Você está na página 1de 13

THE INFLUENTIALS: BACK TO THE

CONCEPT OF OPINION LEADERS!

GABRIEL WEIMANN

Abstract The concept of opinion leadership has been related to


a lingering theoretical and methodological debate. The present

Downloaded from http://poq.oxfordjournals.org/ at New York University on October 12, 2014


study examines a new measure to identify opinion leaders (the
Strength of Personality Scale) developed by the Allensbach Sur-
vey Center in Germany. The results of applying this measure in
Germany and Israel allowed the testing of its validity and effi-
ciency. The findings, however, do not suggest that the influentials
identified by this scale are opinion leaders according to the origi-
nal conceptualization, but fit better the more sophisticated char-
acterization of opinion leadership that stemmed from modifica-
tions of the original concept of opinion leaders.

The mass do not now take their


opinions from dignitaries in
Church or State, from ostensible
leaders, or from books. Their
thinking is done for them by men
much like themselves, addressing
or speaking in their name, on the
spur of the moment. . . .
—John Stuart Mill, On Liberty*

About 100 years after John Stuart Mill had first suggested the impor-
tance of personal influence and the role of influentials, several studies
provided the first empirical evidence of these concepts (Berelson, La-
zarsfeld, and McPhee 1954; Katz 1957; Katz and Lazarsfeld 1955;
Lazarsfeld, Berelson, and Gaudet 1948; Merton 1949). The role of
GABRIEL WEIMANN is Associate Professor of Communication and Chairman of the De-
partment of Sociology at the University of Haifa, Israel. The author gratefully acknowl-
edges the important contributions of Elisabeth Noelle-Neumann, Hans Eysenck, Hans-
Bernd Brosius, and Robert Eisinger. The research was supported by the Alexander von
Humboldt-Stiftung, Bonn, Germany.

1. The full sentence goes on to indicate that this flow of interpersonal influence is
mediated by the newspapers.
Public Opinion Quarterly 55:267-279 © 1991 by the American Ajsocialioa for Public Opinion Reiemrch
Published by The Univenily of ChJago Press / 0033-J62X/91 /O055-O2rt2.5O
268 Gabriel Weimann

interpersonal relations in the flow of information and influence, as


revealed by these studies, caused a growing interest in personal net-
works and in key positions in these networks. Consequently, many
studies focused on influential positions or the "opinion leaders." Sev-
eral hundred opinion leadership studies (see reviews by Rogers and
Shoemaker [1971] and Kopller [1984]) have attempted to identify the
characteristics of opinion leaders in terms of demographic and socio-
economic variables, media exposure, social positions, and personality
traits.2
The measurement of opinion leadership was subjected to empirical

Downloaded from http://poq.oxfordjournals.org/ at New York University on October 12, 2014


criticism and debate (e.g., Gitlin 1978; Katz 1980, 1989; Lin 1973;
Robinson 1976; Weimann 1982). This criticism, combined with the
need for more sophisticated and validated measurement of opinion
leadership, has led to the development of new measurement proce-
dures. One of the recent attempts is the research project conducted
by Noelle-Neumann and the German survey research center, the Al-
lensbach Institut fiir Demoskopie, on a new instrument to identify
opinion leaders, the "Strength of Personality Scale" (PS scale).
The present study is based on applying this scale to different samples
in two societies and testing its validity by various measures, including
a comparison with measures of actual flow of information and influence
based on sociometric mapping of the personal network of a small com-
munity.

Method
The first scale to be called the "Strength of Personality Scale" {Per-
sonalichkeitsstarke) emerged from testing numerous questionnaire
items related to self-perceived levels of personal influence. (The Ger-
man news magazine, Der Spiegel, initiated the development of the new
measure. Interested in what they referred to as "the active consumers
who set standards in their community," managers of Der Spiegel chal-
lenged the Allensbach Institut with the task of developing an instru-
ment that would identify the influentials.) These early scales were
tested and refined after years of pretests with a variety of samples (see

2. Several studies focused on the level of education as it related to opinion leadership


(e.g., Kingdon 1970; Reynolds and Darden 1971; Saunders, Davis, and Monsees 1974;
Summers, 1971), or on gender (e.g., Booth and Babchuck 1972; Eurich 1977; Kingdon
1970) or social class (e.g.. Booth and Babchuck 1972; Summers 1971). Other studies
tried to relate personality traits such as conformity, social insecurity, responsibility,
motivation, anxiety, and emotional stability to the status of opinion leaders (e.g., Brett
and Kernalegun 1975; Robertson and Myers 1969; Saunders, Davis, and Monsees 1974;
Schrank and Gilmore 1973; Summers 1971).
Back to the Concept of Opinion Leaders? 269

Noelle-Neumann 1983, 413-15; 1984 and 1985). The final scale was
established after a factorial reduction of a 34-item questionnaire admin-
istered to a representative sample of 3,542 residents of Germany. The
resulting scale includes 10 items that were later weighted according to
their part-whole correlations with the total scale, with the scores on
this scale ranging from 75 to 149 (see table 1). The respondents were
divided into four approximate quartiles, thus yielding four levels of
personal strength (for a detailed description of the questionnaire, in-
cluding wording of all items and the respondents' breakdowns, see
Noelle-Neumann 1983).

Downloaded from http://poq.oxfordjournals.org/ at New York University on October 12, 2014


The procedure used by the Allensbach Institut to validate the sample
involved the comparison of the scale ratings with the interviewer's
impression (whether the respondent "radiates strength and energy" as
opposed to "being boring," whether the respondent gave the impres-
sion of "being a model for others" and whether he or she appeared
"self-assured" during the interview). The comparison of the results
revealed that interviewers' ratings and the PS scale ratings are highly
correlated (Noelle-Neumann 1985, 11). However, this test is weakened
by the possibility that the interviewers were influenced by the respon-
dent's self-assessment. The need to validate the scale by an "external"
criterion led to the present study.
The scale was administered to two Israeli samples. The first (N =
650) was randomly drawn from the adult, Jewish population of Israel.
The respondents were interviewed at home with a prestructured ques-
tionnaire that included 34 statements, among which were the final 10
PS scale items (the respondent had to choose those cards that carried
statements suited to his or her self-assessment, thus dichotomizing the
answers). Additional questions included sociodemographic character-
istics, measures of community participation, media exposure, func-
tioning as advice givers in various areas, patterns of sociability, and
social involvement. This first sample was used to examine the applica-
bility of the PS scale to another society and to study cross-cultural
differences.
The second sample was an Israeli kibbutz community (TV = 270)
that underwent a sociometric mapping of its personal communication
network and of the flow of information and influence in this network
(for a detailed review of these studies, see Weimann 1982, 1983; also
see Granovetter 1981, 29-30). A sociometric mapping of the personal
communication network was obtained by interviewing every member
of the community: each respondent was asked to list his or her conver-
sational ties with other members (unlimited number of choices) and
was instructed to rate the strength of each tie by three scales (impor-
tance, frequency of contacts, and tenure of the tie), thus operationaliz-
ing Granovetter's definition of ties' strength. Every tie was rated by
270 Gabriel Weimann

T a b l e I . The Personality Strength (PS) Scale and Weighting

Weight
Item Yes No

1. I usually count on being successful in everything I do 13 7


2. I am rarely unsure about how I should behave 14 7
3. I like to assume responsibility 15 7
4. I like to take the lead when a group does things together 17 8
5. I enjoy convincing others of my opinions 15 7

Downloaded from http://poq.oxfordjournals.org/ at New York University on October 12, 2014


6. I often notice that I serve as a model for others 16 8
7. I am good at getting what I want 14 7
8. I am often a step ahead of others 18 9
9. I own many things others envy me for 15 9
10. I often give others advice and suggestions 1? 6
Maximum score 149
Minimum score 75

NOTE: The weighting procedure is detailed in Noelle-Neumann 1985 (5-7).

the two connected persons and the scores given by each member of
the dyad to the same tie were summed, thus adding the element of
mutuality. The sociometric data was arranged in a who-to-whom ma-
trix with each entry (N = 2,511 ties) representing a communication
tie, characterized by its strength. The entire matrix was subjected to a
clique identification procedure, obtained by a computerized sequential
reordering of the matrix (see Richards 1975, 1977; a detailed example
is provided in Rogers and Kincaid 1981, 163-82).
To study the actual flow of use of a tie for communicating informa-
tion, six different items were used, two from each category: "general
news," "consumer information," and "gossip" (for a description of
the items, see Weimann 1982). None of these items was publicized
formally, but were disseminated a month prior to the study by means
of interpersonal communication. To trace the flow, each member was
asked whether he or she was aware of each of the items and to report
on the source of his or her knowledge (who told him or her). Two
decision-making items were used to measure the flow of influence.
These were two issues later put to vote in the assefa (the general
assembly): a decision to allow young members, after their military
service, to spend a year outside the kibbutz, and a decision to accept
or reject the application of a candidate for full membership. Each mem-
ber was asked about consulting, seeking advice, or being influenced
when making up his or her mind regarding these issues. The entries
Back to the Concept of Opinion Leaders? 271

about flow of information and influence were added to the socioma-


trices, in order to identify network positions and their activity in the
flow of information and influence. Finally, we applied the PS scale and
cross-classified this measure with the communication network attri-
butes (position and activity) for each subject.

Findings

The first step involved the cross-cultural comparison of the scale's

Downloaded from http://poq.oxfordjournals.org/ at New York University on October 12, 2014


results by applying several statistical procedures to the data collected
in the German sample (N = 3,542) and the first Israeli sample (TV =
650). A factor analysis of the responses to the 34-item questionnaire
was performed for both samples (using a principal components solution
with varimax rotation). It reveated that, in both samples, the same 10
items used for the PS scale emerged as strongly loaded on one factor,
accounting for about 41% of the variance in each sample. This suggests
stability across social and cultural settings. A split-half reliability test
for both samples yielded coefficients of .78 and .76 for the German and
Israeli samples, respectively. When the 10-item scale itself is factor
analyzed, two distinct factors emerge: a principle one accounting for
75% of the variance (72% in the Israeli sample) that includes seven
items (1, 2, 3, 4, 5, 7, and 10) and a second factor that includes three
items (6, 8, and 9). The first factor seems to measure internal sources
of influenceability, whereas the second describes "external" origins,
derived from comparison with other people (Noelle-Neumann 1985).
The respondents in both samples were divided into four groups ac-
cording to the quartile scores on the PS scale.3 Following the proce-
dure of the Allensbach studies, the relationships between the PS and
sociodemographic characteristics were examined. The breakdowns of
the German and the first Israeli samples, obtained by cross-classifying
PS groupings with various sociodemographic factors, indicate that PS
is related to various sociodemographic variables, and these relation-
ships are very similar for both populations. In terms of gender, there
is a sharp difference between males and females: males score higher
on PS (male-to-female ratio in the "strong" PS level is 1.47 in the
German sample and 1.88 in the Israeli one). PS is also unevenly distrib-
uted by age groups: the highest PS scores are more frequent in the
30-39 age group (in both samples) while the "weak" level is more
frequent in the 60 plus age group (23% of this age group are rated
"weak" in the German sample and 48% in the Israeli one).

3. The distributions of the two samples were different. Consequently, the quartiles do
not have the same ranges.
272 Gabriel Weimann

The strongest distinction is revealed by SES classification: a person


in the higher SES levels has more than three times as much chance to
be ranked as "strong" PS as a person in the lower SES levels. The
weaker PS persons are much more frequent among the lower SES.
Again, this pattern is almost identical for both populations. This rela-
tionship does not indicate direction of causality: opinion leaders may
achieve their leadership through social acknowledgment of their edu-
cation, income, occupation, and popularity, but they may also achieve
higher social positions because of their stronger personality.4
Although the PS scale yields similar results in two different societies,

Downloaded from http://poq.oxfordjournals.org/ at New York University on October 12, 2014


this does not indicate its validity. Is it really a measure of influence-
ability? The application of the PS scale to a social network that was
mapped by its communication links, personal positions, and flow of
information and influence may provide the "external" criteria for test-
ing the scale's validity.
The communication matrix included 2,511 entries connecting the 270
members of the kibbutz. The analysis of the 270 x 270 matrix structure
according to Richard's procedure (see Rogers and Kincaid 1981, 163—
82) revealed 16 cliques. Using the clique identification, we computed
the network position of each individual in his or her clique, by the
number of links with members of his or her clique. The division into
four quartiles (of number of ties) resulted in four levels of network
positions, from "marginals" to "centrals." This measure, originally
gathered for a network analysis (Weimann 1982, 1983), was com-
pared with the results of applying the PS scale to the same respon-
dents. The two measures were highly correlated: a correlation coeffi-
cient of .54 was found between the number of communication links
(in the entire network) and the PS measure, and this correlation was
somewhat higher (.59) when relating PS score to the number of com-
munication links within the clique (the difference between the coeffi-
cients is not statistically significant). Individuals with high strength of
personality are better linked to other individuals in the community and
especially to others in their social group. This relationship is clearer
when the groupings according to the two scales are cross-classified, as
shown in table 2.
Individuals with central positions in their groups are clearly those
with higher levels of PS: 58% of those who are centrally positioned in
the communication network rated as "strong" PS, while only 3% of
the "marginals" rated as "strong" PS individuals. The same applies

4. In the Israeli study, we included a question about military service (obligatory for all
Israeli men and women). The analysis revealed that the military ranking is highly corre-
lated with the PS grouping (-y = .75). Thus, 54% of the officers were rated in the
"strong" PS, while 61% of the nonofficers were rated in the "weak" PS.
Back to the Concept of Opinion Leaders? 273

Table 2. Network Positions and Strength of Personality


(in percentages)

Strength of Personality
Network Position Strong Above Average Moderate Weak N

Central 58 17 14 11 65
High medium 31 28 25 16 68
Low medium 15 28 31 26 68

Downloaded from http://poq.oxfordjournals.org/ at New York University on October 12, 2014


Marginal 3 13 23 61 69

NOTE: X2 = 85.23, df = 9, p < .0001; y = 0.686, p < .0001.

to other network positions, with decreasing PS score according to


lower network position (61% of the "marginals" in their groups are
rated as "weak" on the PS scale, while only 11% of the "centrals"
belong to this category of PS). The strong gamma coefficient of .68
indicates that these two indicators are interrelated. The strong relation-
ship with sociometnc centrality may indicate that the PS scale is not
exclusively a personality measure but an instrument that also reflects
social position. The early studies of Allensbach found the same pattern
(though not by sociometric measures). Thus, for example, the German
surveys revealed that high PS respondents report having more friends
and acquaintances (see Noelle-Neumann 1985, esp. table 10: 67% of
the "strong" PS group reported having many friends and acquain-
tances, compared to only 49% of the "above average," 37% of the
"moderate," and 20% of the "weak"). While one may question the
validity of such self-assessments, the data from the sociomatrix ob-
tained by a mapping of actual communication links in the Israeli kib-
butz community reveal the same relationship.
The final and crucial validity test of the PS scale is its ability to
predict communicative and influential behavior. To do so, measures
of the actual flow of various information and decision-making items in
the kibbutz community were compared with the PS scores. As the
mapping of the communication activity was based on dyadic measure-
ment ("who-to-whom"), the comparison of communicative activity of
individuals according to their PS level was possible. To do so, we
cross-classified each tie activated for the flow of a specific item with
the communicator's PS level. Table 3 presents the communicative ac-
tivity of the four PS groups in terms of their relative share in the flow.
The activity rates are the percentages of the ties activated by each
level of PS in the flow of six information items and two decision-
Table 3. Communicative Activity by Strength of Personality

Average PS
lies Activated"
Non-
Strong Weak Communi- communi-
Item 1 2 3 4 N x2 P cator cator P

Gossip
Item 1 22 31 32 15 425 7.7 n.s. 116.8 114.5 1.08 n.s.
Item 2 24 36 28 22 382 5.6 n.s. 117.1 115.7 0.65 n.s.
General news
Item 1 34 26 24 16 501 6.5 n.s. 120.2 109.8 4.19 <.OO1
Item 2 38 22 22 18 474 9.4 <.05 116.2 109.1 2.18 <.01
Consumer
information http://poq.oxfordjournals.org/ at New York University on October 12, 2014
Item 1 41 28 19 12 375 18.8 <.001 133.4 102.8 12.24 <.OO1
Item 2 39 31 17 13 381 17.6 <.01 130.9 104.6 10.52 <.001
Decision
making
Item 1 48 31 19 2 346 45.2 <.001 137.9 101.5 14.57 <.001
Item 2 46 29 18 7 366 33.2 <.001 138.3 100.6 15.09 <.001

1
Percentage of ties activated by a communicator whose strength of personality is 1,2, 3, or 4.
b
Mest conducted for the differences between the averages under the null hypothesis of no significant difference.
Back to the Concept of Opinion Leaders? 275

making items. The higher the percentage, the more active the individu-
als from the specific PS group were. In other words, a bigger share of
the communication flow is related to communicators from a specific
PS level.
The proportions of ties activated by various levels of PS vary across
items. However, except for the flow of gossip, the most active commu-
nicators are those with higher levels of PS. The majority of ties acti-
vated for the flow of news, consumer information, and influence origi-
nated from individuals with stronger PS. The rates of activation
decreased for lower levels of PS, indicating the strong relation between

Downloaded from http://poq.oxfordjournals.org/ at New York University on October 12, 2014


communicative activity and strength of personality (except for gossip).
This relation is evidently more significant for decision-making items
than for information items. In the case of influence flow (the decision-
making items), almost half of the flow originated from individuals with
"strong" PS (while only 2%-7% of this flow originated from the
"weak" PS group). In other words, the influentials, identified by actual
activity, are mainly those rated as "strong" on the PS scale.
Another comparison presented in table 3 is of PS averages: for each
of the items studied, we compared the average PS for the "communica-
tors" (those who were listed by the respondents as the source of infor-
mation or influence) and the "receivers" (those who were not listed as
sources). The pattern revealed by the averages confirms our previous
finding: the PS averages are significantly higher for the communicators
for most of the information items and for both decision-making items.
This difference is largest for the flow of influence and smallest (and
actually not significant) for the flow of gossip. Thus, individuals with
stronger PS are more active as communicators of news and consumer
information and are clearly the influentials for decision-making items.

Conclusions
From a theoretical perspective, the identification of the "influentials"
should be related to the concept of opinion leaders. The criticism of the
concept of opinion leaders has focused mainly on its methodological
deficiencies and its emphasis on the two-step flow model. However, a
better measurement procedure and a more realistic presentation of the
influentials, regardless of the media effects controversy, should cause
a revision of this concept. Successfully identifying the influentials,
validating the measurement by "external" criteria, and pointing to the
role of these influentials in the flow of interpersonal communication
and influence suggests a "return to the opinion leadership idea." How-
ever, as the findings of both the German and the Israeli studies suggest,
276 Gabriel Weimann

this revised measure of personal influence is independent of the flow


of mass communication.5
Opinion leadership was originally conceptualized as a combination
of personal and social factors. As Katz (1957) noted, being an influen-
tial is related to three attributes: (1) the personification of certain val-
ues (or "who one is"); (2) competence ("what one knows"); and
(3) strategic social location ("whom one knows"). The present study
combines these characteristics by relating them to actual communica-
tive activity. The individuals identified as "strong" on the PS scale
combined personal traits, competence, and social position in their per-

Downloaded from http://poq.oxfordjournals.org/ at New York University on October 12, 2014


sonal network, enabling them to influence others. Thus, it is not merely
a unidimensional measure, but a combination of personal traits with
social network positioning ("It is, however, not enough to be a person
whom others want to emulate, or to be competent. One must also be
accessible" [Katz 1957, 74]). The PS scale's strong correlation with
network positions and communicative activity is not only a method-
ological validation of an empirical instrument, but also provides a theo-
retical linkage between the personal and social dimensions of the opin-
ion leaders concept.
In spite of the similarities between the influential as identified by
the PS scale and the opinion leaders, these terms are not fully inter-
changeable. Four differences emerge from a first comparison. (1) While
opinion leaders were found by Katz and Lazarsfeld (1955) to be fairly
evenly distributed among the different social strata, the influential
(as measured by PS scaling) were concentrated in higher SES levels.
However, later studies of opinion leadership also revealed this unequal
distribution (e.g., Booth and Babchuck 1972; Hamilton 1971; Kingdon
1970). (2) The opinion leaders were found to differ from one subject
to another with very little overlap, while the influential were found
to be active in several areas with considerable overlap (see Noelle-
Neumann 1985, 47). Again, later studies of opinion leaders revealed
some overlap, even in the data used by Katz and Lazarsfeld (Marcus
and Bauer 1964). (3) Opinion leaders were found to be considerably
more exposed to mass media, while this tendency was not significantly
related to the influentials. In fact, it appears that the influentials rely
on personal sources and their special pattern of media consumption,
not merely on higher media exposure.6 This notion of personal influ-

5. The Allensbach Studies as well as the Israeli study found no significant differences
between PS levels in terms of media exposure (e.g., exposure to television and radio
was almost identical for all the PS groups; see Noelle-Neuman 1985, 27).
6. For example, there was no significant difference between PS levels in terms of watch-
ing television or listening to the radio (Noelle-Neumann 1985, table 18) when asked
about leisure activities. In the Israeli sample, the higher levels of PS did not report
higher exposure to newspapers, television, or radio. However, some indication of the
Back to the Concept of Opinion Leaders? 277

ence on opinion leaders was suggested by Katz when revising his two-
step flow model: "Opinion leaders, though more exposed to the media
than nonleaders, nevertheless reported personal influence as the major
factor in their decisions" (Katz 1957, 76). As to the pattern of media
consumption, later studies of opinion leadership related different qual-
ity and form of media consumption to opinion leaders (e.g., Hamilton
1971; Levy 1978; Ostlund 1973). (4) While the opinion leadership con-
cept was based on the leader-follower dichotomy, PS scaling reveals
that influenceability is, rather, a continuous variable. Again, the possi-
bility of a continuum was suggested by later studies of opinion leader-

Downloaded from http://poq.oxfordjournals.org/ at New York University on October 12, 2014


ship (e.g., Lin 1973). Thus, it appears that the influentials are not the
opinion leaders according to the original definition but fit better the
more sophisticated characterization that stemmed from the growing
criticism of early studies on opinion leadership.
From a methodological perspective, the findings of the present study
highlight the efficiency of the new measure. The relationships between
the PS measure and real activity as communicators and influentials
provides the needed validation. The universality of the scale, when
applied to two different societies, further contributes to its usefulness.
Further studies should focus on the relationships between social status
and PS (revealed to be stronger than suggested by early studies of
opinion leaders) and measure the weight of personal traits and social
characteristics that constitute the status of an influential.
Finally, the identification of influentials by means of the PS scale
has important implications for practical-minded scholars of politics,
marketing, advertising, and public opinion in general. The identifica-
tion of influentials and the study of their sources of information and
decision-making processes may have substantive value to those who
combine social research with persuasive communication. The develop-
ment of the PS scale is a crucial first step for such an integration.

References
Berelson, Bernard, Paul F. Lazarsfeld, and W. N. McPhee. 1954. Voting: A Study of
Opinion Formation in a Presidential Campaign. Chicago: University of Chicago
Press.
Booth, Alan, and Nicholas Babchuck. 1972. "Informal Medical Opinion Leadership
among the Middle Aged and Elderly." Public Opinion Quarterly 36:87-94.
Brett, Joyce E., and Anne Kernaleguen. 1975. "Perceptual and Personality Variables
Related to Opinion Leadership in Fashion." Perceptual and Motor Skills
40:775-79.

different quality of media consumption was revealed by the Allensbach studies, espe-
cially when PS ratings were related to reading certain newspapers and specific contents
of the papers; see Noelle-Neumann 1983, tables 271, 27m, and 27n.
278 Gabriel Weimann

Eurich, Carl. 1977. "Beinflussungsstrategien zur Weiterentwicklung des


Meinungsfuhrerkonzeptes." Zeitscrift fur Markt, Meinungs and Zukunftsforschung
20:42-85.
Gitlin, Todd. 1978. "Media Sociology: The Dominant Paradigm." Theory and Society
6:206-53.
Granovetter, Mark S. 1981. "The Strength of Weak Ties: A Network Theory
Revisited." Revised version of a paper presented to the International
Communications Association Meetings (May, 1980), Acapulco.
Hamilton, Herbert. 1971. "Dimensions of Self-Designated Opinion Leadership and
Their Correlates." Public Opinion Quarterly 35:266-74.
Katz, Elihu. 1957. "The Two-Step Flow of Communication: An Up-to-Date Report
on an Hypothesis." Public Opinion Quarterly 21:67-78.
Katz, Elihu. 1980. "On Conceptualizing Media Effects." Studies in Communication

Downloaded from http://poq.oxfordjournals.org/ at New York University on October 12, 2014


1:119-41.
Katz, Elihu. 1988. "Communication Research Since Lazarsfeld." Public Opinion
Quarterly 51 :S25-S45.
Katz, Elihu, and Paul F. Lazarsfeld. 1955. Personal Influence. Glencoe, IL:
Free Press.
Kingdon, John W. 1970. "Opinion Leaders in the Electorate." Public Opinion
Quarterly 34:256-61.
Kflpller, Karlfritz. 1984. Opinion Leaders. Hamburg: Heinrich Bauer Verlag.
Lazarsfeld, Paul F., Bernard Berelson, and Hazel Gaudet. 1948. The People's Choice:
How the Voter Makes Up His Mind in a Presidential Campaign. New York: Duell,
Sloan and Pearce.
Levy, Mark R. 1978. "Opinion Leadership and Television News Uses." Public
Opinion Quarterly 42:402-6.
Lin, Nan. 1973. The Study of Human Communication. Indianapolis: Bobbs-Merrill.
Marcus, Alan S., and Raymond A. Bauer. 1964. "Yes: There Are Generalized
Opinion Leaders." Public Opinion Quarterly 28:628-32.
Merton, Robert K. 1949. "Patterns of Influence." In Communication Research, ed.
P. Lazarsfeld and F. Stanton, 180-219. New York: Harper and Brothers.
Noelle-Neumann, Elisabeth. 1983. Spiegel Dokumentation: Personlichkeitsstarke.
Hamburg: Spiegel Verlag.
Noelle-Neumann, Elisabeth. 1984. The Spiral of Silence. Chicago: University of
Chicago Press.
Noelle-Neumann, Elisabeth. 1985. "Identifying Opinion Leaders." Paper presented at
the 38th ESOMAR Conference, Wiesbaden.
Ostlund, Lyman. 1973. "Interpersonal Communication Following McGovem's
Eagleton Decision." Public Opinion Quarterly 37:601-10.
Reynolds, Fred D., and William R. Darden. 1971. "Mutually Adaptive Effects of
Interpersonal Communication." Journal of Marketing Research 6:441—50.
Richards, William D. 1975. A Manual for Network Analysis. Institute for
Communication Research, Stanford University. Mimeo.
Richards, William D. 1977, "Network Analysis Methods: Conceptual and Operational
Approaches." Paper presented at the fourth annual colloquium on Social Networks,
University of Hawaii, Honolulu.
Robertson, Thomas S., and James H. Myers. 1969. "Personality Correlates of
Opinion Leadership and Innovative Buying Behavior." Journal of Marketing
Research 6:155-67.
Robinson, John. 1976. "Interpersonal Influence in Election Campaigns: Two-Step
Flow Hypothesis." Public Opinion Quarterly 40:304-19.
Rogers, Everett M., and Laurence B. Kincaid. 1981. Communication Network:
Toward a New Paradigm for Research. New York: Free Press.
Rogers, Everett M., and Floyd F. Shoemaker. 1971. Communication of Innovations:
A Cross-Cultural Approach. New York: Free Press.
Saunders, John, Michael J. Davis, and David M. Monsees. 1974. "Opinion
Leadership in Family Planning." Journal of Health and Social Behavior 15:217-27.
Back to the Concept of Opinion Leaders? 279

Schrank, Holly L., and Lois D. Gilmore. 1973. "Correlates of Fashion Leadership:
Implications for Fashion Process Theory." Sociological Quarterly 14:534-43.
Summers, John O. 1971. "Generalized Change Agents and lnnovativeness." Journal
of Marketing Research 8:311-15.
Weimann, Gabriel. 1982. "On the Importance of Marginality: One More Step into the
Two-Step Flow of Communication." American Sociological Review 47:764—73.
Weimann, Gabriel. 1983. "On the Strength of Weak Conversational Ties." Social
Networks 5:245-67.

Downloaded from http://poq.oxfordjournals.org/ at New York University on October 12, 2014

Você também pode gostar