Introduction

Among the quality properties of spaghetti, texture is one of the most important factors affecting pasta quality and consumer acceptance. Dry spaghetti must appeal to the consumer at the point of purchase, whilst cooked spaghetti must meet consumer criteria, such as smooth surface, firmness, etc. (Cole, 1991; Tang et al., 1999; Lee et al., 2002; Sissons et al., 2005). The sensory evaluation of texture is necessary in the production of spaghetti for many reasons: to describe the sensory properties of products, to improve product quality through the production process and to make comparisons with competitors’ products. Furthermore, sensory evaluation is the nearest to consumers’ estimation and therefore it allows the evaluation of overall properties of dry and cooked spaghetti. Therefore, it is considered to be the most reliable method (Latreille et al., 2005). A certain number of studies have been undertaken that involve a comprehensive sensory evaluation of spaghetti texture. Attempts to quantify texture of spaghetti have involved both oral (Voisey et al. 1978a; Kovacs et al., 1997), and non-oral (Voisey et al., 1978b) procedures including visual and tactile evaluations. Various generic descriptive analysis techniques have been used by many researches (Tang et al., 1999; Sözer and Kaya, 2003; Martinez et al., 2007; Sözer et al., 2007) to evaluate textural properties of spaghetti, such as the scoring method (D’Egidio and Nardi, 1996) or/and the method of extensive scoring system (Menger, 1985). Moreover, the method of unstructured line scales was used to record rating of each property (Malcomson, 1991). The International Organization of Standardization (ISO 7340 1985) has developed a standard method using trained panellists for evaluating the textural properties of cooked spaghetti based on the use of reference samples and nine point rating scale. Selected assessors or experts are trained to evaluate sensory properties using different scales and for that reason, they are important factors for describing textural quality of different foods (Stone and Sidel, 1998; Szczesniak, 1998; Szczesniak, 2002). Although evaluation of textural properties performed by trained panel is recognized as a key step in the assessment of spaghetti quality, the use of consumer based measurements is sometimes recommended as an equivalent to the experts (Gonzáles et al., 2001). The consistency of results obtained from different sensory groups of assessors is an issue which has been frequently addressed, particularly within the industry. Previously, many authors compared results from experts and instructed consumer panellists and some of them agreed (Moskowitz, 1996) that both groups showed similar types of results on similar properties, while others recommended the superior capability of experts (Cardello et al., 1982; Guerrero et al., 1997; Hough, 1998).

On the other hand, during the past decade, many multivariate techniques were successfully used in food research to study the interrelationship between multiple data sets and to extract the principal information from massive and complex data (Popper et al., 1988). However, non-parametric data scaled by using contingency tables have not been used to investtigate differentiation of commercial spaghetti evaluated by different panel groups. By using contingency tables for non-parametric data the objectives of this paper were to:

· determine the performance of the panel in relation to the evaluated properties,

• identify differences between the panels,

· determine whether there is a clearly defined boundary between the panels in relation to the observed properties obtained by different sensory methods.

MATERIAL AND METHODS

Samples

Nine samples of commercial wheat spaghetti were purchased in a local food store. The samples were labelled as sample 1 to sample 9. The samples 1-7 were made from durum wheat semolina, whilst samples 8-9 were made from soft wheat farina.

Sensory evaluation

Sensory evaluation was performed by different panel groups: trained and untrained. Within the trained panel there were 5 experts (three female and two male), age between 35 and 45 years. They belong to the sensory panel in charge of assuring the sensory quality control for the Institute of Food Technology (Novi Sad, Serbia), holding weekly sessions for the purpose according to ISO 8586-2 (1994).

An untrained panel consisted of 42 members of staff (29 female and 13 male) from the pasta factory Žitko (Bačka Topola, Serbia). Although previously untrained for the specific task, the staff had a formal training in sensory evaluation and therefore, they were accustomed to the basic concept of the evaluation.

Sensory properties of dry and cooked spaghetti were evaluated by the 5-point category scale with end-points labelled from 1 to 5 as shown in Table 1 (Pestorić et al., 2012).

Both panels evaluated the coded cooked pasta samples by comparison with the control sample for firmness, chewiness, adhesiveness and surface adhesiveness, using 15 cm instructed line scale (anchored at 1.5 cm and 13.5 cm with slight to very, respectively (Kovacs et al. 1997). After evaluation, in order to enable comparison of obtained data and data analysis, the scale was divided into 5 parts (3 cm each) that corresponded to the certain level of quality.

Furthermore, assessors were asked to evaluate the degree of similarity of the samples to the control one regarding firmness, chewiness, adhesiveness, surface adhesiveness of cooked spaghetti, as well as the overall spaghetti quality.

The centre of the scale corresponded to the case when the intensity of sample property was the same as the property of the control sample (commercial sample spaghetti which was endorsed by consensus of sufficient intensity on the score line for the individual characteristic textural properties, which are included in the evaluation of textural quality pasta). The left side of the scale corresponded to lower assessment - “worse”, while the right side corresponded to better assessment - “better” of each property.

In order to obtain the overall quality of spaghetti samples assessors used the hedonic scale with 5 quality grades: 1 – extremely dislike 2 – moderately dislike 3 – neither like nor dislike, 4 – moderately like, 5 – extremely like.

The anchored reference standards, terminology definitions and evaluation techniques were agreed upon by the assessors during training and the formal training before evaluation (Szczesniak, 1963; Civille and Szczesniak, 1973; Muñoz, 1985; ISO 11036:1994; Tang et al., 1999).

All samples were presented to each panel in the single sessions. At the start of each session panellists were given a printed response sheet with written instructions for the tests. The order of sample presentation was completely randomized among assessors, identified with three random numbers. Dry spaghetti was presented on the plastic plates, while cooked spaghetti was presented in 250ml sealed thermal plastic cups and served at room temperature within 30 min after cooking. Spaghetti was cooked in boiling distilled water. The optimal cooking time (al dente point) was determined by compressing the spaghetti strand between two transparent plastic plates at different times. It was reached when the white centre of ungelatinized starch had just disappeared according to the approved AACC Method 16-50 cooking time (AACC, 1995). All properties were evaluated visually, palpatory and gustatory under daylight. Drinking water was provided for palate cleansing between each sample.

Statistical evaluation of data

Statistical analysis comprised the evaluation of spaghetti properties by non-parametric methods according to the frequency of assessments. Non-parametric data were scaled by using contingency tables in order to generate the value for each assessment. Thus, it was enabled to use statistical tests for ratio scales. Generated data were subjected to the Multivariate Analysis of Variance (MANOVA) and Discriminative Analysis. Univariate methods applied included Roy’s test, Pearson’s contingency coefficient (c) and multiple regression coefficient (R). On the basis of discrimination coefficients, the principal properties of spaghetti samples were selected for further analysis. Also, the prediction of homogeneity of assessments was done. Mahalanobis distance indicated the similarity or differences of the panel evaluation in relation to the used sensory method. Statistical analysis of sensory evaluation of panels was conducted using software developed by the Agency "SMARTLINE", Novi Sad (Wilbur, 1984).

Results and Discussion

In accordance with the objectives of the research and methodological approach, the differences in panel performance were observed in the five independent statistical subsets:

· Subset S1 – evaluation of the properties of dry spaghetti using the scoring method,

· Subset S2 – evaluation of the properties of cooked spaghetti using the scoring method,

· Subset S3 – evaluation of the properties of cooked spaghetti using the scale method,

· Subset S4 – evaluation of properties of spaghetti using the control sample,

· Subset S5 – evaluation of the overall spaghetti quality using the hedonic scale method.

Applied MANOVA test (P≤0.05) indicated that there was significant difference between the panels’ evaluation in relation to the subsets S1, S2, S3 and S5 (Table 2). In the case of subset S4, the P-value indicated that there was no significant difference between the panels when the spaghetti samples were compared by control sample.

Furthermore, the differences between the panels in relation to selected properties were tested by Roy’s test. The obtained critical value (p<0.100) have shown the existence of significant differences for all properties within the subsets S1, S2, S3 and S5 (Table 3). As for the subset S4 the obtained critical value was p>0.100, no significant difference between the panels in relation to the assessment of chewiness, stickiness and overall quality was noticed. The critical p-value (p<0.100) obtained for the assessment of firmness and adhesiveness indicated that significant differences between the panels for the observed properties of cooked spaghetti samples existed.

This study showed that the utilization of consumer based measurements is not sometimes recommended as an equivalent to the experts whereas Moskowitz (1996) concluded that both groups are able to show similar types of results on similar attributes on the basis of the comparison of the results from expert and untrained panelists. Also, Worch et al. (2010) showed high similarity between the expert and consumer panels by comparing panels’ performance using the confidence ellipses methodology.

^a p≤0.05

^a c – Pearson’s contingency coefficient (P≤0.01)

^b R – Multiple regression coefficients

^c F – value (Roy’s test)

^a DC – discrimination coefficient

^b Bold values are statistically significant (p≤0.05)

In the evaluation of the degree of similarity between the samples and the control one, the performance of untrained panel does not differ from the performance of a trained expert panel. Untrained panellists can describe the products in a reliable and repeatable way and do not differ from the trained experts. This observation leads to the conclusion that when no cognitive skills are needed the performance of trained and untrained subjects is similar, which was mentioned by González et al. (2001). In this particular case, the use of untrained panel appears to be a good alternative to the trained panel (Worch et al., 2010).

The p-value of c² – test with the confidence intervals of c without zero point indicated either weak or no relationship between the properties within subsets and the panels (Table 3).

By integration of the properties within each subset there were the significant differences between the panels, as well as the clearly defined boundaries between them in relation to the evaluated properties (p<0.001) except for the S4 (p>0.1). Hence, the S4 was excluded from the further observation.

The results of discriminative analysis are shown in Table 4. The most significant difference between the panels within the S1 was recorded for the shape of dry spaghetti evaluated by the scoring method. Also, the significant differences were observed for the fracturability and surface appearance. Within the S2, the most significant differences between the panels were recorded for the surface adhesiveness, adhesiveness and surface appearance of cooked spaghetti evaluated by the scoring method. The adhesiveness, chewiness and surface adhesiveness of cooked spaghetti evaluated by the scale method (S3) were the most discriminative properties between the panels (Table 4). The coefficient of discrimination revealed the existence of significant differences between the panels in rating the overall quality of spaghetti when the hedonic scale method was applied.

From the calculated Mahalanobis distance, the existence of the biggest differences between the panels in relation to subsets S1 and S3, the moderate differences in relation to S2 and S5 and the minimum differences for the subset S4 was confirmed (Table 5).

The obtained results confirmed that S1 (35.604%) and S3 (31.239%) gave the greatest contribution to the performances of the panels. For this reason, this study should be considered comprehensively with all subsets according to their sequence (Table 5).

Accordingly, the biggest differences in the performances of the panel were obtained when the samples of spaghetti had been evaluated using the scoring method and scale. At the same time, the biggest difference in assessment was obtained for shape of S1 (52.59%), adhesiveness of S3 (58.33%) and surface adhesiveness of S2 (58.33%).

It might be that the untrained and trained panel differ because the trained panel has been taught to recognize nuances in these properties such as adhesiveness and surface adhesiveness whereas the untrained have not. Also, this might be due to the fact that experts or trained panel has a better knowledge about type of product such as the shape of spaghetti, through their experience and the training sessions.

* Bold values are statistically significant (p≤0.05)

** Bold values indicate a contribution of a specific property of evaluated quality

As Panel 2 showed higher percent of homogeneity across all sections in relation to the Panel 1, it can be concluded that the pasta samples were more frequently evaluated by higher assessments by Panel 2 in comparison to Panel 1 (Table 6). It is also worth mentioning, that trained panellists are able to extract specific features or attributes from the total percept in order to make overall assessment, unlike the untrained panellist who assess the overall quality on the basis of liking at the moment of assessment.

Conclusions

In general, the performed experiments demonstrated that the trained and untrained panel assessments differ quantitatively due to better increased discriminative ability of the trained panellists and due to the lack of training of untrained panellists. On the basis of the obtained results it can be concluded that clear differences between the trained and untrained panel are obvious except in the case when the properties of spaghetti were evaluated using the control sample.

АCKNOWLEDGEMENTS

This work was financed by the Ministry of Education and Science, Republic of Serbia (Project No. III 46001).

Download full article PDF

DOWNLOAD PDF