American Marketing Association

Advertising-Selling Interactions: An Attribution Theory Experiment

Author(s): William R. Swinyard and Michael L. Ray
Source: Journal of Marketing Research, Vol. 14, No. 4 (Nov., 1977), pp. 509-516
Published by: American Marketing Association
Stable URL: http://www.jstor.org/stable/3151191
Accessed: 22-10-2015 11:54 UTC

REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/3151191?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
Research.

http://www.jstor.org

This content downloaded from 193.0.65.67 on Thu, 22 Oct 2015 11:54:50 UTC
All use subject to JSTOR Terms and Conditions
 WILLIAM
R. SWINYARDand MICHAELL. RAY*

Two common assumptions about advertising-selling interactions are chal-

lenged by a field experiment. The generality of "advertising paves the way
for selling" is questioned by results showing selling-before-advertising to be
more effective than advertising-before-selling. "Salesmen are born not made"
is questioned because the easily learned "labeling" technique markedly
increased selling effectiveness and its interaction with advertising.

Advertising-Selling Interactions: An
Attribution Theory Experiment
___
_

INTRODUCTION The authors propose that a personal sales strategy

An investigation was made of the two principal
components of most promotional budgets-mass
media advertisingand personal selling or contacting.
Their separate effects, some ways in which their
effectiveness can be improved and, most important,
their interactionwere studied.
Research Issues
Although the research was conducted in a social
marketingcontext, the findingsalso suggest directions
for industrial promotion, political campaigns, and
consumer marketing,where advertising is combined
with personal selling. The discussion of the effects
of different sales presentation strategies has still
broader applications. It can be extended to other
promotionaltechniques(such as sampling,premiums,
and offers) which mightcreate the sort of commitment
and self-perceptionachieved by the selling technique
used in this study.
Key questions investigatedwere:
1. Do advertisingand personal selling interact?If so,
is the interaction stronger when the sales calls
precede, or when they follow, the advertising?
2. Does the strengthof an advertising-personalselling
interactiondepend on the type of personal selling
strategyused?
will predispose prospects to be more receptive to
advertisingand will induce more powerful behavioral
responses than would result from either advertising
or personal selling alone or from the simple sum of
their effects. In contrast with conventional wisdom,
the authors also propose that these behavioral re-
sponses will be greater when the sales call precedes
the advertisingthan when it follows.
3. Is a "behavioral influence" strategy in a sales
presentation more effective than a straight sales
approach?
A low pressure strategy in which the prospect is
"labeled" in a manner congruent with a desired
behavior change may be a foundation from which
the prospectwill decide he or she is favorablydisposed
to the salesperson'sargument.For instance, a person
told she is "charitable" (a label) is more likely to
support a charity than a person not so labeled. An
inference from labeling may stabilize the prospect's
behavioral intentions and increase the likelihood of
compliance with the objective of the campaign. In
contrast, the straight sales approach-one in which
a purchase (or other decision) is requested-relies
more on existing favorable attitudes to produce the
compliantbehavior.

*WilliamR. Swinyardis Assistant Professor of Marketing,Ari-
zonaState University,andMichaelL. Rayis Professorof Marketing
and Communication, Stanford University. The research was
supportedin partby an awardto the first authorfrom the American
MarketingAssociationDissertationResearchSupportCompetition.
509
PAST RESEARCH
Interaction Between Advertising and Personal Selling
Attributionand cognitive dissonance theories [3,
7] suggest that personal influence strategieswill have
a pronouncedeffect in increasingreceptivityto subse-

Journalof MarketingResearch
Vol. XIV (November 1977),509-16

This content downloaded from 193.0.65.67 on Thu, 22 Oct 2015 11:54:50 UTC
All use subject to JSTOR Terms and Conditions
510
quentadvertising.Advertisingthus may become more
effective because of the preconditioningeffect of the
personal call. Also, in a more traditional ordering
procedure,the personalcall may be made more effec-
tive because of a preconditioningeffect of the adver-
tising.
Neither of these two orders-ads before selling or
selling before ads-has received specific attention in
the research literature. Eldersveld and Dodge [6]
comparedthe effectiveness of the two communication
approaches in a small study of voting turnout and
attitudechange. Althoughthey do not reportstatistical
significance,they foundthe effect of a single personal
contact to be greater than that of four direct-mail
promotionalitems. Their study has strategy implica-
tions for marketers considering the use of either
personal selling or advertisingbut, because the two
methods were not used simultaneously, no tests of
interactionbetween them were possible.
Business Week [4] reportedan IBM study in which
the separateeffects of advertisingand personalselling
were compared,and accordingto Clayton [5] proprie-
tary studies have concluded that such promotional
tools as product sampling will subsequently lead to
increasedadvertisingeffectiveness.
One "interaction" study, however, stands out-
Morrill's [11] review of the results of more than
100,000 interviews, 1,000 advertising schedules, and
26 product lines. Although Morrill's studies were
correlationaland therefore did not manipulateadver-
tising exposures experimentally, he concludes that
"advertisingacts as a valuable introductionfor the
salesman to his prospective customer." In terms of
the cost-benefits of advertising, Morrillreports that
turningprospects exposed to advertisinginto custom-
ers requiredpersonal selling costs that were from 2
to 28%lower thanthose needed for unexposedgroups.
Morrill'sstudy is a start toward recognitionof an
advertising-personalselling interaction;but because
the study was correlationalratherthan experimental,
thirdvariableeffects and causative directionwere not
controlled.Examinationof different selling strategies
with different exposure levels was not possible with
the data available; nor was it possible to examine
the orderof sellingand advertisingin the total commu-
nication campaign. Note also that Morrill seems to
have assumed that advertisingshould always precede
selling.
Labeling
"Labeling" involves simply telling a person that
he or she has certain tendencies. In an early study,
Bem [2] was led to conclude that being told, "You
likebrownbread,"mayhavethe sameeffect as hearing
oneself say or think, "I like brown bread." Aronson
and Mettee [1] found that randomlyassigned "self-
esteem feedback," which is similar to labeling, had
a predictedeffect on subjects' resistanceto temptation
to cheat in a card game.
This content downloaded from 193.0.65.67 on Thu, 22 Oct 2015 11:54:50 UTC
All use subject to JSTOR Terms and Conditions
JOURNALOF MARKETINGRESEARCH,NOVEMBER1977
Such results can be explained by Kelley's [9]
attributiontheory. His discountingprinciplesuggests
that if the "labeler" is not viewed as manipulating,
and if the "labelee" has no immediately available
contraryevidence, the label cannot be easily discount-
ed and the label will be accepted as the truth. That
is, in a marketingcontext, the prospect would believe
the label to be a true representationof his or her
own inclinationswith regardto the sales proposition
at hand.
Kraut [10] and Steele [14] examined the labeling
construct experimentally,althoughnot strictly within
an attributionframework.Krautlabeled homemakers
as either "charitable"or "uncharitable"after a fund-
raisingsolicitation. His data supportedthe attribution
theory assertion that a "charitable" label would in-
crease subsequent charitabledonations, although re-
sults were not significantfor the "uncharitable"label.
Steele [14] replicatedthis experimentusing a tele-
phone manipulation.His results substantiatethose of
Kraut's"charitable"label; in addition,even the "un-
charitable"label tended to increase subsequent con-
tributions.Attributiontheory could explain even the
latterresultbecause "uncharitable"subjectsmay have
had a history of "discountingcues" (cues providing
contraryevidence) [9] and therefore acted so as to
reinforce that history. In any case, labeling can be
persuasive.
Although anecdotal literature supports the effec-
tiveness of labelingprospective buyers, and the con-
cept has been suggested as a personal sales strategy,
no research has been conducted on its effect as a
selling tool, and certainly no research has examined
its interactionwith advertising.
TEST METHOD
Overview
Data were collected by means of a field experiment
conductedin residentialPalo Alto, California.Female
household residents were personally contacted by a
"Red Cross volunteer" who delivered a persuasive
appeal for the Red Cross Blood Center either with
or without a "charitable"label. (A "small request"
manipulation[8] also was included, but discussion
of this variable is beyond the scope of this report.)
After this contact, respondents received either 0, 1,
2, or 4 mailings of related promotional-advertising
materialfor the Blood Center.
Similarly, a randomlyassigned group of residents
first received the repetitive mailings; they then re-
ceived a personalcontact. A third randomlyassigned
control group of respondents received the mailings
but no personalcontact.
The personal contacts and direct mailings were
completedwithin a two-week period. Duringa period
two days later, all respondents were contacted by
telephonefor collection of the dependentmeasures-
cognition,affect, and behavioralintentionwith regard
ADVERTISING-SELLING
INTERACTIONS
to the Red Cross Blood Center. Data so collected
were examined for the main effects of advertis-
ing/personal selling order, labeling, and the interaction
of these variables with advertising exposure levels.
Experimental Design
The design used was a 20-cell incomplete factorial
with predictor variables of advertising exposure (four
levels: 0, 1, 2, 4 exposures), labeling (two levels:
labeled or not), and personal selling/advertising order
(two levels: selling first or advertising first). Included
in the design was a personal-contact control group
(which received no personal solicitation) and an adver-
tising control group (which received no advertising).
Total size of the sample answering the key behavioral
intention question was 303.
Overview of Procedures
The research was conducted by the following general
procedures.
1. In the "ads follow sales calls" treatment,six female
experimenters, unaware of the hypotheses and
representingthemselves as Red Cross volunteers,
contacted adult female residents at their homes in
residentialPalo Alto. Each "volunteer" delivered
a "persuasive message" recitation, either with or
withouta "charitable"label.
2. The subjects in all treatmentconditions then were
mailed Red Cross promotional-advertisingmaterial
in one of severalrepetitionlevels (withthe exception
of a zero-advertisingcontrol group).
3. For the "sales calls follow ads" treatment, after
the completionof the mailing,the "volunteers"(two
new and four who had done the pre-advertising
contacting) called on post-advertisinghouseholds
andpresentedthe "persuasivemessage" recitations,
either with or without the "charitable"label.
4. Respondentsin all treatmentconditions were then
contacted by telephone for collection of dependent
measures. Male interviewers, representing them-
selves as conducting student research on public
health organizations,administeredthe interviews.
Personal Contacts
Volunteers contacted homes in a manner typical
of volunteer social workers, but were instructed to
speak only with the woman of the house. Once in
contact with the proper respondent, the volunteer
recited the message appropriate to the treatment
condition assigned for the household. No interviewer
discretion was permitted in the assignment, treatment
conditions were rotated systematically, and inter-
viewers were "blind" as to expectations of results.
Messages to be presented in each treatment condi-
tion had been developed through pretesting to achieve
maximum attention at the door with minimum refusal.
They were delivered as follows.
Hello, I'm a volunteerfor the Red Cross Blood Center
here in Palo Alto. I'm not here to ask for money-just
for a minuteof your time. Marchis Red Cross Month,
This content downloaded from 193.0.65.67 on Thu, 22 Oct 2015 11:54:50 UTC
All use subject to JSTOR Terms and Conditions
511
and I am part of a program to help increase the
neighborhood'sconsciousness of the Red Cross.
The Red Cross is very short on blood of all types,
and on volunteers-despite the fact that it's so easy
and convenient to give. You know, the Blood Center
is less thana milefromhere. Again,we wantto increase
a sense of responsibilityaboutthe Red Cross and we've
found that every little reminderhelps.
Added to this message was a labeling treatment
which was also systematically rotated in its presence
or absence. The script for the labeling treatment was:
Label condition: Thankyou for your time in listening
about the Red Cross. I wish more of the people I
met were as interestedin theirfellowmanas you appear
to be.
No label: Thankyou.
Volunteers were instructed to communicate the label-
ing message directly and sincerely to minimize dis-
counting cues.
Several items were left with the respondent. In all
conditions the woman was given a leaflet describing
the facilities and procedures of the blood bank. In
the label condition the volunteer gave the respondent
a small printed card reinforcing the charitable label
(reading, "Those who support the Red Cross gen-
erously help a good cause. Thank you for being a
Red Cross supporter.").
Advertising Exposures
The contents of the direct mailings were based on
the type and quantity of material which the Red Cross
might ordinarily include in such a campaign. Each
envelope contained a "sales" letter, a pamphlet, and
advertising material. Each of the four letters included
appeals for blood donors and Red Cross volunteer
workers. The advertisement was printed on the right
or left margin of the letter itself.
Package versions. Four such direct-mail packages
were prepared. Each contained a different pamphlet
or leaflet. To control for the random effects of the
different letters, advertisements, and pamphlets, the
four letters were rotated among respondents within
each treatment condition, and the four pamphlets were
rotated within each letter condition, with the constraint
that respondents received any given pamphlet only
once. Attention also was given to envelope appear-
ance. Four versions of envelopes were used to maxi-
mize the likelihood that each would be opened.
Timing of mailings. The mailings were sent first
class, metered postage, on March 4, 6, 8, and 10,
1975. Systematic rotation was used to minimize letter
ordering and the effects of recency to the collection
of dependent measures.
In the one-exposure condition, 25% of the treatment
group was sent the advertising on each of the four
mailing dates. In the four-exposure condition each
respondent received the advertising on all of the four
mailing dates. In the two-exposure condition, half of
512 JOURNALOF MARKETINGRESEARCH,NOVEMBER1977

the respondents were sent advertising on March 4
and 8 and the other half were sent the advertising
on March6 and 10.
One-wayanalysisof varianceof datafor early versus
late schedulesconfirmedthat most of the confounding
effects of the mailing timings were controlled. On
the average, each exposure condition received the
same dispersion over the mailingdates, and recency
effects were minimized.
Collectionof DependentMeasures
Dependent measures were collected by telephone
duringthe final three days of the study. To minimize
the chances that respondentswoulddrawa connection
between the solicitations for the Red Cross and the
telephone interview, a cover story was used by inter-
viewers of the sex opposite that of the volunteers.
Respondentswere told that the informationwas being
collected as part of a Stanford University student
project on public health organizations. Interviewers
were experimentally"blind"in thatthey did not know
abouttreatmentconditions and which respondenthad
received which treatment.
Of an initial sample of 480, usable interviews after
up to four callbacks were obtained from 347 for an
overall completion rate of 72.3%. Completion rates
for treatmentconditions ranged from 65.0 to 83.3%.
A chi squareanalysisof cell sizes (excludingthe control
condition)showed no significantvariationacross cells
(chi square = 6.23, 15 d.f., n.s.).
Dependentmeasureswere collected for each of the
Althoughthe datacluster moretightlywhen cell means
are used instead of raw data, hypothesis tests retain
integrity because of the greatly reduced degrees of
freedom (with the consequently higher F-level re-
quirementsto achieve statistical significancelevels).
RESULTS
The Effects of Selling/AdvertisingOrder
A standardbelief in marketingis that "advertising
paves the way for selling." The authorshypothesized
that there would be an interaction effect between
advertisingand selling but, in contrastto the standard
belief, predicted that in this situation the positive
interaction would occur only in the reverse order:
selling before advertising.
The key resultsare shown in Figure1. First examine
the effects of advertisingalone. The advertising-only
controlgroup results are demonstratedby the control
group line at the bottom. The behavioral intention
proportionincreased from 5% to almost 23% over
the three exposure conditions. Tested by means of
a single-variableregression, the slope of the control
group response function is a significant linear trend
over direct mail exposures (regression coefficient >
0, t = 3.241, with 2 d.f., p < .025).
Figure 1
EFFECTSOF SELLING,ADVERTISING,AND THEIRORDER
IPe'rceL
I lI.:lv i I r11 I

levels in the hierarchy of effects [12]-cognition, I 80LcLion

affect, and conation-by using several measures for 80

each level. However, only the measures of conation 75.0

(or, behavioralintentions)are relevantto the hypothe-
ses discussed here. The question on respondents' 70 _
/
behavioralintentionsnamed five social health organi-
zations (among which was the Red Cross) and read
as follows.
60
/
If you were going to do volunteer work for one of i =;1 I s;il.ts cnl't:cLt Iruuips
/' .5
these organizations, which one would be your first 50
I}p- dlvd Lut
cl l il)'.
choice? Second choice? /t' I i II',
1,2. I

Behavioral intentions were coded as the proportion 40

/ ps L-.;lve rt ts i ll-,
p.
of respondentsratingthe Red Cross as first choice. /3r')/ , ,
31.3
30
Analysis Strategy 30.0
Analysis of covariance was the key analysis tool, 26.3 2.'. 7

with treatment means as the data points. The use 20

17.4
I .'./l
of cell means rather than raw data as input values 1*
tOt'l.' lElul
1, . I. t, 1 il11
follows from the difficultyin stabilizingerrorvariance 10 /
fromrawdata values when proportionaldataare used. /
Use of cell means instead of raw data as input
values does not completely eliminate the problem of i.05.0(

heteroscedasticity,but within the range of cell sizes 0 I 2 4

of this study (the 90%rangeamongexperimentalcells
is 6.91 to 14.45), there is little need to move toward Noto: oniH r
:rtil gIr,oup - 1l r 1-pilu
r-s ponii)ullts, poutl-;t
psp A r(ll1 . pir-d(vr
. '1iinn -i
lve.rl i:iqt'
i IX lugi,
i u77':
i)'.
74 r ('potdint xL,
r
aill
the more complex generalized least squares model. s les onit ;i L grolt lps =- 1 51 rs pol i t ; .

This content downloaded from 193.0.65.67 on Thu, 22 Oct 2015 11:54:50 UTC
All use subject to JSTOR Terms and Conditions
ADVERTISING-SELLING
INTERACTIONS
Next look at the results of personal selling contacts
in general (total of both pre-advertisingand post-ad-
vertising sales calls groups, shown by square data
points). At zero advertising and just one sales call,
15.4%claim behavioralor volunteeringintention.This
percentage increases to 28.6, 35.9, and 51.5 over
advertising exposures. The main effect of personal
selling is significantlygreaterthan the control without
selling (F = 42.600, 1 and 4 d.f., p < .003), but there
is no statisticallysignificantinteractionover the expo-
sure conditions. In other words, the slope of advertis-
ing effect does not differ between the sales contact
andcontrol groups. Clearly, in this setting, an interac-
tion between sellingand advertisingdepends on which
comes first.
The hypothesis that only selling before advertising
would producean interactionin this situationis based
on the attributiontheory idea that a personal sales
call can cause an attribution.In other words, the sales
call could form the basis for a self-perception should
the respondent mentally argue, "Why am I listening
here when there is nothing forcing me to? It must
be because I am the sort of person who participates
in causes like this." Such sensitization is expected
to cause each subsequent advertisement to have a
greater incremental effect than it would for a non-
sensitized respondent.
The primary test of this hypothesis is shown in
the results of all sales-calls-before-adsgroups (pre-ad-
vertising) as they compare to the advertising-only
control groups. The hypothesis is strongly supported.
The maineffect of the pre-advertisingselling contacts
is substantial (F = 25.545, 1 and 4 d.f., p < .007).
Note that the combination of a personal contact
followed by a single advertisingexposure appears to
have greater effectiveness than as many as four ad
exposures alone. Note also that at the zero advertising
exposure level the personal contact seems to have
little effect, a result likely due to the relatively long
spanof time between the pre-advertisingsales contact
and measurement. But despite this timing artifact,
when the ads begin to appear, responses begin to
soar. This outcome is visible in the prominentinterac-
tion effects between the two treatment conditions.
These effects are statistically significant(F = 8.636,
with I and 4 d.f., p < .04).
Very different results are predictedfor the post-ad-
vertising sales calls. An interaction is not expected,
becauseadvertisingis not likely to leadto an attribution
of self-perception of being committed. Although ad
exposure has self-perceptionpotential, it seems slight
comparedwith that arisingfrom a personal sales call.
A direct-mailexposure(as used in this research)would
requirethat the letter be opened and that it be carefully
read. A mere advertising impression from a quick
glance at the advertisements would be insufficient.
The likelihood of such careful reading seems much
lower than the likelihood of attention to a personal
This content downloaded from 193.0.65.67 on Thu, 22 Oct 2015 11:54:50 UTC
All use subject to JSTOR Terms and Conditions
513
sales call. Because sensitization may be lower, an
advertising-sellinginteraction is unlikely. Were the
"product" less well known than the Red Cross,
perhapsthe educationalimpact of the ads appearing
before selling would have sufficient effect to cause
an interactionwith personalselling. But in the present
instance it seems likely that advertisingwould have
minimaleffect in sensitizingrespondentsto the subse-
quent sales calls. At most, any interactionis expected
to be nonsignificant.
The post-advertisinginteractioninvolving personal
selling was tested by means of comparisonsbetween
the ads-before-sales-callsgroups(post-groups)and the
control group. Figure 1 shows the predicted main
effects (F = 13.652, with 1 and 4 d.f., p < .021).
The interaction,as predicted, is not significant(F =
1.000, with 1 and 4 d.f., n.s.). Advertising exposure
in the control group has a positive upward effect;
in the selling-after-advertisinggroups it appears to
have scarcely any effect at all. The response slope
for the selling-after-advertisinggroups is not signifi-
cantly positive (t = .418, with 2 d.f., n.s.).
The complete lack of advertisingresponse for the
selling-after-advertisinggroupsis troubling.One would
like to believe that advertisingwould have a measur-
able effect, but the datafail to supportthis hypothesis.
One mightattemptto explainthis lack of advertising
response by pointingout that a saturationeffect-and
perhaps some psychological reactance-might have
occurred.The selling-after-advertising groupsreceived
up to four advertising exposures and then, two or
so days later, received a personal sales contact. This
sales contact may have overcome the effect of adver-
tisingby bringingall advertisinggroupsup to the same
level. The interceptfor the treatmentconditionis high
because of the personalcontact, but the slope is flat.
A check on this explanationwas madeby comparing
resultsof the one-mailingad exposure conditionwhen
the mailings were made early and when they were
made late. A chi squareanalysis over the four timing
positions shows no significant effect (chi square =
1.49, 3 d.f., n.s.). Thus the explanationis weakened
somewhat,andthe superiorityof the selling-firstorder
is demonstratedin this situation.
It seems that careful attention should be paid to
the situation in which advertisingmay pave the way
for a sales call. A marketerusing advertisingbefore
sales calls in a situation similar to the present one,
in which the brand or organizationis already very
well known, would be spending money inefficiently.
In such a situation in the present study the effects
of the two promotionaltools merge at the four-expo-
sure level and advertisingalone is nearly as effective
as advertisingfollowed by personalcontact (chi square
testing response differences at four advertisingexpo-
sures = .223, with I d.f., n.s.).
Naturally, in different situations with a less well-
known sponsor the relationships may be different.
514 JOURNALOF MARKETINGRESEARCH,NOVEMBER1977

But, at the very least, one should not assume that Figure 2
advertisingalways paves the way for the sales call, EFFECTOF LABELINGAND ADVERTISING
nor that such a strategyis always cost-beneficial.
Additional support for the "selling first is best" 'c rCCelt
llchavior:l .
idea is found in a direct comparisonof the pre-adver- InLLel in)

tising and post-advertisinggroup results. Note first 80

thatthis comparisonis confoundedby recency effects.
Personal contacts precede measurements by two
weeks in the pre-groupsand by about two days in 70
the post-groups.
The recency effect is especially noticeable at the
zero advertisingexposure level, where identical per- 110
5b .7
sonal contacts are the only communicationtool and
differ only in proximityto the collection of dependent 50 _ /
measures.The intentionpercentages,pre-group5.0%
and post-group26.3%, show a strong recency effect
(chi square = 3.412, 1 d.f., p < .10).
Despite the recency effect, however, note the pow-
erful upward sweep of the pre-group'sresults from
30/
zero to the one-exposure level. A mere swamping 28.6,
/t. 1 t'td 2 6. 3
effect of the post-selling calls or remindereffect of 11)11,11)1lI)e

advertisingcannot explainthe fact that the pre-adver- 20

_ . / 3 I .7'
tising groups reach a much higher criterionvalue of - /- / //733io I- 19.? .^2III-.

behavioralintentions than either the post-advertising

or controlgroups(at a four-exposurelevel comparison - l1.7 /
of pre and post conditions, chi square = 6.86, with
1 d.f., p < .01). If the pre-postinteractionis examined 1).0

only with regard to one-four exposure conditions it

4
is not significant. But it is significantwhen the zero 0) 1 2

ad condition is included (F = 12.3, 1 and 4 d.f., p

Nuimbll'1 f Adv'l Li ': i ); 'Xl.l4,;r 11f'i

< .025). N t.I l.4 Y I i'ti II ;: t': I ;IIt'Jt,<l

tlildt. [lIi, 1 r-" 2(I,[[[1;it,
t':tp~H,tlt'll:;,, 14.,

This interactiondemonstratesthe wide differences

thatcould come from a promotionalprogram,depend-
ing on just where in the calendarpersonal selling is
positioned with respect to the advertising. Marketing
practitionershave grown to expect that advertising
"warms the prospect up" for the personal sales call
and have often based their promotionalprogramson
this premise. Advertising has educationalvalue and
persuasiveimpact, they mightargue, which open the
door for subsequent sales calls. Though this study
is not an attemptto refute this reasoning,the authors
propose that such an order is not always optimal,
especially in situationswhere the advertiseris already
well known.
Sales Call Quality:LabelingEffects
Researchfindings [10, 14] suggestthatan individual
labeled as charitable may begin thinking of himself
as a more charitable person. Congruent behavior
should follow, as a "charitableperson" should be
more sensitive to charitableappeals, should be more
likely to note and recall them, and should be more
likely to act upon them.
The data for the labeling treatmentare illustrated
in Figure 2, where the control group data also are
plotted for reference. The effect of labelingis signifi-
cant, as predicted (F = 26.308, 1 and 4 d.f., p <
.007). Though labeled groups differ greatly from the
control group (main effect with F = 61.750, 1 and
4 d.f., p < .002), the unlabeled groups show only
moderateimprovementover the control group (main
effect with F = 4.511, 1 and 4 d.f., p < .10). The
results further demonstrate the degree to which in-
teraction between personal selling and advertising is
affected by selling strategy, as interactioneffects are
also significantbetween label and no-label conditions
(F = 9.923, with 1 and 4 d.f., p < .035). Significant
and similar label/no label results also were found
for the pre-advertisinggroups taken alone.
This finding illustrateshow heavily dependent the
success of a sales call is on the techniques used by
the salesperson, and it indicatesthat the benefits from
increasedexpendituresfor sales trainingprograms-
aidingthe salespersonin developingan effective sales
presentationstrategy-may be dramatic.
These data support the concept that personal sales
calls of particulartypes can pave the way for advertis-
ing messages. Not only will an improved selling
strategyhave a main effect, but its effect when used
with subsequent advertisingmay be synergistic. The
sales call strategy seems as importantas whether the
sales call is made at all. Firms using personal selling

This content downloaded from 193.0.65.67 on Thu, 22 Oct 2015 11:54:50 UTC
All use subject to JSTOR Terms and Conditions
ADVERTISING-SELLING
INTERACTIONS
and advertisingmay benefit more from sales training
than firms using personal selling alone.
CONCLUSIONS AND IMPLICATIONS
Advertising-Personal Selling Order Effects
This study demonstratesa situation in which per-
sonal calls do appearto pave the way for advertising
instead of the reverse common assumption. Attempts
to use this finding in marketingpractice, however,
involve problems. An advertisement cannot usually
"close" a sale whereas a personal call can. As a
universalpractice, sales calls before advertisingmay
lead to inefficiencies, but there are some contexts
to which the practicecan be generalized.For instance,
advertising should follow every missionary contact,
trade show contact, detailmancontact, social solicita-
tion, canvasser-all personal sales contacts which do
not includea sales closing. Particularattention should
be paid to contacts in which the prospect has been
allowed to draw dispositional conclusions about his
attentiveness to the contact.
Most sales calls, however, do include attempts to
close the sale. The present research can be applied
here also. Many prospects fail to buy, not because
the sales call was ineffective or the product not
appropriate,but because the timing was wrong, the
funds were not available, other membersof the buying
committee were inaccessible, or for other similar
reasons. Advertisingdirected to these prospects can
have a powerful effect and can lead to "call-in" sales
or more effective subsequent sales calls.
In addition, the similaritiesbetween personal calls
and other promotionalmethods shouldnot be ignored.
Other promotionaltools may provide the base from
which a prospect can draw self-attributions, which
leadthe prospectto increasedsensitivityto subsequent
advertising. Among these tools are product samples
(especially those involving some behavioral commit-
ment, ratherthan free and unsolicited samples), con-
tests, cents-off coupons, and introductory offers.
Thus, usingthese promotionaltools before advertising
may actually have a more positive effect than using
them after advertising.
The importantfactor is that these promotionsform
the basis for an attributionaboutthe self-the prospect
should be led to argue, "Why did I do this? It must
be because I am inclinedtowardthis product." Sensi-
tization such as this will lead to higher advertising
receptivity and purchase response.
Personal Selling versus Advertising
Eldersveld and Dodge [6] compared the effec-
tiveness of advertisingandpersonalsellingin a political
marketingcontext. Although their sample size was
small and confounded by experimentalattrition, and
no significant tests were reported, they did conclude
that a single personal call was more effective in
This content downloaded from 193.0.65.67 on Thu, 22 Oct 2015 11:54:50 UTC
All use subject to JSTOR Terms and Conditions
515
changingbehavior than four direct promotionalmail-
ings.
Theauthors'researchwas not intendedas a re-exam-
ination of the Eldersveld and Dodge study, but it
does tend to supporttheirconclusions. In the post-ad-
vertisinggroup, for instance, where timing consider-
ations permit a comparison, a single personal call
equalledthe effectiveness of four direct mailings(26.3
versus 22.7%volunteeringintention).
This findingspeaksboth well and poorly of personal
sales calls. It establishes that personal selling can be
much more effective than advertising but because a
single personal call typically costs much more than
a single direct-mail promotional package [13], the
study suggests that a marginaleconomic budgeting
approachto promotionalallocations would cause per-
sonal sellingto yield to advertising.It shouldbe noted,
however, that this research comparison is biased in
favor of advertising,because in the present study the
sales callers did not use their power to close the sale,
and the advertising was aided by a stimulated closing
(the telephone interview).
Labeling
The present work demonstrates that a label can
have a powerful effect on behavior. In some cases,
the effects of the label were still being exhibited two
weeks after the labeling treatment was applied, and
presumably persisted well beyond the two-week peri-
od.
Although the research did not include measurement
of self-attributions resulting from the "charitable"
label, the self-perceptionexplanationis a compelling
one. It seems that discountingcues were absent and
that the label was accepted by the respondent as a
valid observation about herself. Thus she "learned"
about herself-"she is charitable"-and acted in a
mannerconsistent with this new understanding.
To a degree, a salesperson may use labeling as a
tool with similar effect. An industrial buyer could
be labeled, "concerned about employee safety," a
retail consumer could be labeled (possibly through
samplingor couponpromotionalmaterial),"interested
in the nourishmentof her family." If the salesperson
can avoid being perceived as manipulative (which
would be a discounting cue), the effect of the label
may be dramaticin increasingthe prospects' receptiv-
ity to the promoted product. Labeling can be a
creative tool for every salesman, and will increase
in effect as discountingopportunitiesare minimized
and sincerity and directness are maximized.
REFERENCES
1. Aronson, Elliot and D. R. Mettee. "Dishonest Behavior
as a Function of Differential Levels of Induced Self-Es-
teem," Journal of Personality and Social Psychology,
76 (January 1963), 31-6.
2. Bern, Daryl J. Beliefs, Attitudes, and Human Affairs.
516
Belmont, California: Brooks/Cole, 1970, p. 54.
3. . "Self-Perception Theory," in L. Berkowitz,
ed., Advances in Experimental Social Psychology, Vol.
6. New York: Academic Press, 1972, 1-62.
4. Business Week. "Advertising Saves Sales Calls," (De-
cember 5, 1959), 69-70.
5. Clayton, Alden. Personal communication, 1975.
6. Eldersveld, Samuel J. and R. W. Dodge. "Personal
Contact or Mail Propaganda? An Experiment in Voting
Turnout and Attitude Change," in D. Katz, D.
Cartwright, S. Eldersveld, and A. McLung Lee, eds.,
Public Opinion and Propaganda. New York: Dryden
Press, 1954, 532-42.
7. Festinger, Leon A. A Theory of Cognitive Dissonance.
Stanford: Stanford University Press, 1957.
8. Freedman, Jonathan L. and S. C. Fraser. Compliance
Without Pressure: The Foot-in-the-Door Technique,
Journal of Personality and Social Psychology, 4 (March
1966), 195-202.
This content downloaded from 193.0.65.67 on Thu, 22 Oct 2015 11:54:50 UTC
All use subject to JSTOR Terms and Conditions
JOURNALOF MARKETINGRESEARCH,NOVEMBER1977
9. Kelley, Harold H. "Attribution in Social Psychology,"
in David Levine, ed., Nebraska Symposium on Motiva-
tion. Lincoln: University of Nebraska Press, 1967.
10. Kraut, R. E. "Effects of Social Labeling on Giving
to Charity," Journal of Experimental Social Psychology,
9 (1973), 551-62.
11. Morrill, John E. "Industrial Advertising Pays Off,"
Harvard Business Review, (March-April 1970), 4.
12. Ray, Michael L. "Marketing Communication and the
Hierarchy-of-Effects," in Peter Clarke, ed., New Models
for Mass Communication Research. Beverly Hills, Cali-
fornia: Sage Publishing, 1973, 147-76.
13. Sales Manpower Foundation. New Cost of Selling Survey
of the Country's Eleven Major Manufacturing Industries.
New York: Sales Manpower Foundation, 1970.
14. Steele, C. M. "Name Calling and Compliance," Journal
of Personality and Social Psychology, 31 (August 1975),
361-9.