Influence of poles on results from reference-based sensory characterization methodologies: Case study with polarized projective mapping consumers
Luis de Saldamando1, Lucía Antúnez1, Ana Giménez1, Paula Varela2, Gastón Ares1*
1 Departamento de Ciencia y Tecnología de Alimentos, Facultad de Química, Universidad de la República. General Flores 2124. CP 11800. Montevideo, Uruguay.
2 Nofima AS, P.O. Box 210, 1431 Ås, Norway
*Corresponding author: Gastón Ares [[email protected]; Telephone: +59829248003; Fax:
+59829241906]
1
2
3
4
5
6
7
8
9
10
11
12
ABSTRACT
One of the main steps related to the implementation of reference-based sensory characterization methodologies is the selection of the references, sometimes referred to as poles. However, research on the influence of the reference selection on the results from these methodologies are scarce. In this context, the aim of the present work was to evaluate the influence of the poles on results from Polarized projective mapping (PPM). Three groups of consumers evaluated two sets of six samples of powdered drinks using three different sets of poles. Results showed that the set of poles did not largely affect sample configurations and consumers' descriptions. However, some differences in sample configurations were identified when the poles did not represent the whole sensory space represented by the sample sets.
PRACTICAL APPLICATIONS
Results of the present work bring light to researchers on how to select poles for a PPM or PSP task. Given the fact that conclusions from PPM with different sets of poles did not largely differ, pole selection seems to be quite flexible. However, it seems advisable to select poles that represent the whole sensory space defined by the samples to be evaluated and that have an intermediate degree of difference among them. Also, more research applications on different categories of samples would be advised, as well as further research on the influence of the poles on results from other approaches for measuring the degree of difference between samples and poles.
Keywords: sensory characterization; PSP; polarized sensory positioning; projective mapping
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
INTRODUCTION
Interest in sensory characterization based on the evaluation of global similarities and differences among samples has increased in the last decade (Ares and Varela 2014). This type of characterization encourages the generation of a synthetic representation of the products, which is inhibited when assessors are asked to focus their attention on specific sensory characteristics (Prescott 1999; Small and Prescott 2005). The most popular methodologies based on the evaluation of global similarities and differences are free sorting and projective mapping, which have been already used for sensory characterization of a wide range of food products of different complexity (Varela and Ares 2012; Valentin et al.
2012).
One of the main disadvantages of free sorting and projective mapping is the fact that all samples should be simultaneously evaluated in the same session. This restricts the number of products that can be evaluated and makes it difficult to compare samples evaluated in different moments in time, which is a common issue in new product development (Ares and Varela 2014).
Reference based-methodologies are based on the comparison of samples with products that are regarded as references, or commonly referred to as poles. The main advantage of this approach is that they enable the comparison of products that are not evaluated in the same session (Valentin et al. 2012). Polarized Sensory Positioning (PSP) is the most relevant example of reference-based methodologies. PSP is based on the evaluation of the degree of difference between samples and each of the poles (Teillet et al.
2010).
Different approaches can be considered to evaluate the degree of difference between samples and the poles (Teillet 2014). Teillet et al. (2010) used unstructured scales ranging from “exactly the same taste” to “totally different taste” when evaluating the taste of mineral waters. In triadic PSP (t-PSP) assessors are asked to indicate to which of the poles a sample is more similar and to which it resembles the least (Teillet et al. 2014). Another approach is
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
PPM combines the main features of Projective mapping and Polarized sensory positioning (PSP) (Ares et al. 2013). Assessors are asked to locate a set of samples on a bidimensional space (a sheet of paper) in which 3 reference samples or poles have been previously plotted. They are asked to try the three poles and then each of the samples.
Assessors have to place the samples in the sheet, taking into account their similarities and differences with the three poles. Assessors are instructed that samples that are perceived as similar should be placed close together, while those that are perceived as different should be located far from each other. After they have located the samples on the sheet a description phase is held to identify the sensory characteristics responsible for the similarities and differences among samples.
As in any reference-based methodology, the key step in the implementation of PPM is the selection of the poles. Published studies have selected poles that represent the main sensory characteristics responsible for the expected similarities and differences among samples of interest. As an example, when studying the sensory characteristics of commercial mineral waters Teillet et al. (2010) selected the poles according to their degree of mineralization, which is the main determinant of the sensory characteristics of mineral water.
Waters with low mineralization are characterized by their metallic and bitter taste, waters with medium mineral content show neutral and fresh taste, while high mineralization provides salty taste. For this reason, the authors selected one with low mineral content, one with intermediate mineralization and a third one with high mineral content. Similarly, de Saldamando et al. (2013) selected poles with three distinct textures (liquid, mousse and cream) when working texture characterization of make-up foundations. However, research in this field is still lacking to provide general recommendations for selecting the poles. de Saldamando et al. (2013) compared results from PSP obtained using two sets of poles that represented the sensory characteristics responsible for the main differences among samples.
They reported that sample configurations were similar, although some differences in conclusions regarding similarities and differences were identified. However, no published study has been found comparing poles with different sensory characteristics.
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
The aim of the present work was to study the influence of the poles on the results from sensory characterizations obtained using Polarized Projective Mapping with consumers, and to compare sets of poles that represent different sensory characteristics.
MATERIALS AND METHODS
In the present work two sample sets of orange-flavoured powdered drinks were evaluated by three groups of consumers using Polarized projective mapping (PPM) with different sets of poles.
Samples
Sixteen samples of commercial orangeflavoured powdered drinks, all of them available in the Uruguayan market were used. A description of the samples is provided in Table 1.
Insert Table 1 around here
Twelve of the sixteen samples were divided into two sets of 6. The other four were used as poles (see PPM description for further details). According to previous studies (Ares et al. 2013; de Saldamando et al. 2013) samples included in Set 1 represented a wider range of sensory characteristics than samples included in Set 2.
Samples were prepared by dissolving the powders in tap water as recommended by the manufacturer on the package. They were stored in a fridge at 10°C until they were served to consumers, within 4 hours from preparation.
Consumers
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
A total of 132 consumers were recruited from the University campus (Universidad de la República, Montevideo, Uruguay) based on their availability and interest to participate.
Their ages ranged from 18 to 60 years old and they were 71% female. Participants signed an informed consent agreement and received a small gift for their participation.
Consumers were randomly divided into three groups of similar size (Groups I, II and III), each of which evaluated the samples using a different sets of poles for each of the sample sets. Chi-square tests indicated that differences in the gender and age frequency distributions of the groups were not significant (p>0.18).
Selection of the poles for the Polarized Sensory Positioning task
In each sample set, the poles selected for Groups I and II represented the main sensory characteristics of the samples. The poles in Groups I and II had similar sensory characteristics. Meanwhile, the three poles included in Group III were more similar among each other than the three poles used for Groups I and II. In this case, the poles represented a narrower sensory space than those in Groups I and II. A description of the poles used by each group is presented in Table 2.
Insert Table 2 around here
The set of poles were selected according to results of previous studies (Ares et al.
2013) and trained assessors' evaluations using Descriptive Analysis. The average intensity of the most relevant sensory characteristics of the three sets of poles considered for each of the sample sets is shown in Table 3.
For Set 1, the set of three poles considered for Group I was selected to represent the three main sensory characteristics: sweetness, sourness and total flavour intensity. Sample P1 represented sour drinks with high total flavour intensity, sample E was characterized by its low flavour intensity and sample F by its sweetness. The poles of Group II also represented these main characteristics and therefore included a sour pole (sample B), a
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
sweet pole with high total flavour intensity (sample C) and one with low total flavour intensity (sample P2). For Group III poles represented sour and sweet drinks (samples P1 and F, respectively) but none of the poles represented drinks with low total flavour intensity. Instead, a pole with intermediate characteristics to those of the other two poles (Sample A) was included (Table 3).
The main characteristics that determined consumers' perception of similarities and differences among samples of Set 2 have been identified in previous studies as off-flavour, total flavour intensity, sweetness and mandarin flavour. As shown in Table 4, consumers in Group I completed the PPM task using one pole with intense off-flavour (sample K), one pole with intense sweetness and mandarin flavour (sample H), and a third pole with low total flavour intensity (Sample P5). The poles in Group II had similar characteristics to those included in Group I: sample H (included in both groups), sample P4 (intermediate off-flavour intensity) and sample G (low total flavour intensity) (Table 3). Meanwhile, the poles included in Group III did not include a pole with intense sweetness and mandarin flavour. This group of poles was composed of a sample with intense off-flavour (sample K), one sample with low total flavour intensity (sample L) and one intermediate sample between these two (sample P4) (Table 3).
Insert Table 3 around here
Consumers received 100 mL of each one of the 3 poles and approximately 30 mL of each sample, which were served in plastic glasses and coded with three-digit random numbers. Consumers were asked to evaluate the poles and the samples, and to place the latter on a white sheet of paper (42cm x 30cm) in which the 3 reference samples or poles had been previously located (Figure 1). The position of the poles on the sheet of paper was determined based on results from previous studies. The position of the poles was the same for all consumers, for both Set 1 and 2.
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
Consumers were explained that they had to place the samples in the sheet according to their similarities and differences, taking into account the places where the references had been previously located. Consumers were told that samples that were placed close to each other were similar and those that were far from each other were different. They were also explained that they had to complete the study according to their own criteria and that there were no right or wrong answers. After positioning the samples consumers were asked to provide a description of the samples by writing on the same sheet.
In the same session consumers evaluated the two sample sets, separated by a 5 minutes break. Presentation order of the sets was balanced across consumers. Still mineral water was available for rinsing between samples. Testing took place in a sensory laboratory in standard sensory booths that was designed in accordance with ISO 8589 (ISO 2007), under artificial daylight and temperature control (22ºC).
Insert Figure 1 around here
Data analysis
The X and Y coordinates of the samples on the sheet were analyzed using Multiple Factor Analysis (MFA) considering each consumer as a separate group of variables (Pagès 2005) and the poles as supplementary individuals. Confidence ellipses were calculated using parametric bootstrapping (Dehlholm et al. 2012).
In each study, comparison of sample configurations from the different consumer groups were performed using the RV coefficient (Robert and Escoufier 1976). A permutation test was used to evaluate the significance of the RV coefficient (Josse et al. 2008).
For each PPM task the words elicited by consumers in the description phase were qualitatively analyzed. First, a search for recurrent terms was performed in which terms with similar meaning were grouped into different categories within each drink. This classification was performed independently by two researchers considering personal interpretation of the meaning of the words and word synonymy as determined by a Spanish dictionary. After
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
individually evaluating the data, a meeting of the researchers was undertaken in order to check the agreement between their classifications. The final categories and their names were determined by consensus between the two researchers, considering their two independent classifications in an open discussion. The frequency of mention of each category was determined by counting the number of consumers who elicited words within each category. Terms mentioned by at least 5% of the consumers were retained for further analysis.
Multiple factor analysis for contingency tables (MFACT) was applied on the frequency tables of the three consumer groups to obtain a representation of terms (Bécue-Bertaut and Pagès 2004).
All statistical analyses were performed in R language (R Core Team, 2013) using FactoMineR (Lê et al. 2008) and SensoMineR (Lê and Husson 2008).
RESULTS
Sample configurations
The set of poles used in the evaluation affected the percentage of variance explained by the first two dimensions of the MFA. As shown in Table 4, for both sample sets the lowest percentage of explained variance was obtained when the poles were closer, representing a narrower sensory space other (Group III).
Insert Table 4 around here
In general, sample configurations were not largely affected by the set of poles. As shown in Table 4, the RV coefficients between sample configurations in the first two dimensions of the MFA were equal or higher than 0.84. For Set 1, which included samples that represented a wide range of sensory characteristics, highly similar sample configurations
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
(RV=0.96) were obtained when poles with similar characteristics were considered (Groups I and II).
Differences in conclusions regarding similarities and differences among particular samples were identified when the evaluations were performed with different sets of poles. In the first set of samples (Set 1) changing the set of poles by another with similar characteristics (Group I and Group II) affected the overlapping of the confidence ellipses of samples A and B, and samples D and E (Figure 2 (a) and (b)). On the other hand, when sets poles with different characteristics were considered sample configurations differed in the relative position of sample F with respect to sample C, as well as in the overlapping of the confidence ellipses of samples A and B,and D and E (Figure 2 (a) and (c)). When both samples A and F were considered as poles (i.e. Group III), the discrimination between samples A and B increased, while sample F was regarded as similar to both samples D and E. For Groups I and II sample F was markedly different from samples D and, E.
Insert Figure 2 around here
Changing the set of poles had a larger influence on sample configurations of Set 2 than on sample configurations of Set 1. With respect to the differences in conclusions regarding similarities and differences among particular samples, considering sets of poles with similar characteristics (Group I and Group II) led to changes in the relative position of samples G and L with respect to samples J, H and I (Figures 3 (a) and (b)). Using poles with different characteristics provided less similar maps (Group III). When the poles represented the whole sensory space (Groups I and II) samples were widely distributed along the first two dimensions of the MFA. Meanwhile, when the poles were highly similar, samples were sorted into three clearly separated groups, as shown in Figure 3 (c).
Insert Figure 3 around here
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
Consumers' descriptions
Consumers' descriptive terms were grouped into categories. For both sample sets, the number of categories elicited by at least 5% of the consumers did not differ between consumer groups, suggesting that the set of poles did not influence the terms consumers used to describe the samples.
A total of 18 and 19 categories were considered for Set 1 and 2, respectively.
Consumers described the samples using terms related to specific sensory characteristics of the drinks (e.g. Sour, Sweet, Orange flavour), terms describing total flavour intensity (e.g.
Concentrated, Diluted, Tasteless) and also terms with hedonic connotation (e.g. Disgusting, Yummy). Sour, Sweet and Diluted were the most elicited categories for describing samples included in Set 1, while Diluted, Sweet and Tasteless were the most relevant categories for Set 2. It is interesting to highlight that although mandarin flavour was considered a relevant sensory attribute for describing the characteristics of the poles, this term was seldom mentioned by consumers for describing samples.
Figure 4 shows results from MFACT performed on the frequency table containing terms used by the three consumer groups to describe samples of Set 1 when different sets of poles were considered. As shown, identical terms from the three consumer groups were located closed to each other, suggesting that they were used similarly. As shown in Figure 4, terms related to the main sensory characteristics responsible for similarities and differences among the drinks (e.g. Sour, Sweet, Concentrated, Diluted, Tasteless, Off-flavour, Mandarin flavour, Orange flavour) were located close to each other in the map, suggesting that they were used in the same way by the three consumer groups who used different poles. Similar results were obtained for the hedonic terms Yummy and Disgusting. This suggests that the set of poles did not affect the way in which consumers used these terms to describe samples. On the other hand, terms with intensity connotation, such as Very sour, Not very sour, as well as terms related to sensory characteristics that did not differ among samples (e.g. Bitter), were located far from each other for the different groups, indicating that they
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
were not used in a stable way by the three consumer groups. Similar results regarding the use of terms were obtained for Set 2 (data not shown).
Insert Figure 4 around here
DISCUSSION AND CONCLUSIONS
The key step for the implementation of reference-based sensory characterization methodologies is the selection of the poles (Varela and Ares 2012; Teillet 2014). The present work studied the influence of the set of poles on results from PPM by considering poles that represented the sensory characteristics that defined consumers' perception of similarities and differences among samples to different extent, as determined in previous studies (Ares et al. 2013; de Saldamando et al. 2013).
The poles for Groups I and II were selected to represent the sensory characteristics responsible for the main similarities and differences among samples. In Set 1 this implied considering samples with distinct sweetness, sourness and total flavour intensity, whereas in Set 2 the main sensory characteristics were off-flavour, total flavour intensity, sweetness and mandarin flavour. Meanwhile, the selection of poles of Group III did not represent one of the sensory characteristics responsible for consumers' perception of similarities and differences among samples in the product category. In Set 1 this was achieved by not including poles with low total flavour intensity, while in Set 2 none of the poles in Group III had intense mandarin flavour (Table 3).
Using different sets of poles which represented the main sensory characteristics responsible for perceived similarities and differences among samples had a minor influence on sample configurations. When the set of poles represented the entire range of variability in the sensory characteristics of the sample set, i.e. the whole sensory space, samples tended to be located within the space defined by the poles (Figures 2a, 2b, 3a and 3b). Therefore, using poles with similar characteristics led to the same conclusions regarding the main
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
similarities and differences among samples. These findings mirrors those reported by de Saldamando et al. (2013) when studying the influence of poles on results from PSP.
According to that study, considering sets of poles with similar characteristics led to minor differences in sample configurations and caused small changes in conclusions regarding similarities and differences among samples. These results suggest that perceived similarities and differences between samples and the poles were defined by the main sensory characteristics of the poles and therefore considering sets of poles that differ in other sensory characteristics does not largely affect results of PPM.
Meanwhile, when the set of poles was replaced by another one which represented a smaller range of variation in sensory characteristics, sample discrimination and conclusions regarding similarities and differences among samples were slightly affected. For Set 1 (Figures 2(a) and 2(c)), sample discrimination for samples markedly different to the poles decreased (samples D and E). For Set 2, sample discrimination decreased through the consideration of a set of similar poles (Figures 2(a) and 2(c)). These results suggest that even if the set of poles does not represent the entire sensory space (i.e., the main sensory characteristics responsible for similarities and differences among samples), the methodology allows the identification of samples with widely different characteristics (i.e. samples D and E in Set 1 and Sample K in Set 2). Similar results have been reported by Teillet et al. (2010) when evaluating mineral waters using PSP. These authors reported that the methodology allowed the identification of a chlorinated water sample, despite the fact that the poles represented different degree of mineralization. However, it should be considered that the ability of the methodology to discriminate among samples that differ in sensory characteristics not represented by the poles is expected to be low.
Results from the present work suggest that the degree of difference among the poles can affect sample discrimination. In Set 1 reducing the degree of difference among the poles led to an increase in sample discrimination within the area of the sensory space that was represented by the poles (samples A, B, C and F) (c.f. Figures 2(a) and 2(c)). Considering
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
better discriminate among samples in that particular area of the sensory space because they focus on a narrower sensory space. However, it should be taken into account that if the poles are too similar to each other the difficulty of the test increases and consumers' ability to evaluate the degree of difference between the samples and the poles could decrease, also the saturation and tiredness because of sample re-tasting could play a role in this case. This effect was observed by comparing sample configurations of Set 2 obtained using Group I and Group III (c.f. Figures 2(a) and 2(c)). In summary, it seems advisable to select poles that represent the whole sensory space defined by the samples to be evaluated and that have an intermediate degree of difference among them, in order to achieve good discrimination without making the task too difficult.
In the description phase of PPM consumers did not describe the samples by comparing them to the poles; instead they provided a brief description of the sensory characteristics of each sample in absolute terms. The terms used by consumers in the descriptive phase of the PPM task were not affected by the use of different sets of poles, nor did the way in which consumers used those terms to describe samples. The terms provided by consumers using different sets of poles were located close to each other in the MFACT, suggesting that there was a good agreement in consumers' descriptions and that consumers' perception of the samples was not affected by the set of poles.
ACKNOWLEDGMENTS
The authors are indebted to Comisión Sectorial de Investigación Científica (Universidad de la República, Uruguay) for financial support, and to Agencia Nacional de Investigación e Innovación (ANII, Uruguay) for the scholarship granted to author Luis de Saldamando. Author P. Varela thanks the financial support from the Norwegian Foundation for Research Levy on Agricultural Products through the research program “Sensory strategies and consumer insight for healthy and palatable food” and FFL and the Research Council of Norway through the RapidCheck project.
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
REFERENCES
ARES, G., DE SALDAMANDO, L., VIDAL, L., ANTÚNEZ, L., GIMÉNEZ, A. & VARELA, P.
2013. Polarized Projective Mapping: Comparison with Polarized Sensory Positioning approaches. Food Qual. Prefer. 28, 510–518.
ARES, G., & VARELA, P. 2014. Comparison of Novel Methodologies for Sensory Characterization. In Novel Techniques in Sensory Characterization and Consumer Profiling (P. Varela and G. Ares, eds.). pp. 365-389. CRC Press: Boca Raton, FL.
BÉCUE-BERTAUT, M. & PAGÈS, J. 2004. A principal axes method for comparing contingency tables: MFACT. Comput. Stat. Data Anal. 45, 481–503.
DE SALDAMANDO, L., DELGADO, J., HERENCIA, P., GIMÉNEZ, A., & ARES, G. 2013.
Polarized sensory positioning: Do conclusions depend on the poles? Food Qual.
Prefer. 29, 25–32.
DEHLHOLM, C., BROCKHOFF, P. B., & BREDIE, W. L. P. 2012. Confidence ellipses: A variation based on parametric bootstrapping applicable on Multiple Factor Analysis results for rapid graphical evaluation. Food Qual. Prefer. 26, 278–280.
ISO. 2007. Sensory analysis: General guidance for the design of test rooms, ISO Standard 8589. International Organization for Standardization: Geneve, Switzerland.
ISO. 2012. Sensory analysis—General guidance for the selection, training, and monitoring of selected assessors and expert sensory assessors. ISO Standard 8586:2012.
International Organization for Standardization: Geneva, Switzerland.
JOSSE, J., PAGÉS, J. & HUSSON, F. 2008. Testing the significance of the RV coefficient.
Comput. Stat. Data Anal. 53, 82–91.
LÊ, S., JOSSE, J., & HUSSON, F. 2008. FactoMineR: An R package for multivariate analysis. J. Stat. Soft. 25, 1–18.
LÊ, S. & HUSSON, F. 2008. SensoMineR: A package for sensory data analysis. J. Sens.
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
PAGÈS, J. 2005. Collection and analysis of perceived product inter-distances using multiple factor analysis: Application to the study of 10 white wines from the Loire Valley. Food Qual. Prefer. 16, 642–649.
PRESCOTT, J. 1999. Flavour as a psychological construct: Implications for perceiving and measuring the sensory qualities of foods. Food Qual. Prefer. 10, 349–356.
R CORE TEAM. 2013. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing: Vienna.
ROBERT, P. & ESCOUFIER, Y. 1976. A unifying tool for linear multivariate statistical methods: the RV coefficient. Appl. Stat. 25, 257–265.
SMALL, D. M., & PRESCOTT, J. 2005. Odor/taste integration and the perception of flavour.
Exp. Brain Res. 166, 345–357
TEILLET, E. 2014. Polarized Sensory Positioning Methodologies. In P. Varela & G. Ares (Eds.), Novel Techniques in Sensory Characterization and Consumer Profiling, pp.
255-269. Boca Raton: CRC Press.
TEILLET, E., SCHLICH, P., URBANO, C., CORDELLE, S., & GUICHARD, E. 2010. Sensory methodologies and the taste of water. Food Qual. Prefer. 21, 967–976.
VALENTIN, D., CHOLLET, S., LELIEVRE, M., & ABDI, H. 2012. Quick and dirty but still pretty good: a review of new descriptive methods in food science. Int. J. Food Sci.
Technol. 47, 1563–1578.
VARELA, P., & ARES, G. 2012. Sensory profiling, the blurred line between sensory and consumer science. A review of novel methods for product characterization. Food Res.
Int. 48, 893–908.
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
FIGURE CAPTIONS
FIGURE 1. GRAPHICAL REPRESENTATION OF HOW THE POLES WERE LOCATED IN THE EVALUATION SHEETS.
FIGURE 2. SAMPLE REPRESENTATION ON THE FIRST AND SECOND DIMENSIONS OF MULTIPLE FACTOR ANALYSIS PERFORMED ON DATA FROM THE EVALUATION OF SET 1 BY THREE CONSUMER GROUPS USING THREE DIFFERENT SETS OF POLES:
GROUP I (A), GROUP II (B), GROUP III (C).
FIGURE 3. SAMPLE REPRESENTATION ON THE FIRST AND SECOND DIMENSIONS OF MULTIPLE FACTOR ANALYSIS PERFORMED ON DATA FROM THE EVALUATION OF SET 2 BY THREE CONSUMER GROUPS USING THREE DIFFERENT SETS OF POLES:
GROUP I (A), GROUP II (B), GROUP III (C).
FIGURE 4. REPRESENTATION OF THE TERMS USED BY CONSUMERS TO DESCRIBE THE SAMPLES, ON THE FIRST AND SECOND DIMENSIONS OF THE MULTIPLE FACTOR ANALYSIS FOR CONTINGENCY TABLES PERFORMED ON DATA FROM THE EVALUATION OF SET 1 BY THE THREE CONSUMER GROUPS WHO USED DIFFERENT SETS OF POLES. EACH TERM IS REPRESENTED BY THE BARYCENTER OF THE EVALUATIONS OF THE THREE GROUPS (CENTRAL POINT) AND THE INDIVIDUAL REPRESENTATION OF EACH OF THE GROUPS.
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
TABLES
TABLE 1. DESCRIPTION OF THE SIXTEEN ORANGE-FLAVOURED POWDERED DRINKS CONSIDERED IN THE TWO SAMPLE SETS.
Sample Main characteristics Sample set
A Contains sugar and vitamins A, C, B2, B3, B6, folic acid 1
B Without sugar 1
C Contains sugar and sweeteners 1
D Contains sugar and sweeteners 1
E Contains sugar and sweeteners 1
F Contains sugar and vitamins A, C, B2, B3, B6 and B9 1
G Contains sugar and sweeteners 2
H Contains sugar and sweeteners 2
I Contains sugar and sweeteners 2
J Contains sugar and sweeteners 2
K Contains sugar, sweeteners and vitamin C 2
L Contains sugar and sweeteners 2
P1 Without sugar 1
P2 Contains sugar and sweeteners 1
P4 Contains sugar and sweeteners 2
P5 Contains sugar and sweeteners 2
439 440 441
442
TABLE 2. DESCRIPTION OF THE POLES FOR THE THREE CONSUMER GROUPS WHO PARTICIPATED IN THE POLARIZED PROJECTIVE MAPPING STUDY.
Group Number of
consumers
Poles (**)
Description
Set 1 Set 2
Group I 44 P1, E, F P5, K, H Poles represented the main sensory
characteristics of the sample set Group II 45 B, P2, C P4, G, H Poles had similar sensory characteristics than
those in Group I
Group III 43 P1, A, F P4, K, L Poles represented a narrower part of the sensory space than those in Groups I and II
443 444
445
TABLE 3. AVERAGE INTENSITY SCORES OF THE MAIN SENSORY CHARACTERISTICS OFSAMPLES CONSIDERED AS POLES BY THE THREE GROUPS OF CONSUMERS, ACCORDING TO DESCRIPTIVE ANALYSIS PERFORMED BY A TRAINED ASSESSORS' PANEL.
Set Pole Group Sweetness Sourness Mandarin
flavour Off- flavour
Total flavour intensity
1
P1 I, III 4.3 a,b 6.5 d 0.5 a 0.3 a 6.5 c
E I 3.6 a 1.3 a 0.4 a 1.5 b 2.3 a
F I, III 6.3 d 2.8 b 0.5 a 0.6 a 6.2 c
B II 3.8 a 7.3 d 0.4 a 0.2 a 6.5 c
C II 7.3 e 2.3 b 7.3 b 0.9 a, b 7.3 d
P2 II 4.6 b 1.4 a 1.0 a 1.4 b 4.0 b
A III 5.2 c 5.6 c 0.6 a 0.7 a 6.8 c,d
2
P5 I 3.6 1.3 a 0.4 a 1.5 a 2.3 a,b
K I, III 4.2 1.0 a 1.1 a 7.6 c 1.8 a
H I, II 6.3 2.3b 4.0 b 0.9 a 5.3 d
P4 II, III 4.6 1.4 a,b 1.0 a 3.4 b 4.0 c
G II 5.6 2.0 b 0.7 a 0.7 a 3.0 b
L III 4.8 3.3 c 1.0 a 0.6 a 4.0 c
Average values within a column and sample set with different superscripts are significantly different according to Tukey's test for a 95% confidence level.
446 447 448
449
TABLE 4. PERCENTAGE OF VARIANCE EXPLAINED BY THE FIRST TWO DIMENSIONS OF THE MULTIPLE FACTOR ANALYSIS FROM POLARIZED PROJECTIVE MAPPING PERFORMED BY THREE CONSUMER GROUPS AND RV COEFFICIENTS BETWEEN SAMPLE CONFIGURATIONS OF THE GROUPS.
Sample set Set of poles (*) Percentage of variance explained by the first two
dimensions of the MFA
RV coefficient between sample configurations Group I Group II Set 1
Group I 74.9% 1 -
Group II 63.3% 0.96 *** 1
GroupIII 55.0% 0.84 *** 0.84 ***
Set 2
Group I 65.6% 1 -
Group II 60.1% 0.87 *** 1
Group III 57.0% 0.87 *** 0.87 ***
(*) The description of the set of poles used by each group is provided in Table 3.
* indicates that the RV coefficient is significant at p≤0.05, ** indicates that the RV coefficient is significant at p≤0.01, while *** indicates that the RV coefficient is significant at p≤0.001.
450 451 452
453
454 455 456
