Faod Quuality and Pwfmu 5 (1994) 173-178 0 1994 Elsetier Science Limited Printed in Great Britain. Al1 rights reserved 09503293/94/$7.00
ELSEVIER
SENSORYEVALUATIONOF'FREERANGE'AND REGULARPORKMEATUNDERDIFFERENTCONDITIONS OFEXPERIENCEANDAWARENESS Peter A. M. Oude Ophuis Department of Marketing and Marketing Research, Wageningen Agricultural University, Hollandseweg 1, 6706 KN Wageningen, The Netherlands
(Received 22 June 1993; accqbted1 November 1993)
advertising claims do not relate to sensory properties. On the other hand, it is obvious that there are other properties as wel1 which are relevant to the consumer. Like most products the image of foods is built upon many characteristics of which sensory attributes are only a part. A considerable part of the foods on the market are fresh foods, and in most instances fresh foods are marketed without reference to a brand image. Advertising campaigns usually concentrate on regional or national promotion of the generic product (e.g. meat, milk or vegetables). When brands are used, this is generally limited to the trade Channel. As brand image does not seem to be present with fresh foods, one could hypothesize that sensory properties are relatively important attributes in the evaluation of these products. Although branding is not common practice with fresh foods, there are certain types of fresh foods that have value added, which can be compared with brand image, because additional information is necessary in the marketing process. Foods that have been produced alternatively, like organic, biodynamic, ecological and ‘free range : are examples of these types of food (Oude Ophuis, 1985). Al1 these products have in common that their unique selling proposition, whether ‘organic’ or ‘free range’, is not directly visible to the consumer. Only additional information wil1 identify the nature of the origin, or the production process, of these foods to the consumer. Another resemblance between these products is their appeal to highly motivated consumers. Since it is presently not easy to acquire these products accidentally, only those who have positive attitudes towards these products wil1 buy them. It can be hypothesized that these positive attitudes may bias consumers’ judgements of the validity of the undocumented claims about superior sensory quality of these products. Product awareness and product experience could thus be factors affecting sensory evaluation of this type of food. In this paper an experiment wil1 be presented where ‘free range’ pork meat was chosen as a vehicle to explore the effects of nonsensory factors on sensory
ABSTRACT An experàment is presented where %_ee range’ pork meat was chosen as a vehicle to explore the effects of nonA sample of 172
sensory factors on sensory maluation.
consumers participated in experimental tasting sessions. Two between-person factors,
prior
labelling, and one within+son were experimentally manipulated. found
experience
and
factor; meat type, Prior experience was
to have significant effects on the evaluation of the
sensory attrìbutes savoury, dry and pleasant on& NO signa$cant
main
factor Howevq
effects were found
fuf
the labelling
many interactions for both labelling and
experience with meat type were found significant, as wel1 as their second order interactions between al1 threefactors. Labelling and prior ex@rience with the product do influence sensory evaluation of pee range’pork favourably for a number of attributes. At least fm the group of consumers that had prior expa’ence
with sfree range’pork, a
strongpositive image may be hypothesized. Kqwords: labelling; image; repeated measures analysis of variante; interactions; free range’ meat.
INTRODUCTION Taste properties are considered to be important attributes of food products. More generally it can be said that not only taste, but also related attributes like texture, colour and smell, together form sensory attributes that characterize foods as unique. Being identifiable product attributes, sensory properties can be successfully employed in marketing strategies for food products. Analysis of food advertising showed that about half of al1 advertising claims are taste oriented (Lord et aZ., 1987). Because al1 foods have taste properties, one can wonder why a higher percentage of food 173
174
P. A. M. Oude Ophuis
evaluation. ‘Free range’ pork is meat that is produced under circumstances which are considered more ‘animal friendly’ than regular production. Pigs have more space per animal in a free range, as compared to regular production where efficiency is the key word and animals are grown in tight boxes. Apart from these circumstances there muy be different treatment concerning feed and veterinary medicine, but this is no automatie consequente. For the present study it is not important if a piti physical differences exist between ‘regular’ and ‘free range’ pork, and consequently in sensory evaluations of the two meats. Rather we are interested if there are changes in evaluation due to information and experience.
INTERACTION BETWEEN COGNITIVE AND SENSORY INFORMATION From psychophysics it is known that the relation behveen physical stimuli and psychological sensations is not a simple one-to-one link, but a complex function that can be affected by external factors. In the marketing environment these extemal factors play a dominant role, and it is, therefore, important to have insight into the mechanisms that can operate in this process. One of the first systematic investigations into the effect of external factors on taste perception was the classic beer brand study of Allison & Uhl (1964). Consumers were generally not able to discern the taste and flavour differences among the various beer brands, but apparently labels and their associations did influence their evaluations. Most scientific research into sensory perception and evaluation of products is performed under conditions where the identity of products is concealed from the subjects (Moskowitz, 1985). With the specific purpose of concentrating on sensory properties, experiments are performed ‘blind’ or even ‘double blind’. In both instances, the identity of the products (e.g. the brand) is not shown to the subjects, and in the last case even the researchers do not know the identity of the products during the experiment. The aim of this procedure is to prevent the sensory information from being confounded by nonsensory information. Perception and preferente based on this confounding information would be distorted. It seems that beliefs and attitudes related to the pro duet can instigate this confounding effect (Moskowitz, 1984). For that reason, sensory research that explicitly announces product identity to subjects should take into account the possible biases that can exist for named products. ‘Free range’ pork is, therefore, a product par excA%nceto study the effects of awareness, because a lot of attitudinal bias can be hypothesized for this type of product.
Previous sensory research in a comparable product category (‘organic’ fresh foods) demonstrated a ‘label’ effect for the word ‘organic’ (Schutz & Lorenz, 1976). Consumer preferences for products labelled organic rated higher than for products labelled commercial. However, in another study, where carrots from different origins (‘organic’ versus ‘regular’) were compared, no significant effect of labelling on sensory evaluation could be demonstrated (Oude Ophuis, 1988). In a study of the effects of brand labelling on ratings of product quality of mineral waters there was an interaction between labelling and product (Nevid, 1981). Neither brand nor labelling as main effect was significant, but the interaction of labelling with brand was significant, thus demonstrating the hypothesized effect that extrinsic cues play a dominant role in quality perception. In another beverage study (Sheen & Drayton, 1988) there was a significant influence of brand label on sensory perception of cola drinks. The reported effect was, in fact, an interaction between label and product. Shepherd et al. (1992) referred to several studies and a review of the area of effects of expectation on hedonic and sensory responses. Their own study of the effects of information on sensory ratings and preferences clearly demonstrated a mediating effect of attitudes. When information was given, both liking and rated likelihood of buying increased for the type of sample towards which subjects had a more generally positive attitude. The objectives of the present study are to re-examine results that are produced from the earlier mentioned studies, i.e. the influence of Iabelling and prior experience on sensory evaluation in this specific product category of ‘free range’ meat. Specific research questions of the present study are: -Does prior experience, and consequently a positive attitude towards ‘free range’ pork, influence sensory evaluation of samples of pork meat? -Does product information (labelling) lead to different sensory evaluations of samples of pork meat? -Are there any differences in sensory evaluation of ‘free range’ pork in comparison with ‘regular’ pork, regardless of effects of prior experience or labelling? Of course, these questions also imply any interactions of these three factors that may exert influence on sensory evaluation of samples of pork meat.
EXPERIMENTAL
METHOD
Design Since it was hypothesized that sensory differences between ‘regular’ and ‘free range’ meat, if any, would be
Sensgr Evaluation of ‘Free Range ’Pork
175
very small, MEAT TYPE was designed as a within-subjectsfactor, i.e. each subject was offered samples of both types of meat for sensory evaluation. With the appro priate statistical analysis this type of repeated measures design is more sensitive to smal1 differences than a full factorial between-subjects design (Tabachnick 8c Fidell, 1989). In combination with crossing the two grouping factors EXPERIENCE and LABEL, this resulted in a three-factor experiment with repeated measuresCase 11 (Winer, 1971). Combination of each of the two levels of EXPERIENCE and LABEL rendered four groups that were al1 observed under both levels of MEAT TYPE.
stimuli, because these parts have a constant, low fat percentage. Additional advantages for sensory testing were the short preparation time and easy partitioning. Since it was not possible to manipulate experimentally the ‘free range’ effect from initial production on, samples were ordered from six different butchers, three for each meat type. This was done to limit the effects of variation due to a range of uncontrollable factors including different feeding regimes and breeds. Al1 samples were prepared under standardized conditions and cooked in unsalted margarine.
Subjects
Al1 subjects attended one of 28 testing sessions, which were organized in the Nutrition and Dietetics College in Nijmegen, The Netherlands. A maximum of eight subjects attended each session. Subjects of both EXPERIENCE conditions were equally and randomly assigned to one of both LABEL conditions. If al1 eight subjects showed up for a session, the two grouping factors were completely balanced with two subjects in each cell. After written and oral instruction, subjects were offered a sample of cooked pork cutlet and asked to cut a piece and taste it while chewing, but not swallowing. After tasting, the sample could be spit out into a spittoon, and the mouth could be washed with water. Bread was available for neutralizing. During tasting, subjects were required to evaluate the sensory characteristics of that particular stimulus on a series of graphical rating scales. The unstructured line scales had a length of 100 mm, and subjects were requested to indicate the perceived intensity of each of 12 attributes on this scale by putting a mark at the position which best reflected their personal perception. The attributes used in this study were derived from a focus group discussion which was held in the preliminary stage of the research project. The following attributes were used (original Dutch equivalents in parentheses) : stringy (draderig), juicy (sappig), bland yla,,), tough (taai), firm (stevig), savoury (hatiig) , sweet (zoet), tender (mals), salt (zout), fat (vet), dry (droog) and pleasant (aangenaam). Scores were measured in millimetres, which served as the basis for the analysis. Subjects were allowed 4 min for each of the six samples to taste and make the sensory evaluation.
In the experimental design, subjects were classified as belonging to either of two groups: prior experience with ‘free range’ meat, or no prior experience with ‘free range’ meat. Actual buying behaviour at a local butcher which only sold ‘free range’ meat was used as an external criterion for selecting the subjects for the group which had prior experience with ‘free range’ meat. Subjects were recruited from this butcher, partly through leaflets with application forms, partly through direct requests for cooperation. Al1 subjects were requested to answer some questions about their actual consumption of ‘free range’ pork, and only those who indicated that they consumed ‘free range’ meat at least once a week were invited to participate in the experiment. Subjects for the group with no prior experience were recruited at three ‘regular’ butchers’ shops, and they were asked to fill out the same application form. Al1 subjects were informed that the sensory testing was about differences between ‘free range’ and ‘regular’ pork, but they did not know any other details about the experiment in advance. Initially, 224 consumers agreed to voluntary participation in the sensory testing, 112 in each group. Al1 subjects were personally invited by a letter, about one week in advance, which contained details about date, time and place, as wel1 as some instructions for conduct before and during the sensory session (Amerine et al., 1965; Jellinek, 1981). Subjects were not paid for their cooperation, and this may have influenced the actual response. Only 172 persons actually showed up at the tasting sessions, 76 in the group with prior experience (41 females and 35 males) and 96 in the group with no prior experience (52 females and 44 males). Ages varied between 16 and 77 years with a mean for the group with experience of 33.9 years and the group with no experience of 29.6 years.
Stimuli Pork meat ‘Schnitzel’)
cutlets of the upper rump (class EAA were chosen to serve as experimental
Sessions and procedure
Analysis Al1 individual attribute ratings of the six meat samples were averaged into two individual scores for the two levels of MEAT TYPE. These average attribute scores were subjected to repeated measures analysis of variante with use of the SPSS MANOVA procedure (Norubis & SPSS, 1992). Univariate results were used because al1 factors were considered to have two levels
176
p. A. M. Ou& @huis
TABLE 1. FRatios from Repeated Measures Analysis of Variante Performed on Each Attribute
Attribute
Betweensubjects effeets and interaction EXPERIENCE
LABEL
EXPERIENCE by LABEL
Stringy
2.25
0.04
0.04
Juicy Bland Tough Firm Savoury Sweet Tender Salt
3.85 3.76 0.00 0.26 7.26** 0.58 0.02 0.39
2.58 2.36 1.22 2.04 0.25 0.12 0.42 0.00
0.44 0.06 0.01 1.90 0.34 0.62 0.17 0.02
Witbin-subjects effect and interaetions MEATTW’E
EXPERIENCE byMEATTYPE
LABEL byMEATTYPE
EXPERIENCE by LABEL byMEATTYPE
0.17 0.26 3.03 0.11 4.91* 5.03 0.05 3.87* a 1.25
9.36**
3.02 6.26*” 3*99* a 5*79* 1.98 4*79* 0.04 14.27***’ 0.00
10*93**
1.06 4*57* 3*99* 1.46 2.01 4.93* 3.35 k3.14** 4.92*
0.73 7*5a** lz
4.03*” 16.15***”
9.00** 17.26***”
1.45 1.11 7.45** a 4*89* a 4*71*” 13.72***” 0.93 11.60**” 2.31
Fat W
0.14 5*54*
1.87 2.94
0.43 0.06
29*2a*** a 4*75*”
Pleasant
8.58**
0.87
0.83
25.38***”
3.16
**p < 0.01; ***p < 0.001. *p < 0.05; “Nat significant against a higher order interaction effect. only, and this approach is more powerful than the multivariate approach when the symmetry conditions are met. The hypothesis that the variante-covariance matrices are equal across al1 levels of the betweensubjects factors can be examined using the multivariate generalization of Box’s M test. Only for the attribute ‘tender’ Box’s M was found significant (p = 0.036), implying that the F ratios from the average univariate results may not be correct. However, because of the very low significante levels for this attribute, as wil1 be shown in the next section, this was not considered a serious problem. Because the within-subjects-factor had only two levels in this design, the test for homogeneity of variances is irrelevant (LaTour 8c Miniard, 1983).
RESULTS Results of the repeated measures analysis of variante are shown in Table 1, which gives an overview of the statistical significante of the experimental factors and their interactions for each attribute separately.
As can be seen from Table 1, F ratios for the betweensubject+effect main LABEL were not significant for any attribute.
Interaction effects It can be seen in Table 1 that significant interactions al1 involved the within-subjects-factorMEAT TYPE. The interaction between the between-sdjects-factors EXPERIENCE and LABEL was not significant for any attribute. If, however, the within-subjects-jktor MEAT TYPE is taken into account, EXPERIENCE and LABEL show significant interactions with this factor, both alone as wel1 as in combination, on a number of attributes. Table 2 provides more insight into the nature of these interactions, because it shows the mean differences between the attribute scores for ‘free range’ and ‘regular’ pork for each of the four cells of the experimental design.
StrIngy * juiiy
Main effects The between-subjects-e$ëctEXPERIENCE demonstrated significant F ratios (p < 0.05) for the sensory attributes ‘savoury’, ‘dry’ and ‘pleasant’ only. Except for these three attributes there was no significant differente in average attribute scores between the groups with and without prior experience with ‘free range’ pork. The direction and magnitude of the EXPERIENCE effect on the various attributes are shown in Fig. 1. On average, over al1 meat samples, the group with prior experience with ‘free range’ pork had higher attribute scores on ‘savoury’ and ‘pleasant’ and a lower score on ‘dry’.
bland
prjor expewnce
tough
“0 prior exper1ence
firm * savoury sweet tender * SlgnlfIcant differente CP < 0 05)
Salt fat * dry *
pleamlt 3
20 40 scores I” mm on graphlc
60 ratlng
80
100
SC&
FIG. 1. Main effect of EXPERIENCE. Average attribute scores (over both meat types) for the groups with (N= 78) and without (N= 94) prior experience of ‘free range’.
Evaluation of ‘FreeRange'Pods
Smq
17’7
TABLE 4. Contrast of MEAT TYPE through Mean Paired Differences between Attribute Scores (Free Bange minus Regular Pork Meat, in mm, Standard Errors in Parentheses) for Each Cel1 of the Experimental Design
Labelled (N= 49)
Ulllabekd (N= 45) Stringy Juicy Bland Tough Firm Savoury Sweet Tender Salt Fat Dry Pleasant
Prior experience
NO prior experience
Attribute
-1.3 1.8 -1.4 -2.5 -4.0 1.9 -3.3 0.8 3.2 -1.0 -3.7 3.1
(3.1) (3.5) (28) (38) (2.7) (2.6) (2.6) (3.7) (2.3) (2.4) (3.3) (3.1)
-3.2 (2.2) 2.9 (2.5) -1.4 (2.5) -6.8 (3.1) 3.9 (2.6) 1.8 (2.0) 1.5 (1.9) 3.6 (2.3) -2.2 (2.2) -4.7 *(2.4) -0.5 (2.5) 2.7 (1.8)
Unlabelled (N= 43)
2.7
(2.7)
-6.2 *(3.1) -0.7 (2.3) 3.2 (3.9) 6.1 *(2.1) 2.0 (2.4) 0.5 (2.0) -1.9 (3.0) 0.5 (2.6) -4.1 *(l.S) 3.1 (3.2) -0.5 (2.6)
Labelkd (N= 35)
-5.0 7.9 -12.3 -10.0 6.1 13.2 -3.4 18.6 5.8 -16.3 -12.8 20.4
(3.2) *(3.0) *(4.1) *(3.6) (3.8) *(3.2) (3.2) *(3.2) *(2.4) *(2.9) *(3.7) *(2.7)
Significante of paired t-tests: *p < 0.05.
The most dramatic differences between ‘free range’ and ‘regular’ pork occur in the group which has prior experience with ‘free range’ pork and has made the sensory evaluation under the labelled condition. This group perceived ‘free range’ pork in comparison with ‘regular’ pork significantly as more juicy, less bland and tough, more savoury and tender, less fat and dry and more pleasant. The effect of the second-order interactions of EXPERIENCE by LABEL by MEAT TYPE is clearly manifest in this group. The significant First-order interactions between EXPERIENCE and MEAT TYPE for firmness, and fattiness are remarkable, since this implies that the group of consumers with prior experience with ‘free range’ pork evaluated ‘free range’ pork on these characteristics differently from ‘regular’ pork, regardless of the effect of LABEL. In practice, Table 2 indicates that the group with no prior experience differed between the two Label conditions whereas the group with prior experience did not.
DISCUSSION It should be noted explicitly that this sensory evaluation pertains only to the samples that were actually used in this experiment. As mentioned before, al1 kinds of circumstantial breeding and feeding factors may influence the sensory quality of pork. These were not under control in this experiment as the average comparison between ‘free range’ pork and ‘regular’ pork was not the primary interest of this experiment. Of course, these results indicate something about the sensory quality of at least some of the ‘free range’ pork that is currently available to consumers in The Netherlands, but for the moment only the relative position of the two product types is important. In this overall comparison between ‘free range’ pork and ‘regular’ pork it
seems that the position of ‘free range’ pork on the sen-
sory continuum is slightly better than that of ‘regular’ pork. At least one combination of attributes suggests a strong attitudinal influence on the sensory judgments. The sensory attributes fat and tender are usually correlated in a technical sense. The more tender meat is, the more fat it wil1 contain. However, where tenderness is considered as a favourable attribute, fattiness is usually considered as a negative attribute. This was clearly demonstrated by the differences in sensory evaluation of both meat types by the group of consumers with prior experience of ‘free range’ pork which was tasted under labelled conditions (Table 2). ‘Free range’ pork, when it is identifiable as such, is considered much more tender and pleasant, but much less fat and dry than ‘regular’ pork.
CONCLUSIONS The present results suggest that labelling and product experience can have significant influence on the sensory evaluation of ‘regular’ pork meat in comparison with ‘free range’ pork. However, the exerted influence was highly context-specific. The effect of labelling in isolation was not significant for any sensory attribute. Thus, the simple fact that consumers were aware of what they were evaluating was not affecting their sensory evaluation. Also, the effect of prior experience with one of the evaluated products was only significant for a few sensory attributes. The most significant changes in sensory evaluation were caused by interactions between both factors, experience and labelling, with the type of meat. Labelling and prior experience with the product did have a favourable influence on the sensory evaluation
178
P. A. M. Oude Ophuis
of ‘free range’ pork for a number
of attributes.
This
of a beer brand study, where the perception of quality was affected by brand image, especially for brands with strong positive images (Jacoby et al., 1971). At least for the group of consumers that had prior experience with ‘free range’ pork, a strong positive image may be hypothesized. These results give support to such a hypothesis and are also in line with the findings of Shepherd et aZ. (1992), which demonstrated differentiation in sensory ratings under the influence of positive attitudes. Finally it can be speculated that contextual elements may be very important in the sensory evaluation of fresh foods. Although for this product category brands are not very common yet, it is clear that a product attribute like ‘free range’ can have the same (or even stronger) impact. effect
was concordant
with
results
ACKNOWLEDGEMENTS The
author
an earlier reviewers
acknowledges version
and is grateful
van Veldhoven,
several useful comments
of this paper
Mirjam
from
on
two anonymous
for the assistance
of Mylène
Bon and Anita te Nijenhuis.
REFERENCES Allison, R. 1. & Uhl, R. P. (1964). Influence of beer brand identification on taste percepti0n.J. MurketingRes., 1,36-9. Amerine, M. A., Pangborn, R. M. & Roessler, E. B. (1965). Pnnciples of Sensory Evaluation of Food, Academie Press, New York. Jacoby, J., Olson, J. C. & Haddock, R. A. (1971). Price, brand name, and product composition characteristics as determinants of perceived qua1ity.J. Appl. Psychol., 55,570-9. Jellinek, G. (1981). Sensorische Lebensmitte@ii@zg: Lehrbuch ftir die Praxis, Siegried, Pattensen.
LaTour, S. A. & Miniard, P. W. (1983). The misuse of repeated measures analysis in marketing research. j Marketing Res., 20,45-57. Lord, J. B., Eastlack, J. 0. & Stanton, J. L. (1987). Health claims in food advertising: is there a bandwagon effect?. J. AdvertisingRes., 27(Z), 9-15. Moskowitz, H. R. (1984). Relative importante of sensory factors to acceptance: theoretical and empirical analyses. J. Food QuaL, 7,75-90. Moskowitz, H. R. (1985). New Directions for Product Testing and Senso7yAnalysis of Foods. Food & Nutrition Press, Westport, CT, pp. 53-73,343-66. Nevid, J. S. (1981). Effects of brand labeling on ratings of product quality. Perceptual and Motor Ski& 53, 407-10. NoruSis, M. J. & SPSS (1992). SPSS for Windows Advanced Stutistics Release 5, SPSS Inc., Chicago, 11~. Oude Ophuis, P. A. M. (1985). Consumption of alternatively produced foods. In Consuw Behaviour Research and Marketing of Agricultural Products, ed. J. E. R. Frijters. Proceedings of the Agrofood workshop organized by the Commission of the European Communities, 13-16 Nov. 1985 in Wageningen, The Netherlands. National Council for Agricultural Research, The Hague, pp. 211-26. Oude Ophuis, P. A. M. (1988). Is sensory evaluation of alternatively produced foods affected by cognitive information and product familiarity? In Food Acceptability, ed. D. M. H. Thomson. Elsevier Applied Science, London, pp. lol13. Schutz, H. G. & Lorenz, 0. A. (1976). Consumer preferences for vegetables grown under ‘commercial’ and ‘organic’ conditi0ns.J. Food Sb., 41(l), 70-3. Sheen, M. R. & Drayton, J. L. (1988). Influence of brand label on sensory perception. In Food Acceptability, ed. D. M. H. Thomson. Elsevier Applied Science, London, pp. 89-99. Shepherd, R., Sparks, P., Bellier, S. & Raats, M. M. (1992). The effects of information on sensory ratings and preferences: The importante of attitudes. Food @al. and Preferact?, 3, 147-55.
Tabachnick,
B. G. & Fidell, L. S. (1989).
Using Multivatiate
Statistics. Harper Collins, New York, pp. 43-7.
Winer, B. J. (1971). Statistical Principles in Experimental Design. McGraw-Hill, New York.