Food Science Text Series The Food Science Text Series provides faculty with the leading teaching tools. The Editorial Board has outlined the most appropriate and complete content for each foodsciencecourseinatypicalfoodscienceprogramandhasidentifiedtextbooksof thehighestquality,writtenbytheleadingfoodscienceeducators. SeriesEditor DennisR.Heldman EditorialBoard David A. Golden, Ph.D., Professor of Food Microbiology, Department of Food ScienceandTechnology,UniversityofTennessee Richard W. Hartel, Professor of Food Engineering, Department of Food Science, UniversityofWisconsin Hildegarde Heymann, Professor of Food Sensory Science, Department of Food ScienceandTechnology,UniversityofCalifornia-Davis Joseph H. Hotchkiss, Professor, Institute of Food Science and Institute for Comparative and Environmental Toxicology, and Chair, Food Science Department, CornellUniversity MichaelG.Johnson,Ph.D.,ProfessorofFoodSafetyandMicrobiology,Department ofFoodScience,UniversityofArkansas Joseph Montecalvo, Jr., Professor, Department of Food Science and Nutrition, CaliforniaPolytechnicandStateUniversity-SanLuisObispo S. Suzanne Nielsen, Professor and Chair, Department of Food Science, Purdue University Juan L. Silva, Professor, Department of Food Science, Nutrition and Health Promotion,MississippiStateUniversity Forfurthervolumes: http://www.springer.com/series/5999 · Harry T. Lawless Hildegarde Heymann Sensory Evaluation of Food Principles and Practices Second Edition 123 HarryT.Lawless HildegardeHeymann DepartmentofFoodScience DepartmentofViticultureandEnology CornellUniversity UniversityofCalifornia–Davis StockingHall,Room106 2003RMISensoryBuilding 14853Ithaca Davis95616 NY,USA CA,USA [email protected] [email protected] ISSN1572-0330 ISBN978-1-4419-6487-8 e-ISBN978-1-4419-6488-5 DOI10.1007/978-1-4419-6488-5 SpringerNewYorkDordrechtHeidelbergLondon LibraryofCongressControlNumber:2010932599 ©SpringerScience+BusinessMedia,LLC2010 Allrightsreserved.Thisworkmaynotbetranslatedorcopiedinwholeorinpartwithoutthewritten permissionofthepublisher(SpringerScience+BusinessMedia,LLC,233SpringStreet,NewYork,NY 10013,USA),exceptforbriefexcerptsinconnectionwithreviewsorscholarlyanalysis.Useinconnection withanyformofinformationstorageandretrieval,electronicadaptation,computersoftware,orbysimilar ordissimilarmethodologynowknownorhereafterdevelopedisforbidden. Theuseinthispublicationoftradenames,trademarks,servicemarks,andsimilarterms,eveniftheyare notidentifiedassuch,isnottobetakenasanexpressionofopinionastowhetherornottheyaresubject toproprietaryrights. Printedonacid-freepaper SpringerispartofSpringerScience+BusinessMedia(www.springer.com) Preface Thefieldofsensorysciencehasgrownexponentiallysincethepublicationofthepre- viousversionofthiswork.FifteenyearsagothejournalFoodQualityandPreference wasfairlynew.Nowitholdsaneminentpositionasavenueforresearchonsensory test methods (among many other topics). Hundreds of articles relevant to sensory testing have appeared in that and in other journals such as the Journal of Sensory Studies.Knowledgeoftheintricatecellularprocessesinchemoreception,aswellas theirgeneticbasis,hasundergonenothinglessthanarevolution,culminatinginthe awardoftheNobelPrizetoBuckandAxelin2004fortheirdiscoveryoftheolfactory receptorgenesuperfamily.Advancesinstatisticalmethodologyhaveacceleratedas well.Sensometricsmeetingsarenowvigorousandwell-attendedannualevents.Ideas likeThurstonianmodelingwerenotwidelyembraced15yearsago,butnowseemto bepartoftheeverydaythoughtprocessofmanysensoryscientists. And yet, some things stay the same. Sensory testing will always involve human participants. Humans are tough measuring instruments to work with. They come withvaryingdegreesofacumen,training,experiences,differinggeneticequipment, sensory capabilities, and of course, different preferences. Human foibles and their associated error variance will continue to place a limitation on sensory tests and actionableresults.Reducing,controlling,partitioning,andexplainingerrorvariance are all at the heart of good test methods and practices. Understanding the product– personinterfacewillalwaysbethegoalofsensoryscience.Noamountofelaborate statistical maneuvering will save a bad study or render the results somehow useful andvalid.Althoughmethodscontinuetoevolve,appreciationofthecoreprinciples ofthefieldisthekeytoeffectiveapplicationofsensorytestmethods. The notion that one can write a book that is both comprehensive and suitable as anintroductorytextwasadauntingchallengeforus.Somemaysaythatwemissed the mark on this or that topic, that it was either too superficially treated or too in depthfortheirstudents.Perhapswehavetriedtodotheimpossible.Nonethelessthe demand for a comprehensive text that would serve as a resource for practitioners is demonstratedbythesuccessofthefirstedition.Itswidespreadadoptionasauniver- sityleveltextshowsthatmanyinstructorsfeltthatitcouldbeusedappropriatelyfor afirstcourseinsensoryevaluation. This book has been expanded somewhat to reflect the advances in methodolo- gies, theory, and analysis that have transpired in the last 15 years. The chapters are now divided into numbered sections. This may be of assistance to educators who maywishtoassignonlycertaincriticalsectionstobeginningstudents.Muchofthe organization of key chapters has been done with this in mind and in some of the v vi Preface opening sections; instructors will find suggestions about which sections are key for fundamentalunderstandingofthattopicormethod.Inmanychapterswehavegone out on a limb and specified a “recommended procedure.” In cases where there are multiple options for procedure or analysis, we usually chose a simple solution over one that is more complex. Because we are educators, this seemed the appropriate path. Note that there are two kinds of appendices in this book. The major statistical methods are introduced with worked examples in Appendices A–E, as in the previ- ousedition.Somemainchaptersalsohaveappendedmaterialsthatwefeltwerenot criticaltounderstandingthemaintopic,butmightbeofinteresttoadvancedstudents, statisticians,orexperienced practitioners.Wecontinue togive referencecitationsat theendofeverychapter,ratherthaninonebiglistattheend.Statisticaltableshave beenadded,mostnotablythediscriminationtablesthatmaynowbefoundbothinthe AppendixandinChapter4itself. One may question whether textbooks themselves are an outdated method for information retrieval. We feel this acutely because we recognize that a textbook is necessarily retrospective and is only one snapshot in time of a field that may be evolvingrapidly.Studentsandpractitionersalikemayfindthatreferencetoupdated websites,wikis,andsuchwillprovideadditionalinformationandnewanddifferent perspectives.Weencouragesuchinvestigation.Textbooks,likeautomobiles,havean element of built-in obsolescence. Also textbooks, like other printed books, are lin- ear in nature, but the mind works by linking ideas. Hyperlinked resources such as websitesandwikiswilllikelycontinuetoproveuseful. Weaskyourpatienceandtoleranceformaterialsandcitationsthatwehaveleftout thatyoumightfeelareimportant.Werecognizethattherearelegitimatedifferencesof opinionandphilosophyabouttheentireareaofsensoryevaluationmethods.Wehave attemptedtoprovideabalancedandimpartialviewbasedonourpracticalexperience. Anyerrorsoffact,errorstypographical,orerrorsincitationareourownfault.Webeg yourunderstandingandpatienceandwelcomeyourcorrectionsandcomments. We could not have written this book without the assistance and support of many people. WewouldliketothankKathy Dernoga forproviding apre-publication ver- sion of the JAR scale ASTM manual as well as the authors of the ASTM JAR manualLoriRothmanandMerryJoParker.Additionally,MarySchraidtofPeryam andKrollprovidedupdatedexamplesofaconsumertestscreeningquestionnaireand field study questionnaires. Thank you Mary. We thank John Hayes, Jeff Kroll, Tom Carr,DannyEnnis,andJianBiforsupplyingadditionalliterature,software,andsta- tisticaltables.GernotHoffmanngraciouslyprovidedgraphicsforChapter12.Thank youDr.Hoffmann.WewouldliketothankWendyParrandJamesGreenforprovid- ingsomegraphicsforChapter10.Additionally,GregHirsonprovidedsupportwith R-Graphics. Thank you, Greg. Additionally, we want to thank the following peo- plefortheirwillingness todiscussthebook inprogress andformaking veryuseful suggestions:MichaelNestrud,SusanCuppett,EdanLev-Ari,ArmandCardello,Marj Albright,DavidStevens,RichardPopper,andGregHirson.JohnHornehadalsobeen very helpful in the previous edition, thank you John. Proofreading and editing sug- gestions were contributed by Kathy Chapman, Gene Lovelace, Mike Nestrud, and MargeLawless. Although not directly involved with this edition of the book we would also like to thank our teachers and influential mentors—without them we would be very dif- ferent scientists, namely Trygg Engen, William S. Cain, Linda Bartoshuk, David Preface vii Peryam, David Stevens, Herb Meiselman, Elaine Skinner, Howard Schutz, Howard Moskowitz, Rose Marie Pangborn, Beverley Kroll, W. Frank Shipe, Lawrence E. Marks, Joseph C. Stevens, Arye Dethmers, Barbara Klein, Ann Noble, Harold Hedrick,WilliamCStringer,RogerBoulton,KayMcMath,JoelvanWyk,andRoger Mitchell. Ithaca,NewYork HarryT.Lawless Davis,California HildegardeHeymann Contents 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.1 IntroductionandOverview . . . . . . . . . . . . . . . . . . . 1 1.1.1 Definition . . . . . . . . . . . . . . . . . . . . . . . 1 1.1.2 Measurement . . . . . . . . . . . . . . . . . . . . . 3 1.2 HistoricalLandmarksandtheThreeClasses ofTestMethods . . . . . . . . . . . . . . . . . . . . . . . . . 4 1.2.1 DifferenceTesting . . . . . . . . . . . . . . . . . . . 5 1.2.2 DescriptiveAnalyses . . . . . . . . . . . . . . . . . 6 1.2.3 AffectiveTesting . . . . . . . . . . . . . . . . . . . 7 1.2.4 TheCentralDogma—AnalyticVersus HedonicTests . . . . . . . . . . . . . . . . . . . . . 8 1.3 Applications:WhyCollectSensoryData? . . . . . . . . . . . 10 1.3.1 DifferencesfromMarketingResearchMethods . . . 13 1.3.2 DifferencesfromTraditionalProduct GradingSystems . . . . . . . . . . . . . . . . . . . 15 1.4 SummaryandConclusions . . . . . . . . . . . . . . . . . . . 16 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 2 PhysiologicalandPsychologicalFoundations ofSensoryFunction . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 2.2 ClassicalSensoryTestingandPsychophysicalMethods . . . . 20 2.2.1 EarlyPsychophysics . . . . . . . . . . . . . . . . . 20 2.2.2 TheClassicalPsychophysicalMethods . . . . . . . . 21 2.2.3 ScalingandMagnitudeEstimation . . . . . . . . . . 23 2.2.4 CritiquesofStevens . . . . . . . . . . . . . . . . . . 25 2.2.5 EmpiricalVersusTheory-DrivenFunctions . . . . . 25 2.2.6 ParallelsofPsychophysicsandSensory Evaluation . . . . . . . . . . . . . . . . . . . . . . . 26 2.3 AnatomyandPhysiologyandFunctionsofTaste . . . . . . . . 27 2.3.1 AnatomyandPhysiology . . . . . . . . . . . . . . . 27 2.3.2 TastePerception:Qualities . . . . . . . . . . . . . . 30 2.3.3 TastePerception:AdaptationandMixture Interactions . . . . . . . . . . . . . . . . . . . . . . 30 2.3.4 IndividualDifferencesandTasteGenetics . . . . . . 33 2.4 AnatomyandPhysiologyandFunctionsofSmell. . . . . . . . 34 2.4.1 AnatomyandCellularFunction . . . . . . . . . . . . 34 ix x Contents 2.4.2 RetronasalSmell . . . . . . . . . . . . . . . . . . . 36 2.4.3 OlfactorySensitivityandSpecificAnosmia . . . . . 37 2.4.4 OdorQualities:PracticalSystems . . . . . . . . . . 38 2.4.5 FunctionalProperties:Adaptation,Mixture Suppression,andRelease . . . . . . . . . . . . . . . 39 2.5 Chemesthesis. . . . . . . . . . . . . . . . . . . . . . . . . . . 41 2.5.1 QualitiesofChemestheticExperience . . . . . . . . 41 2.5.2 PhysiologicalMechanismsofChemesthesis . . . . . 42 2.5.3 Chemical“Heat” . . . . . . . . . . . . . . . . . . . 43 2.5.4 OtherIrritativeSensationsandChemical Cooling . . . . . . . . . . . . . . . . . . . . . . . . 44 2.5.5 Astringency . . . . . . . . . . . . . . . . . . . . . . 45 2.5.6 MetallicTaste . . . . . . . . . . . . . . . . . . . . . 46 2.6 Multi-modalSensoryInteractions . . . . . . . . . . . . . . . . 47 2.6.1 TasteandOdorInteractions . . . . . . . . . . . . . . 47 2.6.2 IrritationandFlavor . . . . . . . . . . . . . . . . . . 49 2.6.3 Color–FlavorInteractions . . . . . . . . . . . . . . . 49 2.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 3 PrinciplesofGoodPractice . . . . . . . . . . . . . . . . . . . . . . . 57 3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 3.2 TheSensoryTestingEnvironment . . . . . . . . . . . . . . . . 58 3.2.1 EvaluationArea . . . . . . . . . . . . . . . . . . . . 59 3.2.2 ClimateControl . . . . . . . . . . . . . . . . . . . . 62 3.3 TestProtocolConsiderations . . . . . . . . . . . . . . . . . . 63 3.3.1 SampleServingProcedures . . . . . . . . . . . . . . 63 3.3.2 SampleSize . . . . . . . . . . . . . . . . . . . . . . 63 3.3.3 SampleServingTemperatures . . . . . . . . . . . . 64 3.3.4 ServingContainers . . . . . . . . . . . . . . . . . . 64 3.3.5 Carriers . . . . . . . . . . . . . . . . . . . . . . . . 65 3.3.6 PalateCleansing . . . . . . . . . . . . . . . . . . . . 65 3.3.7 SwallowingandExpectoration . . . . . . . . . . . . 66 3.3.8 InstructionstoPanelists . . . . . . . . . . . . . . . . 66 3.3.9 RandomizationandBlindLabeling . . . . . . . . . . 66 3.4 ExperimentalDesign. . . . . . . . . . . . . . . . . . . . . . . 66 3.4.1 DesigningaStudy . . . . . . . . . . . . . . . . . . . 66 3.4.2 DesignandTreatmentStructures . . . . . . . . . . . 69 3.5 PanelistConsiderations . . . . . . . . . . . . . . . . . . . . . 72 3.5.1 Incentives . . . . . . . . . . . . . . . . . . . . . . . 72 3.5.2 UseofHumanSubjects . . . . . . . . . . . . . . . . 73 3.5.3 PanelistRecruitment . . . . . . . . . . . . . . . . . 74 3.5.4 PanelistSelectionandScreening . . . . . . . . . . . 74 3.5.5 TrainingofPanelists . . . . . . . . . . . . . . . . . 75 3.5.6 PanelistPerformanceAssessment. . . . . . . . . . . 75 3.6 TabulationandAnalysis . . . . . . . . . . . . . . . . . . . . . 75 3.6.1 DataEntrySystems . . . . . . . . . . . . . . . . . . 75 3.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 Contents xi 4 DiscriminationTesting . . . . . . . . . . . . . . . . . . . . . . . . . 79 4.1 DiscriminationTesting . . . . . . . . . . . . . . . . . . . . . . 79 4.2 TypesofDiscriminationTests . . . . . . . . . . . . . . . . . . 80 4.2.1 PairedComparisonTests . . . . . . . . . . . . . . . 80 4.2.2 TriangleTests . . . . . . . . . . . . . . . . . . . . . 83 4.2.3 Duo–TrioTests . . . . . . . . . . . . . . . . . . . . 84 4.2.4 n-AlternativeForcedChoice(n-AFC)Methods . . . 85 4.2.5 A-Not-Atests . . . . . . . . . . . . . . . . . . . . . 85 4.2.6 SortingMethods . . . . . . . . . . . . . . . . . . . . 87 4.2.7 TheABXDiscriminationTask . . . . . . . . . . . . 88 4.2.8 Dual-StandardTest . . . . . . . . . . . . . . . . . . 88 4.3 ReputedStrengthsandWeaknessesofDiscrimination Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88 4.4 DataAnalyses . . . . . . . . . . . . . . . . . . . . . . . . . . 89 4.4.1 BinomialDistributionsandTables . . . . . . . . . . 89 4.4.2 TheAdjustedChi-Square(χ2)Test . . . . . . . . . . 90 4.4.3 TheNormalDistributionandtheZ-Test onProportion . . . . . . . . . . . . . . . . . . . . . 90 4.5 Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92 4.5.1 ThePoweroftheStatisticalTest . . . . . . . . . . . 92 4.5.2 Replications . . . . . . . . . . . . . . . . . . . . . . 94 4.5.3 Warm-UpEffects . . . . . . . . . . . . . . . . . . . 97 4.5.4 CommonMistakesMadeintheInterpretation ofDiscriminationTests . . . . . . . . . . . . . . . . 97 Appendix:ASimpleApproachtoHandlingtheA,Not-A, andSame/DifferentTests . . . . . . . . . . . . . . . . . . . . 98 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99 5 Similarity,EquivalenceTesting,andDiscriminationTheory . . . . 101 5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 5.2 CommonSenseApproachestoEquivalence . . . . . . . . . . 103 5.3 EstimationofSampleSizeandTestPower . . . . . . . . . . . 104 5.4 HowBigofaDifferenceIsImportant? DiscriminatorTheory . . . . . . . . . . . . . . . . . . . . . . 105 5.5 TestsforSignificantSimilarity . . . . . . . . . . . . . . . . . 108 5.6 TheTwoOne-SidedTestApproach(TOST) andIntervalTesting . . . . . . . . . . . . . . . . . . . . . . . 110 5.7 ClaimSubstantiation . . . . . . . . . . . . . . . . . . . . . . . 111 5.8 ModelsforDiscrimination:SignalDetectionTheory . . . . . . 111 5.8.1 TheProblem . . . . . . . . . . . . . . . . . . . . . . 112 5.8.2 ExperimentalSetup . . . . . . . . . . . . . . . . . . 112 5.8.3 AssumptionsandTheory . . . . . . . . . . . . . . . 113 5.8.4 AnExample . . . . . . . . . . . . . . . . . . . . . . 114 5.8.5 AConnectiontoPairedComparisonsResults ThroughtheROCCurve . . . . . . . . . . . . . . . 116 5.9 ThurstonianScaling . . . . . . . . . . . . . . . . . . . . . . . 116 5.9.1 TheTheoryandFormulae. . . . . . . . . . . . . . . 116 5.9.2 ExtendingThurstone’sModeltoOther ChoiceTests . . . . . . . . . . . . . . . . . . . . . . 118