Data on sensory profile of green Spanish-style table olives studied by Quantitative Descriptive Analysis

This article contains processed data related to the research published in “Sensory profile of green Spanish-style table olives according to cultivar and origin” [1]. It provides information on the physicochemical characteristics of the analysed samples and the results of the multivariate analysis used in the above-commented paper. Particularly, it includes: i) the values of pH, titratable acidity, combined acidity, and NaCl for batches according to samples, ii) the scores given to each descriptor by the panelists according to samples, iii) the histogram of the overall scores for descriptor, iv) the boxplot of descriptors over samples, v) the effect of samples and contribution of panelists to the interaction sample∙panelist, vi) correlation between the panelists and the whole panel, vii) panelist performance, viii) panel repeatability, ix) sensory profile of samples (spider graph), x) adjusted means for descriptor according to samples, xi) prevalence of descriptors on samples, xii) product effect as assessed by p-value.


Specifications table
The data are available with this article

Value of the data
The data include the physicochemical characteristics of diverse batches of green Spanish-style table olives prepared from the main cultivars (M, G, and H) from diverse origins, where they were also subjected to processing. The information is important for the characterization of the evaluated samples.
The data matrix, containing the scores assigned by panelists to each descriptor according to samples, is critical for assessing the possible influences of cultivars and cultivation origins.
The multivariate data analysis presented here is an ideal tool for the evaluation of the green Spanish-style table olive profiles, according to the origin and cultivar. The data analysis could include ANOVA and multivariate analysis. Results found here encourage the realisation of similar studies with other cultivars and countries.
As deduced from data, only a few descriptors were enough to describe significant differences among treatments. However, ANOVA and multivariate analysis were useful to properly assess the panel performance and panelists scoring, as well as the discrimination among samples. The methodology may apply to future studies.

Data
The data cover the physicochemical characteristics of the samples to be analysed (Table 1), the results of the evaluation of the diverse samples by the sensory analysis ( Table 2, Figs. 1-4), the statistical study of them (ANOVA and Multivariate), and the main results with respect to panel and panelists' performance (Tables 3-6, Figs. 5 and 6), as well as particular aspects related to the discrimination power (Figs. 7 and 8). All these data are related to a previously published research (1).

Experimental design, materials and methods
The olives were harvested from the second half of September to the last week of October 2014. They belong to the Manzanilla (M), Gordal (G), and Hojiblanca (H) cultivars grown in different areas of Fig. 1. Histogram of the overall scores given by the panelists to the different descriptors. Table 1 Physicochemical characteristics of the studied samples.

Sample Batch pH
Titratable acidity (g lactic acid/L brine) Combined acidity (Eq/L brine) Salt (g NaCl/ L brine) Notes: GU, (cultivar, Gordal; origin. Utrera (Sevilla)); GA, (Gordal; Arahal (Sevilla)); MAG, (Manzanilla, Alcalá de Guadaira (Sevilla)); MP, (Manzanilla, Posadas (Córdoba)); MAm, (Manzanilla; Almendralejo (Badajoz)); HC, (Hojiblanca; Casariche (Sevilla)); HAI, (Hojiblanca; Alameda (Málaga)); HE, (Hojiblanca; Estepa (Sevilla)). Andalusia and Extremadura (Spain): Manzanilla, Alcalá de Guadaira (Sevilla), Posadas (Córdoba), and Almendralejo (Badajoz); Gordal, Utrera (Sevilla) and Arahal (Sevilla); Hojiblanca, Alameda (Málaga), Estepa (Sevilla), and Casariche (Sevilla). By combining cultivar and origin, the samples were coded as MAG, MP, MAm, GU, GA, HAl, HE, and HC, respectively. The olive samples were elaborated in duplicate batches as green Spanish-style, using polyethylene containers (5.2 kg olives and 3.4 L brine), in the pilot plant facilities of Instituto de la Grasa. They received similar processing treatments. The fruits were debittered with a lye solution (1.90-2.10 g/L) which penetrated 2/3 of the flesh, followed by one washing with tap water for 11-17 h. After washing, brining was carried out by adding a 11% NaCl solution into the containers. The evolution of the spontaneous fermentation was monitored by controlling the pH, titratable acidity, combined acidity, NaCl, and sugar in the brine [2]. After 5 months, there was no residual sugar left, and the process was considered finished. By then, the physicochemical characteristics (Table 1) were considered normal for the style and cultivar and appropriate for the subsequent sensory analysis. This was accomplished by a panel composed of 9 men and 6 women, recruited because of their major role in the implementation of the Sensory Analysis Method for Table Olives [3] and their high level of training due to participation in the habitual sensory analysis of table olives for decades [4]. The panelists were familiarised with the Quantitative Descriptive Analysis (QDA) techniques by training them for 1 h twice a week for two Boxplots of descriptors' scores over samples. GA, (cultivar, Gordal; origin, Arahal (Sevilla)); GU, (Gordal; Utrera (Sevilla)); HAI (Hojiblanca; Alameda (Málaga)); HC, (Hojiblanca; Casariche (Sevilla)); HE, (Hojiblanca; Estepa (Sevilla)); MAG, (Manzanilla; Alcalá de Guadaira (Sevilla)); MAm, (Manzanilla; Almendralejo (Badajoz)); MP, (Manzanilla; Posadas (Córdoba)). months. After agreement concerning the descriptors to be used and training on the characteristics of the QDA, the panel was used for table olive evaluation. The green Spanish-style table olive samples were taken directly from the fermentation vessels and presented to panelists in a randomised order, contained in standard glasses [5], and coded with three randomly chosen digits. The panelists were asked to mark the intensity of the different descriptors in their corresponding scales. Between tests, the panelists were provided with tap water to cleanse the palate. The scores of the attributes were measured with the exactitude of one decimal point and the results tabulated.
The data were analysed using the SensoMineR v.1.07 software [6]. The program was designed and programmed in R language [7] and collects classical methods usually applied when analysing sensory data as well as others directly conceived in the developers' laboratory. SensoMineR provided the results of the analyses of variance (ANOVA) models and numerous easy to interpret graphical outputs also generated a virtual panel, by bootstrapping techniques, for the multivariate analysis and construction of product confidence ellipses. The Spider graphs were produced using Panel Check V1.4.2 software [8].