Says Who ? : Epistemic Authority Effects in Social Judgment

Text-only Preview

In M.P. Zanna (Ed.)Advances in Experimental Social Psychology (vol. 37, in press).
New York: Academic Press

Says Who? : Epistemic Authority Effects in Social Judgment
© Arie W. Kruglanski
University of Maryland
Amiram Raviv, Daniel Bar-Tal, Alona Raviv, Keren Sharvit
Shmuel Ellis
Tel-Aviv University
Ruth Bar
Bar Research Inc.
Antonio Pierro and Lucia Mannetti
University of Rome, “La Sapienza”

Says Who? : Epistemic Authority Effects in Social Judgment 2

In press in M. P. Zanna (Ed.). Advances in Experimental Social Psychology, Vol. 37.
We are indebted to Carol Dweck, Mario Mikulincer and Tory Higgins for comments
on an earlier draft.
This chapter features the concept of ascribed epistemic authority (Kruglanski, 1989)
offered as a unique perspective on source effects in social judgment. In contrast to prior
approaches that viewed the source of communication as external to the self, we assume
that both the self and external sources may be assigned different degrees of epistemic
authority in different domains, and that this determines how individuals process
information, make decisions and undertake actions. The present framework traces the
socio-developmental aspects of epistemic authority assignments, and considers individual
differences in the distribution of authority assignments across sources. From this
perspective, we claim a central role in human judgment to the information’s source, and
the assessment of its epistemic authority is seen to constitute an essential preliminary
phase in individuals’ approach to information.

© Arie Kruglanski

Says Who? : Epistemic Authority Effects in Social Judgment 3

A critical aspect of human social functioning concerns people’s informational
dependence on others (Kelley & Thibaut, 1968). As we negotiate our way through the
labyrinths of interpersonal relations and task exigencies we encounter a continuous flow
of information in the form of communications, advice, exhortations and pleas from a
variety of sources. These pose the ubiquitous question of whom to (informationally) trust,
and whose statements to discount, or regard with suspicion. The issue has profound
implications not only for one’s personal dealings with one’s social and physical
environments but also for the workings of society itself. In this vein, the political scientist
Robert Putnam (2000) commented on the declining trust in government characteristic of
contemporary American society and on the dangers this poses to its communal
functioning. In other words, for many Americans the government has ceased to represent
an informational source to be relied on, encouraging political disengagement and civic
© Arie Kruglanski

Says Who? : Epistemic Authority Effects in Social Judgment 4
The significance of source variables for individuals’ reactions to the information
given has not been lost on social psychologists. To the contrary, such variables have
figured prominently in major theories of communication and persuasion beginning with
the classic works of the Yale group (Hovland, Janis & Kelley, 1953; Hovland & Weiss,
1951). This continues with the currently influential dual-mode theories of persuasion
(Petty & Cacioppo, 1986; Chaiken, Lieberman & Eagly, 1989) and the parametric
unimodel (Erb et al., 2003; Kruglanski & Thompson, 1999a,b; Pierro, Mannetti,
Kruglanski & Sleeth-Keppler, 2004). Yet these various treatments have been limited in
scope and have not addressed the full array of issues that source characteristics may raise
in the domain of social judgment.
The present article introduces a broader perspective on source effects framed from
the subjective standpoint of the information’s recipient. We first review the treatment of
source effects in several major models of persuasion. We then introduce the present,
framework focused on the concept of “epistemic authority,” and review empirical
research conducted under its aegis. A final discussion highlights the unique properties of
the present approach and considers its implications for the place of source effects in
notions of information processing and human judgment.
The Early Work: Hovland, Janis & Kelley’s (1953) Learning Theory Paradigm of Source
Hovland et al.s (1953) conception of persuasion and attitude change rested on the
learning theory paradigm dominant at the time. From that perspective, communicators’
influence derives from the incentives they provide. These can be tangible or intangible
and based on such source characteristics as expertness, trustworthiness, similarity, or
© Arie Kruglanski

Says Who? : Epistemic Authority Effects in Social Judgment 5
likeability. In turn, these (and other possible characteristics) are learned through
experiencing the consequences of accepting or rejecting a given source’s influence, and
are generalized to other similar sources. In general, Hovland et al. (1953) suggest that the
learned characteristics of the source determine individuals’ motivation to attend to and
comprehend the message, as well as accept its implications (McGuire, 1969). Within this
conceptual framework, Hovland et al. (1953) focused in particular on the variable of
communicator credibility, that they parsed into its two constituents, expertness and
trustworthiness, the latter referring to the communicator’s intent to convey valid
information. Early empirical studies (e.g, Hovland and Weiss, 1951) indeed showed that
messages ascribed to high credibility sources tend to be accepted to a greater extent than
ones ascribed to sources of low credibility.

An issue that occupied researchers of source effects concerned a separation and
comparison of the two components of credibility namely expertise and trustworthiness
(Cohen, 1964; Hovland et al., 1953). In an empirical investigation of this problem,
Kelman and Hovland (1953) exposed high school students to a communication
advocating a more lenient treatment of juvenile delinquents. This communication was
attributed to one of three different sources: (1) a well informed and intentioned judge, i.e.,
a source possessing a high degree of expertise as well as of trustworthiness, (2) a member
of the studio audience, i.e., a source of intermediate presumed expertise and
trustworthiness, and (3) a juvenile delinquent on bail, charged with drug dealings and
other shady business, i.e. a source of high presumed familiarity (and in this sense
“expertise”) in the domain, yet of low trustworthiness. As might be expected, the
message delivered by the judge had significantly greater persuasive impact than one
© Arie Kruglanski

Says Who? : Epistemic Authority Effects in Social Judgment 6
delivered by the juvenile delinquent with the studio audience member’s message exerting
an intermediate impact, closer in magnitude to that of the judge than to the delinquent.
Hovland et al. (1953) interpreted these findings as suggesting that trustworthiness may
play a greater role in attitude change than expertness. Subsequent research confirmed the
significant role that the communicator’s intent may play in attitude change (e.g., Allyn &
Festinger, 1961; Weiss & Fine, 1955; 1956).
Though the distinction between expertise and “trustworthiness” (understood as the
intent to speak the truth) is of interest, the attempt to separate the two and pit them
against each other could be misleading. For the question always turns on whom to
(informationally) trust (the ubiquitous “says who?” question in the title of this chapter).
In this sense expertise, intent as well as a possible host of other learned source
characteristics (e.g., membership in one’s ingroup, age, gender) drive the acceptance of
communications through the overall variable of credibility.1 In our subsequent analysis,
we refrain from analytically disentangling the different source characteristics that could
contribute to a source’s epistemic authority, and instead address the attribution of such
authority as a Gestalt.
Source Effects from the Dual Mode Perspective: The Elaboration Likelihood Model

An important shift in the way persuasion effects in general and source effects in
particular were approached by persuasion researchers was occasioned by the work of
Petty and Cacioppo (1986) on the Elaboration Likelihood Model (ELM), and Chaiken’s
(1979) work on the Heuristic-Systematic Model (HSM). Conceptually, this shift signaled
a break from the neo-behavioristic paradigm adopted by Hovland et al. (1953) and a

1 For instance, one might evolve a lay theory that even well intentioned experts cannot be trusted because of
their arrogance and tendency to operate through the blinds of their misconceptions.
© Arie Kruglanski

Says Who? : Epistemic Authority Effects in Social Judgment 7
move toward the cognitive Zeitgeist as part of the “cognitive revolution” that has been
transforming the psychological science as a whole. Specifically, both of these dual mode
theories regarded persuasion as essentially a cognitive activity wherein conclusions are
reached on the basis of more or less extensive processing (or “elaboration”) of various
types of information available in the persuasive context.

Beside its novel conceptual perspective on source effects, the dual-mode
paradigm was based on the recognition that prior research on this topic yielded a
disappointing crop of empirical findings. As Petty and Cacioppo (1986, p.125) put it:
“…although it might seem reasonable to propose that by associating a message with an
expert source agreement could be increased (e.g., see Aristotle’s Rhetoric), the
accumulated contemporary research literature suggested that expertise effects were
considerably more complicated…Sometimes expert sources had the expected
effects…sometimes no effects were obtained…, and sometimes reverse effects were
noted.. Unfortunately, the conditions under which each of these effects could be obtained
and the processes involved in producing these effects were not at all apparent…”
In the ELM, source characteristics are often considered as “cues” that is as
“peripheral” signals likely to impact persuasion under conditions of a low “elaboration”
likelihood, that is, in circumstances where an individual’s motivation and/or capacity to
process information are low. For instance, in the classic study by Petty, Cacioppo and
Goldman (1981) source expertise was manipulated orthogonally to message argument
quality. Cross-cutting both, the researchers manipulated participants’ issue involvement,
considered as a determinant of elaboration likelihood. It was found that source expertise
had persuasive effects only under conditions of low issue involvement but not under
© Arie Kruglanski

Says Who? : Epistemic Authority Effects in Social Judgment 8
conditions of high issue involvement. By contrast, message argument quality had
persuasive effects only under high, but not under low, issue involvement. These findings
were interpreted to demonstrate the cue function that source characteristics may often
fulfill, and according to which they may exert impact under low (but not under high)
elaboration likelihood conditions.

It is noteworthy that in the ELM, source characteristics may fulfill functions other
than that of a “cue.” In particular, they may constitute “message arguments” as when a
physically attractive person advertises a beauty product (her or his appearance serving as
proof of the product’s effectiveness) (Petty & Cacioppo, 1986). Finally, according to the
ELM source characteristics may serve a motivating function, prompting extensive or
restricted elaboration of the information. Knowing that a source is expert or that he/she
represents a majority position (Mackie, 1987) may motivate recipients to pay close
attention to its message and to take it seriously. Similarly, knowing that the source is
inexpert might reduce the recipients’ consideration of the message.
Source Effects in the Heuristic Systematic Model
In the HSM, source characteristics are regarded as “heuristic” information, related
to simple and general rules of thumb or “heuristics,” such as “expertise implies
correctness,” “friends can be trusted,” or “majority opinions are valid.” Knowing that a
source is a friend, an expert, or member of a majority may then prompt an acceptance of
her or his recommendations through an application of the corresponding heuristic. Much
like the ELM, the HSM too distinguishes between two separate and qualitatively distinct
modes of persuasion: The heuristic mode depicted above (likely to be adopted when the
recipients’ cognitive and motivational resources are limited) and a systematic mode in
© Arie Kruglanski

Says Who? : Epistemic Authority Effects in Social Judgment 9
which the message information is carefully considered (likely to be adopted when the
recipients’ cognitive and motivational resources are ample) (Chaiken, Liberman, &
Eagly, 1989). Furthermore, the heuristic mode is assumed to afford lesser judgmental
confidence than the systematic mode; hence it is assumed to be opted for when lower
confidence threshold is deemed sufficient for the recipient’s purposes. When it is
insufficient, that is, when the issue is important enough to require considerable
confidence, the systematic mode is assumed to “kick in” instead.
Source Effects in the Unimodel

Both the ELM and the HSM stress the qualitative distinctions between the two
modes of persuasion (i.e., the peripheral and central modes in the ELM and the heuristic
and systematic modes in the HSM). By contrast, a recently proposed “unimodel” (Erb et
al., 2003; Kruglanski & Thompson, 1999a,b, Kruglanski et al., 2004; Pierro, Mannetti,
Kruglanski & Sleeth-Keppler, 2004; Chun, Spiegel & Kruglanski, 2002) stress the
commonalities between those modes. In these terms, both “peripheral cues” and
“message arguments” function as evidence for various conclusions that recipients may
draw from information in the persuasive context. Some such conclusions may refer to
actual recommendations espoused by the communication source, or flowing from the
message arguments. Other conclusions may relate to the desirability of paying close
attention to the source’s pronouncements, e.g., because of its expertise, prestige, or power
(Fiske, 2004) versus ignoring them because of the source’s low standing on these
particular dimensions.
According to the unimodel, the evidential function of information derives from a
fundamental syllogistic structure. For instance, a person may subscribe to a (major)
© Arie Kruglanski

Says Who? : Epistemic Authority Effects in Social Judgment 10
premise that “experts are correct” (or, to the belief that “if expert, then correct”) and then
assume that a position expressed by a source was valid because “she is an expert” (a
minor premise). Similarly, a person may assume that “anything that promotes a clean
environment should be supported” (a major premise), and then proceed to accept the
statement that “electrically powered automobiles should be supported” because of the
belief that “electrically powered automobiles preserve a clean environment” (a minor
premise). Note that whereas the former statement about expertise represents a peripheral
or heuristic “cue,” and the latter represents a “message argument,” both are seen to
function in a fundamentally identical syllogistic manner. The same would be true of a
case where a source’s characteristic, e.g., physical attractiveness, functioned as a
“message argument,” that is, as a “minor premise” that in conjunction with a major
premise whereby “physical attractiveness of a cosmetic product’s user attest to its
effectiveness,” yields the conclusion that the product, indeed, is effective. Finally, the
case where a source characteristic (e.g., ‘social power’) prompts an individual to closely
attend to the source’s argument may be based on a conviction (major premise) that “the
views of powerful persons are worth considering.” Thus, the unimodel stresses the
similarities between source and message effects based on their similar syllogistic
functioning as evidence for various conclusions.
On the Uniqueness of Source Effects: The Concept of Epistemic Authority
In contrast to the foregoing emphasis on commonalities that source-effects may
share with other “peripheral cues” (in the ELM), other “types of heuristic information”
(in the HSM), or other minor premises (in the unimodel) the present chapter highlights
ways in which source effects are unique. As shown subsequently, this framework offers
© Arie Kruglanski