close

Вход

Забыли?

вход по аккаунту

?

How to modeT persuasive message with concept testing

код для вставки
SMtlon 1тАФPag┬л ^
_Marl(ctin3 News
ltayi3,iM3
How to modeT persuasive message with concept testing
THE OPTIMUM PERSUASIVE
two differeot executions ^n final-ad
atea to the reader unless he learned
cated
amount of cashi" Gra'Ss said. "Data
message for an a4vertisement can be form) for each ofthe fio^r concepts.
that conflectijm.' These are called"" collection was carried out musing a
modeled using concept testing proce"Our first7stcp was t<x perfofm a "тАвprimary' connections," Grass said.
simple procedure, 'remote fnotiyated
dures, according to Robert C. Grass^
content analysifr-of each execution
"Secondary connections" or suplearning,' which provides lntorrnation
group manager. Corporate Marketand prepare a list of'c6nnections* for porting inforhration items are given ^- about the maximum teaching potening Research Division, E, I. Du Pont
each one," he said. "A connection is a
rating of 2 and "tectiary" connections,
tial of an ad.
De Nemours & Co. Inc., Wilmington тАв
piece of TnesSage-oriemediiFT^ffid^seitems^Tiich thp reacfei^ really
"WORKING FROM A.UST4>f
Del.
.
formation that a read.er might learn doesn't have to learn, are assigned a
appropriate respondentsтАФin this
At the AMA's 14th Attitude Reindependently of other ad content? rating of 3.
search Conference in Rancho Mirage,
For examplj;, the brand name of tKe
"The client4s4hтВмn^asked to convene^ case. <:otton growersтАФa sample ofthe
target audience is recruited by teler
Calif., he described the procedure
product is a coniiection~as is the nanfe the individual rdlers to resolve differphone to' participate in the study.
usingi^s examples ad concepts for a
of4he manufacturer. We-^ll these ences. Once conserisus is reached, the
Jiwy
arf^ld that thc^xJvflLieceiveaiL
new agrichemical product caHed Blast.
lists 'connection'sheets.'"
connection sheets are returned ttj,
ad in the mail, and an appointment ia.
The herbicide is intended to be used
The lists, or "connection sheets," form the basis of the rese.arch procemade to call them after they have reon cotton, bun;an also be used on . jare_tl^ submitted tothe ┬гHent, who is dure," he said. '
ceived
the ad.
* ,._
corn/ It doesn't hurt crops, but con- asked to check them for accuracy.. Jhe
Figures 1 and 2 are sample connec"An
ad
is
th┬лL
m^ed
in
a
specialtrols' 46 different weeds and stays in
.client sends copies of each list to afl
tion sheets for the Residual and Broad
ly designed envek>pe^te┬лach responthe soil a long time. These^features ^embersofthe team who will exam- ^Spectrum concepts. Each is defined
^
,dent. Each envelope also contains a
allowed the product to be positioned
ine the results of the research. Grass
by the items rated I on the list.
*
pressure-sensitive
label bearing the re┬╗
in a numbgr <?f iways, >yhich formed
said. Each>person rates the items o ┬╗
"As the dependent variables for
.
turn
addi
r^nd~a
return envelope.
the basis for tne four
тАФ
concepts 'Hit l i s t s : тАФ тАФ ^
evaluating the concepts, it was de^
t
the
appropriate
time, the resubmitted for testing. Grass said.
"A Ri\TING OF 1 -is assigned to cided to use a five-point buying-intenspondent
Is
recontacted
by
phone, inThe concepts were named Residuan item if'the client considers that the tion scale and product preference exstructs^ how to open the envelope
concept represented by the execution тАв pressed as a choice between receiving
al, Broad Spectrjim, Safety, and Scienand remoVe the a d ^ d the labels. He
tific Breakthrough, Du Pontprepared ^wpuld not be successfully communi- the product as a gift or an equivalent
is then given instructions'to study the
ad while the interviewer waits on the
phone.
^
FIGURE 1: Connection Sheet for Realduil con"When he indicates that he has
FIGURE 2: Conn┬лction ShMt for Broad Spectrum
studied the ad long enough, he is
cept. "Biasr heriMcide print ad, "New Blast goes
concept Blast herbidde print ad. "New Blast:
asked a short series of questions about
first flush;"
HnaNy, there's a herMcMe..."
--^
^ e ^d's appearance and format, Theimportthc*
Importance
respondent is asked to reinsert the ad
RaMnfl
CofltwcUon
Rating
Connecllon
in the mvelope, reseal it, and affix the
j
return-postage ^nd return-address
t
Product a d m t l M d It c ^ M Blast
Product adverilted it called Blast
label.
1
Product It тАв lMfMcid#.
"He can no longer refer to the ad
Product It madt by Du Pont
". -3 .
Productit made by Du Pont
without destroying the envelope. The
Product can tit ut#d on cotton.
ProducLcan be uted on cotton.
тАв тАвтАв '
1 тАв
Product can b* uMErsfTCom.
fnterviewer then administers a recogProduct can be uted on coni.
Product controlt w┬л┬лdt all th┬л way
1
1
Product controlt 46 Afferent vveeds,
nition-recall interview.
through hanrMt
2
Product controit weedt all the way
"The respondent IS told that he will
' 2
Product It iww.
.
'
through harvett
be read a series of statements, that
Product controlt w t t d t btttar than
Product It new.
some of these statements will have
any other h┬лrbidd┬л.
Product controlt weedt better than
been contained in or implied by the
Product
controlt
|imtonw┬л┬лd.
any
other
herbicide.
" 3
ad,
while others will not. When he
Product controlt vthwtteaf.
1
Product controlt Jimtonweed.
3
hears eaeh statement, he is to say
Product controto coddvbur.
1
Product controit velvetteaf.
Product provldet early control plut
1
whether he is sure the statement was
Pro<luct controlt cocRtalbur.
long-ttinn retfduti contool.
Product conb'oit muttardt.
made or implied by the ad, whether he
Product rmuitt In тАвatler cropping,
3
2
Product klllt weedt on contact.
is sure it was not made or implied by
ProductMl^uItt In lett problernt at
3
2
Product providet early control piut
the ad, or whether he is not sure either
harvett
long-term retldual control.
way.
Product it a whoie new ttandard of
"When this procedure is completed,
weed control.
the respondent is.reread only thosr~
FIGURE 4:
Product contrcit weedt by acting
statements which he was sure were
Grid fragmentior learning of connections.
on t h e T o o i n i weii at contact with
made or implied by the ad, and in each
foilage.
RESlDUAt
SPECTRUM.
Other herbiddet attack weedt only by
case he is asked whether he believes it,
killing the top growth you can i
disbelieves it, or is not sure. Finally,
1. Product called
the respondent is asked thejntentionBlatt
to-buy of other persuasive measure,
and the inteiview is concluded.
2. Product It
FIGURE 5: Variation In execution effeetivenesi,
herbicide
heedless to say, any returned enveтАв
% Complete
lopes, vn^ich show signs of tampering
^ . Uted on cotton
Learnert/
are discard^ from the sample of inBelievert
terviews."
a
7. Controlt 46
CONCEPT A
(12)
Execution t
Execution 2
6. Controit weedt
through hanrett
CONCEPT B
13. Controit
<
JlmtonweedтДв"-^ тАв"
Execution 1
Execution 2
CONCEPT C
16. Controit
muttwdt
-тАв: ,,-_,.
(46)
1 Execution 1
Execution 2
21. Lett problemt
arharvett
(66)
50
42
Execution 1
Execuflon 2
FiGURE 6: Variation in concept effectiveness
(Base: Concept leamers/believers oniy)
FiGURE 7:
Roie of ieaming in concept effecth^ness.
Mean tcale rating
Nkellhoodof
uting product*
CONCEPT A
44
CONCEPT D
MEAN SCAtE RATING,
LIKELIHOOD OF USED PRODUCT*
0.96
CONCEFTB
Learned/
CONCEPT C
Did Not Leam/
1.12
.
Concept
Total
Sample
' Concept A
0.96
0.93
Concept B
0.67
0.93
Concept C
1.12
1.07
ConceptD
1.21
1.09
CONCEPTD
*Flve-point buylng-intentlon scale (-2 to -тАвтАв2)
F i G U R C 8 t 3 u y i n g d n t ┬л i t i o n relatkmships^of indlviduai M┬╗nnectlons.^
Buylng-intantion
Level of
Said or impHed
relationship
tlgnlflIn concept-execution
(X')
cance
--y
---y'^
' . ^1 ^ , '
Cennectidfr
1.02
'~
"'тАФ-тАв тАФ= -'-'- "'- тАв
^ ;,_l:,^":-;
тАв
' тАв.
. * .
тАв
% of total
pertuatlve
Impact
daMvarad
CONCEPT A
Connection 41
Execution 1
Execution 2
7
Connection 16.
CONCEPTT
Connection 31
114
D
No
D
\
2.49
0.20
Connection 37
2.01
0.20
Connection 39
1.99
0.20
ConnecOcmMi
0.20
IM'
CONCEPT d
Execution 1
Execution 2
CONCEPT D
Execution 1
Execution 2
44.5
тАвIt
Statement
BellevaMe
Thje
Don't
Know
False
D
D
D
Product adverVted it
Called Blatt
cr
4. Product It a herbicide.
5. Product Is ma┬лle by
Monsanto.
6. Product Is made by
Du Pont
"^
7. Produce can be uted
6. Product eai be uted
on cotton.
Execution 1
Connection 23
Connection
Yes
Product advertited I t
called Bravo.
Connection 22
wonnecnon
Statement Said/
lmpll#d in Ad
9: Comparlaon^of ccm^ept-'executior^s as
deliverers of the "idear concept
-^--yтАФ-'-'
Connection
4
CONTINUED ON PAGE 21
.
FIGURE 3:
Recognltlon-recali aheet for Blast concept-execution teat
Concept
*Flve-point buying-intentton tcaie (-2 to тАв 2 )
Connection 36
ConnecUon
'
tion, whether that-connection existed
in an execution by implication dr by
explicit statement, and what degree of
belief was attached to each connection
when it was leartied," he said.
Figure 4 shows a portion of a: grid
with one column of data for the Residual concept and the other for the
Broad Spectrum concept. ItjcontaiM
jtist efght of the SO recoiputionrnecall
sutQments. The, boxed numbers
resetit primary conhections inili
concept, the numbers in parentheses
represent connections that were not
explicitly stated in the execution, but
were present by implication. Grass
said.
"Note that for statements No. 1, 2,
and 3, both executions were able to
commuhicate thejbasics of the concepts equally wellтАФthat is, brand
name, generic identification, and cro^application," he said.
"Statement No. 7, which is a primary element of the Spectrum concept, wasii't particularly well taught
by that execution. Indeed, in the Residual execution, where this information was present only by implication,
12% of the readers thought this state"inent was being made.
"Statement No. 8 is an example of
information taught well by both exe=cutions, but which is considered only
to be part of the Residual concept.
Statement No. 13 is the reverse case.
Statements No. 16 and 2 fare exampies of information explicitly statedand taught weUtp one concept execu>
tien but ako taught appreciably well
by implication in the other execution."
AS SHOWN IN FIGURE S,
the executions varied in their ability to
communicate. The differences were
most pronounced in Concepts A and
C.
"Based only on those respondents *who learned and believed each cohr
cept, we can then Took at-the differ- тАв/
ences between concepts with respect
to their abilities to influence the persuasive measure," Grass saTd. "Figure
6 shows the likelihood of using the
product, as measured on a five-point
scale.
"Responses to the second depen"ttem variable showed similar results.
Examining these data told us that
^IGURE
Connection 33
Connection
THE NEXT STEP was to compile <┬л recognition-recall list of statements that included all \he information from all eight executions. Since a
given respondent only saw one ad,
many ofthe statements on the list had
not been explicitly stated in his ad.
The list ako contained a number of
statements that were obviously
wrong. Grass said.
Figure 3 shows a portion of a recognition-recall sheet, administered to
al|^ respondents regardless of which
concept they saw. "With these dau,
we constructed a grid showing all
eight concept executions, how much
learning occurred for each connec-
ftobeitC.
D
D
D
D
D
D
D
D "D
Produerwon't hold
back your crop, no
matter wfien you
n
10. Product controit 46
different'
weedt aii the w
through
May 13, I M S
Marfccting News
Selectrare (Populations with iQW-cost,\
statisticaliy efficient mijitipitcitysampiing
MARKET
DEVELOPMENT INC ^
Research und Counsel
SepnenL
A
RESEARCH TECHN^Q^UE
important arid growing markets." '
. called multiplicity sampling can be
Emphasizing that "rare populaused by marketers to obtain information" i$ a relative term, Dutka cited
. tion about highly segmented-^iarkets,
Providing valid, reliable DATA and linguistic^
some comparisons: 6.1 niillion U.S.
according to Sqlemon DUtka, chairhouseholds have microwave ovens,
cultural and behavioral COUNSEL at every phase
. , manr Audits & Surveys Inc., New
equal to all the households in Czecho-;
York.
otlflarketipgto the Hfepaprc consumer. -y"
Slovakia; I.S million U:S. households
Also.knowik as network sampling
had a first child m 1980, more than all
or the extended sampling unit, multi;
the households in Irelandvand the 6.7
RESEARCH SPECIALTIES:
.'
1__L.,
┬╗ plicity sampling is more efficient anff'
million U.S. households with videoтАв Quantitative and qualitative, studies
тАв.
pi^ctical in selecting rare population
cassette recorders nearly equals all oT^
тАв Hispanic validated design, scales, tabulation 'тАв /
тАв
- sainples than conventional sampling
- the households in East Germany.
'..
тАв
Tightly
controlled
data
collection
procedures
proe┬лlures, he told the 29th Annual
тАФ. "From ^another -standpnint, the
тАФ ' . тАФ ~ тАФ I i Consumei,"advertisiny, rnerchandisii'ig research
Advertising Research Foutidatiori
2.8% of our population-^yhich jogs
Tactical analysis, interpretation,
J Conference in New Yoi
equal the, population of Switzerland,,
rec6mmenda(ion&
"Whilf this appwi
the 9.2% who are teenagers would
тАв Social-class relevant sampling and analysis
fully constructed, and is more compopulate Romania, and the i2%~with
plex from a mathetnaticaF standpoint
тАв National regional, local sampling techniques
high serum cholesterol surpass by.
Ihan current sampling techniques, it
four million tne population of Cana- тАвтАв
COUNSELING AREAS
' , *
produces higher yields in locating the
da," he %aid.
тАв
11
years'U
S.
and
Latin
American
lihguisfio, cultural, and
"Solomon
Dutka
spedfied population with fewer
Multiplichy sampling substantially
behavior,
analytic
experience
.
тАв screening interviews^ and therefore
increases the probability of finding the
tended to a network of consumers or
тАв тАв 15 years U.S. and Latin Amencan marketing, marketrng
lower costs," Dutka said.
desired
characteristic among a particuhouseholds.
Multiplicity sampling was first u$ed
research experience
┬╗. _
-lar universe over con>┬лnti()nal sam-*.
Fof example, a network may conтАФby the. National Center for Health
pling
procedures,'
Dutka
said.
HowтАв
National
and
local
advertising/sales
promotion tactics and
sist of all households containing
Statistics as a statistically efficient
ever,
network
structures
ntiay
be
unevaluation
proced'ures
,
тАв
_i)lood relatives, or, in industrial remeans of sampling individtials with
ambiguously
defined.
B Expe/lise in nutritional, food, toiletry. autOTpharmaceutt-search^ra fletwork may consist of ail"rare diseases. Instead of the sampling
"An individual-sampled from any
plants of a given corporation.
cal, soft drink, beer/liqgor products among Hjspanics
unit consisting of a single individual
- household in the network then also
тАв Pr"oduct testing and new product development
"SAMPLING
POPULATIONS
or, household, the samplingi is extestifies as to the size of the network
in Hispanic high-volume product categories
with rare characteristics has long been and the existence of t h e i a i e trait
a problem, and in the past it often (las among the immediate family," he
'- - тАв Cultural awareness training tp existing distribU'
been prohibitively expensive for mar- said. "Network sampling produces
tion channels '
keters," he said. "With the increasing тАФsignificant gains over conventional.
U Completely bicultural and hilingnal
methods through reductions in samFROM PAGE 20 segmentation of markets, the ability
to study small or rare populations is pling variance and costs."
7801 Mission Center Court, Suite 243 тАв San Diego, CA 92108
becoming more essential, especially in
C and D, as verbalized by
_ (714)298 2612
view of the fact that some represent
. the client,idb a better job than Concepts A and B, significant at the 90%
confidence level.
"Had we used the total sample for
this evaluation. We would have come
out with the same suggested order, but
without statistical significance. Separating out those┬╗who did not learn
and believe the concept suggested a_
different rank order (Figure 7). Thus,
it is apparent that sometlung else may
be going on in this learning process
that affjects persuasion other than,
what we found verbalized in the
client's concepts."
Up to this point, the research results
could be used to recommend a concept to the client. But Grass said he
nterf to know if there anfvcn b
concept than any of tMe four alternatives tested. J
"So the next step was to cast aside
aU the preconceived concepts we had
tested and throw all tjhe data from all
respondents from all the samples into
the same pot," he said. "Using^ variety of statistical procedures, we then
looked at the relationship between
learning plus believing a given connection and its effect on the dependent
persuasive measure. We chose to look
at the simple chi-square relationships,
all positive, for all 50 coniiecTionii and
found 16 of them with weights significant at the 80% confidence level or
better (see F4gure 8).
"THIS APPROACH LED to
the construction of a theoretical 'ideal'
concept that should result in maximum persuasiveness. Such a concept
is independent of execution. However, we can now work our way down
this list of connections, rank-or^^
dered by weight, to build a persuasive
'menage that includes as many elements as it seems practical, to transmit to the target audience.
"At this point, the client may decide
to prepare new executions . ^ this
'ideal' concept. Alternatively, he may
wish to know which of the existing
executions creates the greatest persuasive impact in terms of the new 'ideal'
concept. If we represent 100% persua^Computer-Assisted Telephone Surveys
sive impact as the theoretical case, we
gg ComputefcAssisted
p
p
n express the achievement of a given
Lavigge
Telephone
^erlng details that normally slow down the
тАвexecution as a percentage of this theoSurveys cpniblne the reach and convenience of
intenviewing: phone numbers to be called
X
retical maximum.
WATS
line
inteFviewing
with
the
speed
and
skip pattems, rotation of altemative parts of ques"When we perform this calculation
accuracy
of
computer
control.
That
combination
tions or answers. Plus, once thd question is
(Figure 9) for the eight executions
tested using all 16 connections listed
means your research results can be delivered
answered,^he cornputer can tabulate immediearlier, w e ^ d that six of the eight
In a matter of days, or even hoursтАФnot weeks.
ately, and can produce a running report at ariy
executions perforn^ at the same relaAnd
C.A.T.S.
are
flexible,
too:
the
questions
stage of the research,^k)mputer output in^your
tive level of persuasive impact, but the
asked
are
related
to
information
secured
earlier
other twoтАФboth from the original
t i i d i
hard copy printouts cjn be
C.bncept CтАФperlormat a i i f ^
in the inteiyiew aoa fhe questionnaire can be
delivered to you just hours after the final interview.
ly higher level.
modified for later interviews to take advaritage
To learn more about our Computer-Assisted
, "Whereas both Concepts C and D
of what Is learned. .
appear <n he favored over the-others,
Telephone Surveys, or any of our other marketthe executions of Concept C are clearThe computer aSisi^ts our interviewers by rememing research services, contact any of our offices.
ly to be preferred on the basis of a//the
information learned and believed
The client may be able to generate
"hew executions of the ideal concept.
but if he wants to run with what he'┬л~
' an Equifax Company
got, the jexectitioos of Concept C are
to S. Riverside Plaza
9 Corporate Square
240 Stockton
his best bet.'
ChlcaQ6,lL80┬л┬╗7
AtlantfcQi;35329v
On Union
(312)726.0866 - - - - im^W^^&(^^^^^^ ^^
(800)6219006
900)241-0643
Grass
For fast results, call on CATS * - тАФ
Документ
Категория
Пионер
Просмотров
25
Размер файла
8 795 Кб
Теги
1/--страниц
Пожаловаться на содержимое документа