Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

The Use of Taste Response Tests in Market Research

Author(s): Gideon Hadary


Source: Journal of Marketing, Vol. 10, No. 2 (Oct., 1945), pp. 152-155
Published by: American Marketing Association
Stable URL: http://www.jstor.org/stable/1246325
Accessed: 23-02-2016 19:10 UTC

REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/1246325?seq=1&cid=pdf-reference#references_tab_contents

You may need to log in to JSTOR to access the linked references.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.

American Marketing Association is collaborating with JSTOR to digitize, preserve and extend access to Journal of Marketing.

http://www.jstor.org

This content downloaded from 165.123.34.86 on Tue, 23 Feb 2016 19:10:11 UTC
All use subject to JSTOR Terms and Conditions
THE USE OF TASTERESPONSETESTSIN
MARKET RESEARCH
GIDEON HADARY
EDITOR'S NOTE: The author, now in the Army, was score is used in establishingthe value of
formerly .gricultural Economist, Office of Distribution,
War Food Administration.
a commodity (i.e., butter) even though
the consumer need not necessarily be
ELIABLE information concerning con- able to distinguishbetweenthe different
sumer preferences and acceptance is grades, and there is no relationshipbe-
prerequisite for the success of most food tween the "value" (objective or nutri-
merchandising programs. Such informa- tive) and the scoreof the commodity.2
tion is usually obtained from taste re- Initial taste responseor flash tests are
sponse tests. Various techniques may be the most commonly used technique in
used in such tests. Principally these con- ascertainingconsumer preferencesand
sist of (I) judging by trained experts; acceptance.Samplesof severalitems are
(2) initial taste tests indicating the sub- usuallypresentedto a largegroupof con-
jective response of a large number of sumers who are asked to indicate the
consumers; and (3) observations of productwhich they prefer.
actual takings of a sample of consumers Admittedly, prolonged consumption
over an extended period of time. tests are the most reliable techniqueof
In spite of the importance of consumer determiningconsumerfood preferences.
taste preference tests, reports on the However, this procedure is the more
comparative merits of these research costly and difficultone; hence the least
techniques are conspicuous by their commonlyused.
absence in marketing literature. For It is the purposeof this paperto exam-
example, in the period between I936 and ine the findingsof initial taste tests in the
I944, nothing pertaining to the subject light of their comparabilitywith pro-
was published in THE JOURNAL OF longed consumption patterns, and to
MARKETING.1 present some of the problemsrelated to
Product judging by experts is not a the use of taste responsetests in market
market research technique. It is com- research.
monly the first step in product develop- CHOCOLATE MILK BUTTERFAT CONTENT
ment. Such judging is based on arbi-
TESTS
trarily defined organoleptic concepts of
a "perfect" product. The food tested is Several tests were conducted by the
compared by specially trained technolo- author at the Universityof Wisconsinto
gists with the "perfect" and scored ac- ascertain consumers' butterfat prefer-
cordingly. In many instances such a ences in chocolate milk drinks.3These
tests consistedof both initial taste tests
1 Several reports summarizing taste tests were pub-
and prolongedconsumption tests. The
lished, but these did not discuss the matter of tech-
niques as such. Cf. R. Franzen and D. Teilheit, "A 2 In fact it requires intensive training and prolonged
Method for Measuring Product Acceptance," 7ournal experience to become an expert qualified to designate
of Marketing, Vol. 5, No. 2, October I940, pp. I56-i6I, the grades of many foods.
and M. Sellers, "Pre-testing of Products by Con- 3 Gideon
Hadary and Asher Hobson, "Consumers
sumers," 7ournal of Marketing, Vol. 6, No. 4, April Want Butterfat in Chocolate Milk," What's New in
1942, pp. 76-80. Farm Science, December 1943, pp. 34-35.
152

This content downloaded from 165.123.34.86 on Tue, 23 Feb 2016 19:10:11 UTC
All use subject to JSTOR Terms and Conditions
THE JOURNA,LOF MARKETING 153

procedureused and the findingsare out- same basic meals but have a choice of
lined in the following. drinks.
Initial Taste Tests. Two samples of Two kindsof chocolatemilk,one made
chocolate milk were given to 841 con- from whole milk, having 3.5 per cent
sumersin Madison,Wisconsin,in the fall butterfat and one made from defatted
of I942.4The cups containingthe sam- milk, having I.5 per cent butterfat,were
ples were marked conspicuously with placed at the end of the servicecounter,
either a square (1) or a circle (o). The in half-pint glasses stacked on trays
consumerswere asked to taste the two markedwith a square (D) or circle (0).5
samples and record their preferences, The order in which the two chocolate
without knowing in what respect the milks were placed was rotated several
samples differed. times duringthe week. These dinersdid
One of the samples was made with not know in which way the samplesdif-
partially defatted milk, having I to 2 feredexcept for the designatingsymbols.
per cent butterfat, and the other with The sampleswere served for a period
whole milk, having 3.5 per cent butter- of four consecutiveweeks. In this period
fat. All other factors (such as color, in- the combinedaverageper capita choco-
tensity of flavoring,viscosity and sweet- late milk consumptionat the noon meal
ness) were controlled.These tests were was 1.42 glasses. This consisted of .85
conducted with large groups of college glass, or 60 per cent, of chocolate milk
students duringthe lunch meal or at the made from whole milk, and .57 glass, or
end of a social meeting. 40 per cent, of chocolate milk made from
It was found that 500, or 59 per cent, defatted milk. Statistical tests indicated
of the respondentspreferredthe whole that a point of stability in the relative
milk product; 284, or 34 per cent, the takings of the two drinks was reached at
partially defatted product;and 57, or 7 the end of the second week.
per cent, indicatedno preference. Comparison of the Two Techniques.
Analyses were made by sex and age Fifty-nine per cent of the interviewees
groups,showing that the
essentially same preferred the whole milk product in the
relative preferencesfor the two drinks initial taste test. On the average, 60 per
existed in the differentgroups. cent of the chocolate milk consumed dur-
Prolonged Consumption Tests. Un- ing the prolonged consumption test was
limitedquantitiesof chocolatemilk were made of whole milk.6 These data show a
given to approximately450 male stu-
dentswith theirnoonmealin a university 5 It is claimed that, all other things being the same,
the geometric form used to designate samples has an
dormitorycafeteria.The studentseating effect on consumer choices. This is said to be particu-
in this dormitory do not pay for each larly pronounced as between common and uncommon
meal separately,paymentbeingmade in forms, i.e., circle vs. hexagon. Since both the square
advanceof the school term. Chargesfor and circle are common geometric figures it is believed
that the symbols used in this test did not introduce a
the meals do not vary with the quantity bias in the preferences.
of food taken, and all diners obtain the 6 The data in the initial taste tests are given as three-
way preferences (i.e., whole, deffated, neither). Where-
as in the prolonged test a two-way comparison is made.
4 A total of
2,07I people were interviewed in the If allowances were made for the 57 samplers in the initial
initial taste tests. Of these, 1,230 sampled defatted taste test who did not express a definite choice, then
products having less than one or more than two per over 63 per cent would be found to prefer the whole
cent fat. This report is confined to the 84I students who milk product. The samplers in the prolonged consump-
tested the I-2 per cent butterfat product vs. the whole tion test probably consumed an equal quantity of the
milk product. two milks.

This content downloaded from 165.123.34.86 on Tue, 23 Feb 2016 19:10:11 UTC
All use subject to JSTOR Terms and Conditions
154 154 THE YOURNAL OP MARKETING
7OURNAL OF

high degreeof agreementin the findings investigate consumer preferences for


of the two techniques. cheese. This investigation consisted of
This high degree of agreementis at- both initial taste tests and prolonged
tributed,in the main, to the fact that in consumption tests. In one instance,
both the initial taste test and in the pro- preferences of store customers and farm-
longedconsumptiontest the interviewees ers at a picnic, and in the other, prefer-
were selected from the same "parent" ences of college students, nurses, physici-
universe.The need for extreme caution ans and surgeons, were obtained.
in the selectionof samplesis paramount A canvass of 4,293 consumers was
in order to assure comparabilityin the made to determine their preferences for
findingsof the two techniques. different varieties and ages of cheese. All
The commercial application of the marks of identification were removed
findingsof the butterfatcontent tests as from the cheese and insofar as possible,
reported in this paper is somewhat every factor influencing choice, excepting
limited.It is recognizedthat the reaction those reflected through taste, were
obtained in tests conducted in institu- eliminated. The cheese was given to the
tions arenot comparablewith those con- consumers without cost to tlhem.
ductedin households.However,fromthe The tests were conducted among cus-
point of view of techniques,the elimina- tomers of retail grocery stores in the fol-
tion of all other factors,includingprice,7 lowing cities: Madison and Milwaukee,
and the controlled selection of inter- Wisconsin; Atlanta, Georgia; Charlotte,
viewees, are necessaryfor a comparison North Carolina; Cleveland, Ohio; and
of the findingsof the two approaches. Pittsburgh, Pennsylvania. Customers of
From this point of view the procedures these stores were presented with two
used are believed to have considerable samples of cheese of similar shape. Each
merit. sample represented cheese of a different
Although a point of stability was ob- kind or age. The customers were asked
served in the prolonged consumption to state their preferences after eating the
phase of the butterfat content test, it is two pieces.
possible that further exposure of the Natural mild American and processed
students to the drinks would have re- American were paired against each other
sulted in differentfindings. in eleven stores in six cities. Of 2,II2
CONSUMER PREFERENCES FOR CHEESE8
samplers, 794, or 34 per cent, preferred
the natural mild cheese, and 1,318, or 66
An intensive study was conducted by per cent, preferred the processed cheese.
Asher Hobson and M. A. Schaars to The comparisons between samples of
natural mild and natural aged American
7 Price is a very important factor affecting consumer
cheese served together in nine stores to
acceptance. Several observations suggest that insofar
as foods are concerned, "price" might determine the 1,302 customers, established that 604,
"value" consumers attach to products in contrast to the or 48 per cent, preferred the mild cheese,
classical economic concept of "price" being a function of whereas 698, or 52 per cent, preferred
"value." An initial taste response test comparing three
identical samples of wine which were designated as cost- the aged cheese.
ing %2.00, $4.oo and $6.oo per gallon, respectively, Natural aged American and processed
indicated an overwhelming preference for the sample
8 The data presented in this section are drawn from a
purportedly costing $6.oo per gallon. Although impor-
tant, the effect of price must be eliminated in a theoreti- study published as Research Bulletin I28 by the
cal examination of the relationship between the two Agricultural Experiment Station, University of Wis-
techniques. consin, October I935.

This content downloaded from 165.123.34.86 on Tue, 23 Feb 2016 19:10:11 UTC
All use subject to JSTOR Terms and Conditions
JOURNAL OF MARKETING
THE JOURNAL 155
155

Americanwere served togetherin seven in the prolongedconsumptiontests. Of


stores in five cities, to 879 samplers. the average daily consumption, 34.5
Threehundredand forty-eight,or 39 per per cent was natural mild cheese; I7.0
cent, expresseda preferencein favor of per cent, natural aged; and 48.5 per cent,
natural aged cheese, and 53I, or 6I per processed.
cent, preferredthe processedcheese.
Prolonged Consumption Test. Aged CONCLUSIONS
American,mildAmerican,and processed In the chocolate milk study close
Americancheese were served daily with agreement was observed in the findings
the meals for fifteenweeks to a groupof of the two methods; however, in the
approximately I o Short Course stu- cheese study the findings of the two were
dents at the University of Wisconsin, at variance. Perhaps the difference is due
some twenty-five supervisingnurses of to the manner in which the selection of
the Wisconsin General Hospital, and interviewees in the two cases was made.
around ten physicians and surgeons at The need for extreme caution in selecting
the Wisconsin GeneralHospital. a representative sample in the taste test
In some instances, the cheeses were is of prime importance in a comparison
identifiedby "x," "y" and "z" designa- of the techniques.
tions, and in others by differentcolored Although the chocolate milk study
doilies.9 The samples were sliced into showed an excellent agreement between
cubes, and an unlimited supply was the two techniques, there are no other
available at all times to all diners. Al- case studies available which verify these
though there was some variation be- findings, in order to permit of drawing
tween the three groups,on the whole it definite conclusions as to the theoretical
was found that the average daily per comparability of the two testing pro-
capita cheese intake was .044 pound.Of cedures. Therefore, it is impossible to
this amount,on the average.007 pound pass upon the validity of initial taste
constituted aged Americancheese; .015 tests conclusively.
pound, mild American;and .022 pound, Extreme caution must be exercised in
processedAmericancheese. planning and conducting taste response
Comparisonof Two Techniques.From tests. The symbols (geometric forms, al-
the preferencesexpressedin the initial phabetical letters, and numbers) or the
taste tests it follows that 27.3 per cent price identification used to designate the
of the customerspreferrednatural mild samples affect the preferences of con-
cheese; 30.3 per cent, natural aged sumers. Procedures applicable to one
cheese; and 42.4 per cent, processed type of commodity need not necessarily
Americancheese.10 apply to others. Considerable additional
The relativepreferencesweredifferent information and case studies are needed
9 Generally, all other things being the same, con- before it will be possible to isolate the
sumers will tend to prefer samples marked "x" (the factors affecting consumer taste prefer-
unknown) over "y" and "z." Similarly, consumers have ences.
been observed to express a preference for identical
foods designated by certain colors as against others.
The indiscriminate use of numbers also affects con- 10These percentages were calculated in accordance
sumers' taste preferences; products designated with with the method developed in connection with a study
Roman numerals should never be compared with those in consumer preference rating. George H. Brown and
with Arabic numerals. Samples marked with numbers Gideon Hadary, "Beverage Preferences of Industrial
having certain associations, i.e., "57," "66," and "606" Workers," The 7ournal of Business, April I944, pp. I I I-
will introduce a bias in the responses. I 17.

This content downloaded from 165.123.34.86 on Tue, 23 Feb 2016 19:10:11 UTC
All use subject to JSTOR Terms and Conditions

You might also like