MODELING THE EVOLUTION OF THE PERFORMANCE OF A SENSORY PANEL: A MIXED-MODEL AND CONTROL CHART APPROACH

A sensory panel is often used to profile the same type of product with the same set of attributes for many years. We are interested in characterizing the evolution of the performance of such a panel (and its panelists) over time. This article presents a methodology based on a mixed‐model approach that takes into account the evolution of both panel and panelist in the same model. At the panel level, linear and quadratic evolutions of the performance are tested. At the panelist level, the method allows detection of whether some panelists perform better than others, and whether this difference remains the same or evolves over time. This mixed‐model approach is followed by a graphical representation using a control chart method to identify occasional outliers. Data used to illustrate this methodology are eight sensory profiling data sets collected on ready‐made frozen meals between 1997 and 2001 (every 6 months). The performance index chosen as an example in this study is the individual repeatability measured by standard deviation over replicates.     

GRAPES: A METHOD AND A SAS® PROGRAM FOR GRAPHICAL REPRESENTATIONS OF ASSESSOR PERFORMANCES

GRAPES computes individual and global analyses of variance for sensory profiling data, consisting of several sessions in which all the panelists gave scores to all the products for a number of attributes. The fitted model takes into account the session effect. GRAPES summarizes the results by means of graphical assessor scatterplots which allow to check and to compare panelist performances, such as the way of using scale, the reliability, the discrimination power and the agreement with the panel. In addition, GRAPES detects the outliers for each of these criterion. The usefulness of GRAPES for the panel leader will be demonstrated using texture and flavor profiling of 4 restructured steaks by 12 assessors for 15 attributes. The SAS® program GRAPES, available by e-mail from the author, requires the SAS/BASE®, SAS/STAT®, SAS/GRAPH® and SAS/IML® softwares.     

IMPLEMENTATION OF CONSUMER TASTE PANELS

A taste panel consisting of residents from the local community was developed. In this paper, the benefits of a consumer panel over an in-house panel are discussed. Recruitment procedures, panelist training and performance monitoring are outlined.     

INFLUENCE OF PANEL INCONSISTENCY ON THE OUTCOME OF SENSORY EVALUATIONS FROM DESCRIPTIVE PANELS

Different types of panelist by treatment interaction are explored to determine how they influence the outcomes of discrimination tests. The study compares the situations where panelists are considered as fixed or random effects over the range of most testing conditions for small panels (5–15 panelists) that replicate their judgements. Magnitude interaction and nonperceivers or nondiscriminators have minor effects on test outcomes. Cross-over interaction increases the chances for a type II error, especially when panelists are considered as random effects. False discrimination increases the chances for a type I error when panelists are considered as fixed effects. Applications of methods to reduce the chances for these errors in the testing for differences among treatments are discussed.     

EFFECT OF SAMPLE NUMBERS ON THE SENSORY ACUITY OF A TRAINED GROUND BEEF TEXTURE PANEL

Trained sensory panelists evaluated 8,12 or 18 samples of ground beef at each meeting to determine the effect of sample numbers on panelist perception of certain sensory properties. Six ground beef formulations were selected to represent a broad range of fat and connective tissue content. Panelists evaluated samples for tenderness, connective tissue amount and juiciness. Sensory ratings were affected by formulation, anticipation of sample numbers and actual number of samples (sample x session). Although panelists discriminated among formulations presented in the last six samples somewhat differently from the way they discriminated among the same formulations presented in the first six samples, there was little evidence that the panel became less sensitive.     

THE DOLLAR VALUE OF PRODUCT QUALITY: THE EFFECT OF PRICING VERSUS OVERALL LIKING ON CONSUMER STATED PURCHASE INTENT FOR PIZZA

This paper investigates the combined effects of price and product quality (overall liking) on the acceptance of restaurant pizza pies. The results reveal that when a panelist evaluates both liking and purchase intent (with price attached to the purchase intent question), product quality is a significantly stronger determinant of purchase intent than is the stated item price. For instance, a 10% increase in product quality (viz., liking) has the same effect on purchase intent as a 20% reduction in item price. The results are surprising, because for other categories investigated by the same method price is a significantly greater determinant of purchase intent. In terms of predictability, 57% of the variation in purchase intent ratings is accounted for by liking, 24% of the variation is accounted for by price, and the remaining 19% is due to error.     

A PANEL PERFORMANCE PROCEDURE IMPLEMENTED IN R

Monitoring performance is essential for the efficient use of a sensory panel both during training and while carrying out product assessments. We present a concise procedure to monitor panel performance based on classical statistical methods. The program includes tests for the ability to discriminate between products, repeatability of assessments, scale use, agreement between panelists and a principal component analysis map of panelist means across attributes. The algorithm is implemented in R – a state‐of‐the‐art, freely available statistical software package. The program output is summarized in graphs and tables. This easily applicable panel performance procedure is aimed at improving sensory practice, especially in areas where the use of highly complex systems is not feasible.     

COMPARISON OF METHODS FOR ANALYZING REPLICATED PREFERENCE TESTS

Preference testing is commonly used in consumer sensory evaluation. Traditionally, it is done without replication, effectively leading to a single 0/1 (binary) measurement on each panelist. However, to understand the nature of the preference, replicated preference tests are a better approach, resulting in binomial counts of preferences on each panelist. Variability among panelists then leads to overdispersion of the counts when the binomial model is used and to an inflated Type I error rate for statistical tests of preference. Overdispersion can be adjusted by Pearson correction or by other models such as correlated binomial or beta‐binomial. Several methods are suggested or reviewed in this study for analyzing replicated preference tests and their Type I error rates and power are compared. Simulation studies show that all methods have reasonable Type I error rates and similar power. Among them, the binomial model with Pearson adjustment is probably the safest way to analyze replicated preference tests, while a normal model in which the binomial distribution is not assumed is the easiest.     

USE OF SENSORY EVALUATION FOR MEASURING DEODORIZING EFFECTS OF AN AIR CONDITIONER

The application of sensory methodology for measuring deodorizing effect of an air conditioner equipped with electric plasma was introduced. Deodorizing effect was measured using chemical and sensory methods at different time (0, 30 and 60 min) and mode (control, blowing and cooling) of an air conditioner. Smoke from a roll of cigarette in a closed room was used as a source of odor and the concentrations of acetic acid and ammonia were measured as odorous chemical components. As one of the sensory methods triangle test was used and as a first step to obtain deodorizing effects by triangle test, the threshold of each panelist was obtained as the log dilution ratio of odor concentration at which the difference from odorless air was detected. The odor concentration at each time and mode was calculated using the threshold of the panel and the deodorizing effect was obtained on the basis of the odor concentration. In addition to a triangle test, scaling methods such as category scaling or magnitude estimation were used to measure deodorizing effect of an air conditioner. Deodorizing effects by scaling methods were calculated based on odor intensity with time at each mode. The regression analysis was done between the efficacy of deodorizing effect by sensory test and those by acetic acid and ammonia, the R² values of the regression equations for triangle test, category scale, and magnitude estimation were 0.84, 0.72 and 0.69, respectively. Deodorizing effect by triangle test explained the decrease of acetic acid and ammonia better than those by category scaling or magnitude estimation while high cost and time consuming labor involved in triangle tests reduced the merit. The results of this study demonstrated that various sensory methods could be used to measure deodorizing effect of air conditioners and further researches on fast and reliable methods are needed to establish the official procedures.     

AN OBJECTIVE NUMERICAL METHOD OF ASSESSING THE RELIABILITY OF TIME-INTENSITY PANELISTS

A measure of the reliability (T-IR) of time-intensity measurements was developed based on the concept of standard deviation as a measure of panelist variability. The T-IR measure was applied to time-intensity data collected from 10 panelists evaluating the sweetness of 4 model sweetener solutions on horizontal and vertical time-intensity line orientations. T-IR scores showed that the panelists were similarly reliable across the sweeteners and orientations. As well, independent of scale orientation, responses to sweeteners were similarly reliable. The T-IR measure can be used to maintain a high level of performance by monitoring time-intensity panelists. T-IR also provides an objective method of selecting panelists for time-intensity panels.     

Use of genotyping by sequencing data to develop a high‐throughput and multifunctional SNP panel for conservation applications in Pacific lamprey

Next‐generation sequencing data can be mined for highly informative single nucleotide polymorphisms (SNPs) to develop high‐throughput genomic assays for nonmodel organisms. However, choosing a set of SNPs to address a variety of objectives can be difficult because SNPs are often not equally informative. We developed an optimal combination of 96 high‐throughput SNP assays from a total of 4439 SNPs identified in a previous study of Pacific lamprey (Entosphenus tridentatus) and used them to address four disparate objectives: parentage analysis, species identification and characterization of neutral and adaptive variation. Nine of these SNPs are F_ST outliers, and five of these outliers are localized within genes and significantly associated with geography, run‐timing and dwarf life history. Two of the 96 SNPs were diagnostic for two other lamprey species that were morphologically indistinguishable at early larval stages and were sympatric in the Pacific Northwest. The majority (85) of SNPs in the panel were highly informative for parentage analysis, that is, putatively neutral with high minor allele frequency across the species’ range. Results from three case studies are presented to demonstrate the broad utility of this panel of SNP markers in this species. As Pacific lamprey populations are undergoing rapid decline, these SNPs provide an important resource to address critical uncertainties associated with the conservation and recovery of this imperiled species.     

Assessing the importance of genotype x environment interaction for root traits in rice using a mapping population. I: a soil-filled box screen

Altering root system architecture is considered a method of improving crop water and soil nutrient capture. The analysis of quantitative trait loci (QTLs) for root traits has revealed inconsistency in the same population evaluated in different environments. It must be clarified if this is due to genotype × environment interaction or considerations of statistics if the value of QTLs for marker-assisted breeding is to be estimated. A modified split-plot design was used where a main plot corresponded to a separate experiment. The main plot factor had four treatments (environments), which were completely randomized among eight trials, so that each treatment was replicated twice. The sub-plot factor consisted of 168 recombinant inbreed lines of the Bala × Azucena rice mapping population, randomly allocated to the seven soil-filled boxes. The aim of the trial was to quantify QTL × environment interaction. The treatments were chosen to alter partitioning to roots; consisting of a control treatment (high-soil nitrogen, high light and high-water content) and further treatments where light, soil nitrogen or soil water was reduced singly. After 4 weeks growth, maximum root length (MRL), maximum root thickness, root mass below 50 cm, total plant dry mass (%), root mass and shoot length were measured. The treatments affected plant growth as predicted; low nitrogen and drought increased relative root partitioning, low-light decreased it. The parental varieties Bala and Azucena differed significantly for all traits. Broad-sense heritability of most traits was high (57–86%). Variation due to treatment was the most important influence on the variance, while genotype was next. Genotype × environment interaction was detected for all traits except MRL, although the proportion of variation due to this interaction was generally small. It is concluded that genotype × environment interaction is present but less important than genotypic variation. A companion paper presents QTL × environment analysis of data.