Abstract
A recent trend in descriptive sensory evaluation methodology has been the application of rapid evaluation techniques. The ease in use makes the techniques extremely easy to implement by industry and university environments. Thus, one might not consider validity in the choice of method. The overall aim of this thesis is to compare and evaluate selected rapid evaluation techniques for sensory profiling. Method variations have been suggested for evaluations in product development and quality control, and method insight is provided. The thesis includes three original studies, designed as a consequence of the current practices and needs faced in the industry.
Study I compared applicability and validity of rapid methods across several panels of trained assessors. Two rapid approaches were introduced for the evaluation of foods. The first method, ‘Free Multiple Sorting’, allows subjects to perform ad libitum free sortings, until they feel that no more relevant dissimilarities among products remain. The second method, ‘partial Napping’ (PN),is the concept of restricting the responses of Napping to a specific modality, directing perception and still allowing a holistic approach to products. The new methods were compared to the Flash Profile, Napping and conventional descriptive profiling. Furthermore, an approach for applying confidence ellipses to Multiple Factor Analysis (MFA) results from the statistical package R were suggested for the graphical validation and comparisons. PN allowed repetitions on different sensory modalities and showed these repetitions to be closely related to conventional profile results. Semantic comparisons showed large differences between methods and had the closest relations found between the two conventional profiles. Comparisons of the methods’ practical differences highlighted the time advantage of the rapid approaches and their individual differences in the number of attributes generated.
Study II had a twofold purpose. First, it simulated real-life product tasting situations as they often take place in consensus. Second, it evaluated variations in projective mapping frame geometry and restrictions on the reported semantics. Two rapid descriptive evaluation techniques were proposed to represent a consensus evaluation. One of the approaches, ‘consensus attribute rating’ (CAR), allows a group of assessors to rate products on a list of pre-selected attributes. The other approach, ‘consensus Napping’, allows a group of assessors to project products according to an agreed consensus placement on a paper sheet. Evaluations were performed either by groups of experienced sensory assessors or by product experts. Compared with conventional profiling techniques, the evaluations showed significant correlations between some product configurations, but no consistent and systematic similarities. On average, product expert groups had less in common with the reference profiles than the trained panellist groups and the semantic descriptions of products varied to a large degree for all groups. Hence, consensus profiling with untrained assessors should not be used for the purpose of considering consistency between panels, while assessors trained in the product may perform more reliably. As for projective mapping variations of frame geometry, evaluations in a rectangular or a round geometry were compared. Assessors used the horizontal and vertical dimensions differently when varying the projective frame. A rectangular frame would guide the assessor to act more two-dimensionally and emphasise the longer dimension as more important, while a round frame might facilitate a more holistic approach. The rectangular approach is especially suitable for PN with restrictions in reported semantics, while global Napping (GN) with no restrictions could be advantageous when performed with both geometric frames.
Study III showed that it was possible for naïve assessors to differentiate sound clips with projective mapping. Observation studies and interviews of assessors showed that they applied different projection strategies when placing their samples. Clear groupings of strategies were made and consisted of one-way linear projections, two-way L or T shaped projections, dispersed projections, lumpy projections, categorical projections and categorical double linear projections. The one- and two-dimensional projection strategies dominated more than half of the projective mappings in the study, while the more theoretically ideal dispersed projections accounted for one sixth. The projection strategies were thought to be affected by the complexity of the sample set, but most of all by the assessor’s level of method experience.
Based on the studies and literature review, a new mind-set explaining assessor types, training and measurements, is presented. The appropriate use of rapid descriptive methodology is proposed and, based on the findings and evaluations of the studies, a concept scale that combines holistic and analytic assessor responses is proposed for future evaluations.
Study I compared applicability and validity of rapid methods across several panels of trained assessors. Two rapid approaches were introduced for the evaluation of foods. The first method, ‘Free Multiple Sorting’, allows subjects to perform ad libitum free sortings, until they feel that no more relevant dissimilarities among products remain. The second method, ‘partial Napping’ (PN),is the concept of restricting the responses of Napping to a specific modality, directing perception and still allowing a holistic approach to products. The new methods were compared to the Flash Profile, Napping and conventional descriptive profiling. Furthermore, an approach for applying confidence ellipses to Multiple Factor Analysis (MFA) results from the statistical package R were suggested for the graphical validation and comparisons. PN allowed repetitions on different sensory modalities and showed these repetitions to be closely related to conventional profile results. Semantic comparisons showed large differences between methods and had the closest relations found between the two conventional profiles. Comparisons of the methods’ practical differences highlighted the time advantage of the rapid approaches and their individual differences in the number of attributes generated.
Study II had a twofold purpose. First, it simulated real-life product tasting situations as they often take place in consensus. Second, it evaluated variations in projective mapping frame geometry and restrictions on the reported semantics. Two rapid descriptive evaluation techniques were proposed to represent a consensus evaluation. One of the approaches, ‘consensus attribute rating’ (CAR), allows a group of assessors to rate products on a list of pre-selected attributes. The other approach, ‘consensus Napping’, allows a group of assessors to project products according to an agreed consensus placement on a paper sheet. Evaluations were performed either by groups of experienced sensory assessors or by product experts. Compared with conventional profiling techniques, the evaluations showed significant correlations between some product configurations, but no consistent and systematic similarities. On average, product expert groups had less in common with the reference profiles than the trained panellist groups and the semantic descriptions of products varied to a large degree for all groups. Hence, consensus profiling with untrained assessors should not be used for the purpose of considering consistency between panels, while assessors trained in the product may perform more reliably. As for projective mapping variations of frame geometry, evaluations in a rectangular or a round geometry were compared. Assessors used the horizontal and vertical dimensions differently when varying the projective frame. A rectangular frame would guide the assessor to act more two-dimensionally and emphasise the longer dimension as more important, while a round frame might facilitate a more holistic approach. The rectangular approach is especially suitable for PN with restrictions in reported semantics, while global Napping (GN) with no restrictions could be advantageous when performed with both geometric frames.
Study III showed that it was possible for naïve assessors to differentiate sound clips with projective mapping. Observation studies and interviews of assessors showed that they applied different projection strategies when placing their samples. Clear groupings of strategies were made and consisted of one-way linear projections, two-way L or T shaped projections, dispersed projections, lumpy projections, categorical projections and categorical double linear projections. The one- and two-dimensional projection strategies dominated more than half of the projective mappings in the study, while the more theoretically ideal dispersed projections accounted for one sixth. The projection strategies were thought to be affected by the complexity of the sample set, but most of all by the assessor’s level of method experience.
Based on the studies and literature review, a new mind-set explaining assessor types, training and measurements, is presented. The appropriate use of rapid descriptive methodology is proposed and, based on the findings and evaluations of the studies, a concept scale that combines holistic and analytic assessor responses is proposed for future evaluations.
| Originalsprog | Engelsk |
|---|
| Forlag | Department of Food Science, University of Copenhagen |
|---|---|
| Antal sider | 151 |
| Status | Udgivet - 2012 |