TY - JOUR

T1 - Comparison of different ways of handling L-shaped data for integrating sensory and consumer information

AU - Asioli, Daniele

AU - Nguyen, Quoc Cuong

AU - Varela, Paula

AU - Næs, Tormod

N1 - Funding Information: The authors would also like to thank for the financial support received from the Norwegian Foundation for Research Levy on Agricultural Products FFL, through the research program “FoodSMaCK, Spectroscopy, Modelling and Consumer Knowledge” (2017-2020). Special thanks go to Hilde Kraggerud (Tine, Norway) for the support with the sample materials and to Stefan Sahlstrøm (Nofima) for his help with the milling procedure. The author Quoc Cuong Nguyen acknowledges the support of time and facilities from Ho Chi Minh City University of Technology (HCMUT), VNU-HCM for this study. Publisher Copyright: © 2021 Elsevier Ltd

PY - 2022

Y1 - 2022

N2 - Different approaches for handling L-shaped data are compared for the first time in a study conducted with Norwegian consumers. Consumers (n = 101) valuated eight different yoghurt profiles varying in three intrinsic attributes such as viscosity, particle size, and flavour intensity following a full factorial design. Sensory attributes, consumers’ liking ratings, and consumer attributes were collected. Data were analysed using two different approaches of handling L-shaped data: approach one used two-step Partial Least Square (PLS) Regression using L-shaped data including the three blocks such as sensory attributes, consumers’ liking ratings, and consumer attributes, while approach two was based on one-step simultaneous L-Partial Least Square (L-PLS) Regression model of the same three blocks of data. The different approaches are compared in terms of centering, step procedures, interpretations, flexibility, and outcomes. Methodological implications and recommendations for academia and future research avenues are outlined.

AB - Different approaches for handling L-shaped data are compared for the first time in a study conducted with Norwegian consumers. Consumers (n = 101) valuated eight different yoghurt profiles varying in three intrinsic attributes such as viscosity, particle size, and flavour intensity following a full factorial design. Sensory attributes, consumers’ liking ratings, and consumer attributes were collected. Data were analysed using two different approaches of handling L-shaped data: approach one used two-step Partial Least Square (PLS) Regression using L-shaped data including the three blocks such as sensory attributes, consumers’ liking ratings, and consumer attributes, while approach two was based on one-step simultaneous L-Partial Least Square (L-PLS) Regression model of the same three blocks of data. The different approaches are compared in terms of centering, step procedures, interpretations, flexibility, and outcomes. Methodological implications and recommendations for academia and future research avenues are outlined.

KW - Consumers

KW - L-shape data

KW - Method comparison

KW - One-step L-PLS

KW - Two-step PLS

KW - Yoghurt

U2 - 10.1016/j.foodqual.2021.104426

DO - 10.1016/j.foodqual.2021.104426

M3 - Journal article

AN - SCOPUS:85116904950

VL - 96

JO - Food Quality and Preference

JF - Food Quality and Preference

SN - 0950-3293

M1 - 104426

ER -