HUNGARIAN JOURNAL OF INDUSTRIAL CHEMISTRY VESZPREM Vol. 30. pp. 235 - 239 (2002) A NEW VBA SOFTWARE AS A TOOL OF FOOD SENSORY TESTS Z. K6KAI, J. HESZBERGER2, K. KOLLAR-HUNEK2, G. KOLLAR1 (Sensory Laboratory, Postharvest Dept., Szent Istvan University, Hll18. Budapest, Menesi ut 45, HUNGARY 1 Postharvest Dept., Szent Istvan University H1118. Budapest, Menesi ut 45, HUNGARY 2 Dept of Chemical Information Technology, Budapest University of Technology and Economics, H1521 Budapest, Pbox.91. HUNGARY) Received: October 17, 2002 This paper was presented at the gth International Workshop on Chemical Engineering Mathematics, Bad Honnef, Germany, September 18-212002 Sensory testing is an essential tool for introducing new products to the market. To achieve reliable sensory data several factors should be controlled from the experimental design, through the coding of the samples to the proper conduct of the analysis. To improve the efficiency of sensory testing computer support of the process is necessary. The target of our project was to develop software, which supports the profile analysis testing method. The selection of the platform (Microsoft Excel) was motivated by its widespread use and easy accessibility. The preparation of the score sheets is performed on the panel leader's computer; afterwards the copies are moved to the panellists' Pes. The software can handle unstructured scales, structured scales, category scales and text fields. Data collection is followed by complex data analysis and graphic presentation. Keywords: food sensory testing, apple profile analysis, visual basic for excel software Introduction Sensory analysis is an essential tool to achieve successful marketing strategy in the food sector. Though sensory quality is not the only key to meet consumer demands (price strategy, promotion, point of sale and other product characteristics are also essential), sensory data can be utilized in several marketing aspects. Sensory testing is generally considered to be subjective) as it relies on human individuals, instead of 'more accurate' equipments of other research fields (e.g.: analytical chemistry, rheology, etc.). Since no reliable model exists for transforming analytical and other instrumental data to perceive sensory quality [9], it is necessary to reduce the objective character of sensory testing. Due to accelerated research on this field there are several solutions for providing reliable sensory data for decision making [1]. One of the basic principles is, that sensory testing should be considered similarly to instrumental testing. From the practical point of view it means: • proper experimental design; Contact information; E-mail: zkokai@omega.kee.hu • understanding the limitation of our 'equipments' (in our case: the assessors) • and suitable statistical analysis of the data. These requirements mentioned above are really just the most basic ones. The importance of these principles is indicated by the fact that already 24 ISO standards deal with different aspects of sensory testing. Quality oriented sensory research in this way might mean more tasks to deal with, so information technology has a huge impact in saving time and energy. Sensory testing Before proceeding with the exact details of our research. one more question should be discussed. Sensory testing [5, 6] can be performed with: • naive assessors (consumers)~ • trained and selected assessors and • experts 236 These different groups should be considered as different tools (like we have different instruments in an analytical laboratory). Consumer tests usually focus on preference, and ask the question 'Which product do you prefer?'. In this case it is very important that the people involved should be representative to the target population [10]. Testing with selected assessors or experts helps to answer the question 'Why is one product preferred, and why is the other one rejected?'. If we use the wrong tool for answering the question (e.g. experts or selected assessors for 'Which product do you prefer?') the results will be invalid. In our research we worked mainly with the second group mentioned above (selected assessors). Tools of information technology in Sensory Laboratory of SziU The Sensory Laboratory of Szent Istvan University, Budapest, Hungary has a specially designed sensory booth system, which was established in accordance with the relevant ISO standards [3]. During the recent period a Local Area Network (LAN) was built in the lab, providing the possibility of PC based testing. This solution greatly improves the efficiency of sensory evaluation. The criti~al point of such a system is the software applied. Several software systems are known for supporting sensory analysis. Since these software are sold in a moderate number of copies, even the academic prices are considerably high. This motivated us to find a solution more suitable to the possibilities of the Hungarian academic sector. Finally it has been decided to develop an own software system in Visual Basic for Applications (VB A [7 ,8]) for our specific neeris [ 4]. While the- staff of the Sensory Laboratory (SziU) provided L~e know-how of sensory testing, the Department of Chemical Information Technology, (BUTE) provided all support on the field of information technology. Since one of the current researches was dealing with the method of profile analysis of apples, in the first step this software module was developed. Sensory testing methods can be divided into three main groups: • difference testing methods; • ranking methods • and descriptive methods. Profile analysis [2] belongs to the group of descriptive methods. These kinds of procedures require trained assessors or experts. The testing session has several steps, which indicates the time demand of the analysis. The nature of the method requires some coUective work of the assessors in one phase of the analy.~is. YBAProject {Profil_anal.xls) EI ·•~:Y Microsoft Excel Objects . ; ·~ ThisWorkbook ~ -~ Worksheetl (Basic_data) :. ···~ Worksheet2 (Scoresheet) . ~ Worksheet3 (Stat_eval) · ... ~ Worksheet4 (Diagrams) Forms r ·Em Category _scale ..§ Descriptive_scale ,- ··Em Identification_of _eval_method ' § Name _of _samples \' ·· § Scoresheet_editor : i.. Em Unstructured_line_scale B._{;;:,~ Modules l·· ·-4 al_Init ~ ... 4 a2_Main · · 4 a3 _Cell _protection r .. -4, bl_Stat_evaluation L.4 b2_Diagram_creator Fig. I Main parts of the profile analysis supporting VBA software The VBA software created Our VBA software consists of 5 modules, 6 user forms, and works on 4 Excel worksheets, as it is shown on the Fig.l. Among the modules the al, a2, a3 marked ones create the score sheets for the assessors, including the protection of the cells which should remain unchanged during the sensory testing procedure. The subroutines of these modules call the forms, and they fill out the first two worksheets (Basic_data and Scoresheet). The bl and b2 marked modules supervise the data collection from the filled out score sheet-files, make statistical evaluation in the third worksheet (Stat_eval) and create the diagrams in the fourth worksheet. The usage of the software we show in the next section on a real apple profile analysis. A real apple proide analysis using the software In the first step of profile analysis the assessors get the samples, and they are asked to create a list of sensory attributes, which they consider important. In this step the assessors work individually. The second step is the group discussion, when the assessors decide which attributes should remain in the final evaluation system. The discussion is supervised and helped by the panel leader. For each sensory attribute the group has to choose an evaluating method {e.g. unstructured scale, category scale, descriptive evaluation, etc.}. Our software makes it easy to choose the evaluating method and specify the further details {Figs.2-5). In this step the screen can be projected to help the work of the group. Title of scwesheet: Number of attributes? (max 30) j 10 ~ Number of samples? (max 6) l 4 ~ Number of a5sesments? (max 20) Fig.2 The score sheet editor ldenhlacahon of the evaluahon method Attribute's number . • • I There are altogether 0 attributes ' deflneci . I Currently edited attrib ute 's 1 number: I.-3-- Fig.3 Setting the evaluation method Using the Score sheet editor form the assessors specify the title of score sheet, and the number of attributes, samples and assessments. Each attribute has a sequence number and a code corresponding to their type (evaluation method). On the Figs.4-5 we show the way of setting up two different scales - unstructured line and category scales. In both c~ses we specify attribute s name, for the first type we gtve the legends at the start and at the end of the scrollbar belonging to the unstructured line, for the s~cond one the names of the categories are going to be gtven. When all the attributes are defined and the evaluation methods are specified, the software asks to type in the name of the samples (varieties) into a form and thereafter creates a block design for the presentation sequence of the samples. A randomly generated, three- di~it number code is also assigned to each sample (Fzg.6). These techniques are essential to avoid psychological faults during testing. 237 Unstructured line scale (, 'EJ Attribute's name: I Red col or Fig.4 Unstructured )jne scale Category scale £1 Attribute's name Lenticell spots Components of category scale -- Components' names few acceptable Fig.5 Setting up a category scale H K -,_..;. · Sample's name Ida red Jonathan Topred sample 10 A 8 c 1 8 A 0 2 c 0 A 3 0 c 8 4 A 8 0 5 8 A c L Golden 0 c 8 A c 0 0 3-dl~it s.:~m lecodes 82 356 289 7 675 967 34 167 342 7 861 7~ 829 718 492 584 1 351 21 396 9 609 :J) 219 Fig.6 Creating the test design automatically With choosing Create coresheet from datasheet' (see Fig.J) the software creates the core sheets for each assessor. Then the score sheets are copied to the PC in the sensory booths and the as e or i ready to te t. Thi step of the testing means individual work again (Fig.7). 238 Table I Sensory test of scab resistant apple varieties using Jonathan as control variety Green flesh White flesh Yell ow flesh Hardness Juicness Peel Sweet taste Sour taste Odour Aroma Taste+ Aroma Releika Remo Resi Rewena Reglindis Jonathan 44 53 76 88 65 71 70 57 60 62 62 50 53 59 41 74 53 86 78 78 65 65 74 52 56 66 58 59 50 61 54 65 54 44 59 74 86 73 76 52 64 58 65 64 41 47 59 75 69 71 58 53 50 51 52 47 65 56 78 71 76 46 67 56 59 62 S C . D E_ _F. G H ··'·····. J · Szent lstvin University. Posth:~rvest Oepanmer£ !}t:)!"~-···-, Assessor's [)code- ,Sampes' codes Attrbies 1235 5$.4 Sensory L01bor:rtory ~ ·. }4 Apple profile an01lysis ~~-"...~ 182 356 Fig.7 Filling out the score sheets When the testing session is finished, the data are collected from each PC. The online filling of the questionnaires means, that the time-consuming data input from paper based questionnaires can be skipped. By the VBA macros of the module bl_Stat_evauationl (see Fig.l) the individual data are collected in one worksheet (Stat_eval), and statistical analysis takes placa Every attribute is analysed for significant differences~ and for the better understanding~ the results are represented in diagrams (Figs.8 and 9). After the t1rst experiences in the usage of our new software in laboratory (testing) circumstances we tried to use it on the Hortus Hungaricus exhibition for the evaluation of the sensory testing of scab resistent apple varieties. organized by the Postharvest Club. This was a different field where we could use the soft\vare. In the circumstances of the Hortus exhibition it was not possible to use computer network and online sensory test data input. We created the score sheets by the software. but in this case we made paper based (nar<:lcopy) questionnaires from the printed out score sheet. On the exhibition the experts filled out these questionnaires, and we could make a real-time evaluation by a laptop. After evaluating the first group of assessors. for the latter visitors we could show a presentation about the results. \Ve investigated six varieties: five resistant apples: Replica~ Reno. Resin. Rowena and Regrinds. As control variety we used the B c D E F 1GT ffi Red colour 1?-: 100 19' 80 20 ~ 60 21 40 22 ~ 20 23' 0 :ill !dared Jomrtan Golden Parmen 25 i "'26"1 Yellow colour 27 i 1~! I " ~- ~s·~ 29. }0; I I I I 31 . 32-~ 33 I 34·: ida red Jonatan Golden Parmen Fig.8 Graphical presentation of the results, according to the properties White flesh Fig.9 Graphical presentation of the results, according to the apple samples Fig.JO Comparison of apples· sensory profiles. Jomithan (left) and Rewena (right) well known and in Hungary preferred Jonathan. From our result we have got in a short way the fact there seems to be no significant difference between some re- appJes and on the market preferred Jonathan. (Table 1, Fig.JO) The ,profile analysis" on the Hortus exhibition was of course only a first attempt to use our software in non- laboratory circumstances. We made ranking tests with more than 200 assessors as well. Releika Jonathan Rewena Resi Remo Reglindis Table 2 Comparison of apples, ranking Releika Jonathan Rewena Resi Remo no no 1% 1% 40 no 5% 1% 67 27 no 1% 128 88 61 no 167 127 100 39 211 171 144 83 44 Reglindis 1% 1% 1% 5% no As one can see on Table 2, the evaluation of simple ranking by Friedman test doesn't show significant differences between Releika, Jonathan and Rewena. Discussion The first experiences with the VBA based sensory analysis supporting software showed, that the time demand of both the preparation and testing step can be considerably reduced. Online questionnaires mean no data input is necessary from paper-based questionnaires. Data analysis and report making is almost real time. Some details of the software will be developed (sample code printing, etc.). In our future work we plan to develop similar software modules for other testing methods (difference testing, ranking and other descriptive methods). A database system managing the different data is also planned to be built. Acknowledgements The authors wish to express their gratitude to the Postharvest Club, to the Wink LtD (Vasarosnameny) to Beata Kapolna and Rita Szabo (SziU) The work has been supported by the Hungarian National Research Foundation (OTKA, grant # T030241!99). 239 SYMBOLS BUTE Budapest University of Technology and Economics LAN Re-apple SziU VBA Local Area Network Disease (scab) resistant apple variety Szent Istvan University Visual Basic for Application REFERENCES 1. CRAMWINCKEL A. B., MAZIJK-BOKSLAG and D. M- y AN: Voedingsmiddelen-technologie, 1990, 23, 14 2. ISO 11035: 1994 Sensory analysis - Identification and selection of descriptors for establishing a sensory profile by a multidimensional approach 3. ISO 8589: 1988 Sensory analysis - General guidance for the design of test rooms 4. K6KAI Z., HESZBERGERJ, KOLLAR-HUNEK K., SZABO R. and KOLLAR G.: Proceedings of MKN '02, Veszprem, Hungary 2002, 1, 177 (in Hungarian) 5. MSZ: ISO 8586-1: 2000 Sensory analysis- General guidance for the selection, training and monitoring of assessors - Part 1: Selected assessors 6. MSZ: ISO 8586-2: 2000 Sensory analysis - General guidance for the selection, training and monitoring of assessors - Part 2: Experts 7. Visual Basic Bookmark, www.vb- bookmark.com/vba.html 8. Visual Basic Home, rnsdn.microsoft.corn/vbasic 9. WORK T. M., BUSHWAY R. J, PERKINS L. B. et ai.: Fruit Varieties Journal, 1994, 48(1),14 10. ZECH J.: Obstbau, 1989, 14(5), 209 Page 236 Page 237 Page 238 Page 239 Page 240