & Ngapo, T. M. (2001). Meat appearance: Pork chops. A tool for surveying consumer preferences. INRA, France, ISBN: 2-73800976-X. Fortomaris, P., Arsenos, G., Georgiadis, M., Banos, G., Stamataris, C., & Zygoyiannis, D. (2006). Effect of meat appearance on consumer preferences for pork chops in Greece and Cyprus. Meat Science, 72, 688–696. Frederickson, B. L., & Kahneman, D. (1993). Duration neglect in retrospective evaluation of affective episodes. Journal of Personality and Social Psychology, 65, 45–55. Grunert, K. G. (1997). What’s in a steak? A cross-cultural study on the quality perception of beef. Food Quality and Preference, 8, 157–174. Mela, D. J. (2001). Why do we like what we like? Journal of the Science of Food and Agriculture, 81, 10–16. Ngapo, T. M., Martin, J.-F., & Dransfield, E. (2004). Consumer choices of pork chops: Results from three panels in France. Food Quality and Preference, 15(4), 349–359. Ngapo, T. M., Martin, J.-F., & Dransfield, E. (2007a). International preferences for pork appearance. I: Consumer choices. Food Quality and Preference, 18(1), 26–36. Ngapo, T. M., Martin, J.-F., & Dransfield, E. (2007b). International preferences for pork appearance. II: Factors influencing consumer choice. Food Quality and Preference, 18(1), 139–151. Rozin, P., & Tuorila, H. (1993). Simultaneous and temporal contextual influences on food acceptance. Food Quality and Preference, 4, 11–20. SAS (1999). SAS Users Guide: Statistics. Version 8.1. Cary: SAS Institute Inc. Steenkamp, J.-B. E. M. (1989). Product quality: An investigation into the concept and how it is perceived by consumers. Assen, Gorcum [Cited by, Grunert, K. G. (1997). What’s in a steak? A cross-cultural study on the quality perception of beef. Food Quality and Preference, 8, 157–174]. Turner, M., & Collison, R. (1988). Consumer acceptance of meals and meal components. Food Quality and Preference, 1, 21–24. Varey, C., & Kahneman, D. (1992). Experiences extended across time: Evaluation of moments and episodes. Journal of Behavioural Decision Making, 5, 169–186. Verbeke, W., De Smet, S., Vackier, I., Van Oeckel, M. J., Warnants, N., & Van Kenhove, P. (2005). Role of intrinsic search cues in the formation of consumer preferences and choice for pork chops. Meat Science, 69, 343–354.
Contents lists available at ScienceDirect
Meat Science journal homepage: www.elsevier.com/locate/meatsci
Do pig farmers preferences bias consumer choice for pork? Response to critique of the pork preference studies T.M. Ngapo a,*, J. Fortin a, J.-F. Martin b a b
Agriculture and Agri-Food Canada, Food Research and Development Centre, 3600 Bd Casavant Ouest, Saint-Hyacinthe, Québec, Canada J2S 8E3 Unité de Maladies Métaboliques et Micronutriments, INRA, Theix, 63122 St-Genès-Champanelle, France
a r t i c l e
i n f o
Article history: Received 30 October 2009 Received in revised form 8 March 2010 Accepted 11 March 2010
Keywords: Consumers Preferences Pork Choice Colour Fat Marbling Drip
a b s t r a c t Québec consumers and pig farmers selected their preferred chop from 16 images that had been modified to give 16 treatments: two levels each of fat cover, colour, marbling and drip. The selection process was repeated eight times from different groups of chops. Fat cover (47% preferred lean) and colour (44%, light red) were the most frequently chosen characteristics. No significant differences were observed between farmers and consumers preferences (v2 test, P < 0.05). Two preference-based clusters were found; 41% preferring dark red, lean meat and 59%, light red, lean meat, without marbling or drip. Choice-based clusters showed no significant links with either individual socio-demographic items, including pig farmer as occupation, or the three socio-demographic-based clusters observed (v2 test, P < 0.05). No evidence was found to suggest that the choices of pig farmers differed from those of consumers and, therefore, inclusion of pig farmers in consumer panels would not bias consumer choice for pork. Crown Copyright Ó 2010 Published by Elsevier Ltd. All rights reserved.
1. Introduction Using photographic images varying systematically in appearance, a programme was started in France to identify the most important characteristics of fresh pork which determine consumer choice and to show how consumer segmentation in choice relates to socio-demographic differences (Ngapo, Martin, & Dransfield, 2004). This study was undertaken in 23 countries (Chen, Guo, Tseng, Roan, & Ngapo, 2010; Cho et al., 2007; Cipolli, Silveira, Ngapo, & Dransfield, 2003; Fortomaris et al., 2006; Ngapo, Martin, & Dransfield, 2007a, 2007b; Verbeke et al., 2005) and data collection sites in many countries included agricultural shows. These shows provided a cross-section of the general public with great variation in socio-demographic attributes including consumer occupation. Regardless of this demonstrated variability, when the results of this study have been presented, comments have invariably been raised about sample bias due to high participation of pig farmers. While of no consequence to the international study, these comments evoke the question ‘‘do pig farmers really have a different preference for pork than the general public?” After all, pig farmers are merely consumers of pork, who happen to raise pigs.
* Corresponding author. Tel.: +1 450 768 3300; fax: +1 450 773 8461. E-mail address: [email protected] (T.M. Ngapo).
The entire Canadian panel in the international study comprised 248 Quebecers and 802 Albertans, from which a representative sample of 200 consumers was selected (Ngapo et al., 2007a, 2007b). An additional 50 Québec pig farmers were surveyed as a part of the Québec panel, but due to concerns of potential sample bias these responses were not used. These responses provide an opportunity to determine if pig farmers’ preferences differ from those of consumers in Québec, the aim of the present study.
2. Materials and methods 2.1. Pork chops The method and chop characteristics are described in detail by Ngapo et al. (2004). Briefly, photographs of 16 commercial pork chops were computer-modified to give two levels of each of the characteristics: fat cover, colour, marbling and drip. The resulting 256 (2 2 2 2 16) images have been published as a book (Dransfield, Martin, Miramont, & Ngapo, 2001) which can be used as a tool for analysing the importance of those factors in consumer choice. The book is comprised of six series of which series 1 + 2, 3 + 4, and 5 + 6 each contain all 256 images. A series constitutes 16 (A4) pages or eight double-pages. Every double-page contains the 16 different chop shapes and each chop represents one of the combinations of the four characteristics studied. Therefore, every
0309-1740/$ - see front matter Crown Copyright Ó 2010 Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.meatsci.2010.03.022
789
T.M. Ngapo et al. / Meat Science 85 (2010) 788–791
double-page contains a complete set of all 16 combinations of the two levels of each of the four characteristics. Both the order of representation of the characteristics with respect to the chop shape and the position of the chops in a double-page are randomised. It is important to note that the chop shape was not a factor studied. 2.2. Consumers Consumers were pork eaters older than 15 years of age and chosen at random. Consumers were surveyed in July and August of 2002 in Québec at agricultural shows and workplaces. In total 298 consumers from Québec completed the survey, including 50 pig farmers. 2.3. Selection Each consumer selected their preferred chop from each doublepage showing the 16 combinations of two options of each of the characteristics of colour, fat cover, marbling and drip. The selection was repeated from eight different double pages which showed the same 16 appearance characteristics and chops, but in different combinations. The consumers also completed a short questionnaire asking basic socio-demographic and purchase- and eating-behaviour information (Table 1). The questionnaire was an abridged version of that used in the international survey (Ngapo et al., 2007b). Each new consumer was given a series in the order 1–6 so that all series were used approximately equally throughout a survey period.
consistently selected. The results were quantified by the definition that if P6 of 8 choices for one consumer are the same for a given characteristic, the choice is a ‘real’ choice (P < 0.14). If <6 choices are the same, selection for the given characteristic is considered to be ‘inconsistent’. This test assumes a binomial distribution of the results (P = 0.5). Significant differences in the number of choices were observed using the v2 test (P < 0.05). The number of characteristics used to form consistent choices was then obtained by calculating the percentage of consumers using 0, 1, 2, 3 or all 4 characteristics. For each consumer and using all choices, the sum of the number of times each combination of characteristics was selected is calculated. Using the sums for each combination, a contingency table is constructed comprising the consumers and their choices. Using this contingency table, a correspondence analysis using all 15 dimensions was undertaken accounting for 100% of the variability. The coordinates of each consumer obtained in the 15 dimensions of the correspondence analysis were used as the basis for cluster analyses. Firstly, a hierarchical cluster analysis using Ward distance was undertaken using the SAS CLUSTER procedure (SAS, 1999). The number of clusters to be retained was selected by considering the ‘distance’ between clusters and the profile of the resulting graph. A disjoint cluster analysis was then carried out using the SAS FASTCLUS procedure (SAS, 1999) forcing the consumers into the different clusters. Links between the consumer choice-based clusters and questionnaire items were determined using v2 test. Similarly to the choice data, correspondence and cluster analyses were undertaken to define clusters based on the questionnaire items.
2.4. Data analyses and statistical methods 3. Results and discussion 2.4.1. Analysis of choices The choices consumers made were divided into three categories for each characteristic; in the first two categories the consumer actually chose one of the two levels of the given characteristic, whereas for the third category the given characteristic was not
3.1. Consumers The socio-economic questionnaire composition and responses are given in Table 1. There were proportionately more consumers
Table 1 Questionnaire composition with responses attained for the entire consumer sample, Québec consumers and pig farmers. Question
What is your age (years)? Gender? Marital status? How many people live in your household?
How often do you eat meat?
Response options
<35 years P35 years Female Male Single Married 1 2 3 4 5+ Every meal, everyday Once/day, everyday Several times/week Once/week or less
Consumers (number) Quebecers
Farmers
All
Quebecers
Farmers
84 213 189 107 87 198 27 132 45 46 34 50 116 116 14
66 181 173 73 81 154 25 116 37 33 23 38 88 107 13
18 32 16 34 6 44 2 16 8 13 11 12 28 9 1
28 71 63 36 29 66 9 44 15 15 11 17 39 38 5
27 73 70 29 32 63 10 47 15 13 9 15 35 43 5
36 64 32 68 12 88 4 32 16 26 22 24 56 18 2
6 168 72 151 93 50 20 37 55 105 190 60
3 44 3 37 13 11 2 17 19 25 39 14
3 70 25 63 36 57 21 18 25 43 76 25
2 68 29 61 38 54 22 15 22 42 77 24
6 88 6 74 26 85 15 34 38 50 78 28
The following make reference to fresh pork and not to sausages, ham or other processed pork products How often do you eat pork? Everyday 9 >Once/week 212
Consumers (%)
All
In the percentage columns, given as a percent of those who claim to have changed their frequency of consumption of pork.
790
T.M. Ngapo et al. / Meat Science 85 (2010) 788–791
Table 2 Selection of visual chop characteristics by Quebec consumers and pig farmers according to consistency of choice (P6 out of 8 choices deemed a selection for a given level of characteristic; <6 deemed inconsistent). Selection (% consumers)a
Selection (number of consumers)
a
All
Quebecers
Farmers
All
Quebecers
Farmers
Colour
Dark Light Inconsistent
64 130 104
57 103 88
7 27 16
21 44 35
23 42 35
14 54 32
Fat cover
Fat Lean Inconsistent
20 141 137
17 116 115
3 25 22
7 47 46
7 47 46
6 50 44
Marbling
Marbled Not marbled Inconsistent
31 74 193
24 64 160
7 10 33
10 25 65
10 26 65
14 20 66
Drip
Drip No drip Inconsistent
29 96 173
27 80 141
2 16 32
10 32 58
11 32 57
4 32 64
No significant differences in selection were observed between farmers and Quebecers for any given level of a characteristic by v2 test (P < 0.05).
less than 35 years of age in the group of farmers (36%) than the group of Quebecers (27%). More than two thirds of the Quebec consumers were female compared to less than a third of the farmers. A third (32%) of the Quebecers were single, whereas almost all (88%) of the farmers were married. Almost a half of the Quebec consumers lived in two-person households and about 12% in each of households of 1, 3, 4 or 5+ persons, whereas a third of the farmers lived in two-person households, about a quarter in households of 4 or 5+ persons and 16% in three-person households. Few lived in one-person households reflecting the small proportion of single farmers. Almost all respondents eat meat at least several times a week of which 80% of the farmers eat meat at least once a day compared to 50% of the Quebec consumers. Frequency of pork consumption was also higher in the farmers, with 94% eating pork more than once a week compared to 70% of the Quebecers. A quarter of the farmers claimed to have changed their pork consumption recently of which most (85%) have increased their consumption. In comparison, 38% of the Quebecers have changed their pork consumption recently, of which about half have increased their consumption. With responses from more than three quarters of both farmers and Quebec consumers, the most cited reason for liking pork is its taste. 3.2. Consumer choices The number of choices observed for all four characteristics are presented in Table 2. The more selected choices are the same for Quebecers and farmers with averages being 44% light red colour, 47% lean fat cover, 25% no marbling and 32% no drip. This analysis also shows that most of the choices for marbling (65%) and drip (58%) can be considered random based on the inconsistent selection for a given level of characteristic. While a greater proportion of the producers (54%) than consumers (42%) appeared to prefer the light red pork, none of the choices of consumers and producers were significantly different by v2 test (P < 0.05). Of the characteristics studied, the number that the consumers and producers used to make their selection, based on the ‘real’ and ‘inconsistent’ choices above, was similar for farmers and consumers with, on average, 19% using one characteristic, 41% using two and 23% using three. Only 7% used all four characteristics, and another 7% of consumers did not use any. It is notable that twice as many producers (14%) as consumers (7%) did not use any of the characteristics studied to make their choices. A 15-dimensional correspondence analysis projecting the consumers and producers depending on their choices of the 16 different combinations showed no consensus for any particular
Table 3 Percent of consumers selection within chop characteristics for each cluster with significant differences (shown in bold) as determined by v2 (P < 0.05) compared to results of the entire panel. Clusters (% consumers) 1
All (% consumers) 2
Colour
Dark red Light red Inconsistent
53 0 47
0 73 27
21 44 35
Fat cover
Fat Lean Inconsistent
11 36 53
4 55 41
7 47 46
Marbling
Marbled Not marbled Inconsistent
19 16 65
5 31 64
10 25 65
Drip
Drip No drip Inconsistent
12 29 59
8 35 58
10 32 58
41 121
59 177
Consumers in cluster (%) Consumers in cluster (number)
combination of the four characteristics studied. The first dimension, accounting for 16.7% of the total variability, corresponded to colour, the second, accounting for 10.5% of the total variability, to fat cover and the third accounting for 8.4% of variability, to drip. 3.3. Choice-based clusters Cluster analyses were undertaken to group respondents with similar strategies for pork chop selection (Table 3). The clusters can be defined as follows: Cluster 1 of 121 consumers (41%): preferred dark red, lean meat. Cluster 2 of 177 consumers (59%): preferred light red, lean meat, without marbling and without drip. The strongest preferences differentiating the two clusters were those of colour with 53% of respondents in cluster one preferring dark red meat and 73% in cluster two preferring light red meat. The distribution of the individual questionnaire items over the two choice-based clusters of farmers and consumers combined was tested by v2 (P < 0.05) and no differences in distributions between the clusters were observed.
T.M. Ngapo et al. / Meat Science 85 (2010) 788–791
Correspondence and cluster analyses were undertaken to derive clusters based on the socio-demographic and purchase- and eating-behaviour items, items from the questionnaire. The hierarchical cluster analysis showed three different clusters. These clusters would be of interest if a significant link was observed with the choice-based clusters. However, no significant link was observed between the choice-based and questionnaire-based clusters when assessed by v2 test (P < 0.05) and therefore the questionnaire-based clusters were not further investigated. 3.4. Occupation impact Studies demonstrating a significant effect of consumer occupation on food preferences are few. One might question as to how a consumer’s occupation could influence food preferences, yet on many occasions when presenting consumer preference data obtained at agricultural shows, comments from the audience have been raised about sample bias due to high participation of pig farmers. Questioning these scientists, economists and marketers as to the reasoning behind the supposition that pig farmers might have different preferences for pork chops than other consumers, generally evoked responses of the belief of an acquired superior knowledge of meat quality as a result of producing the animals from which the meat is attained. While this logic is questionable, it is not unreasonable to think that pig farmers might consume more pork than the average consumer, which, through greater experience with the product, might in turn influence their preferences. Many models exist to predict consumer behaviour of which experience plays an important role. For example, Steenkamp (1989) and Anderson (1994) presented frameworks analysing the quality perception process for food products based on the assumption that qualities sought in a food product by a consumer are always experience and credence qualities. Grunert (1997) expanded this model to include the determinants of experience as opposed to expected quality and applied this model to meat. While there are numerous such models, few studies on the impact of experience on consumer choice of foods are reported, but rather studies tend to focus on the ‘‘before purchase” part of the models excluding experience, or describe the impact of experience at consumption. This is no surprise considering the complexity of the relationship between experience and expectation (Frederickson & Kahneman, 1993; Mela, 2001; Varey & Kahneman, 1992) and the difficulty in measuring the former which is a consequence of intensity and duration of experience, and is subject to distortions in memory (Rozin & Tuorila, 1993). One study that demonstrates the influence of experience with a product on the perception of that product is that of Turner and Collison (1988) who undertook 285 hedonic acceptance evaluations of whole meals and their components. Results showed that customers who were either catering staff or catering students, and therefore directly involved in the food industry, scored lower than other customers. In this study, the occupation of the consumers directly impacted on experience with the food product studied which in turn impacted on preference. In the present study, 94% of pig farmers claimed to eat pork more than once a week demonstrating that this occupational group has greater experience with the product than the Quebec consumers (70% claimed the same consumption frequency). However, no differences in visual preferences of pork chops were observed. Of the four characteristics studied, the most important in both Quebec consumers and farmers choices of pork chops were fat cover (with 47% of the entire panel preferring lean fat cover) and colour (44% preferring light red meat).
791
4. Conclusions Two clusters of preferences for pork chops were identified from a survey of Quebec consumers that included 50 farmers. These clusters showed no significant links with either individual sociodemographic items of the consumers, including farmer as occupation, or the three socio-demographic-based clusters identified on the whole sample. The most important characteristics in Quebec consumers and farmers choice of pork chops were fat cover and colour. Using this novel choice system of within-consumer preference replication and image manipulation, no evidence was found to suggest that pig farmer’s choices of pork differ from that of consumers in Quebec and inclusion of pig farmers in the consumer panel would not bias consumer choice for pork. Acknowledgements The authors would like to acknowledge Égide, France and the Ministère de Relations Internationale (MRI), Québec, Canada for the travel grant accorded to Mr. J.-F. Martin. The authors also thank Dr. E. Dransfield for his invaluable role in this project and without whom the study would not exist. References Anderson, E. S. (1994). The evolution of credence goods: A transaction approach to product specification and quality control. MAPP working paper no. 21. The Aarhus school of business, Aarhus [Grunert, K.G. (1997). What’s in a steak? A crosscultural study on the quality perception of beef. Food Quality and Preference, 8, 157–174]. Chen, M. T., Guo, H. L., Tseng, T. F., Roan, S. W., & Ngapo, T. M. (2010). Consumer Choice of Pork Chops in Taiwan. Meat Science, 85(3), 555–559. Cho, S., Park, B., Ngapo, T., Kim, J., Dransfield, E., Hwang, I., et al. (2007). Effect of meat appearance on South Korean consumers’ choice of pork chops determined by image methodology. Journal of Sensory Studies, 22, 99–114. Cipolli, K. M. V. A., Silveira, E. T. F., Ngapo, T. M., & Dransfield, E. (2003). Consumer preferences of pork chops – The Brazilian survey. In Proceedings 5th Pangborn sensory science symposium: A sensory revolution (p. 187), 20–24 July 2003, Boston, USA. Dransfield, E., Martin, J.-F., Miramont, J., & Ngapo, T. M. (2001). Meat appearance: Pork chops. A tool for surveying consumer preferences. INRA, France, ISBN: 2-73800976-X. Fortomaris, P., Arsenos, G., Georgiadis, M., Banos, G., Stamataris, C., & Zygoyiannis, D. (2006). Effect of meat appearance on consumer preferences for pork chops in Greece and Cyprus. Meat Science, 72, 688–696. Frederickson, B. L., & Kahneman, D. (1993). Duration neglect in retrospective evaluation of affective episodes. Journal of Personality and Social Psychology, 65, 45–55. Grunert, K. G. (1997). What’s in a steak? A cross-cultural study on the quality perception of beef. Food Quality and Preference, 8, 157–174. Mela, D. J. (2001). Why do we like what we like? Journal of the Science of Food and Agriculture, 81, 10–16. Ngapo, T. M., Martin, J.-F., & Dransfield, E. (2004). Consumer choices of pork chops: Results from three panels in France. Food Quality and Preference, 15(4), 349–359. Ngapo, T. M., Martin, J.-F., & Dransfield, E. (2007a). International preferences for pork appearance. I: Consumer choices. Food Quality and Preference, 18(1), 26–36. Ngapo, T. M., Martin, J.-F., & Dransfield, E. (2007b). International preferences for pork appearance. II: Factors influencing consumer choice. Food Quality and Preference, 18(1), 139–151. Rozin, P., & Tuorila, H. (1993). Simultaneous and temporal contextual influences on food acceptance. Food Quality and Preference, 4, 11–20. SAS (1999). SAS Users Guide: Statistics. Version 8.1. Cary: SAS Institute Inc. Steenkamp, J.-B. E. M. (1989). Product quality: An investigation into the concept and how it is perceived by consumers. Assen, Gorcum [Cited by, Grunert, K. G. (1997). What’s in a steak? A cross-cultural study on the quality perception of beef. Food Quality and Preference, 8, 157–174]. Turner, M., & Collison, R. (1988). Consumer acceptance of meals and meal components. Food Quality and Preference, 1, 21–24. Varey, C., & Kahneman, D. (1992). Experiences extended across time: Evaluation of moments and episodes. Journal of Behavioural Decision Making, 5, 169–186. Verbeke, W., De Smet, S., Vackier, I., Van Oeckel, M. J., Warnants, N., & Van Kenhove, P. (2005). Role of intrinsic search cues in the formation of consumer preferences and choice for pork chops. Meat Science, 69, 343–354.