consumer-oriented quality improvement of tomatoes in indonesia

20
Euphytica 71 : 161-180,1993 . ©1993 KluwerAcademicPublishers .PrintedintheNetherlands. Consumer-orientedqualityimprovementoftomatoesinIndonesia Howtoconstructanideotype?Howtoassessqualityproblems?* OliviervanLieshout LembangHorticulturalResearchInstitute,JlTangkubanPerahu,Lembang,Indonesia187AvenidaNortey9 CallePoniente,#604,SanSalvador,ElSalvador Received27October1992 ;accepted17September1993 Keywords: tomato, Lycopersiconesculentum, Indonesia,ideotype,qualityproblems,consumerperceptions, consumerpreferences,price-qualitytrade-off Abstract ConsumerdemandsbecomeincreasinglystronginWesterncountries .Productqualityhasproventobean importantmarketingtooltocompeteeffectivelyonthesesaturatedmarkets .Marketsindevelopingcountries aremostlyproducerdominated .Notmuchattentionispaidtoproductquality.Agriculturalresearcherstradi- tionallyareproducer-oriented,ratherthanconsumer-oriented .ResearchontomatoesinIndonesiahas shown,thateveninproducerdominatedmarkets,consumershavespecificwishesregardingquality,which shouldbetakenintoconsiderationinordertodevelopdemandandtoenhancefarmers'profitability .This papershedslightonaresearchmethodologytotranslateconsumerpreferencesintoanideotype,apractical productprofileconsistingofoptimallevelsofrelevantphysicalandchemicalproperties .Thisideotypeshould beusedtoobjectivelyevaluateagriculturalresearch,aimedatimprovingproductquality .Inaddition,alistof qualityproblemsofcurrentsupplyispresented .Thisprioritylistservesasaguidelineforresearchplanning . Finally,amethodispresentedtofindouttowhatextendconsumersarewillingtopayahigherpricefor improvedquality. Introduction Qualityimprovementofagriculturalproductsisof- tenassociatedwithasaturatedmarket,wherecon- sumershaveamplechoiceandrejectproductsper- ceivedtobeofpoorquality .Thishasforcedpro- ducerstobecomeconsumer-oriented .Inmanyde- velopingcountries,asituationofrelativescarcity prevails.Qualityimprovementisprimarilypro- *Thispaperisasummarizedversionofaresearchreportwith thesametitle.TheinterestedreadercanobtainacopyfromDr . G .Grubben,CPRO,PO.Box16,6700AAWageningen,TheNe- therlands,referringtoLEHRI/ATA-365InternalCommunica- tionNo .46 . ducer-orientedandaimedatimprovingproductsin suchaway,thattheycanbeproducedprofitablyin sufficientlylargequantities .However,evenincon- ditionsofrelativescarcity,itisstilltheconsumer, whoultimatelyusesher/hisownperceptionsand beliefsofqualitytoevaluateaproduct .Supplyofa (new)agriculturalproduct,whichcorrespondsbet- terwithconsumerpreferences,willenablefarmers toraisepricesandincreaseprofitability. Agoodexampleistheintroductionin1985ofa newvarietyoftomatoinIndonesia .Theseso-called `Taiwan'tomatoescurrentlyholdamarketshareof 50%inthetownssurveyed .Theyaresoldata25% higherprice,whileproductioncostshaveremained

Upload: olivier-lieshout

Post on 06-Jul-2016

214 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: Consumer-oriented quality improvement of tomatoes in Indonesia

Euphytica 71 : 161-180, 1993 .©1993 Kluwer Academic Publishers . Printed in the Netherlands.

Consumer-oriented quality improvement of tomatoes in IndonesiaHow to construct an ideotype? How to assess quality problems?*

Olivier van LieshoutLembang Horticultural Research Institute, Jl Tangkuban Perahu, Lembang, Indonesia187 Avenida Norte y 9Calle Poniente, #604, San Salvador, El Salvador

Received 27 October 1992 ; accepted 17 September 1993

Key words: tomato, Lycopersicon esculentum, Indonesia, ideotype, quality problems, consumer perceptions,consumer preferences, price-quality trade-off

Abstract

Consumer demands become increasingly strong in Western countries . Product quality has proven to be animportant marketing tool to compete effectively on these saturated markets . Markets in developing countriesare mostly producer dominated . Not much attention is paid to product quality. Agricultural researchers tradi-tionally are producer-oriented, rather than consumer-oriented . Research on tomatoes in Indonesia hasshown, that even in producer dominated markets, consumers have specific wishes regarding quality, whichshould be taken into consideration in order to develop demand and to enhance farmers' profitability . Thispaper sheds light on a research methodology to translate consumer preferences into an ideotype, a practicalproduct profile consisting of optimal levels of relevant physical and chemical properties . This ideotype shouldbe used to objectively evaluate agricultural research, aimed at improving product quality . In addition, a list ofquality problems of current supply is presented . This priority list serves as a guideline for research planning .Finally, a method is presented to find out to what extend consumers are willing to pay a higher price forimproved quality.

Introduction

Quality improvement of agricultural products is of-ten associated with a saturated market, where con-sumers have ample choice and reject products per-ceived to be of poor quality. This has forced pro-ducers to become consumer-oriented . In many de-veloping countries, a situation of relative scarcityprevails. Quality improvement is primarily pro-

* This paper is a summarized version of a research report withthe same title. The interested reader can obtain a copy from Dr .G. Grubben, CPRO, PO. Box 16,6700 AA Wageningen, The Ne-therlands, referring to LEHRI/ATA-365 Internal Communica-tion No . 46 .

ducer-oriented and aimed at improving products insuch a way, that they can be produced profitably insufficiently large quantities . However, even in con-ditions of relative scarcity, it is still the consumer,who ultimately uses her/his own perceptions andbeliefs of quality to evaluate a product . Supply of a(new) agricultural product, which corresponds bet-ter with consumer preferences, will enable farmersto raise prices and increase profitability.

A good example is the introduction in 1985 of anew variety of tomato in Indonesia . These so-called`Taiwan' tomatoes currently hold a market share of50% in the towns surveyed . They are sold at a 25%higher price, while production costs have remained

Page 2: Consumer-oriented quality improvement of tomatoes in Indonesia

162

A - CAUSAL RELATIONSHIPS :

instrumental psycho-physical perceptual

QUALITY

OBJECTIVE

SUBJECTIVE

PERCEIVEDINSTRUMENTS* PROPERTIES

ii PERCEPTIONS

0 QUALITY :

MARKETING* grading* pricing* promotion* advertising* distribution

POST-HARVEST* harvest time* storage* cooling* transport

CULTURAL PR .* plant prot .* fertilizers* training* croppingsystems

BREEDING* cultivars/hybrids

4

NON-PRODUCT* packaging* outlet* origin* price

PRODUCT* color* gloss* shape* volume* skin damage* penetration* ribs* thickness* color flesh* pits* locular gel* core lenght* water cont .* solids* sugars* acidity* and others

1

COGNITIVE* keepability* ease to usehealthiness

* vitamins

EXTERNAL* appearance* firmness* size* aroma

INTERNAL* appearance* pits-gel

--- ORGANOLEPTIC* tomato-like* juiciness*sweeUsour

EXPECTEDQUALITY

EXPERIENCEDQUALITY

Note: Dotted lines are relationships not under investigation in this paper

about equal to those of local varieties grown undersimilar conditions . The success of `Taiwan' toma-toes is partly due to the fact that the product ex-celled on important consumer criteria and was,therefore, able to command a higher price .

International, national and to a lesser degree pri-vate agricultural research institutes, both in devel-oping and developed countries, are traditionallyproducer-oriented rather than consumer-oriented .They have little experience with consumer research

(e.g ., Stevens & Rick, 1986b ; Ameriana et al ., 1991 ;Potts & Chilver, 1992) . Partly this could be justifiedby the need to assist farmers to increase production,partly it is so because in the opinion of many agri-cultural technicians consumer preferences are diffi-cult to measure, they seem highly subjective andsensitive to changes in `fashion' (Buurma, 1992) .

This paper intends to reveal consumer demandsof tomatoes and translate important subjective con-

B - METHOD OF INVESTIGATIONS :

OPTIMAL USE OF

phase II OPTIMAL FRUIT phase I OPTIMAL CONSUMERINSTRUMENTS * PROFILE QUALITY

Page 3: Consumer-oriented quality improvement of tomatoes in Indonesia

sumer preferences into a practical product profileconsisting of objectively improvable properties.

Research objectives

The Lembang Horticultural Research Institutenear Bandung, West Java, Indonesia has startedconsumer-oriented quality research in 1989 . Prod-ucts surveyed were garlic and shallots (Ameriana etal., 1991). Indonesia shows a growing demand forhigher quality tomatoes (Van Lieshout, 1992b) . Thetask at hand was to reveal consumer criteria and toimprove the quality of local tomato varieties. Twodistinct household uses exist : raw, for fruit or as dishdecoration and cooked as a spice, so-called 'bum-bu' . Generally speaking a higher quality tomato iswanted for fruit use. Spice tomatoes are used insmall quantities (half a tomato per meal) and aresquashed before being cooked, therefore, smalldamages can be removed .

The research consisted of two phases :Phase I: a representative consumer research with si-multaneous laboratory analyses, aimed at findinganswers on the following questions :a. what is the optimal consumer quality?b. what is the ideotype or optimal product profile?c. what are the current quality problems?d. what are consumers willing to pay for better

quality?Phase II: production research aimed at eliminatingcurrent quality problems . The ideotype, or optimalproduct profile should be used for quality evalua-tion of field trials and (on farm) experiments . Thekey question is :e. what is the optimal use of instruments?Phase II is currently under execution . This paperdeals with the first phase only. It focuses on quanti-fying the optimal fruit profile for tomato used rawas fruit or in salads. A simultaneous study, was donefor tomato used as spice (Ameriana, 1992). In addi-tion, a list of quality problems of current tomatosupply was to be made.

The research model

The model presented in Schedule 1, is adapted fromthe quality perception research model of Steen-kamp & Van Trijp (1988). See also Wolters & VanGemert (1990) .

When purchasing a product, a consumer has cer-tain quality expectations . At the time of consump-tion, it is actually experienced : did the product pos-sess the characteristics, which were liked, or werethose disliked absent? When expectations are satis-fied a consumer will be strengthened in her/his be-lief, that s/he is buying the right product and this willincrease repeat purchases. On the other hand, if theexperience turns out dissatisfactorily, the consumerwill look for alternatives .

The model distinguished three types of causal re-lationships (Part A) . Both expected quality and ex-perienced quality can, to a large extent, be ex-plained by what consumers know - cognitive fac-tors - and by what they see, feel, smell and/or taste -sensory factors (perceptual relationships) .

In turn, cognitive and sensory factors are influen-ced by product properties - color, shape, sugar con-tents - and by non-product properties - packing,

origin of the product, outlet, price, brand name -(psycho-physical relationships) .

To improve quality expectation and quality expe-rience, production methods - cultivar used, culturalpractices, post-harvest handling - and marketingtools - grading, packing, propaganda, pricing - haveto be defined, which improve objective properties(instrumental relationships) .

The method of investigation works in the reverseorder (see B in research model) . What is the opti-mal consumer quality? What is the optimal productprofile? What is the optimal use of quality instru-ments?

Measurement

An important issue in quality research is measure-ment. This section deals with how perceptual andpsycho-physical relationships were measured, howan ideotype was constructed and quality problems

163

Page 4: Consumer-oriented quality improvement of tomatoes in Indonesia

164

formulated. Finally a method to measure willing-ness to pay is outlined .

Perceptual relationshipsMost sensory analyses on food products take placeunder controlled, standardized circumstances andare carried out with the help of panels (Meilgaard etal., 1987) . Panels have proven to be very usefulwhen detailed aromatic descriptions of flavourwere required (Martens, 1984 ; Vesseur, 1990; Janse,

1991). The argument for using panels is that analyt-ical descriptions by consumers would not be relia-ble, because they fall outside their natural behavior

(Koster,1990) . Panel training is expensive and time-

consuming .In our opinion, for simple sensory evaluation of

food aspects, like shape, color, sweet- or sournessand taste intensity, panels are not required . Con-sumers can be asked to give sensory descriptionsthemselves (Gormley & Maher, 1987 ; Steenkamp &Van Trijp, 1988; Steenkamp, 1989 ; Eytan, 1990) .

When a consumer is assessing a product, a multi-tude of aspects must be evaluated and related to itsoverall quality. With multiple regression analysisthese relationships can be quantified . However,there are two problems .

First, consumers have restricted information pro-

cessing capacity. It is unlikely, that all sensory as-pects are evaluated separately. Earlier research hasshown, that usually three to four more general di-mensions are being used (Wieringa & Van Raay,1987; Steenkamp, 1989; Hobson et al ., 1990; Schutz,

1990). With principal component analysis the un-derlying impressions can be extracted from the sen-sory descriptions, given by consumers . Selection offactors is based upon the 'eigenvalue' criterium .The eigenvalue is the sum of the squared factorloadings and must exceed one, only then the factorexplains more variance than the composing varia-bles themselves (Hair et al ., 1987) . Significant factorloadings over 0 .50 were underlined and only thesedetermined the name of the common denominator.Factor scores replaced the original sensory scores .Factor analyses were carried out separately for ex-ternal, internal and organoleptic aspects and alsoseparately for each varietal type to see whether or

not perceptual differences existed between thetypes .

Secondly, perceptions are inter-correlated . Thiscan be due to natural causes (a tomato which is ripe,

is also red) or because consumers interpret two as-pects similarly (a tomato with a green shoulder andone with a gloss are both indicators for healthiness,thus they are correlated, although there exists nocausal relationship) . Multiple regression analysislooks for a solution with as few explaining variablesas possible, which will lead to exclusion of relevantvariables with high intercorrelations . This is unac-ceptable when making an optimal product profile,which should incorporate all relevant properties .

Therefore,(a) single correlations between perceived quality

and all sensory dimensions and cognitive as-pects were calculated: the larger the correla-tion, the more significant the influence on per-ceived quality ;

(b) standardized partial regression coefficients(beta-values or path coefficients) were calculat-ed, using stepwise multiple regression analysis,to quantify, which dimension contributed mostto the explanation of perceived quality, takinginto account intercorrelations . The beta-valueof a variable is the contribution to explainingvariance of a dependent variable, after adjust-ing for the variance already explained by previ-ous variables. The higher (the absolute valueof) the beta-value, the more important the vari-able (Pedhazur, 1982) .

Psycho-physical relationshipsSensory and cognitive aspects, which contributedsignificantly to perceived quality, were `translated'into physical and chemical product properties, ob-jectively measured in the laboratory . For this trans-lation simultaneous measurement of sensory andobjective aspects was indispensable . With this,three problems arose .

First, some sensory properties were difficult tomeasure instrumentally, because laboratory analys-es were prohibitively complicated and expensive, orbecause the amount of product available for theanalysis was too little, since all destructive tests hadto be carried out with the remainder of the evaluat-

Page 5: Consumer-oriented quality improvement of tomatoes in Indonesia

ed tomato . This particularly concerned organolep-tic properties: crispiness, toughness of skin, juicin-ess and above all aromatic aspects of flavor andodor. Other properties which were difficult to mea-sure were gloss (degree of reflection) and skinsmoothness .

A solution was to train a mini panel of three lab-oratory staff to describe non-measurable propertiesin a standardized way. As much as possible nationalor international standards were used . In Indonesiastandards for fresh fruit and vegetables are virtuallynon-existent and of no practical use (Van Lieshout& Santika, 1991) . In the absence of standards newones had to be developed . Properties for which nosatisfactory standard description could be devel-oped had to be left unmeasured .

Secondly, psycho-physical relationships are gen-erally weak and correlations low, because percep-tions are not only influenced by product properties,but also by factors like education, socio-economicclass and by random circumstances, which naturallyoccur at a market place. Of practical importance isthe statistical significance of the correlation. To re-duce the probability of excluding properties whichmight be of influence, correlation coefficients wereconsidered significant at p < .10 (Steenkamp & VanTrijp, 1988) .

Thirdly, product properties are intercorrelated .As with perceptual relationships above, single cor-relations were first calculated . Significant correla-tions were cross-checked and validated by earlierresearch or in absence of literature, by expert opin-ion of a plant physiologist . Subsequently, beta-val-ues were calculated for those properties, whichwere considered causal . The higher the beta-valuethe more impact on the perceptual dimension andhence on perceived quality . R Z was also calculatedto establish how much percent of variance in a per-ceptual dimension could be explained by the causalproperties under analysis .

Quality differences between varietal typesVarietal differences were calculated for expectedand experienced quality and for each relevant per-ceptual dimension using the average factor scoresper type. Analysis of variance and a mean ranking

165

test (LSD test; p < 0.05) was used to differentiatebetween means.

Classification of propertiesBefore establishing the optimal fruit profile and de-termining existing quality problems of the varietaltypes a classification was made of product proper-ties in first class, second class and third class. Aproperty was called first class when it was consid-ered better than average by a majority of consum-ers. A property was called third class when it wasconsidered worse than average by a majority of con-sumers. When no majority existed a property wascalled second class .

The optimal fruit profileThe optimal fruit profile, or ideotype, is defined asthe hypothetical product, which is composed of firstclass properties only. Only those properties weredeemed important, which had a significant causalrelationship with an important perceptual dimen-sion. It should be stressed that dimensions with thegreatest impact on quality have the highest priority .

In order to establish the optimal level of such aproperty, psycho-physical relationships were cross-tabulated . Two rows contained the factor scores ofthe perceptual dimension : in the upper row thescores better than average, in the lower row thescores worse than average . The columns containedthe laboratory values. Subsequently, percentageswere calculated through division of frequencies bycolumn totals .'

The optimal level of a property (say, sugar con-tents) was determined by whether or not a majorityof consumers existed, who considered the productbetter (i .e., sweeter) than average at a particular

' The width of each column is important in this respect . A smallerclass, say with steps of 0.1 units, increases irregularities, due tofewer observations per column . A wider class, say with steps of0.5 units, reduces the precision . To reduce the arbitrary nature ofdividing into classes, the divider should be as small as possible,but taking into account a minimum column total . To calculatethis total, it must be noted, that with cross tabulation the expect-ed cell frequency should not be less than 5 . In this research rowtotals were approximately 100 (above and below average) andthe sample total was 200, it follows that the column total must beover 10 .

Page 6: Consumer-oriented quality improvement of tomatoes in Indonesia

166

Table la. External sensory attributes and corresponding laboratory analyses

Sensory Perception

I

External Properties

description

A

Appearance of tomatogeneral appearanceripenessfreshnessdescription of colorevenness of color

shape

sizegloss

B

Texture of tomatofirmnessskin smoothness

C

Odor (smell)strength typical odor

strong-weakstrength smell of rot

strong-weak

level. Beyond that optimal level the property inquestion was called first class .

Because of the heterogeneity of supply and de-mand, a minimum of 60% of the consumers wasdeemed sufficient to be called a majority.

Quality problemsQuality problems are third class properties . Whichthird class property to work on in a quality improve-ment program is outlined below.

First, it was calculated how much percent of eachtomato type was considered to be of `poor' or `very

Laboratory analysis

measurement on 5-point scale

range (seenotes)

good-bad

physical damage (1)

0- 3ripe-unripe

ripeness class (2)

30-100% redfresh-unfresh

days after harvest

0- 5 daysnice-not nice

CBT color chart (3)

3- 12even-uneven

green shoulder

no-yessunscalding

no-yesnice-not nice

LEHRI shape chart (4)

1- 10sphericity index (5)

60-175number of ribs

0- 10small-big

volume (6)

30-250 cm'glossy-dull

intensity of gloss (7)

1- 5

soft-hard

manually felt firmness (8)

1- 5smooth-coarse

not measured

not measurednot measured

1) degree of physical damage is calculated (mini-panel) as a sum of scores, whether (score = 1) or not (score = 0) the tomato posessed thefollowing damages : bruises, spots, insect damage, shrivels or broken skin, scars or growth cracks, blossom and rot .2) ripeness class: - pink 30-60% pinkish or red . - light red 60-90% red, - red over 90% red.3) CBT = Central Bureau for Auctions of Horticultural Products in Holland . CBI color chart: 1-2 green/yellow, 3-5 yellow-orange, 6-8orange/red, 9-10 red, 11-12 dark red .4) Lehri shape chart : 1-3 oblate, 4--6 round, 7-10 elongated (pear, oval) .5) Sphericity index = (/2(al + a2) * a3 * a4)' /a4 * 100, where al = largest diameter (size measured in centimeters) .a2 = smallest diameter. a3 = length of the diagonal . a4 = height of the tomato .6) volume = 1/6 tc * al * a2* a4 .weight in grams.7) intensity of gloss: average score of the measurement by three panelists (after having previously tested (try-out) 100 tomatoes) :1 = very dull, 2 = dull, 3 = moderate, 4 = glossy, 5 = very glossy .

8) firmness manually felt: average score of the measurement by three panelists (after having previously tested (try-out) 100 tomatoes) :1 = very soft, 2 = soft, 3 = moderate, 4 = hard, 5 = very hard .

poor' quality. Secondly, the problem levels of bothperceptual dimensions and properties were deter-mined, by establishing whether or not a majority ofconsumers considered the quality poorer than aver-age at a particular level of a property . This was donein the same way as for the optimal level . Thirdly,after having established the problem levels the per-centage of third class tomatoes was calculated . Thelarger the percentage, the greater the problem .Again it shouls be noted, that properties influen-cing dimensions with a greater impact on perceivedquality must have priority .

Page 7: Consumer-oriented quality improvement of tomatoes in Indonesia

Willingness to payIn order to assess how much more they were willingto pay for better quality (or alternatively how muchless for poorer quality), consumers were asked to

Table lb. Internal sensory properties and laborory analyses

Sensory Perception

II

Internal Properties

description

D

Appearance inner partgeneral appearancecolor of fleshquantity of fleshquantity of pitsquantity of locular gellength of core

Laboratory Analysis

Sensory Perception

Laboratory Analysis

measurement on 5-point scale

range (seenotes)

orderly-messy

number of loculi

2- 10nice-not nice

LEHRI color chart (1)

1- 9little-much

thickness of wall (2)

1- 12 mmfew-many

visible pits (3)

1- 3little-much

visible Jocular gel (4)

1- 3short-long

length of core

2- 45 mmcore length/height

1-100%E

Texture inner partfirmness of flesh

soft-firm

penetrometer (5)

10-350 mmF

Odor (smell) tomatostrength of aroma

strong-weak

not measured

167

indicate what price they were prepared to pay fortomatoes of the same quality, they just evaluated .The procedure followed was the Gabor and Gran-ger method (Steenkamp & Van Trijp, 1988) . This

1) Lehri fruit flesh color chart : 1-3 green/yellow, 4-b yellow/orange, 7-9 red/dark red .2) thickness of fruit wall : average of 3 measurements of outer fruit wall, top, center and bottom end .3) visible pits : average score of the measurement by three panelists (after having previously tested (try-out) 100 tomatoes) :1 = less than average, 2 = average, 3 = more than average .4) visible locular gel : average score of the measurement by three panelists (after having previously tested (try-out) 100 tomatoes) :1 = less than average, 2 = average, 3 = more than average .5) penetrometer : average score of 3 penetrations: near top, in the middle and near bottom end, using a 50 gram weight and 10 secondsafter plummeting.

Table 1c . Organoleptic properties of tomato

measurement on 5-point scale

range (see notes)

not measuredtotal soluble solidsugar*acid (1)glucose + fructosetotal titrable acidpHsugar/acid ratio

not measurednot measuredmoisture contents

notes : 1) sugar * acid ='J(sugar content)' + (acid content)'2) All other chemical tests applied in this research are commonly used (see Stevens, 1979 ; Janse, 1991) .3) The procedures used are thoroughly evaluated and described in Kartika, 1991 .

III Organoleptic Properties description

G Flavour/Aromastypical tomato taste strong-weaktastefulness strong-weak

sweetness sweet-not sweetsourness sour-not sour

tartness tart-not tartH Texture/Consistency

crispiness of flesh crispy-mealyrubbery skin tough-softjuiciness juicy-dry

3.1% - 6.8%0 .6 - 3.00.4% - 3.10 .20 - 0.803 .70 - 4.860.58 - 8.00

92% -96%

Page 8: Consumer-oriented quality improvement of tomatoes in Indonesia

1 68

Table 2 . Some characteristics of three types of tomatoes in Indonesia

method takes into account, that price is a cost on thehousehold budget, but also an indication for qual-ity: if it is too low in the eyes of the consumer, (s)hemay think the quality is (too) low and reject theproduct. On the other hand there exists a maximumprice, above which (s)he is not prepared to pur-chase. In order to simulate the real purchasing sit-uation as much as possible, the Gabor-Grangermethod was applied on expected quality, i.e ., after

having evaluated the product from the outside.

Eight prices in random order were given to the re-spondent, asking whether or not (s)he was willing toregularily buy tomatoes of the current quality atthat price . If not, (s)he was asked why not : was theprice too high or was the person of the opinion thatat such a price no reasonable quality could be had .Prices ranged from Rp 400 to Rp 1,800 per kilogramand were adapted to average minimum and maxi-mum prices year round per market . Intervals of Rp200 per kilogram were used, which was considered

to be a reasonable price difference .

Research design

The crux of the research is simultanuous measure-ment of consumer perceptions and laboratory val-ues. Each consumer evaluated one tomato and thesame tomato was measured objectively in the lab-oratory. Non-destructive tests were carried out onehour before and destructive tests within four hoursafter consumer evaluation . Two hundred respon-dents, representative in respect of age and incomedistribution in Javanese towns, were interviewed atsix different market places and in two supermarketsin Bandung, West Java and in Yogyakarta, CentralJava. The research took place in January 1992 . Theconsumer gave sensory descriptions for twelve ex-ternal, eight internal and eight organoleptic attri-

description

firm, oval to round, well keepablejuicy, oval to oblate, sweetsoft, pale, oblate, hard skin, sour

butes, as well as four cognitive aspects, using 5 or6-point Lickert scales. In addition, the consumerevaluated overall quality twice : expected quality atthe beginning of the interview, comparing the justreceived tomato with a representative sample ofabout fifty tomatoes, available at that particularmarket and experiended quality at the end of theinterview having thoroughly investigated and eatenthe tomato. Finally, socio-economic and user char-acteristics were asked. Table 1 reports the consumerattributes and the corresponding laboratory analys-

es carried out .

Selection of tomatoesAs a result of differences in altitude, availability ofvarieties and technical know-how between produc-tion areas an enormous natural variation exists insize, color, shape, firmness, damages, maturity, sug-ar and acidity levels . Globally three different typesof tomatoes could be discerned . See table 2 .

`Taiwan' varieties, which are modern hybridsfrom a Taiwan based breeding company, and High-land tomatoes are best grown at altitudes over 500

m a.s .l. . Lowland tomatoes are grown under 200 ma.s .l. Highland and Lowland contain both local va-rieties and officially released varieties from

AVRDC selections (Van Marrewijk, 1992) .Not always all types were available at all markets .

Therefore selection of tomatoes was manipulatedin such a way, that all three types were being eval-uated by respondents in all markets . Only ripe to-matoes with over 30% red coloring were used :green mature tomatoes, breakers and turning wereexcluded, because they are not usually consumedfresh.

tomato type market share price weight

1 . `Taiwan' varieties 50% Rp 1,000/kg 70 gr2. Highland varieties 30% Rp 800/kg 80 gr3 . Lowland varieties 20% Rp 500/kg 40 gr

Page 9: Consumer-oriented quality improvement of tomatoes in Indonesia

Results

1. The optimal consumer quality of fruit tomato

Perceptual dimensionsWith factor analysis four external, two internal andthree organoleptic perceptual dimensions werefound. Tables 3 to 5 report factor matrices . The ex-ternal dimensions used by consumers, were : `outerappearance', combining color and color distribu-tion, gloss, shape and skin smoothness; `ripeness',grouping perceived ripeness, firmness, keepability,ease of use (squashing, slicing) ; 'healthy/aromatic',combining perceived strength of aroma, healthi-ness and vitamin A and C contents ; and `size' .

The internal dimensions distinguished by con-sumers were `inner appearance', composed of or-derly appearance, nice color of flesh, strong aromaand much flesh, and 'pits-gel', referring to the per-ceived amount of pits, gel and the core length .

The first organoleptic dimension was named`typical tomato taste' . At one pole the typical flavor

Table 3 . Factor matrix for external sensory tomato aspects afterVarimax rotation (n = 200)

Table 4 . Factor matrix for internal sensory tomato aspects afterVarimax rotation (n = 200)

Factor name :

orderly inner appearancenice color of fleshstrong aromamuch fleshfew pitsno locular gelshort corefirm fleshcumulative variance

169

characteristics: tasteful, aromatic, crispy . At theother pole the off-flavors : mealy, tart, little taste .The second dimension was called 'juicy/no toughskin', combining sensory juiciness and skin tough-ness: the tougher the skin, the drier the tomato . Thelast factor yielded a 'sweet/not sour' dimension .

This needs some elaboration . Earlier researchhas shown that each of the four sweet/sour combi-nations has its own flavor characteristic : low sweet-ness/low acidity produces a tasteless, insipid toma-to. Contrarily a high sweetness/high acidity tomatois well-tasting and fruity, sweet/sourish. The othercombinations are low sweetness/ high acidity, whichgives a tart taste and high sweetness/low acidity,which gives a bland taste (Grierson & Kader, 1986) .The last combination (tart versus bland) coincidedbest with the dimension found . Apparently, Indo-nesian consumers associated high sweetness with

Table5 . Factor matrix for organoleptic tomato aspects after Vari-max rotation (n = 200)

Factor name: Outerappearance

Well

Healthy/ Sizekeepable/ aromaticripe

nice color 0 .74 0 .18 0.19 -0.08smooth skin 0 .69 0 .01 0.03 -0.08fresh 0 .66 0 .20 0 .24 -0.04even coloring 0 .64 -0.16 -0.18 -0.15niceappearance 0.63 -0.01 0 .32 0.00gloss 0 .61 -0.12 -0.06 -0.10nice shape 0 .53 0 .04 -0.38 -0.29ripe 0 .44 0 .66 0.02 -0.07easy to use -0.04 0 .65 0.01 0.11firm -0.48 -0.58 -0.13 -0.02well keepable -0.10 -0.57 -0.16 0.20strong aroma -0.01 0 .01 0.74 0.20healthy 0 .39 -0.14 0.51 -0.20a lot ofvitamines 0 .17 0 .18 0.65 -0.20small size -0.23 0 .06 -0.08 0.71no smell of rot -0.28 -0.17 0.16 -0.49cumulativevariance

26% 37% 45% 51%

Factor name : Typical tomato Juicy/ Sweet/not sourtaste tough skin

tasteful 0 .77 0 .10 0.05aromatic flavor 0.63 0 .14 -0.07crispy flesh 0 .52 -0.08 0.19tart -0.51 0 .11 -0.21juicy -0.00 0 .77 -0.15tough skin 0 .03 -0.76 -0.08sweet 0 .37 -0.09 0.70sour 0 .08 0 .07 -0.89cumulativevariance 23% 40% 53%

Inner appearance Pits/gel

0.87 0 .120.82 -0.050.76 -0.220.60 0 .300.14 0 .76

-0.05 0.69-0.04 0 .46

0.27 0 .3031% 49%

Page 10: Consumer-oriented quality improvement of tomatoes in Indonesia

170

Table 6 . Single correlations (r) and beta-values (3) between ex-pected quality of tomato and cognitive and external sensory fac-tors- the underlined word indicates the preferred direction . (n =200, p < 0 .05, R2 = 0 .41)

External Sensory Dimension

r = (3outer appearance (nice/not nice)

0.59healthy/aromatic (healthy/strong/not healthy/weak 0 .25ripeness (rte/unripe

0.05 (ns)size (big/small)

0.04 (ns)

note : single correlations and beta-values are equal, because thefour factors are independent (orthoganal) . ns = not significant .

low acidity and vice-versa. The intercorrelation of r_ .378 (p < 0.001) confirmed this .

Separate factor analyses for each type of tomatoyielded similar results. It could thus be concludedthat consumers looked at tomatoes irrespective ofthe type (Van Lieshout, 1992a) .

Expected qualityQuality expectations of tomato were related toboth sensory and cognitive aspects (Table 6) . Over40% of variance could be explained: quality expec-tations were raised by nice outer appearance and byperceived healthiness . The other two factors, per-

ceived ripeness and size did not contribute to qual-ity expectations . However, ripeness and size as well

Table 7. Single correlations (r) and beta-values ((3) between ex-perienced quality of tomato and external, internal and organo-leptic dimensions - the underlined word indicates the preferreddirection (n = 200, p < 0 .05, R2 = 0 .60)

note : ns = not significant .

as outer appearance played significant roles withtomato used for 'bumbu' (Van Lieshout, 1992a) .

Experienced qualityExperienced quality correlated significantly with 6out of 9 dimensions (Table 7) . This makes clear, thatquality experience is indeed a complex issue . How-ever, of total variance 60% could be attributed tovariance in inner appearance, outer appearance,perceived sweet-sourness and typical tomato taste .

Looking at the beta-values, it appears, that typ-ical tomato taste is now excluded, because of its ve-ry high correlation with inner appearance (r =0.583). In other words if a tomato looks well fromthe inside, its typical tomato taste is strong . If on theother hand the tomato looks bleak and messy, withthin fruit walls and no aroma, its taste is perceivedas weak, insipid . The optimal tomato has a nice in-ner and outer appearance, is sweet, not sour, has lit-tle pits-gel substance and looks healthy with astrong aroma and high vitamin contents.

Quality differences between varietal typesThe average scores for expected and experiencedquality of tomatoes are given in Table 8 `Taiwan'varieties were considered to be of above average togood quality. The average score for experiencedquality was less favorable (-0.30) than for expectedquality. In other words `Taiwan' varieties could notmeet the high expectations raised by external per-ceptions and cognitive beliefs . On average High-land varieties were of a more than reasonable qual-ity, yet of similar experienced quality as `Taiwan' to-matoes: above average. Lowland varieties werefound bad for fresh fruit consumption in terms ofexpected and experienced quality . For the lattertypes no significant difference between expectedand experienced quality was found .

Table 9 shows differences in perceptions betweenthe three types of varieties . With regards to expect-ed quality, `Taiwan' ranks first on outer appearancewich is the principle dimension . Highland varietieswere perceived the healthiests ; Lowland varietiesscored last on both dimensions .

Quality experience (which is dominated by innerand outer appearance, sweetness and strong tomatotaste) is similar for `Taiwan' and Highland toma-

External Sensory Dimension r G3outer appearance (nice/not nice) 0 .58 0.25healthy/aromatic (healthy/strong/

not healthy/weak) 0 .31 0.14

ripenes (ice/unripe) -0.11 (ns) -0.03 (ns)size (big/small) 0 .04 (ns) 0.08 (ns)

Internal Sensory Dimensioninner appearance (nice/not nice) 0 .68 0.35amount of pits-gel (little/much) 0 .21 0.16

Organoleptic Dimensionsweet-not sour (sweet-not sour/not

sweet-sour) 0 .47 0 .21typical tomato taste (strong /weak) 0 .42 0.08 (ns)juice-skin (juicy-not tough/dry-

tough) - 0.07 (ns) - 0 .06 (ns)

Page 11: Consumer-oriented quality improvement of tomatoes in Indonesia

Table 8. Expected and experienced quality of tomato varieties used for fruit in Indonesian towns : 1 = very bad, 3 = reasonable, 6 = very

good (n = 200)

note : different letters indicate significantly different means between varieties (LSD-test : p < 0 .05) .

toes . Only on perceived amount of pits and gel,Highland varieties scored less . Again Lowland to-

mato experience is behind on all dimensions . Thisranking fully explains the differences between va-rieties in expected and experienced quality (Table

8) .

2. The optimal fruit profile

The optimal tomato has high expected and experi-enced quality. In the previous section it was de-scribed how consumers arrived at their qualityopinion, which perceptions are important . This sec-tion deals with a translation of perceptions into ob-jective parameters, a classification of relevant par-ameters into first, second and third class levels andfinally the construction of the optimal fruit profile .

Psycho-physical relationshipsSignificant single correlations between sensory and

cognitive dimensions and product properties are re-ported in Table 10 . Most product properties influen-ced more than one dimension, although not all cor-relations, had a causal basis. Causal relationshipswere either validated by literature (e.g ., Stevens,

Table 9. Perceptual differences between three types of tomato varieties in Indonesia . (n = 200, p < 0.05)

171

1979; Goodenough & Atkin, 1981 ; Janse, 1987 ;

Gould, 1983; Hobson & Kilby, 1984; Stevens, 1985 ;Blanc, 1986 ; Diez et al ., 1986; Stevens & Rick, 1986a ;Van Vuurst-de Vries, 1986 ; Jordan et al ., 1990; Gori-ni & Testoni, 1990) or by Indonesian tomato expertsat LEHRI, Lembang, Indonesia (Sinaga, Asgar,Budi Jaya, Purwati, Sumiati) .

Table 11 shows the beta-values of the significantcausal relationships between properties and per-ceptual dimensions significantly affecting expectedand experienced quality .

Pl-outer appearance ; 37% of variance could beexplained. It was primarily enhanced by a full redskin color. Further volume and gloss contributedpositively. The significant quadratic effect of spher-icity indicates, that the optimal shape is round (in-dex 90-120); elongated (<_ 90) and oblate (>_ 120) to-matoes are preferred less. The signifiicant quadrat-ic effect of firmness indicates, that when a tomato ishard (manually felt firmness class 4) its outer ap-pearance is best; when a tomato is perceived mod-erately hard (class 3) or very hard (class 5) its ap-pearance is considered moderate ; when a tomato issoft or very soft (classes 2 and 1 respectively) its out-er appearance is poor. Severity of visible damagescontribute negatively. Ripeness and color of flesh,

note : Letters indicate preferred ranking according to the consumer : a = highest, c = lowest . Equal figures indicate no significantperceptual difference existed between the types of varieties (LSD test ; p < 0 .05) .

variety average expectedquality

average experienced difference expected-experienced qualityquality

`Taiwan varieties 4 .44 a 4 .14 a - 0 .30 (p = 0 .09)Highland local varieties 3 .79 b 3 .83 a 0.04 (n .s .)Lowland local varieties 1 .88 c 1 .86 b 0.02 (n .s .)All varieties supplied 3 .72 3 .58 -0.14 (n .s .)

P1 outerappearance

P3 healthy/aromatic

P5 innerappearance

P6 pits & gel P7 strong taste P9 sweet/not sour

`Taiwan' 0 .45 a 0 .06 b 0.34 a 0.27 a 0 .19 a 0.31 aHighland 0 .04 b 0 .30 a 0.30 a -0.28b 0 .28 a 0 .12 aLowland - 0 .78 c - 0.33 c - 0.79 b -0.24b -0.52b -0.61b

Page 12: Consumer-oriented quality improvement of tomatoes in Indonesia

172

Table 10 . Significant single correlations between product properties and perceptual dimensions. Causal relationships are underlined .Properties with more than one causal relationship are marked with an asterisk . (n = 200, p < 0.10)

EXTERNAL PROPERTIES

despite having a high single correlations, no longercontributed, when skin color was adjusted for .

P3-perceived healthiness/strength of aroma ofthe tomato could be related to its gloss and having agreen shoulder, as well as to its moisture contents .With these laboratory measurements only 7% ofvariance could be explained.

P5-inner appearance perceptions, the most im-portant dimension, can be improved significantly

by a full red color of flesh and a thick fruit wall . It isfurther enhanced by reduction of the number of loc-uli and the severity of visible damages .

P6-perceived amount of pits and gel is correlatedto the amount of pits and gel measured by the mini-panel. It should be noted, that despite being signif-icant at p < .001, R2 is relatively low, 10% . This iscould be due to inaccuracies of measurement . An-other reason could be, that neither absolute corelength nor relative core length (length as a percent-

age of the distance between stem and corona) couldbe related to this dimension .

P7-typical tomato taste (tasteful, aromatic, cris-py, not tart) can be made stronger by an improvedred color, by a thicker fruit wall and by high valuesof sugar*acid. Reduction of the severity of visibledamages also improves the taste . Total soluble sol-

ids dropped out of the equation, because of the highcorrelation with sugar x acid (r = 0 .60, p < .001) . It

must be noted, that all beta-values are relativelylow (p < .10) and consequently R 2 is only 11% . This isprobably due to the fact, that aromatic componentshave not been measured .

P9-perceived sweetness/not sourness could onlybe explained for 16% . It is improved by a high sugar/acid ratio, a high pH, high solids contents . Even af-ter having adjusted for sugar/acid ratio, ripeness im-proved the taste, and hence the quality of fruit to-mato .

1 volume 0 .27 0.34 0.192 sphericity -0.38 -0.38 -0.16 -0.16 -0.363 color of skin 0.41 0.31 0.18 0 .20

* 4 ripeness 0 .41 0 .32 0.22 0.145 ribs -0.28

* 6 visible damages -0.20 -0.17 -0.187 firmness manual -0.23 0 .22 0.29 0.248 green shoulder 0 .16 0 .19 0.12 0.17

* 9 gloss 0 .38 0 .18INTERNAL PROPERTIES

0.42 0 .38 0 .22 0.25* 10 color flesh* 11 wall thickness 0 .34 0 .38 0.14 0 .19 0.3212 loculi -0.28 -0.23 -0.18 -0.14 -0.3913 penetration -0.2414 pits -0.24 -0.2215 Jocular gel -0.14 -0.1216 core length 0 .17 -0.22 0 .23 0 .20

CHEMICAL PROPERTIES0.23 0 .31 0 .23 0 .23* 17 sugar%

* 18 sugar/acid 0 .25 0 .13 0 .19 0 .2819 acid% -0.18 -0.17 -0.2420 pH 0.14 0 .26

* 21 soluble solids 0 .12 0 .14 0 .1922 sugar*acid 0 .26 0 .21 0 .1823 moisture% -0.16

P1 better P3 healthier/ P5 better P6 less pits & P7 stronger P9 sweeter/outer app. aromatic inner app. gel

taste less sour

Page 13: Consumer-oriented quality improvement of tomatoes in Indonesia

It should be noted, that sugar contents should beas high as possible (within the ranges measured, i .e .,0.4-3.1%). But acidity has an optimum : to achieve

173

Table 11. Significant beta-values and R 2. Perceptual dimensions regressed stepwise on properties with causal relationships (n = 200, p <

Note : a significant quadratic effect indicates, that the relationship is not lineair, but curvi-lineair, with an optimum value (or range) .*** p < 0 .001, ** p < 0 .01, * p < 0 .10 .

high values of sugar x acid, total titrable acids mustbe high, but to obtain high values of sugar/acid ra-tio, total titrable acids must be low. However, a cur-

Table 12. Percentage of consumers considering the TASTE of tomato SWEETER or LESS SWEET than average and correspondinglevels of SUGAR CONTENTS measured by the percentage of glucose + fructose (see also footnote 2)

0.10)

PI betterouter app.

P3 healthier/aromatic

P5 betterinner app .

P6 less pits & P7 stronger P9 sweeter/less sourgel taste

EXTERNAL PROPERTIESI volume 0.15*2 sphericity - 0.19**

spheri quadratic 0 .20**3 color of skin 0.35***4 ripeness 0.14*6 visible damages - 0.15* - 0.14* - 0.14*7 manual firmness 0.17**

firmn quadratic 0 .118 green shoulder 0 .13*9 gloss 0.16** 0.15*INTERNAL PROPERTIES10 color flesh 0 .26*** 0 .14*11 wall thickness 0 .26*** 0 .1112 loculi 0.1113 penetration14 pits -0.29***15 locular gel - 0.21**CHEMICAL PROPERTIES18 sugar/acid ratio 0 .22**20 pH 0.23**21 soluble solids 0.13*22 sugar*acid 0 .13*23 moisture% -0.15*R2 37%*** 7%** 34%*** 10%*** 11%*** 16%***

Classification of sugar contents

Third class Second class First class n

Range of Sugar 0.40 0 .70 1 .00 1 .10 1 .20 1 .30 1 .40 1 .50 1 .60 1 .70 1 .80 1 .90 2.20Contents (%) 0.69 0.99 1 .09 1 .10 1 .39 1 .49 1 .49 1 .59 1 .69 1 .79 1 .89 2.19 3.49Considering TasteSweeter 4 6 7 8 7 1 7 8 12 11 9 7 14 99

26.7% 33.3% 36.8% 50.0% 46.7% 8.3% 43.8% 57 .1 83 .3 73.3% 69.2% 42.8% 73.6%Considering TasteLess Sweet 11 12 12 8 8 11 9 6 2 4 4 9 5 101

73.3% 66.7% 63.2% 50.0% 53.3% 91 .7% 56.2% 42.9% 16.7% 26.7% 30.8% 57.2% 46.4%Number ofConsumers 15 18 19 16 15 12 16 14 12 15 13 16 19 200

Page 14: Consumer-oriented quality improvement of tomatoes in Indonesia

174

Table 13 . Classification of tomato properties accoding to consumer perceptions of relevant sensory dimensions and cognitive factors . Theproperties marked with an asterisk are used to compose the optimal fruit profile

P1 - OUTER APPEARANCE•

color (CBT chart)•

volume (cm')•

sphericity index•

visible damages•

number of ribs•

gloss•

firmness (manually tested)•

penetration (mm/50 g/10 sec)'

P3 - HEALTHY/AROMATICgloss

green shoulder•

moisture contents

Note 1. n .a . means that no 60% majority of consumers existed, who liked or disliked tomato at any level of that property .2 . penetration has been included, because it is an easily verifiable measurement for firmness, it basically serves to calibrate the mini-panel's manual testing .

vi-linear effect between acidity and taste qualitycould not be found. This is probably so, because theeffect of acidity is greater in sugar/acid ratio than itseffect in sugar x acid .

Classification of propertiesOf the significant causal relationships above, theoptimal level and the problem level of physical andchemical properties were determined . Table 12shows an example of how this quality classificationwas arrived at .

When sugar contents was higher than 1.6%, anaggregated majority of 68% of consumers consid-ered the tomatoes sweeter than average and conse-quently of better quality. This is the optimal level .Sugar contents above 1.6% is therefore first class .

Tomatoes with a sugar contents below 1.1% wereconsidered less sweet, hence worse than average byan aggregated 67% of consumers. The problem lev-el is 1.1% . A tomato with less than 1.1% sugar is

INNER APPEARANCEcolor of flesh (LEHRI-chart) > 6

<_ 3thickness fruit wall (mm)

?6 mm

< 2 mmnumber of loculi

2-3

? 7visible damages

0

? 3AMOUNT OF PITS & GELvisible pits

< 2

? 3locular gel

1

n.a .TYPICAL TOMATOTASTEglucose + fructose contents ? 2%

<- 1 %color of flesh (LEHRI-chart) >- 7

<- 3visible damage

0

>- 2.5thickness fruit wall (mm)

n.a .

< 3 mmP9 - SWEET/NOT SOUR

sugar/acid ratio

>- 4.5

< 2.5glucose + fructose contents >_ 1.6%

< 1 .1•

total titrable acids <_ 0.3% >- 0 .6%•

pH >_4.2 <4.0•

total soluble solid contents ? 4 .6 <- 3.6

called third class. In between 1 .1% and 1 .6% a toma-to may be referred to as second class .

Columns 7 and 13 are illustrative for problemsdue to small samples (see footnote number 2) . Thepsycho-physical relationship is weakened by this ir-regularity (r = 0 .23). However, it must be remem-bered, that the importance lies with the statisticalsignificance of the relation . This significance was al-ready tested in Table 11 (p < .01). We can thereforesafely conclude, that the classification is valid, de-spite the irregularities .

The optimal fruit profileIn Table 13 all classifications of relevant propertiesare given . Class boundaries did not always coincide .The optimal sugar contents for a typical tomatotaste is 2%, for a sweet/not sour taste 1 .6% is suffi-cient. Which level should be taken as optimal? Theanswer is 2%, because both will be first class . A sim-ilar note can be made for the problem level .

optimal problem optimal problemlevel level level level

P5 ->_ 9 <_ 5

*?80 <_ 40

*90-120 ? 120

*0 ? 2

n.a.' ? 9

P6 -> 4 < 2

*4 < 2

*<--100 ? 160

P7

> 4 n.a .yes n .a .<- 94% > 96%

Page 15: Consumer-oriented quality improvement of tomatoes in Indonesia

3. Quality problems

Table 14 shows a ranking of quality problems of per-ceptual dimensions and a ranking of product prop-erties .

`Taiwan' tomatoesOf the `Taiwan' tomatoes researched 28% was con-sidered too sour or insufficiently sweet . This couldbe explained by the fact, that 19% of the investigat-ed `Taiwan' tomatoes had a sugar/acid ratio lessthan 2.5. This was due to high acidity (over 0 .6%),

which occurred in 18% of the tomatoes, rather than

Table 14b. Ranking of product properties of three tomato types, according to the percentage in class III

to low sugar level (less than 1 percent), which wasonly encountered in 5 % of all cases .

The second quality problem was that 22% of thetomatoes were considered to have a poor outer ap-pearance (albeit being the best of the three varietaltypes researched) . This could be related to the fol-lowing class III properties : 24% had visible damag-es, 22% were of a yellow/orange color, 18% had ableak flesh color and 16% were pink (not fully ripe) .

The third problem, a lot of pits and gel (20%) wasclearly due to a lot of locular gel (34%) and not totoo many pits (6%) .

The fourth problem was that 20% of `Taiwan' to-

175

Table 14a. Ranking of perceptual dimensions of three tomato types, according to the percentage of tomatoes considered to be of `poor' or`very poor' quality when experienced for fruit

`Taiwan' n = 99 Highland n = 42 Lowland n = 59

1 a lot of locular gel 34% 1 unfirm manually 38% 1 low sugar/acid ratio 862 many visible damages 24% 2 a lot of locular gel 33% 2 unfirm manually 61%3 yellow/orange color 22% 3 many visible damages 24% 3 oblate shape 58%4 low sugar/acid ratio 19% 4 thin fruit wall 21% 4 thin fruit wall 545 bleak color of flesh 18% 5 high penetration 19% 5 high titrable acidity 51%6 low soluble solids 18% 6 low pH 14% 6 low glucose + fructose 467 high titrable acidity 18% 7 bleak color of flesh 14% 7 no gloss 45%8 not fully mature 16% 8 yellow/orange color 14% 8 bleak color of flesh 41%9 low pH 14% 9 low soluble solids 14% 9 yellow/orange color 39%

10 small 14% 10 low sugar/acid ratio 12% 10 not fully mature 39%11 high moisture contents 10% 11 small 10% 11 many visible damages 37%12 unfirm manually 9% 12 many pits 10% 12 many loculi 37%13 no gloss 9% 13 oblate shape 10% 13 low pH 36%14 thin fruit wall 6% 14 high moisture contents 5% 14 small 34%15 high penetration 6% 15 low glucose + fructose 5% 15 many pits 24%16 many pits 6% 16 not fully mature 5% 16 low soluble solids 20%17 low glucose + fructose 5 % 17 no gloss 2% 17 high penetration 17%18 oblate shape 4% 18 high titrable acidity 0% 18 high moisture contents 10%19 many loculi 3% 19 many loculi 0% 19 a lot of locular gel 5%

`Taiwan' n = 99 Highland n = 42 Lowland n=59

poor experienced quality 28% poor experienced quality 33% poor experienced quality 90%

1 not sweet/too sour 28% 1 a lot of pits and gel 43% 1 not sweet/too sour 69%2 poor outer appearance 22% 2 not healthy/not aromatic 24% 2 poor inner appearance 443 a lot of pits and gel 20% 3 poor inner appearance 13% 3 poor outer appearance 41%4 not healty/not aromatic 20% 4 poor outer appearance 12% 4 insipid taste 36%5 insipid taste 11% 5 not sweet/too sour 12% 5 not healthy/not aromatic 30%6 poor inner appearance 6% 6 insipid taste 10% 6 a lot of pits/gel 21%

Page 16: Consumer-oriented quality improvement of tomatoes in Indonesia

176

Rpl.400

Rpt,200

0 Rpl.000

N

Pl • Rp600

Rp400

Rp800

Rp200 -_ i._ .

"too expensive"

"acceptable price range"

"too cheap"

very poor (n = 32)

reasouable (u =14')

good (ii=64)poor (u=46)

above average (n=21)

very good (n=23)Expected Quality of Fruit Tomato

Fig. 1. Highest and lowest acceptable price for different levels of expected quality of fruit tomato .

matoes are not perceived as healthy and aromatic .The only significantly related property with a prob-lem level was moisture contents (over 96% mois-ture) in 10% of the tomatoes. As stated before, per-ceived healthiness/aroma could not be attributedfurther, because the tests carried out on aromaticsubstances were limited .

From this analysis the following priority list canbe made :•

improve taste by:reduction of the acidity level, ultimately to 0 .3% .•

improve the outer appearance by:reducing visible damages to zero, both pre- andpost-harvest damages ;enhancing red color of both skin and flesh to at-tain red value 9 on the CBT scale .•

improve internal consistency by :reduction of locular gel to `a little bit', which isconsidered optimal .

Highland tomatoesAccording to consumers 43% of the Highland to-matoes surveyed, had many pits and a lot of loculargel, which is not liked when used for fruit . As with`Taiwan' tomatoes this is primarily attributable to

too much locular gel (33 %) and not to too many pits(10%) .

Healthiness/Aroma was considered poor/weakfor 24% of Highland tomatoes . Like in the case of`Taiwan' tomatoes, this problem could not be relat-ed to laboratory parameters, since only 5% of theHighland tomatoes had a high moisture contents .

From this analysis the following priority list canbe made :

To increase firmness by :reducing locular gel to `a little bit', which is con-sidered optimal .

Lowland tomatoesPerceived sweetness is the most hampering factorfor Lowland tomato quality: 69% of consumers ex-perienced it as sour, insufficiently sweet . This wasdue to a low sugar/acid ratio (86% of investigatedtomatoes!), which was affected by both high acidity(51%) and low sugars (46%) .The second problem was poor inner appearance

(41%) . This could be related to a thin fruit wall(54%), bleak flesh color (41%) and many loculi(37%) .The third problem was poor outer appearance

(40%), due to their oblate shape (58%), visible

Page 17: Consumer-oriented quality improvement of tomatoes in Indonesia

damages (37%), yellow/orange/pink color (39%)and insufficient ripeness (39%) .

From this analysis the following priority list canbe made :•

improve taste by:increasing sugar/acid ratio by both reducingacidity and increasing sugar contents ;harvesting tomatoes when they are fully mature .•

improve inner appearance by :enlarging thickness of fruit wall ;enhancing red color of flesh to full red ;reduction of the number of loculi ;minimizing visible damages, both pre- and post-harvest damages .•

improve outer appearance by :enhancing red color of skin to full red ;creating a round tomato .

4. Willingness to pay

What is a reasonable price range for fruit tomatoes ingeneral?Fig. 1 shows, that a price of upto Rp 1,000 per kilo isacceptable for a majority of Indonesian consumers .At higher prices the purchase intention rapidly de-

wOy

4-r

100%

40%

0%

willing to purchase

Rp400 Rp600

clines, due to a growing percentage of consumersconsidering the price too expensive . It is interestingto note, that price is indeed seen as a quality indica-tor: at Rp 400 per kilo over 40% of the consumersthink that the quality is too poor. Hence a tomatoprice between Rp 400 and Rp 1,000 is deemed rea-sonable by a majority of consumers .

What is an acceptable price range in relation to ex-pected quality?The results above can be related to the scores of ex-pected quality. It can be assumed, that quality influ-ences willingness to pay. Fig. 2 shows the averagemaximum price acceptable to consumers (aggregat-ed over all varietal types) and the average accept-able minimum price per quality level . It appears,that especially the upper level of the acceptablemaximum price is influenced by perceived quality :for very poor perceived quality the average maxi-mum price is about Rp 700 per kilo, for reasonablequality it is Rp 1,000 and for very good quality con-sumers are willing to pay an average of Rp 1,200 perkilo. The average minimum price hardly changes . Itfluctuates between Rp 500 and Rp 600 per kilo . Itcan thus be concluded, that the range of acceptabil-ity increases with improved quality.

In Bandung and Yogyakarta (n = 200)

R p800 Rp1,400Rpl,000

Rp1,200Price per Kilo

V2J not willing, quality too low )d not willing, too expensive

Fig. 2. Percentage of consumers willing or not willing to pay fruit tomatoes at different prices per kilo .

Rp1,600 Rpl,800

177

Page 18: Consumer-oriented quality improvement of tomatoes in Indonesia

1 78

Table 15. Result of stepwise regression analysis on maximumprice willing to pay per kilo tomatoes and perceived expectedquality and consumers' income level as explaining variables (n =200)

Note : expected quality levels: 1 = very low, . . ., 6 = very high .income levels : 1 = very low, . . ., 7 = very high .

What is the extra value of improved quality andwhich factors influence this?The maximum price consumers were willing to payfor the tomatoes of the quality they just evaluated,was explained with multiple regression analysis, us-ing perceived quality and socio-economic and usercharacteristics of the consumer as independent var-

iables. Table 15 shows the results. There existed a

very significant relationship between the maximumprice consumers were willing to pay and perceivedquality. For each `step', e.g., from `reasonable' to`good', consumers were willing to pay about 10%more, or an average of Rp 87 per kilo extra . In addi-tion to perceived quality, income level of the con-sumer positively influenced willingness to pay .Higher income consumers would not mind payingabout Rp 20 extra per kilo of tomatoes for each in-come step. The interpretation can also be reversed :if quality or income are decreased by one step theconsumers were willing to pay Rp 87 and Rp 21 perkilo less .

5. Conclusion

It can be concluded that tomato quality, as judgedby consumers, can be explained by their subjectiveperceptions of external, internal and organolepticproperties. With this study it has been shown thatthese subjective perceptions are indeed objectivelymeasurable. It has also been shown that consumersare prepared to pay a higher price for improvedquality. This opens up commercial possibilities forbreeders and growers .

The aforementioned success of `Taiwan' toma-toes is fully supported by the results of this study .Their superior perceived quality explains that, de-spite their higher price, `Taiwan' tomatoes are wellsold. They possess, in the eyes of the consumers,better characteristics than Highland and Lowlandtomato varieties: better outer and inner appear-ance, a strong, sweet taste and little locular gel . Nev-ertheless, `Taiwan' tomatoes can still be improvedfurther. It appeared that the quality experience wassignificantly lower than quality expectations . Tosafeguard success in the future `Taiwan' breedersand growers must tackle the major problem relatedto quality experience : a sweeter taste by reducingthe acidity level.

An issue not addressed in this study is the grower .Although, it is strongly recommended that agricul-tural researchers, especially breeders, acquire moreexperience with product quality research orientedat the final consumer, it is obvious that the commer-

cial success of breeding companies lies in the handsof their clients : the growers.

Therefore it would be particularly interesting tobreeding companies to carry out additional re-search among the `seed consumers' . What are theirquality expectations when purchasing certainbrands of seed? Which objectively measurable fac-tors influence their cognitive and sensory percep-tions? How does the quality of the full grown prod-uct influence the farmer's perceptions of seed qual-ity?

It is hoped that this paper has made clear tobreeders, agronomists and post-harvest techniciansthat consumer preferences are indeed objectivelymeasurable and can serve as a guideline for breed-ing programs and other agricultural research .

Acknowledgements

This research was only possible thanks to the col-laboration of my fellow researchers Mrs . MiekeAmeriana, Mr . Budi Jaya, Mrs. Etti Purwati andMr. Ali Asgar of the Lembang Horticultural Re-search Institute at Lembang, Indonesia . It wasfunded by the Directorate General of InternationalCooperation of the Ministry of Foreign Affairs of

explaining variables range b beta

proba-bility

expected quality 1-6 Rp 87 0.53

< 0.001income level 1-7 Rp 21 0.14 0.019intercept Rp 592 < 0.001

Page 19: Consumer-oriented quality improvement of tomatoes in Indonesia

the Netherlands . Our sincere thanks to Gijsbert van

Marrewijk, Vakgroep Plantenveredeling, Agricul-tural University, Wageningen ; Jan-Benedict Steen-kamp, Vakgroep Marktkunde en Marktonderzoek,Agricultural University, Wageningen ; Nol Verhaag,Agricultural Economic Institute, The Hague . Final-ly, we acknowledge the use of an excellent statisticalsoftware package, MSTAT-C, version 1 .4 by RussellD. Freed and Scott P. Eisensmith of Michigan StateUniversity, USA .

References

Ameriana, M ., W.G. Koster & A . Asgar, 1991 . Grading dalampemasaran bawang merah . Internal Communication LEHRI/ATA 395, No. 27 .

Ameriana, M., 1992. Perbaikan Kualitas Tomat untuk Bumbu .LEHRI paper (unpublished) .

Blanc, D ., 1986 . The Influence of cultural practices on the pro-duction in protected cultivation with special reference to to-mato production . Acta Horticulturae, No . 191 : 85-98 .

Buurma, J .S ., 1992. Personal Communication, LEHRI, Lem-bang, Indonesia.

Diez, et al ., 1986 . Effect of hollowness and vascular necrosis ontomato quality. Acta Horticulturae, No . 191 : 229-235 .

Eytan, E ., 1990 . Tomato tasting test. Acta Horticulturae, No . 259 :79-82.

Goodenough, P.W. & R .K . Atkin, 1981 . Quality in stored & pro-cessed vegetables and fruit: proceedings of a symposium heldat Long Ashton Research Station, University of Bristol, 1979 .Academic Press, London .

Gormley, R. & M. Maher, 1987. Tomato flavour: taking tastetests to the supermarket . The Grower-UK, No. 107, Vol . 21 :29-31 .

Gorini, F & A . Testoni, 1990. The relation between colour andquality of vegetables . Acta Horticulturae, No. 259 : 31-60.

Gould, W.A., 1983 . Tomato production, processing and qualityevaluation 2nd ed .) . AVI, Westport .

Grierson, D. & A .A. Kader, 1986. Fruit riping and quality. In :Atherton, J .G. & Rudich, J . (Ed .), The Tomato Crop. Chap-man and Hall, London, pp. 241-280.

Hair, J .E, R .E. Anderson & R . Tatham, 1987. Multivariate DataAnalysis . MacMillan Publishing Co., New York .

Hobson, G .E . & P Kilby, 1984 . Rapid assessment of tomato com-position during high quality fruit production and distribution .Acta Horticulturae, No . 163 : 47-54 .

Hobson, G.E ., C. Gough & C. Townley, 1990 . Measuring con-sumer reaction to the flavour of fresh tomatoes Acta Horticul-turae, No . 259 :107-116.

Janse, J . & Rudich, J . 1987. Relatie tussen suikergehalte en houd-baarheid bij tomaat? Prophyta-NL, No . 41, Vol . 3 : p 80.

Janse, J ., 1991. Werkwijze in smaakonderzoek bij tomaat op

179

PTG. Fax sent on 24 Sept ., 1991, from PTG, Naaldwijk, TheNetherlands .

Jordan,J.L ., R.L. Shefelt, J .C. Garner & J.N. Variyam, 1990 . Esti-mating the value of internal quality aspects to consumers . Ac-ta Horticulturae, No . 259 :139-144.

Kartika, I .,1991 . Pengaruhan varietas, penambahan tepung mai-zena dan lama pengetalan terhadap mutu pasta tomat. Mas-ter's Thesis . Agricultural Institute, Bogor, Indonesia .

Koster, E .P, 1990. Sensory analysis and consumer research inproduct development . Acta Horticulturae, No. 259:13-23.

Lieshout, J .A.O. van & A. Santika, 1991 . Quality Standardiza-tion of fresh vegetables . Internal Communication LEHRI/ATA 395, No.34 .

Lieshout, J .A.O. van, 1992a . Consumer Oriented quality im-provement for tomato used for fruit . Internal CommunicationLEHRI/ATA 395, No . 46 .

Lieshout, J.A.O. van, 1992b. Consumption of fresh vegetables inIndonesia . Internal Communication LEHRI/ATA 395, No .48 .

Marrewijk, G .A.M. van, 1992 . Personal Communication, LEH-RI, Lembang, Indonesia .

Martens, M ., 1984 . Quality and quality evaluation . Acta Horti-culturae, No. 163 :15-30 .

Meilgaard, M ., G . V. Civille & T. Carr, 1987 . Sensory EvaluationTechniques . CRC Press Inc ., Boca Raton, Florida:

Pedhazur, E.J., 1982 . Multiple Regression in Behavorial Re-search (2nd ed .) . Holt, Reinhart and Winston, Inc ., Fortworth,Texas, USA .

Potts, M. & A.P. Chilver, 1992. Personal Communication, CIP/LEHRI, Lembang, Indonesia .

Schutz, H.G ., 1990 . Measuring the relative importance of senso-ry attributes and analytical measurement to consumer accept-ance . Acta Horticulturae, No . 259 :173-174 .

Steenkamp, J .E.B.M. & J .C.M. van Trijp, 1988 . Onderzoek naarkwaliteitperceptie als research guidance voor produktontwik-keling . Wageningse Economische Studies No . 10, Landbou-wuniversiteit, Wageningen, The Netherlands .

Steenkamp, J .E .B.M ., 1989 . Product Quality. Van Gorkum, As-sen, The Netherlands .

Stevens, M.A ., 1979 . Tomato Quality: potential for developingcultivars with improved flavor . Acta Horticulturae, No. 93 :317-329.

Stevens, M .A .,1985 . Quality of tomatoes . In : Pattee (Ed .), Eval-uation of the Quality of Fruits and Vegetables . AVI, Westport,pp. 20-36 .

Stevens, M.A. & C.M. Rick, 1986a . Genetics and Breeding . In:

Atherton, J .G. & Rudich, J . (Ed.), The Tomato Crop. Chap-man and Hall, London, pp. 35-110.

Stevens, M.A. & C.M. Rick, 1986b. The future of the field crop .In : Atherton, J .G. & Rudich, J . (Ed .), The Tomato Crop .Chapman and Hall, London, pp . 559-580 .

Vesseur, W., 1990. Tomato tasting and consumer attitude . ActaHorticulturae, No . 259 : 83-89 .

Vuurst-de Vries, R .G . van der, 1986. Smaak en aangenaamheidin relatie tot fysico-chemische parameters van vier rassen to-

Page 20: Consumer-oriented quality improvement of tomatoes in Indonesia

180

maten. Internal communication . Sprenger Instituut, Wagen-

Wolters, C .J . & L.J . van Gemert, 1990. Towards an integratedingen, The Netherlands .

model of sensory attributes, instrumental data and consumerWieringa, B . & FF van Raay, 1987 . Consumenten gedrag . Stenfert

perceptions of tomatoes . Part I . Relation between consumerKroese BV, Leiden, The Netherlands .

perception and sensory attributes. Acta Horticulturae, No.259:91-106.