Professional Documents
Culture Documents
19 4 337
19 4 337
are available. Among the sources which influence a very wide arc of the
individual's behavior with very little respect for the boundaries of subject
matter are enduring networks of social contacts. Thus, for example, in
politics, neither the newspaper editor nor even the ward heeler appear to
affect the individual's vote as powerfully as his parents, his spouse, or his
ethnic and religious loyalties.1 .'
But personal inmience is not the only aspect of interpersonal relations
which is relevant to an understanding of the flow of mass communications
and their effects on individual decisions. The extent of an individual's
integration in a group may be an important variable, for example. Thus,
one study has shown that children who are relatively well integrated in
groups of peers tend to make quite, different use of adventure stories on
the radio than children who are relatively isolated from such contacts.8
Or, an earlier study has shown that Allied propaganda to German troops
did not take effect until the soldier was cut off from his intimate, inter-
personal ties—when his own small unit of peers and non-commissioned
officers broke up. 4
If "belonging or not belonging" is a key variable, so is belonging to one
group rather than another. An individual is.ordinarily reluctant to depart
from the norms of his particular group, unless the departure itaelf receives
some form of group support, and a communication aimed at influencing
his thoughts or actions may therefore fail. When changes occur, it is
usually only when the individual perceives that his group approves, or
that support comes from a dissident sub-group, or from an outside group
toward which the individual sees himself moving or whose presumed
standards he accepts.8
In our study, however, we first interviewed all the local doctors we could
and, by means of three sociometric questions, determined the network
of social relations among them without reference to any one particular
decision. The channels of information and influence which entered into
a recent drug decision were also ascertained in all interviews, and com-
pared with the doctor's position in the social network. Only then did we
superimpose on the social network the flow of adoptions of a new prac-
tice, as indicated by the precise sequence of dates on which each respond-
ent introduced a certain new medication. This latter information was ob-
tained through an examination of the prescription files of local pharmacies.
shares an office with another may feel less hesitant about the risk of trying
out new drugs in general, and not merely because of anything the other
doctor may have said about any particular drug; or, a doctor whose
friendship is not reciprocated by other doctors may develop quite a dif-
ferent relationship to other rhgnnpU of information—he may become
more friendly with the pharmaceutical reipuman, for example—than the
doctor who is well integrated in the medical community. Thus, a doctor's
association with other doctors may serve (1) as an important source of
information and influence on a particular innovation, and (2) as an
important determinant of his response to innovations in general, and to
information and influence wnanaring from other sources.
For this reason, the socdometric data will be related to doctors' behavior
in two basically distinct ways. The first of these is the examination of
certain characteristics which differentiate individuals who have received
many and few sotiometric choices. What, for instance, is the use made of
Each respondent was asked how he had learned about two of the drugs
he had recently adopted. Table 1 cross-tabulates the replies with the
popularity of the doctor among his colleagues. An interesting pattern
TABLE 1 *
DRUG-TALK CHOICES AND CHANNELS EMPLOYED IN DECISIONS ON TWO RECENT
DRUG ADOPTIONS
Numbtr of Drug-Talk Choitts Rtcavtd
Mont One, Two ThxttorMort
Mail and periodical! from drug houses 30% 18% 21%
Articles in journals 10 39 32
Detail men (salesmen) 40 25 21
Colleagues 15 15 16
Meetings 0 0 11
Other rninnflii 5 3 0
100% 100% 100%
(Number of fhannMn) (20) (34) (19)
T h e percentages are based on the total number of channels mentioned, which exceeds
the number of doctors in each category, since many doctors reported on two decisions, and
the average number of channels per decision was 1.88.
emerges. Journal reading plays a much larger role in the drug adoptions
of doctors who receive one or more designations than among the isolates.
This conforms to the hypothesis of the "two-step flow of communications,"
which states, in essence, that messages originating outside of the individ-
ual's face-to-face group do not impinge on him directly, but are mediated
by a few members of his group, who expose themselves to messages from
the outside world more than their confreres.
The source of this hypothesis is in mass media research, where it has
been found that opinion leaders are in closer contact with the mass media
than their followers. But doctors can keep in touch with the outside
world in other ways than through the printed media—e.g., by attendance
at medical meetings in other cities. The two-step flow hypothesis therefore
leads one to expect that out-of-town meetings as well as journals would
be more important in the decisions of the most popular drug discussion
partners than in those of their less popular colleagues. Table 1 bears out
this expectation: only those who receive three choices or more report that
what they learned at meetings entered into their decisions. It would seem
that the star does indeed serve as a relay: he incorporates what he has
learned at the meetings into his own decisions and, it may be assumed,
into his conversations with others.
The drug decisions of the sociometric isolate, by contrast, are marked
by heavier reliance on commercial sources—direct mail advertising from
the pharmaceutical companies and the "detail man" as the drug house
representative is called. These sources, to be sure, also come from out of
town. What differentiates them from journals and meetings is that, being
commercial sources, they carry less prestige, and that it requires less effort
to keep up with them. Evidently the two-step flow hypothesis does not
apply to all channels from the outside world. Some of the channels do
impinge on most members of their intended audiences directly, and are
not disproportionately utilized by the opinion leaders.
Incidentally, the differential importance which the detail man has for
the social isolate is not limited to his role as a purveyor of information
about new medicines. Two of the four doctors who received no friendship
choices volunteered the information that they very often Calked with the
detail man on subjects other than drugs: "Well get off the track sometimes
on economics, politics, family affairs," said one. Another relatively isolated
doctor corhmented on the detail men as follows:
They are helpful—they know mil the doctor* in the communities around here and
give you all Che dirt and gossip and incidental news about what is going on amongst
the doctors in this community.
This pharmaceutical salesman evidently serves as a near-professional
companion for men who are relatively cut off from informal contacts with
other physicians.
A final observation from Table 1 yields a surprise. The classic model of
the two-^tep flow of communication provides that it is the rank-and-file
members who are influenced by the opinion leaders, while the latter
are influenced by the mass media. But in Table 1 "colleagues" are
mentioned exactly as frequently by opinion leaders as by the rest. The
most likely explanation is that the opinion leaders may themselves
turn to colleagues of even higher status, and that it may take three
or four steps, perhaps rather than two, before a level of leadership is
reached where dependence on personal contacts is markedly decreased.8
Revision of the model to allow for multistep flow of communications
would thus seem in order. A second revision has already been pointed out:
•Regarding the relatively low overall frequency of mentions of "colleagues" in Table
1, see note 12 below.
that messages from the outside world need not come through the printed
word or other mass media. In the case of physicians, we have seen that
attendance at society meetings may take over this function. As the multi-
step flow hypothesis is applied to still other publics, it may be found that
the opinion leader can keep in touch with relevant parts of the outside
world in still other ways. Finally, the model may not apply to channels
of low prestige and unusually easy accessibility.
One more finding, still tentative, regarding the channels of information
used in recent drug decisions is especially worthy of attention. Medical
problems differ widely in the degree to which their answers are well-
structured and generally established. Table 1 includes the reports of
doctors concerning their adoptions of a variety of different drugs. We can,
however, divide these into two broad classes: one class of drugs is appli-
cable to certain acute conditions which call for immediate action and
present a very small number of alternate methods of treatment; success
or failure is visible in a day or two. The other class of drugs is applicable
to chronic conditions where many dozens of treatments compete, and the
effectiveness of therapy is very difficult to gauge. We find that colleagues
constitute only seven per cent of the channels reported as leading up to
the adoption of a drug for the acute conditions but represent 22 per cent
of the channels concerned with the chronic diseases. In other words, the
role of colleagues increases substantially in the relatively more ambiguous
situation. In a sense, this recalls the studies of suggestion and influence by
Sherif and others: the role of personal influence is more important in the
relatively more unstructured situation.
TABLE 2
UtE OF CHANNELS OP COMMUNICATION IN DECISIONS REGARDING ACOTE AND
CHRONIC CONDITIONS*
TABLE 3
DATE or FIRST PRESCRIPTION OF A SPECIFIED NEW DKUO BY EACH DOCTOR*
Our original intention was simply to correlate early or late use of the
new drug with characteristics of individual doctors. But after Table 3 was
drawn up, we were intrigued by the strange alternation of slow periods
and spurts which it shows: first, three pioneers adopt the new drug at
three-week intervals. Then, suddenly, during a period of only eleven days
in February seven doctors start writing prescriptions for the new drug.
During the next three months, only four doctors start prescribing the
drug, at widely spaced intervals. Then, again, three doctors initiate use
of the drug within a period of only five days in mid-May. No further
doctors appear on the table after this, although our data are complete for
an additional month and a half. There thus remain six doctors who either
do not prescribe the drug at all, or do so only at an unknown time very
much later than any of their colleagues.10
How is this alternation of slow and fast periods to be accounted for?
One possibility is, of course, that the fast periods are those when there
is more need for this particular drug in the community—times of epi-
demics, for instance. However, the slow and fast periods do not coincide
with the known seasonal variations in the incidence of diseases for
which this drug is indicated. A second possible explanation for the alter-
nation of slow and fast periods is that the two fast periods correspond to
two sales campaigns that may have been carried on by the pharmaceutical
companies involved. This may be the case but fails to explain why certain
doctors responded to the first campaign, while others did not.
The third possibility is that the alternations can be accounted for by
social relations. During the initial slow period a few hardy individuals
try out the new product—perhaps watched with interest by their more
conservative colleagues. Each spurt then represents the almost simul-
taneous adoption of the drug by a well-integrated group of physicians.
This hypothesis, unlike the other two, would not only explain why spurts
occur at all, but also why it is the particular doctors who participate in
each spurt that act in unison—e.g., why Drs. A, B, C and D act simul-
taneously, but not Drs. D, E, and F.
We therefore decided to bring together the data on first prescription
dates with the data on social groupings. The result is Sociogram II which
is identical with Sociogram I, but adds information on the date of each
doctor's first prescription for the new drug. Different shadings denote the
time periods during which each doctor first prescribed it. Triangles indi-
cate doctors following specialties not normally prescribing drugs through
pharmacies. Solid circles indicate other doctors who did not prescribe
the drug at all during the period covered by the survey. Let us now trace
the flow of the innovation through this map of friendships among the
physicians. Two questions will occupy us as we do so: (a) are doctors who
adopt the drug on successive dates in contact with one another? (b) is each
spurt of adoptions located within one of the cliques?
"Data collected elsewhere mbsoquent to the termination of thii surrey make it likely
that most of these "diehards" did at least try out the new drug some time during 1954.
LEGEND
Flraf ftpttfOt* Ik* * . « :
O tafon I/IS/M Cr
tint » * • parted*)
(1) Initial slow period. The first three doctors are symbolized by blank
circles. Their initial prescriptions took place on December 1, December
22, and January 10.
(2) First spurt. Seven physicians represented by single hatched circles
adopted the drug between February 1 and 11. Four of them are in direct
contact with a pioneer (Le., a blank circle), and the remaining three have
second-order contact with a pioneer.
At the end of this February spurt, three out of four doctors in Clique C
have used the drug; so have six out of eight doctors in Clique A; but not
a single member of Clique B. (Surgeons and other non-prescribing special-
ists are excluded.) Even more striking than this contrast between the
proportions of each clique who have used the drug by February 11 is
another fact. One might think that the remainder of Cliques A and C
would follow suit after some lag—perhaps two or three weeks later. But
in fact only one of them ever filed a prescription for the new drug with
the pharmacies reporting to us throughout the period covered—i.e.,
through the end of June. In other words, after the first pioneering period
and the February spurt, only real diehards still hold out in Cliques A and
C.
(3) Second Slow Period. Several more doctors (recorded as cross-hatched
circles) adopt the drug during the next few weeks; their dates are widely
dispersed over a long period of time (March 4, March 29, April 9, and
April 16).
(4) Second Spurt. Then suddenly, between May 14 and 18, three doctors
in Clique B, represented by dotted circles prescribe the drug leaving only
two doctors as non-prescribers in Clique B. Again, these two are not merely
a little behind their colleagues, but they are "diehards," having no pre-
scriptions for the new drug on record throughout the period covered by
our survey.
We have now traced the sequence of prescriptions of a new drug
through our map of friendships among the doctors in this city, much as
doctors themselves do when they trace the "epidemiology" of an infectious
disease on the map of a city. What we have seen can be summarized in
three propositions:
(1) That half or more of the members of each clique who are ever to
adopt the drug do so within a few days of one another.
(2) That for each of the three cliques, it is possible to state a cut-off date
such that those members who have not adopted the drug by that date do
not adopt it at all during the survey period. Only one respondent deviates
from this generalization.
(3) That no one (except the three pioneers) adopts the drug unless he
has a direct sociometric contact with a doctor who adopted it before him.
Three respondents deviate from this generalization, and it is therefore
restated as follows: that drug adoptions on any particular date are more
frequent among doctors who are in direct sociometric contact with others
who have already adopted the drug, than among doctors who lack such
contact.
Tables 4 and 5 constitute quantitative expressions of the empirical
deviations from these three propositions. Table 4 shows no deviations
from the first proposition. It also shows that the one doctor who actually
deviated from the second proposition and prescribed the drug after the
cut-off date for his clique constitutes but 14 per cent of the deviations that
could have occurred. Table 5 tabulates the deviations from the third
TABLE 4
CLIQUES AND SPURTS* OF DRUG ADOPTIONS
A 2/10 6 2 4 67% 3
B 5/18 4 1 3 75 2
C 2/11 4 1 2 50 V 25% 50%
All Cliques 14 4 9 64% 1 7% 14% 5
Neutrals 3 1
Total 17 6
•"Spurt" is defined as the seven days ending with the cut-off date. Doctors following
specialties not normally prescribing drugs through pharmacies are excluded from Table 4.
TABLE 5
ADOPTIONS OF THE N E W DRUO AMONO DOCTORS W H O , ON THE D A Y THE
ADOPTION OCCURRED, WERE AND WERE NOT IN DIRECT SOCIOHETRIC CONTACT
WITH ANOTHER DOCTOR W H O HAD ALREADY ADOPTED I T *
Number of Doctors whs, on the date indicated, had not yet adopted the new drug and:
Feb. 1 0 6 6 1 7 8 14
Feb. 5 2 5 7 0 6 6 13
Feb. 6 1 6 7 0 4 4 11
Feb. 10 1 5 6 1 3 4 10
Feb. 11 1 4 5 0 3 3 8
Mar. 4 0 4 4 1 2 3 7
Mar. 29 1 3 4 0 2 2 6
Apr. 9 1 2 3 0 2 2 5
Apr. 16 1 2 3 0 1 1 4
May 14 1 1 2 0 1 1 3
May 18 2 0 2 0 0 0 2
After 0 0 0
Total num-
bers 11 38 49 3 31 34
Total per
cent 22% 9%
•Doctors who never adopted the drug during the period covered by the survey, as well
as doctors following specialties not normally prescribing drugs through pharmacies, are
excluded from Table 5, as are the three earliest adopters. The results are not substanti-
ally altered when the length of the intervals between the dates in Table 5 is taken into
account and the three earliest adopters are added.
influential did not have significantly higher adoption rates than non-
leaders; in "high adoption" areas, however, the leaders were far ahead-18
In summary, we have found that the bringing together of independently
established information on social relationships, on innovating behavior,
and on the use of channels of information in decision-making supports the
hypotheses of the role played by face-to-face contacts in mediating inno-
vations from the world outside.
We have seen that the notions of the importance of interpersonal
relations are applicable to a case of decision-making among specialists
concerning matters based on scientific findings where well-recognized
expert sources of knowledge exist. This, we noted, is particularly the
case when decisions are required for relatively ambiguous situations.
Finally, we have found it necessary to propose amendments for the
model of the two-step flow of communications: by considering the possi-
bility of multi-step rather than two-step flow; by noting that sources other
than printed publications may be the channels to the outside world main-
tained by the opinion leaders; by noting that the model may not apply
to channels of low prestige and unusually easy accessibility; and by
differentiating various kinds of leadership, especially by emphasizing the
differential roles of the innovator or pioneer on the one hand and the
opinion leader or arbiter on the other.
"See E. A. Wilkening, "Informal Leaden and Innovators in Farm Practices," Rural
Sociology, Vol. 17, pp. 272-275; H. F. Lionberger, "Some Characteristics of Farm
Operators Sought as Sources of Farm Information in a Missouri Community," Rural
Sociology, Vol. 18, pp. 327-338; C. P. Marsh and A. L. Coleman, "Farmers" Practice
Adoption Rates in Relation to Adoption Rates of Leaders," Rural Sociology, Vol. 19,
pp. 180-181. These three studies are excellent representatives of a research tradition of
the greatest importance for undents of communication.