research

May07

SourcePersuasionClark.pdf

Home >Psychology homework help >research

Research article

Perceptions of source efficacy and persuasion: Multiple mechanisms for source effects on attitudes

JASON K. CLARK1*, ABIGAIL T. EVANS1 AND DUANE T. WEGENER2

1University of Iowa, USA; 2Ohio State University, USA

Abstract

When communicators are perceived as likely to bring proposed outcomes to fruition, they have source efficacy. Although

perceptions of source efficacy are common in persuasion settings, this construct has received little direct research attention. The

present research explored how source efficacy may impact persuasion in different ways at different levels of motivation to process

messages. Across three experiments, participants encountered message arguments of varying quality from a source manipulated

to be relatively efficacious or inefficacious. When motivation to process the message was low, source efficacy served as a

peripheral cue (Experiment 1). When motivation was high, efficacy information learned before the message biased processing of

ambiguous messages (Experiment 2), but source efficacy learned after the message affected the amount of confidence people had

in their message-related thoughts (Experiment 3). These effects of source efficacy were distinct from effects of perceived source

expertise/credibility. Copyright # 2011 John Wiley & Sons, Ltd.

INTRODUCTION

In many settings, the qualities of a persuasive message source

can determine the extent to which their communication is

influential (see Briñol & Petty, 2009b). Relative to other source

characteristics, one factor that has received little direct

research attention is the perceived efficacy or effectiveness

of the message source. Perceptions of efficacy would relate to

how an individual or a group is viewed in terms of being able to

bring about what he, she, or they propose. Perceptions of

source efficacy could influence persuasion in many common

situations such as encountering speeches from political

candidates, letters to the editor of a newspaper, and co-

workers advocating changes at the office. However, little is

known concerning the various ways that perceptions of

efficacy might influence attitudes.

Compared to ineffectual sources, highly efficacious

advocates should be viewed as more likely to use persuasive

techniques (e.g., provide valid arguments) that increase the

likelihood of success with their proposal. Expectations

associated with effectiveness likely contribute to many effects

of traditional persuasion variables related to source expertise/

credibility, trustworthiness, and power. However, perceptions

of efficacy need not and often do not co-occur with these other

characteristics. For instance, consider efficacy in relation to

perceptions of source credibility or expertise. In some cases,

we might view elected public officials (e.g., Senators,

Presidents; high efficacy) as having substantial ability to

bring outcomes to fruition (e.g., policy changes, executive

orders), but we may also believe that they possess little

expertise on the topic (e.g., energy conservation, stem cell

research). In contrast, some communicators may be viewed as

having a great deal of expertise in a particular domain (e.g.,

nuclear physicists, social psychologists), but they may have

little ability to directly influence public policy or other

outcomes (low efficacy). Similar contrasts between efficacy

and other constructs can also be drawn. Therefore, considering

these differences, it is worthwhile to directly explore how and

under what circumstances perceived efficacy may affect

persuasion.

SOURCE EFFICACY AND POTENTIAL MECHANISMS OF PERSUASION

Consistent with prominent models of attitude change (e.g., the

Elaboration Likelihood Model, ELM, Petty & Cacioppo,

1986), source efficacy may influence attitudes in different

ways depending on the message recipient’s motivation and

ability to actively process the central merits of an issue.

Organizing effects of source efficacy according to the level of

cognitive processing is important because attitudes that result

from relatively higher levels of thinking tend to be more

durable and directive of behavior (see Petty & Krosnick,

1995). In an initial investigation, Clark and Wegener (2009)

examined effects of source efficacy when initial motivation to

think was relatively moderate (not constrained to be low or

high). In these studies, participants received a message that

was designed to be moderate in terms of personal relevance (or

issue involvement) and source efficacy was found to influence

the extent to which arguments were carefully scrutinized. In

European Journal of Social Psychology, Eur. J. Soc. Psychol. 41, 596–607 (2011)

Published online 26 January 2011 in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/ejsp.787

*Correspondence to: Jason K. Clark, Department of Psychology, University of Iowa, Iowa City, IA 52242, USA. E-mail: [email protected]

particular, high levels of source efficacy increased processing

of messages that disagreed with message recipients’ pre-

message attitudes (compared with low levels of source

efficacy), but high levels of source efficacy decreased

processing of messages that agreed with message recipients’

pre-message attitudes (compared with low levels of source

efficacy).

However, effects on amount of processing are only one

potential type of influence. At more extreme levels of

motivation or ability to think, a growing body of research

suggests that source factors hold the potential for distinctly

different types of influence (see Briñol & Petty, 2009b; Petty &

Wegener, 1998). The present research investigated source

efficacy effects via three of these other potential mechanisms.

Specifically, we examined the possibility that source efficacy

may (1) serve as a cue/heuristic when motivation to process is

low, (2) bias the direction of effortful message processing

when motivation is high and message arguments are

ambiguous, and (3) affect the confidence people have in their

message-related thoughts when motivation is high and source

efficacy information is encountered after the message. For

each of the proposed roles for source efficacy, the persuasive

effects should be driven by perceptions that efficacious sources

are likely to present information that is more valid and

compelling compared to advocates that lack efficacy. It is

interesting to note that previous persuasion studies have also

been based on similar assumptions (i.e., that expert sources or

perhaps majority or powerful sources are likely to provide

valid, compelling, or otherwise ‘‘correct’’ information). Yet,

those studies have not assessed these presumed mediating

beliefs about the sources or their messages.

Across the experiments, we employed methods and

procedures that have been widely used in the persuasion

domain. Participants received a persuasive message that varied

in terms of the quality of arguments and also received efficacy

information regarding the source. Post-message attitudes and

message-related thoughts were measured in order to assess the

relative influence of source characteristics versus the central

merits of the persuasive appeal (see Petty & Cacioppo, 1986,

for discussion of this method). In the sections to follow, we

discuss each potential mechanism in more detail prior to the

experiment that tests that mechanism.

EXPERIMENT 1: SOURCE EFFICACY AS A

PERIPHERAL CUE

When motivation or ability to process information is lacking,

many source factors (e.g., expertise, Chaiken & Maheswaran,

1994; Petty, Cacioppo, & Goldman, 1981; attractiveness,

Chaiken, 1980) have been shown to serve as simple cues that

directly influence attitudes. For example, Petty, Cacioppo, and

Schumann (1983) varied the celebrity status of a spokesperson

as well as the personal relevance of a consumer product (low

vs. high) and strength of product features (unambiguously

weak vs. unambiguously strong). When the personal relevance

of the product was low (and participants were relatively

unmotivated to process the message), participants reported

more favorable attitudes toward the product when the endorser

was described as a celebrity rather than a non-celebrity.

Moreover, consistent with celebrity status serving as a cue in

low-motivation conditions, the strength of product features in

the advertisement had no influence on evaluations. In contrast,

when the product was highly relevant (and participants were

motivated to think), the quality of product features had a

substantial influence, and the celebrity status of the endorser

had no significant impact on attitudes.

These findings suggest that message recipients may rely on

perceptions associated with the likeability or attractiveness of

a source as a basis for their attitude toward an advocacy. In the

case of source efficacy, expectations regarding the likely

strength of an appeal (rather than perceived attractiveness)

should hold a similar potential to be used as a peripheral cue.

Compared to ineffectual sources, efficacious advocates should

be viewed as more likely to make valid, compelling arguments.

Thus, when insufficient motivation or ability keeps message

recipients from closely scrutinizing the quality of the

(relatively complex) available arguments, message recipients

may rely on these (simpler) source perceptions as a shortcut for

determining that the advocacy probably has merit. In contrast,

when recipients are motivated and able to scrutinize messages,

the quality of the unambiguous message content should have

the most influence on post-message attitudes, and source

efficacy should have little impact.

Thus, for Experiment 1, we predicted that source efficacy

would have a stronger influence on post-message attitudes

when motivation to think carefully about the message is

relatively low rather than high. In addition, we expected this

effect of efficacy to be mediated by beliefs that the source

would provide valid arguments in support of the advocated

position.

Method

Participants and Design

Two hundred fourteen Indiana University undergraduates

participated in exchange for partial course credit in their

introductory psychology classes. Participants were randomly

assigned to a 2 (source efficacy: low, high)� 2 (personal

relevance: low, high)� 2 (argument quality: weak, strong)

between-participants factorial design.

Procedure

Upon arrival to the lab, participants were seated at a computer

station. They were then told that their task would be to assess

factors related to the readability of a written communication.

Following this cover story, participants completed a 20-item

survey in which one question pertained to attitudes toward

nuclear power plants (1¼ definitely opposed to 9¼ definitely

in favor) and the remaining items served as filler. Upon

completion of the survey, participants received information

that manipulated the efficacy of a message source (low vs.

high) followed by information that varied the personal

relevance of the forthcoming communication (low vs. high).

Participants were then given a set of compelling (strong) or

specious (weak) arguments advocating the development of

additional nuclear power plants in the United States. Following

Source efficacy and persuasion 597

10990992, 2011, 5, D ow

nloaded from https://onlinelibrary.w

iley.com /doi/10.1002/ejsp.787 by Florida International U

niversity, W iley O

nline L ibrary on [08/05/2023]. See the T

erm s and C

onditions (https://onlinelibrary.w iley.com

/term s-and-conditions) on W

iley O nline L

ibrary for rules of use; O A

articles are governed by the applicable C reative C

om m

ons L icense

this message, participants reported their post-message

attitudes, engaged in a thought-listing task, and responded

to manipulation checks. After these measures, participants

were thanked for their contributions and debriefed.

Independent Variables

Source Efficacy Participants received one of two source

descriptions designed to manipulate perceptions of efficacy. In

the low (high) efficacy conditions, participants were told:

‘‘Over the past 10 years, Mr. Saunders has been very

unsuccessful (successful) in getting his recommendations

(on a variety of issues) approved by governing bodies.’’

Personal Relevance Immediately after the source descrip-

tion, participants learned the topic of the advocacy (i.e.,

nuclear power plants) and received information that varied the

personal relevance of the message (adapted from Haugtvedt &

Wegener, 1994). In low personal relevance conditions,

participants were told that new nuclear power plants were

being considered for the relatively distant US states of Arizona

and New Mexico. However, in conditions of high personal

relevance, participants were informed that the plants were

being proposed for their own state (Indiana) and a neighboring

state (Illinois).

Argument Quality Participants received one of two

versions of a message advocating the development of more

nuclear power plants. The message consisted of either four

weak arguments or four strong arguments (approximately 300

words; adapted from Clark, Wegener, & Fabrigar, 2008b). For

example, one weak argument stated that nuclear power plants

are more desirable because they are more aesthetically

pleasing than conventional power plants. In the strong version,

however, one argument stated that additional nuclear plants

could help reduce the emissions of greenhouse gases by

replacing a number of conventional power plants. An analysis

revealed that participants held largely negative pre-message

attitudes toward nuclear power plants (M¼ 3.83, SD¼ 1.92;

lower than the scale midpoint of 5, t(213)¼�8.93, p< .001).

Thus, the nuclear power message was counter-attitudinal for a

majority of participants.

Dependent Measures

Post-message Attitudes After reading the message,

participants reported their attitudes on six measures. The first

five items used the stem ‘‘Nuclear power plants are:’’ (1¼ bad,

useless, harmful, negative, unnecessary, to 9¼ good, useful,

beneficial, positive, necessary). The sixth item was: ‘‘Building

nuclear power plants in the United States is a good idea’’

(1¼ strongly disagree to 9¼ strongly agree). Responses to

these six measures were highly reliable (a¼ .97) and were

averaged to form a post-message attitude index.

Cognitive Responses Following completion of the atti-

tude measures, participants completed a thought-listing task in

which they listed a maximum of eight thoughts that came to

mind while reading the message (see Wegener, Downing,

Krosnick, & Petty, 1995, for specific thought-listing instruc-

tions). Upon listing eight thoughts or after 3minutes elapsed,

participants were prompted to rate each of their thoughts. Each

listed thought was presented sequentially by the computer and

was coupled with the following choices: Positive, negative,

neutral, or unrelated to nuclear power plants. The overall

favorability of each participant’s self-rated thoughts was

indexed by subtracting the number of negative thoughts from

the number of positive thoughts and dividing this difference by

the total number of thoughts listed.

Expected Message Validity After the thought-listing task,

participants reported their pre-message expectations concern-

ing the cogency of the arguments. These items were ‘‘After

reading the description of William Saunders’ past experience

but before reading the message,. . .’’: (1) ‘‘To what extent did

you expect the message to have strong arguments?’’ and (2)

‘‘to what extent did you expect the message to make a

compelling case in favor of nuclear power? (1¼ not at all to

9¼ very much so).’’ Responses on these measures were

reliable (a¼ .92) and scores were averaged to form an index of

expected message validity.

Source Efficacy Manipulation Check Perceptions of source

efficacy were assessed via two scaled items. These items were

‘‘After reading the description of William Saunders’ past

experience but before reading the message,. . .’’: (1) ‘‘How

likely did you believe that he would be effective in helping

bring about more nuclear power plants?’’ and (2) ‘‘ how likely

did you believe that he would successfully achieve his goal of

building more nuclear plants? (1¼ not at all likely to 9¼ very

likely).’’ These ratings were highly correlated (a¼ .91) and

were thus averaged to form an index of perceived source

efficacy.

Perceptions of Source Credibility In addition to measur-

ing views of efficacy, two items targeted perceptions of source

credibility. These questions were ‘‘After reading the descrip-

tion of William Saunders’ past experience but before reading

the message,. . .’’: (1) ‘‘How likely did you believe he was an

expert source?’’ and (2) ‘‘how likely did you believe he was a

credible source? (1¼ not at all to 9¼ very much).’’

Participants’ responses on these measures were averaged to

form an index of perceived source credibility (a¼ .92).

Personal Relevance Manipulation Check Participants

answered two questions targeting personal relevance of the

nuclear power issue. These scales were paired with the

following statement: ‘‘The issue of building new nuclear

power plants is:’’ (1¼ not at all relevant to me, not at all

important to me, to 9¼ very relevant to me, very important to

me). Responses were averaged to form an index of perceived

personal relevance (a¼ .89).

Results

Expected Message Validity

A 2(source efficacy: low, high)� 2 (personal relevance: low,

high)� 2 (argument quality: weak, strong) Analysis of

Variance (ANOVA) revealed a significant main effect of the

efficacy manipulation. Participants who received the high

efficacy source reported that they expected the message to

contain more compelling arguments (M¼ 5.69, SD¼ 2.12)

than participants who received the low efficacy source

(M¼ 4.51, SD¼ 1.91), F(1, 206)¼ 18.80, p< .001,

598 Jason K. Clark et al.

10990992, 2011, 5, D ow

nloaded from https://onlinelibrary.w

iley.com /doi/10.1002/ejsp.787 by Florida International U

niversity, W iley O

nline L ibrary on [08/05/2023]. See the T

erm s and C

onditions (https://onlinelibrary.w iley.com

/term s-and-conditions) on W

iley O nline L

ibrary for rules of use; O A

articles are governed by the applicable C reative C

om m

ons L icense

h2partial¼ .08. Of less importance, there was also a main effect

of personal relevance (with higher expectations of message

validity when relevance was low rather than high), F(1,

206)¼ 4.99, p¼ .027, h2partial¼ .02. No other effects

approached significance (ps> .11).

Source Efficacy Manipulation Check and

Perceived Source Credibility

Indices of perceived source efficacy and perceived credibility

were submitted to separate three-way ANOVAs. On the index

of source efficacy, a robust main effect of the efficacy mani-

pulation emerged, F(1, 206)¼ 58.89, p< .001, h2partial¼ .22.

Participants who received the high efficacy description viewed

the source as more effective (M¼ 5.50, SD¼ 2.06) than

participants who received the low efficacy information

(M¼ 3.34, SD¼ 2.02). No additional effects were significant,

ps> .34. On perceived source credibility, a similar main effect

of the efficacy manipulation was found such that higher levels

of credibility were perceived when the source was high

(M¼ 5.52, SD¼ 2.02) rather than low in efficacy (M¼ 4.05,

SD¼ 1.91), F(1, 206)¼ 18.75, p< .001, h2partial¼ .08 (no

additional significant effects were found, ps> .15). Percep-

tions of efficacy and credibility were correlated, r¼ .66,

p< .001. However, effects of source efficacy on attitudes were

not due to effects of the efficacy manipulation on perceptions

of credibility.

Perceived Personal Relevance

Results of a three-way ANOVA showed a significant main

effect of the personal relevance manipulation such that the

nuclear power topic was perceived as more relevant when

proposed plants were in participants’ home state and a

neighboring state (M¼ 5.55, SD¼ 2.28) rather than distant

states (M¼ 4.66, SD¼ 2.37), F(1, 206)¼ 8.02, p¼ .005,

h2partial¼ .04. In addition, a robust main effect of argument

quality (Mweak ¼ 4.78 [SD¼ 2.35] vs. Mstrong¼ 5.44

[SD¼ 2.34], F[1,206]¼ 4.47, p¼ .036, h2partial¼ .02) and a

marginal main effect of source efficacy (Mlow ¼ 4.81

[SD¼ 2.40] vs. Mhigh¼ 5.41 [SD¼ 2.30], F[1,206]¼ 3.57,

p¼ .060, h2partial¼ .02) emerged. No other effects approached

significance (ps> .21).

Post-message Attitudes

A three-way Analysis of Covariance (ANCOVA) was performed

on the post-message attitude index with pre-message attitude

(M¼ 3.83, SD¼ 1.92) and perceived source credibility

(M¼ 4.62, SD¼ 2.04) serving as a covariates. Consistent

with personal relevance influencing motivation to process

(e.g., Petty & Cacioppo, 1979), the quality of the message

arguments significantly influenced post-message attitudes

when personal relevance was high (adjusted means, Mweak

¼ 4.37 [SE¼ .24] vs. Mstrong¼ 5.69 [SE¼ .24], F[1,204]¼ 15.42, p< .001, h2partial¼ .07), but not when it was low

(adjusted means, Mweak ¼ 4.78 [SE¼ .23] vs. Mstrong¼ 4.99

[SE¼ .24], F< 1), Personal Relevance�Argument Quality,

F(1, 204)¼ 5.46, p¼ .020, h2partial¼ .03 (see Figure 1).

However, the crucial test of our first hypothesis is whether

the manipulation of source efficacy influenced attitudes more

when personal relevancewas low rather than high (the opposite

pattern as for the argument quality manipulation). Consistent

with this prediction, a robust Personal Relevance� Source

Efficacy interaction emerged, F(1, 204)¼ 3.92, p¼ .049,

h2partial¼ .02 (see Figure 1). The efficacy of the source tended

to influence attitude favorability when personal relevance was

low (adjusted means, Mlow ¼ 4.57 [SE¼ .24] vs. Mhigh¼ 5.20

[SE¼ .25], F[1,204]¼ 2.58, p¼ .109, h2partial¼ .01), but not

when relevance was high (adjusted means, Mlow ¼ 5.19

[SE¼ .25] vs. Mhigh¼ 4.87 [SE¼ .23], F< 1). Main effects of

both covariates (pre-message attitude: F[1,204]¼ 56.90,

p< .001, h2partial¼ .22; perceived source credibility: F[1,204]¼ 16.54, p< .001, h2partial¼ .08) and argument quality

(F[1,204]¼ 10.25, p¼ .002, h2partial¼ .05) were also found.

Thought Favorability and Thought Mediation

A three-way ANCOVA was conducted in which pre-message

attitude (F[1,204]¼ 18.42, p< .001, h2partial¼ .08) and perceived

source credibility (F[1,204]¼ 13.65, p< .001, h2partial¼ .06)

served as covariates. Similar to post-message attitudes,

argument quality significantly influenced the favorability of

High Personal Relevance

3.9

4.4

4.9

5.4

5.9

Source Efficacy

Po st

-M es

sa ge

A tti

tu de

Weak

Strong

Low Personal Relevance

3.9

4.4

4.9

5.4

5.9

Low High

Source Efficacy

Po st

-M es

sa ge

A tti

tu de

Weak

Strong

Figure 1. Adjusted mean post-message attitude as a function of personal relevance, source efficacy, and argument quality in Experiment 1

Source efficacy and persuasion 599

10990992, 2011, 5, D ow

nloaded from https://onlinelibrary.w

iley.com /doi/10.1002/ejsp.787 by Florida International U

niversity, W iley O

nline L ibrary on [08/05/2023]. See the T

erm s and C

onditions (https://onlinelibrary.w iley.com

/term s-and-conditions) on W

iley O nline L

ibrary for rules of use; O A

articles are governed by the applicable C reative C

om m

ons L icense

thoughts when personal relevance was high (adjusted means,

Mweak¼�.38 [SE¼ .08] vs. Mstrong¼�.13 [SE¼ .08],

F[1,204]¼ 5.71, p¼ .018, h2partial¼ .03), but not when

personal relevance was low (adjusted means, Mweak ¼�.11

[SE¼ .07] vs. Mstrong¼�.09 [SE¼ .08], F< 1), Personal

Relevance�Argument Quality, F(1, 204)¼ 2.42, p¼ .122,

h2partial¼ .01. Main effects of the personal relevance (adjusted

means, Mlow ¼�.10 [SE¼ .05] vs. Mhigh¼�.25 [SE¼ .05],

F[1,204]¼ 4.24, p¼ .041, h2partial¼ .02) and argument quality

manipulations (adjusted means, Mweak ¼�.24 [SE¼ .05] vs.

Mstrong¼�.11 [SE¼ .05], F[1,204]¼ 3.36, p¼ .068,

h2partial¼ .02) also emerged.

Attitudes that result from high levels of elaboration should

be based on participants’ assessments of the central merits of

the attitude object (Petty & Cacioppo, 1986). Therefore, the

favorability of listed thoughts should be influenced by the

argument quality manipulation (as in the ANCOVA results)

and post-message attitudes should be based on those thoughts

when high levels of elaboration occurred. This should result in

thought favorability mediating argument quality effects on

attitudes in conditions of high personal relevance. Because of

the moderate size of the current sample, this possibility was

examined using bootstrapping procedures outlined by Shrout

and Bolger (2002; see also Preacher & Hayes 2004, 2008). The

bootstrap analyses treated the obtained data within each

relevance condition as the population and randomly drew 5000

samples of size equal to the condition, with replacement.

Estimates of the indirect effect of argument quality on post-

message attitudes via thought favorability (controlling for pre-

message attitude and perceived source credibility) were

calculated for each sample and used to generate bias corrected

confidence intervals for the indirect (mediated) effects. Results

of these analyses showed that thought favorability significantly

mediated the influence of argument quality on post-message

attitudes when personal relevance was high (estimated mean

indirect effect¼ .40, BC CI 95: .0938–.8195), but not when

personal relevance was low (estimated mean indirect

effect¼ .04, BC CI 95: �.2712 to .4128). In regressions

using data from the high relevance conditions, the impact of

argument quality decreased (from the original value of

b¼ 1.32, t[103]¼ 4.14, p< .001, r2¼ .14), but remained

significant (b¼ .93, t[102]¼ 3.20, p¼ .002, r2¼ .09) when

thought favorability (b¼ 1.56, t[102]¼ 5.60,

p< .001, r2¼ .23) was included in the model.

Source Efficacy and the Mediating Role of Message

Validity Expectations

As reported earlier, the source efficacy manipulation influ-

enced expectations of message validity. Do expectations of

message validity mediate the cue effects of source efficacy

under low-relevance conditions? To test this prediction, we

conducted mediated moderation regression analyses that

followed procedures specified by Muller, Judd, and Yzerbyt

(2005; see also Wegener & Fabrigar, 2000). If expectations of

message validity are at least partly responsible for source

efficacy effects on attitudes, then including a Personal

Relevance�Expected Message Validity term in the model

should decrease the impact of the Personal Relevan-

ce� Source Efficacy term (reported in the ANCOVA results).

A centered regression (including all terms from the ANCOVA

analyses and all interaction terms with expected message

validity and perceived credibility each as a substitute for

source efficacy) showed that the Personal Relevance� Source

Efficacy term decreased (from the original value of b¼�.94, t[204]¼�1.99, p¼ .049, r2¼ .02) to non-significance,

b¼�.59, t(197)¼�1.17, p¼ .245. In contrast, the Personal

Relevance�Expected Message Validity interaction predicted

post-message attitudes, b¼�.30, t(197)¼�1.99, p¼ .048,

r2¼ .02, above and beyond the Personal Relevance� Source

Efficacy term.

Bootstrapping analyses (similar to those previously

described) showed that expected message validity mediated

effects of the source efficacy manipulation on attitudes when

personal relevance was low (estimated mean indirect

effect¼ .29, BC CI 95: .0499–.7412) but not when personal

relevance was high (estimated mean indirect effect¼ .04, BC

CI 95: �.0243 to .2988). Also, when perceived source

credibility was treated as the mediator instead of expected

message validity, source credibility failed to serve as a

mediator (in low relevance conditions, estimated mean indirect

effect¼ .02, BC CI 95: �.0257 to .1840). Thus, cue effects of

the source efficacy manipulation appear to be at least partly

due to beliefs that the source would provide compelling

arguments and not because of perceptions that the efficacious

source was also highly expert or credible per se.

Discussion

Experiment 1 provided evidence that perceptions of source

efficacy can be used as a peripheral cue to persuasion when

people are relatively unmotivated to think carefully about the

central tenets of an advocacy. That is, when the personal

relevance of the nuclear power message was low, post-message

attitudes were more favorable when source efficacy was high

rather than low—regardless of the quality of the message

arguments. Also, efficacy had no impact on the valence of

participants’ thoughts in response to the message, supporting

characterization of the effects as relatively non-effortful or

cue-based. In conditions of high personal relevance, however,

the efficacy of the source did not influence post-message

attitudes. Rather, the favorability of attitudes and message-

related thoughts was determined by the quality of the message

arguments. In addition, analyses showed that the effects of

source efficacy were not due to perceptions of source

credibility. However, participants expected the high efficacy

source to present more valid, compelling arguments than the

low efficacy advocate and these perceptions accounted for the

cue effects on persuasion.

EXPERIMENT 2: SOURCE EFFICACY BIASING

EFFORTFUL MESSAGE ELABORATION

Consistent with the multiple roles and biased processing

postulates of the ELM, some persuasion variables have been

shown to hold a biasing potential at high levels of message

elaboration, especially when information about the target is

relatively ambiguous (e.g., Chaiken & Maheswaran, 1994;

600 Jason K. Clark et al.

10990992, 2011, 5, D ow

nloaded from https://onlinelibrary.w

iley.com /doi/10.1002/ejsp.787 by Florida International U

niversity, W iley O

nline L ibrary on [08/05/2023]. See the T

erm s and C

onditions (https://onlinelibrary.w iley.com

/term s-and-conditions) on W

iley O nline L

ibrary for rules of use; O A

articles are governed by the applicable C reative C

om m

ons L icense

Petty, Schumann, Richman, & Strathman, 1993; see also

Wegener, Clark, & Petty, 2006). For example, Chaiken and

Maheswaran (1994) used message arguments that were

ambiguous (not clearly strong or weak) or unambiguous

(clearly strong or weak). Under high-motivation conditions,

source credibility had no effect on attitudes or on thoughts

about the attitude object (a consumer product) when the

arguments were unambiguous (as in Petty et al., 1981, 1983).

However, when arguments were ambiguous, a credible source

was more persuasive than a non-credible source, and effects of

source credibility were mediated by thought favorability (i.e., a

credible source led to more favorable thoughts than a non-

credible source, and thought favorability influenced post-

message attitudes; for similar effects involving mood of the

message recipient, see Petty et al., 1993).

These past findings suggest that source efficacy could also

bias processing of ambiguous message arguments. As noted

earlier, sources that possess substantial efficacy should be

expected to present cogent arguments, whereas ineffectual

sources may be viewed as likely to offer more specious

information. Thus, when the strength of an appeal is not clearly

strong or weak, activation of these beliefs could tend to push

effortful message-related thinking in a direction that is

expectancy-consistent (e.g., to give the ‘‘benefit of the doubt’’

to an efficacious source, but to look harshly on a source

believed to be ineffective).

Experiment 2 used relatively ambiguous (mixed) message

arguments and conditions of high motivation to test the

hypothesis that source efficacy will bias the direction of

message-related thinking and subsequent attitudes. Further-

more, these biases in processing should be due to expectations

that the efficacious source is more likely to present cogent

information compared to an ineffectual source.

Method

Participants and Design

Seventy-seven undergraduates at the University of Alabama

participated and received partial course credit. Participants

were randomly assigned to one of two conditions (low source

efficacy or high source efficacy).

Procedure

Procedures were similar to Experiment 1 with the following

exceptions. The attitudes of experimental interest concerned

comprehensive exams for college seniors and participants

received a message advocating these exams (pre-message

attitude, M¼ 4.84, SD¼ 2.45; difference from scale midpoint

[5], p¼ .578). Prior to receiving the message, the efficacy of

the source was varied using the same manipulation as

Experiment 1. Moreover, all participants were told that the

exams were being considered for their institution the following

academic year—in order to create high personal relevance

(see, Petty & Cacioppo, 1979; Petty et al., 1981).

The message contained relatively ambiguous arguments

that were developed by modifying a previously created

message in favor of the exams (for the original strong and weak

arguments, see Petty & Cacioppo, 1986). This original

message stated that the grade point average at another

institution has increased by ‘‘31%’’ (since implementing the

exams), graduates obtain starting salaries of ‘‘$3000–4000’’

higher, and ‘‘have a 55% greater chance of landing a top job’’

compared to graduates from schools without the exams. In

contrast, the more ambiguous version used in the current study

stated that GPA increased by ‘‘1%’’ but that this increase was

not ‘‘statistically meaningful,’’ that starting salaries were

‘‘$100–$200 higher,’’ and that students ‘‘might have a slightly

greater chance of landing a job.’’ Following procedures

specified by Petty and Cacioppo (1986), these ambiguous

arguments were pretested to ensure that they elicited more

favorable cognitive responses compared to previously created

weak arguments and elicited less favorable thoughts than

previous strong arguments.

After reading the message, participants completed a series

of dependent measures that were the same as those used in

Experiment 1, except that any reference to ‘‘nuclear power

plants’’ was replaced with ‘‘senior comprehensive exams.’’

Upon completion of these measures, participants were

debriefed and thanked.

Results

Expected Message Validity

Responses on the two items targeting message validity

expectations (a¼ .91) were averaged to form a single index.

As in Experiment 1, a one-way ANOVA revealed that

participants in the high source efficacy condition expected a

message that contained more compelling arguments

(M¼ 6.04, SD¼ 1.65) than participants who received the

low efficacy source (M¼ 4.83, SD¼ 2.03), F(1, 75)¼ 8.16,

p¼ .006, h2partial¼ .10.

Source Efficacy Manipulation Check and Perceived

Source Credibility

Separate indices of perceived source efficacy (a¼ .89) and

credibility (a¼ .91) were created by averaging scores from the

two scaled measures that corresponded to each construct. Each

index was submitted to independent one-way ANOVAs

wherein manipulated source efficacy was the sole factor. As

expected, a significant effect of the efficacy manipulation was

found on the index of perceived efficacy (Mlow¼ 4.71

[SD¼ 1.92] vs. Mhigh¼ 6.11 [SD¼ 1.57]), F(1, 75)¼ 12.20,

p¼ .001, h2partial¼ .14. On the index of perceived source

credibility, a similar effect was observed (Mlow¼ 4.68

[SD¼ 2.02] vs. Mhigh¼ 5.70 [SD¼ 1.65]), F(1, 75)¼ 5.84,

p¼ .018, h2partial¼ .07. As in Experiment 1, perceptions of

efficacy and credibility were correlated, r¼ .54, p< .001, but

effects of source efficacy on attitudes went above and beyond

effects of the efficacy manipulation on perceived credibility.

Perceived Personal Relevance

An index of perceived personal relevance of the comprehen-

sive exam issue was calculated by averaging responses to the

Source efficacy and persuasion 601

10990992, 2011, 5, D ow

nloaded from https://onlinelibrary.w

iley.com /doi/10.1002/ejsp.787 by Florida International U

niversity, W iley O

nline L ibrary on [08/05/2023]. See the T

erm s and C

onditions (https://onlinelibrary.w iley.com

/term s-and-conditions) on W

iley O nline L

ibrary for rules of use; O A

articles are governed by the applicable C reative C

om m

ons L icense

two-scaled items (a¼ .90). The results of a one-way ANOVA

showed no differences as a function of the source efficacy

manipulation (Mlow¼ 7.19 [SD¼ 1.86] vs. Mhigh¼ 7.07

[SD¼ 2.14]), F< 1, p> .78. Consistent with creating high

levels of personal relevance, perceptions of relevance

(M¼ 7.13, SD¼ 1.99) were significantly higher than the

midpoint (5) of the nine-point scale, t(76)¼ 9.37, p< .001.

Post-message Attitudes

The six measures of post-message attitudes (a¼ .96) were

averaged to form a single index. A one-way ANCOVA was

conducted wherein pre-message attitude (M¼ 4.84, SD¼ 2.45)

and perceived source credibility (M¼ 5.18, SD¼ 1.90) served as

covariates. In support of predictions regarding biased processing,

source efficacy significantly influenced attitudes. Participants

were more favorable toward senior comprehensive exams when

source efficacy was high (Madjusted¼ 6.06, SE¼ .28) rather than

low (Madjusted¼ 5.24, SE¼ .27), F(1, 73)¼ 4.23, p¼ .043,

h2partial¼ .06. The covariate of pre-message attitude also

related significantly to post-message attitudes, F(1,

73)¼ 32.10, p< .001, h2partial¼ .31.

Thought Favorability and Thought Mediation

Favorability of thoughts was indexed as in Experiment 1. A

marginally significant main effect of source efficacy emerged

from an ANCOVA, F(1, 73)¼ 3.40, p¼ .069, h2partial¼ .05.

Consistent with post-message attitudes, thoughts tended to be

more favorable when the efficacy of the message source was

high (Madjusted¼ .17, SE¼ .12) rather than low

(Madjusted¼�.14, SE¼ .11). More importantly, however, the

results of a bootstrapping analysis (which controlled for pre-

message attitude and perceived source credibility) offered

additional evidence in support of our biased processing

prediction. The favorability of participants’ thoughts signifi-

cantly mediated the effect of source efficacy on attitudes,

estimated mean indirect effect¼ .42, BC CI 95: .0058–1.0043.

Specifically, in regression analyses, the impact of source

efficacy on post-message attitudes fell to non-significance

(b¼ .40, t[72]¼ 1.19, p¼ .240) when thought favorability

(b¼ 1.39, t[72]¼ 5.92, p< .001, r2¼ .32) was included in the

model. In support of the hypothesis, these findings suggest that

knowledge of the source’s efficacy served to color participants’

message-related thoughts and these biased cognitions in turn

produced differences in post-message attitudes.

Source Efficacy and the Role of Message Validity

Expectations

Similar to the cue effects of source efficacy, differences in

anticipated message validity are predicted to account for the

biased processing effect of efficacy. This hypothesis was

supported by the results of regression-based mediational

analyses that controlled for pre-message attitudes and

perceptions of source credibility. Specifically, the impact of

the source efficacy manipulation on post-message attitudes

decreased (from the original value of b¼ .82, t[73]¼ 2.06,

p¼ .043, r2¼ .05) and fell to non-significance, b¼ .58,

t(72)¼ 1.49, p¼ .141, when message validity expectations

were included in the model as a predictor. Furthermore, these

message validity perceptions were found to be a strong

predictor of attitudes, b¼ .32, t(72)¼ 2.76, p¼ .007, r2¼ .10

(in bootstrapping analyses, estimated mean indirect

effect¼ .24, BC CI 95: .0009–.7700).

A regression that contained identical predictors of the

favorability of participants’ thoughts revealed a similar

mediational pattern. The direct effect of source efficacy was

non-significant (b¼ .23, t[72]¼ 1.39, p¼ .168) when message

validity perceptions were included in the model, whereas the

impact of validity perceptions was robust (b¼ .10,

t[72]¼ 2.02, p¼ .047, r2¼ .05; in bootstrapping analyses,

estimated mean indirect effect¼ .07, BC CI 93: .0015–.2186).

In addition, when perceptions of source credibility were treated

as the mediator, the indirect effects were non-significant (on

attitudes, estimated mean indirect effect¼�.08, BC CI 95:

�.4248 to .0176; on thought favorability, estimated mean

indirect effect¼�.01, BCCI 95:�.0933 to .0276). Thus, biased

processing effects of source efficacy are also directly accounted

for by beliefs that the source would provide compelling

arguments, not because of perceptions that the efficacious

source was also expert or credible.

Discussion

Experiment 2 provided evidence that source efficacy can

influence persuasion when motivation and ability to process

are high and message arguments are relatively ambiguous. In

particular, the high efficacy source elicited more favorable

thoughts and attitudes toward comprehensive exams compared

to the low efficacy advocate. Additional analyses demonstrated

that these effects were not due to perceptions of source

credibility. Rather, differences in source efficacy were linked

with different expectations regarding the validity of the

message and these perceptions accounted for the biased

processing effect of efficacy on persuasion.

EXPERIMENT 3: SOURCE EFFICACY

VALIDATING THOUGHTS

Recent research has identified another distinct way that

variables can influence high-thought persuasion—by influen-

cing the amount of confidence message recipients have in their

thoughts. That is, thoughts held with greater confidence should

have more impact on attitudes (Petty, Briñol, & Tormala,

2002). In support of this hypothesis, research has identified a

variety of factors that can determine thought confidence in

response to persuasive messages including the credibility of

the message source (e.g., Briñol, Petty, & Tormala, 2004), the

mood of message recipients (Briñol, Petty, & Barden, 2007),

and the extent to which message recipients feel powerful

(Briñol, Petty, Valle, Rucker, & Becerra, 2007). For instance,

Briñol and colleagues (2004) provided source credibility

information after reading a message and found that people had

greater confidence in their thoughts when they learned that the

source was credible rather than non-credible (see also Tormala,

Briñol, & Petty, 2006, 2007). Moreover, differences in thought

602 Jason K. Clark et al.

10990992, 2011, 5, D ow

nloaded from https://onlinelibrary.w

iley.com /doi/10.1002/ejsp.787 by Florida International U

niversity, W iley O

nline L ibrary on [08/05/2023]. See the T

erm s and C

onditions (https://onlinelibrary.w iley.com

/term s-and-conditions) on W

iley O nline L

ibrary for rules of use; O A

articles are governed by the applicable C reative C

om m

ons L icense

confidence determined attitude favorability such that more

confidence in thoughts led to greater influence of those

thoughts on resulting attitudes. Validation effects have been

shown to occur primarily when the confidence-inducing factor

is introduced after a message and when substantive message

processing has occurred (see Briñol et al. 2004; Briñol & Petty,

2009a). When encountered before an advocacy, the same

variables act as peripheral cues (if elaboration likelihood is

low, see Experiment 1) or bias message processing (if

elaboration likelihood is high and message arguments are

ambiguous, see Experiment 2) while having little effect on

confidence (see Briñol, Petty, Valle, et al., 2007; Tormala et al.,

2007).

Across these and other investigations, self-validation

research has demonstrated that message recipients must first

produce thoughts about a message before they experience

confidence or doubt in those cognitions. In addition,

persuasion variables that have been identified as influencing

thought confidence are those that plausibly trigger perceptions

of accuracy, validity, and trust (for a review, see Briñol & Petty,

2009a). That is, if a source is one that should present the best

available arguments, then one can be confident that presented

positive features of the advocated position are, in fact, positive

and that weak, specious support for the advocacy means that

there are no truly compelling arguments available. On the other

hand, if the source is less effective, then one may doubt

whether the advocated position would have the desirable

consequences attributed to it (when strong arguments are

given) or whether better support might exist (when weak

arguments are given). If source efficacy is linked to perceptions

that information is likely to be accurate, then efficacious

sources should elicit greater confidence in message-related

thoughts compared to advocates that lack efficacy. Further-

more, having more rather than less metacognitive confidence

should result in attitudes that are based more on one’s thoughts

about a message.

Thus, in Experiment 3, the efficacy of a source is learned

after message recipients have processed the tenets of an

advocacy. We predicted that participants would be more

confident in their previous message-related thoughts once they

later learned that a source is efficacious rather than

inefficacious. Furthermore, we expected this elevated confi-

dence to be associated with stronger influences of message-

related thoughts on attitudes toward the issue.

Method

Participants and Design

One hundred nineteen University of Alabama undergraduates

participated and received partial course credit. Participants

were randomly assigned to a 2 (argument quality: Weak,

strong)� 2 (source efficacy: Low, high) between-participants

factorial.

Procedure

The procedure and materials were similar to those used by

Briñol et al. (2004). Upon arrival to the lab, participants were

told they would receive information concerning the household

use of phosphate-based detergents and that legislators were

reviewing proposals designed to encourage the use of such

products. Tomotivate careful thinking, participants were asked

to pay close attention to the upcoming message because they

were part of a small group of participants taking part in this

important policy research (see Petty, Harkins, & Williams,

1980; Tetlock, 1983). Participants then received a message in

favor of using phosphate-based detergents. After the message,

participants completed a thought-listing and were then given

efficacy information about the message source. Subsequently,

participants reported their attitudes, responded to scales of

thought confidence, rated the valence of each listed thought,

and completed manipulation checks.

Independent Variables

Argument Quality Participants received a message con-

sisting of either weak or strong arguments advocating

phosphate-based detergents. These weak and strong versions

were taken directly from materials used in past research (see

Tormala et al., 2006). Strong arguments focused on the lower

cost of phosphate detergents over non-phosphate detergents

and claimed that the former are more environmentally friendly.

Weak arguments highlighted that phosphate detergents come

in more attractive packaging and are less scented compared to

non-phosphate detergents.

Source Efficacy Consistent with past self-validation research,

the efficacy of the source was manipulated after receipt of the

message and assessment of cognitive responses (see Briñol &

Petty, 2009a). The efficacy manipulation (low vs. high) was

identical to that used in the current Experiments 1 and 2.

Dependent Measures

Thought-listing Immediately after reading the message,

participants engaged in the same thought-listing task used in

Experiments 1 and 2.

Post-message Attitudes After receiving the source effi-

cacy information, participants reported their attitudes toward

phosphate detergents on six semantic differential scales. Each

scale ranged from 1 to 9 and contained one of the following

pairs of anchors: Negative–positive, bad–good, unfavorable–

favorable, against–in favor, harmful–beneficial, and foolish–

wise. Responses were averaged to form an index of post-

message attitude (a¼ .94).

Thought Confidence Following the attitude measures,

participants were asked to think back to the thought listing

completed earlier and to report the confidence they had in the

validity of their listed thoughts. These nine-point items were:

‘‘Overall, how much confidence do you have in the thoughts

you listed?’’ (none at all to very much), ‘‘Overall, how valid

would you say your thoughts are?’’ (not at all valid to

extremely valid), ‘‘How certain are you that the thoughts you

had while reading the message were ‘correct’?’’ (not at all

certain to very certain), and ‘‘How certain are you that of all

the possible thoughts that one might have about the message

and phosphate detergents, your thoughts generally reflected

the ‘right’ way to think and feel about what you saw?’’ (not at

all certain to very certain). An index of thought confidencewas

formed by averaging responses to these measures (a¼ .88).

Source efficacy and persuasion 603

10990992, 2011, 5, D ow

nloaded from https://onlinelibrary.w

iley.com /doi/10.1002/ejsp.787 by Florida International U

niversity, W iley O

nline L ibrary on [08/05/2023]. See the T

erm s and C

onditions (https://onlinelibrary.w iley.com

/term s-and-conditions) on W

iley O nline L

ibrary for rules of use; O A

articles are governed by the applicable C reative C

om m

ons L icense

Thought-rating Next, participants completed a thought-

rating task that was identical to that used in previous

experiments. Furthermore, overall favorability of thoughts

was calculated the same as in Experiments 1 and 2.

Source Efficacy Manipulation Check Perceptions of

source efficacy were assessed via two scales that were

identical to those used in Experiments 1 and 2 (except that the

attitude object was now ‘‘phosphate detergents’’). Responses

were averaged into a single index (a¼ .95).

Perceptions of Source Credibility Perceived source credi-

bility was measured on the same two scales as in Experiments

1 and 2 (a¼ .91).

Results

Source Efficacy Manipulation Check and Perceived

Source Credibility

The indices of perceived source efficacy and source credibility

were submitted to independent 2 (Argument Quality: weak

vs. strong)� 2(Source Efficacy: low vs. high) ANOVAs.

On perceptions of efficacy, main effects of the efficacy

manipulation (Mlow¼ 4.54 [SD¼ 2.22] vs. Mhigh¼ 6.60

[SD¼ 1.91]), F(1, 115)¼ 31.48, p< .001, h2partial¼ .22, and

argument quality were found (Mweak¼ 4.99 [SD¼ 2.14] vs.

Mstrong¼ 6.15 [SD¼ 2.37]), F(1, 115)¼ 9.81, p¼ .002,

h2partial¼ .08. In addition, an unexpected tendency emerged

such that argument quality had a greater influence on

perceptions when the source was high (Mweak¼ 5.68

[SD¼ 1.91] vs. Mstrong¼ 7.52 [SD¼ 1.41]) as opposed to

low in efficacy (Mweak¼ 4.30 [SD¼ 2.17] vs. Mstrong¼ 4.77

[SD¼ 2.27]), Argument Quality� Source Efficacy, F(1,

115)¼ 3.46, p¼ .065, h2partial¼ .03.

ANOVA results on perceptions of source credibility showed

a similar pattern. The source efficacy manipulation created

differences in perceived credibility, (Mlow¼ 4.27 [SD¼ 1.87]

vs. Mhigh¼ 5.39 [SD¼ 2.23]), F(1, 115)¼ 9.09, p¼ .003,

h2partial¼ .07. Also, the manipulation of argument quality

produced differences in perceived credibility (Mweak¼ 4.43

[SD¼ 2.21] vs. Mstrong¼ 5.23 [SD¼ 1.96]), F(1, 115)¼ 4.68,

p¼ .033, h2partial¼ .04. However, the Argument Qual-

ity� Source Efficacy interaction did not approach signifi-

cance, F< 1, p¼ .362. As in Experiments 1 and 2, perceptions

of efficacy and credibility were correlated, r¼ .62, p< .001.

Thought Favorability

A two-way ANCOVA was performed in which the manipula-

tions of source efficacy and argument quality were between-

participant factors and perceived source credibility served as a

covariate (M¼ 4.79, SD¼ 2.10). Similar to past self-validation

research, the analysis revealed a main effect of argument

quality on thought favorability (adjusted means,Mweak ¼ .014

[SE¼ .08] vs. Mstrong¼ .60 [SE¼ .08]), F(1, 114)¼ 27.79,

p< .001, h2partial¼ .20. No significant main effect of source

efficacy and no Argument Quality� Source Efficacy inter-

action were found (Fs< 1, ps> .49). This is consistent with

participants engaging in substantive processing of the message

and reporting the content of thoughts they had prior to learning

the efficacy of the source. The perceived credibility of the

source was correlated with thought favorability (F[1,114]¼ 6.20, p¼ .014, h2partial¼ .05).

Thought Confidence

Consistent with hypotheses, a significant main effect of

efficacy emerged from a two-way ANCOVA (controlling for

perceived credibility, M¼ 4.79, SD¼ 2.10), F(1, 114)¼ 4.98,

p¼ .028, h2partial¼ .04. Participants reported greater confi-

dence in thoughts when the source was high (Madjusted¼ 6.43,

SE¼ .23) rather than low (Madjusted¼ 5.71, SE¼ .22) in

efficacy. The argument quality manipulation tended to

influence thought confidence such that ratings were higher

when message arguments were strong (M¼ 6.31, SE¼ .22)

rather than weak (M¼ 5.83, SE¼ .22), F(1, 114)¼ 2.22,

p¼ .139. No additional effects approached significance

(ps> .25).

Post-message Attitudes

As predicted, a two-way ANCOVA (controlling for perceived

credibility) yielded a significant Argument Quality� Source

Efficacy interaction on the index of post-message attitude, F(1,

114)¼ 3.95, p¼ .049, h2partial¼ .03 (see Figure 2). When

source efficacy was high (and thought confidence was

relatively high), the quality of the message arguments had a

greater influence on post-message attitudes (adjusted means,

Mweak¼ 5.49 [SE¼ .23] vs. Mstrong¼ 7.60 [SE¼ .26], F

[1,114]¼ 41.28, p< .001, h2partial¼ .27) compared to when

source efficacy was low (and thought confidence was relatively

low; Mweak¼ 5.74 [SE¼ .25] vs. Mstrong¼ 6.92 [SE¼ .22], F

[1,114]¼ 11.15, p¼ .001, h2partial¼ .09). A robust main effect

of argument quality (F[1,114]¼ 46.57, p< .001, h2partial¼ .29)

and an effect of the perceived credibility covariate

(F[1,114]¼ 7.81, p¼ .006, h2partial¼ .06) also emerged.

Mediation Analyses

Because participants were instructed to think carefully about

the message, we expected that attitudes about phosphate

detergents would be based, at least in part, on assessments of

5.0

5.5

6.0

6.5

7.0

7.5

8.0

Low High

Source Efficacy

Po st

-M es

sa ge

A tti

tu de

Weak

Strong

Figure 2. Adjusted mean post-message attitude as a function of argument quality and source efficacy in Experiment 3

604 Jason K. Clark et al.

10990992, 2011, 5, D ow

nloaded from https://onlinelibrary.w

iley.com /doi/10.1002/ejsp.787 by Florida International U

niversity, W iley O

nline L ibrary on [08/05/2023]. See the T

erm s and C

onditions (https://onlinelibrary.w iley.com

/term s-and-conditions) on W

iley O nline L

ibrary for rules of use; O A

articles are governed by the applicable C reative C

om m

ons L icense

the central merits of the appeal. Mediation analyses which

controlled for perceptions of source credibility showed that the

favorability of participants’ thoughts significantly mediated

the effect of argument quality on attitudes. A simultaneous

regression showed that the impact of argument quality

decreased (from the original value of b¼ 1.58,

t[116]¼ 6.57, p< .001, r2¼ .27), but remained a robust

predictor of post-message attitudes (b¼ 1.02, t[115]¼ 4.21,

p< .001, r2¼ .14) when thought favorability (b¼ .97,

t[115]¼ 5.25, p< .001, r2¼ .19) was included in the model.

Furthermore, a bootstrap analysis revealed that this media-

tional pattern was significant, estimated mean indirect

effect¼ .56, BC CI 95: .2888–.9612.

However, our primary prediction regarding validation of

thoughts suggests that this link between thoughts and attitudes

should also be determined by the efficacy of the message

source. When the source was high in efficacy, participants’

attitudes should have been more reliant on cognitive responses

because of greater thought confidence compared to conditions

in which the source was low in efficacy. We tested this

assertion using mediated moderation regression procedures

similar to those used in Experiment 1. In an initial model, post-

message attitudes were regressed on centered terms for

thought favorability, source efficacy and their interaction

(controlling for the main effect of perceived source credibility

and a Thought Favorability X Source Credibility interaction).

A robust Thought Favorability� Source Efficacy interaction

showed that favorability of thoughts was more predictive of

attitudes when source efficacy was high rather than low,

b¼ .81, t(114)¼ 2.41, p¼ .017, r2¼ .05. However, if confi-

dence is responsible for source efficacy effects, then a Thought

Favorability�Thought Confidence term should reduce the

impact of this distal Thought Favorability� Source Efficacy

interaction on attitudes. A second analysis examined the

impact of the proposed mediator on post-message attitudes.

This test included centered terms for thought favorability,

thought confidence, and their interaction and showed that

thought favorability was a stronger predictor of attitudes at

higher levels of thought confidence, Thought Favorabil-

ity�Thought Confidence interaction, b¼ .26, t(114)¼ 2.79,

p¼ .006, r2¼ .06.

In order to pit the mediator against the distal predictor (and

control for perceptions of source credibility), we ran a model

that included main effects of thought favorability, source

efficacy, argument quality, perceived credibility, and thought

confidence. Importantly, interaction terms were also included

that paralleled the Thought Favorability� Source Efficacy

predictor (involving argument quality as a substitute for

thought favorability and perceived credibility for source

efficacy). This regression revealed that the Thought Favor-

ability� Source Efficacy interaction on attitudes decreased to

non-significance, b¼ .11, t(106)¼ .29, p¼ .773. However, the

Thought Favorability�Thought Confidence interaction

remained a strong and significant predictor of attitudes,

b¼ .27, t(106)¼ 2.93, p¼ .004, r2¼ .07. A bootstrapping

analysis using the Thought Favorability�Thought Confi-

dence term as a mediator reproduced the regression results, but

also showed that this mediational pattern held significance,

estimated mean indirect effect¼ .20, BC CI 95: .0013–.7169.

This pattern of results supports the predicted role of source

efficacy on self-validation of thoughts. A stronger relation

between participants’ thoughts and attitudes was found when

the source was high as opposed to low in efficacy and this

effect was accounted for by differences in thought confidence

that were experimentally induced by the manipulation of

source efficacy.

Discussion

Experiment 3 supported the prediction that source efficacy can

validate people’s thoughts in response to a persuasive message.

When learning that the source of a previously scrutinized

message was high in efficacy, participants felt more confident

in their message-related thoughts, and the cogency of the

message arguments had a greater impact on attitudes compared

to conditions of low source efficacy. In addition, mediation

analyses suggested that attitudes were based more on message-

related thoughts when source efficacy was high rather than low

and that this effect was driven by differences in thought

confidence.

As previously discussed, these observed effects on

confidence should occur primarily under certain conditions.

Past research has shown that when little thought is dedicated to

a message, encountering a potential confidence-inducing

variable after a message has little effect on confidence (e.g.,

Briñol et al., 2004; Petty et al., 2002). Also, even at high levels

of motivation and ability to think, the sequence of activation is

crucial. For example, Tormala et al. (2007) found that when the

credibility of a message source was manipulated prior to

message receipt, source credibility biased the valence of

participants’ thinking but it did not influence thought

confidence. In contrast, when research participants learned

about source credibility after the message, credibility influenced

confidence in the thoughts people had generated in response to

the message. Consistent with these and other investigations of

self-validation, Experiment 3 participants were induced to think

carefully about the advocacy, and efficacy information about the

source was presented only after the message.

GENERAL DISCUSSION

Compared to many other characteristics of persuasive message

sources, differences in perceived efficacy have received little

direct research attention. This is surprising when one considers

the vast number of situations where persuasive communi-

cations are received from sources that may vary in terms of

their potential to facilitate proposed outcomes (e.g.,

politicians, co-workers, activist groups). In some of these

situations, message recipients may be motivated to carefully

attend to tenets of the source’s message, but at other times,

message recipients may be relatively unmotivated to do so.

Thus, the aim of the current research was to examine some of

the ways that perceptions of source efficacy can influence

persuasion at different levels of motivation to process a

persuasive message. Experiment 1 demonstrated that knowl-

edge of a source’s efficacy could serve as a peripheral cue or

heuristic to persuasion when motivation to think is relatively

low. In contrast, Experiment 2 showed that efficacy could bias

the valence of motivated participants’ message-related

cognitions when the tenets of a message were relatively

ambiguous and source efficacy information preceded the

Source efficacy and persuasion 605

10990992, 2011, 5, D ow

nloaded from https://onlinelibrary.w

iley.com /doi/10.1002/ejsp.787 by Florida International U

niversity, W iley O

nline L ibrary on [08/05/2023]. See the T

erm s and C

onditions (https://onlinelibrary.w iley.com

/term s-and-conditions) on W

iley O nline L

ibrary for rules of use; O A

articles are governed by the applicable C reative C

om m

ons L icense

message. Lastly, the findings of Experiment 3 suggested that

source efficacy can serve yet another distinct role. When

source efficacy information is learned after a message has been

processed (and, therefore, message recipients have previously

generated thoughts about the persuasive appeal), the level of

efficacy can determine the amount of confidence people have

in their thoughts. The resulting level of confidence then

determines the extent to which those thoughts impact the

attitudes that people formulate about the advocated object or

position. Each of these roles occurs at lower (cue effects) or

higher (biasing processing, validating thoughts) levels of

motivation than were used in previous research, where source

efficacy influenced the extent to which people thought

carefully about the content of a persuasive appeal (Clark &

Wegener, 2009).

Implications and Future Directions

The current research has a number of implications and should

open doors to many possibilities for future research. One

direction for future inquiry could be to explore other distinct

ways that source efficacy can influence persuasion. For

example, some variables can serve as a central merit of the

advocacy when amount of thinking is high and the variable is

construed as a central dimension of the object or issue (see

Petty & Cacioppo 1986). For instance, differences in source

attractiveness have been shown to operate as arguments in

studies where a message focused on a beauty product (Petty &

Cacioppo, 1983) or the public image of a restaurant (Shavitt,

Swan, Lowrey, & Wänke, 1994). Perceptions of efficacy may

also hold this potential. However, the current findings suggest

that efficacy may need to serve as a more central dimension of

an issue to find such effects. Recall that, in contrast to the

attractiveness findings of Petty and Cacioppo (1983) and

Shavitt et al. (1994), no effects of efficacy were found on

attitudes or thoughts in the current Experiment 1 when

processing was high and the cogency of arguments was clearly

strong or weak. It seems plausible, however, that if the efficacy

of a source is more centrally related to an advocacy (e.g., a

political candidate discussing strategies he/she would use to

implement changes if elected) that efficacy could influence

persuasion across levels of argument strength (and thus, serve

as an additional central merit).

As previously described, Clark and Wegener (2009) found

that when motivation to think was relatively moderate, source

efficacy influenced the extent of message processing differ-

ently, depending on whether the position of a communication

was relatively pro- versus counter-attitudinal. Although the

position of a message has been shown to produce similar

processing effects with several other persuasion variables (e.g.,

see Baker & Petty, 1994; Clark, Wegener, & Fabrigar 2008a;

Wegener, Petty, & Smith, 1995), there is little research that

supports a moderational role for message position at more

extreme levels of processing motivation and ability (as

reflected in the current research). We did explore this

possibility in supplementary analyses of the current exper-

iments that measured the valence of pre-message attitudes (i.e.,

Experiments 1 and 2). Regression analyses that treated pre-

message attitudes (and all interaction terms) as centered

continuous predictors showed no evidence of moderation by

message position (ps> .2). These null findings may have been

driven by factors related to the samples (e.g., few Experiment 1

participants had favorable pre-message attitudes toward

nuclear power, small sample size in Experiment 2). Alter-

natively, the absence of moderation may have been due to the

different mechanisms that could operate at extreme versus

moderate levels of processing motivation. Clark and Wegener

(2009) found that concern about the outcomes of a proposal

accounted for the interactive effects of message position and

source efficacy on processing. It is plausible that concern-

based motives play a different, or less influential, role when

motivation or ability to process is constrained. These

possibilities could be examined in future research.

Our examination of source efficacy may also have

implications for many of the traditional source variables in

persuasion, including power (e.g., Festinger & Thibaut, 1951;

Raven & French, 1958), majority/minority status (e.g., Crano

& Chen, 1998; Moscovici, 1980; see Wood, Lundgren,

Ouellette, Busceme, & Blackstone, 1994), and source

credibility (e.g., Chaiken & Maheswaran, 1994; Petty et al.,

1981). Each of these factors may influence perceptions of

source efficacy, and these perceptions may be the (previously

hidden) ingredient that drives effects of those variables. For

example, perceptions of efficacy can drive processing of

persuasive messages when motivation is relatively moderate

(with high source efficacy increasing processing of counter-

attitudinal messages, but low source efficacy increasing

processing of pro-attitudinal messages, e.g., Clark &Wegener,

2009). Similar effects have been shown for majority/minority

sources (e.g., Baker & Petty, 1994; Martin & Hewstone, 2003),

although perceptions of efficacy have not been directly

addressed. It could also be that perceptions of source efficacy

might be responsible for at least some cue, biased processing,

or validation effects of variables such as source power,

majority status, or expertise.

Consistent with the current research, perceptions of efficacy

might be associated with beliefs that the source is likely to

provide compelling arguments. If so, these beliefs could be key

parts of cue, biased processing, or thought validation effects.

When considering distal variables like power, majority/

minority status, and expertise, however, it is clear that there

may be differences across these variables in the extent to which

they are related to beliefs that the source will provide valid

arguments. Although perceptions of credibility may be

consistently associated with such beliefs (as in the current

research), perceptions of source power or majority status may

be more loosely or even not at all connected to such

perceptions, at least in some settings. This would make it

particularly interesting to consider other potential reasons for

people to support what an efficacious (or powerful, majority, or

expert) source advocates. Efficacy (and power and majority

status) also brings with it the possibility that the advocacy is

simply more likely to be established. Thus, some form of

wishful thinking (McGuire & McGuire, 1991) could add to

beliefs about argument validity to drive cue effects (when

motivation or ability to process is minimal), biased processing

effects (when motivation and ability are high, but message

arguments are ambiguous), or thought validation effects (when

motivation and ability are high, but message recipients learn

about the source characteristics after, rather than before, the

message).

606 Jason K. Clark et al.

10990992, 2011, 5, D ow

nloaded from https://onlinelibrary.w

iley.com /doi/10.1002/ejsp.787 by Florida International U

niversity, W iley O

nline L ibrary on [08/05/2023]. See the T

erm s and C

onditions (https://onlinelibrary.w iley.com

/term s-and-conditions) on W

iley O nline L

ibrary for rules of use; O A

articles are governed by the applicable C reative C

om m

ons L icense

We look forward to these and other future research

directions. We hope that this future work will further

investigate the interesting relations among source character-

istics and how those characteristics combine to influence the

effectiveness of those sources in various persuasion settings.

ACKNOWLEDGEMENTS

We thank the following individuals for their assistance with

data collection: Lauren Eilmann, Courtney Guy, Bradley

Hargett, W. Taylor Monson, Daniel Rincon-Cruz, E. Keith

Shackelford II, Adam Share, and Erica Shin. A preliminary

report of Experiment 1 was presented at the 2009 meeting

of the Society for Personality and Social Psychology and

Experiment 3 was presented at the 2009 meeting of the

Midwestern Psychological Association.

REFERENCES

Baker, S. M., & Petty, R. E. (1994). Majority and minority influence: Source- position imbalance as a determinant of message scrutiny. Journal of Personality and Social Psychology, 67, 5–19.

Briñol, P., & Petty, R. E. (2009a). Persuasion: Insights from the self-validation hypothesis. In M. P. Zanna (Ed.), Advances in Experimental Social Psy- chology (Vol. 41, pp. 69–118). New York: Elsevier.

Briñol, P., & Petty, R. E. (2009b). Source factors in persuasion: A self- validation approach. European Review of Social Psychology, 20, 49–96.

Briñol, P., Petty, R. E., & Barden, J. (2007). Happiness versus sadness as a determinant of thought confidence in persuasion: A self-validation analysis. Journal of Personality and Social Psychology, 93, 711–727.

Briñol, P., Petty, R. E., & Tormala, Z. L. (2004). Self-validation of cognitive responses to advertisements. Journal of Consumer Research, 30, 559–573.

Briñol, P., Petty, R. E., Valle, C., Rucker, D. D., & Becerra, A. (2007). The effects of message recipients’ power before and after persuasion: A self- validation analysis. Journal of Personality and Social Psychology, 93, 1040–1053.

Chaiken, S. (1980). Heuristic versus systematic information processing in the use of source versus message cues in persuasion. Journal of Personality and Social Psychology, 39, 752–766.

Chaiken, S., & Maheswaran, D. (1994). Heuristic processing can bias sys- tematic processing: Effects of source credibility, argument ambiguity, and task importance on attitude judgment. Journal of Personality and Social Psychology, 66, 460–473.

Clark, J. K., &Wegener, D. T. (2009). Source entitativity and the elaboration of persuasive messages: The roles of perceived efficacy and message discre- pancy. Journal of Personality and Social Psychology, 97, 42–57.

Clark, J. K., Wegener, D. T., & Fabrigar, L. R. (2008a). Attitude accessibility and message processing: The moderating role of message position. Journal of Experimental Social Psychology, 44, 354–361.

Clark, J. K., Wegener, D. T., & Fabrigar, L. R. (2008b). Attitudinal ambiva- lence andmessage-based persuasion:Motivated processing of proattitudinal information and avoidance of counterattitudinal information. Personality and Social Psychology Bulletin, 34, 565–577.

Crano, W. D., & Chen, X. (1998). The leniency contract and persistence of majority and minority influence. Journal of Personality and Social Psychology, 74, 1437–1450.

Festinger, L., & Thibaut, J. (1951). Interpersonal communication in small groups. Journal of Abnormal and Social Psychology, 46, 92–100.

Haugtvedt, C. P., & Wegener, D. T. (1994). Message order effects in persua- sion: An attitude strength perspective. Journal of Consumer Research, 21, 205–218.

Martin, R., &Hewstone,M. (2003).Majority versus minority influence:When, not whether, source status instigates heuristic or systematic processing. European Journal of Social Psychology, 33, 313–330.

McGuire,W. J., &McGuire, C. V. (1991). The content, structure, and operation of thought systems. In R. S. Wyer, & T. K. Srull (Eds.), Advances in social cognition (Vol. 4, pp. 1–78). Hillsdale, NJ: Erlbaum.

Moscovici, S. (1980). Toward a theory of conversion behaviour. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 13, pp. 209–239). New York: Academic Press.

Muller, D., Judd, C.M., &Yzerbyt, V. Y. (2005).Whenmoderation is mediated and mediation is moderated. Journal of Personality and Social Psychology, 89, 852–863.

Petty, R. E., Briñol, P., & Tormala, Z. L. (2002). Thought confidence as a determinant of persuasion: The self-validation hypothesis. Journal of Personality and Social Psychology, 82, 722–741.

Petty, R. E., & Cacioppo, J. T. (1979). Issue-involvement can increase or decrease persuasion by enhancing message-relevant cognitive responses. Journal of Personality and Social Psychology, 37, 1915–1926.

Petty, R. E., & Cacioppo, J. T. (1983). Central and peripheral routes to persuasion: Application to advertising. In L. Percy, & A. Woodside (Eds.), Advertising and consumer psychology (pp. 3–23). Lexington, MA: D.C. Heath.

Petty, R. E., & Cacioppo, J. T. (1986).Communication and persuasion: Central and peripheral routes to persuasion. New York: Springer-Verlag.

Petty, R. E., Cacioppo, J. T., & Goldman, R. (1981). Personal involvement as a determinant of argument-based persuasion. Journal of Personality and Social Psychology, 41, 847–855.

Petty, R. E., Cacioppo, J. T., & Schumann, D. (1983). Central and peripheral routes to advertising effectiveness: The moderating role of involvement. Journal of Consumer Research, 10, 135–146.

Petty, R. E., Harkins, S. G., & Williams, K. D. (1980). The effects of group diffusion of cognitive effort on attitudes: An information processing view. Journal of Personality and Social Psychology, 38, 81–92.

Petty, R. E. & Krosnick J. A. (Eds.), (1995). Attitude strength: Antecedents and consequences. Mahwah, NJ: Erlbaum.

Petty, R. E., Schumann, D. W., Richman, S. A., & Strathman, A. J. (1993). Positive mood and persuasion: Different roles for affect under high- and low-elaboration conditions. Journal of Personality and Social Psychology, 64, 5–20.

Petty, R. E., & Wegener, D. T. (1998). Attitude change: Multiple roles for persuasion variables. In D. Gilbert, S. Fiske, & G. Lindzey (Eds.), The handbook of social psychology (4th ed., pp. 323–390). New York: McGraw- Hill.

Preacher, K. J., & Hayes, A. F. (2004). SPSS and SAS procedures for estimating indirect effects in simple mediation models. Behavior Research Methods, Instruments, & Computers, 36, 717–731.

Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behavior Research Methods, 40, 879–891.

Raven, B. H., & French, J. R. P. (1958). Legitimate power, coercive power, and observability in social influence. Sociometry, 21, 83–97.

Shavitt, S., Swan, S., Lowrey, T. M., & Wänke, M. (1994). The interaction of endorser attractiveness and involvement in persuasion depends on the goal that guides message processing. Journal of Consumer Psychology, 3, 137– 162.

Shrout, P. E., & Bolger, N. (2002). Mediation in experimental and nonexperi- mental studies: New procedures and recommendations. Psychological Methods, 7, 422–445.

Tetlock, P. E. (1983). Accountability and complexity of thought. Journal of Personality and Social Psychology, 45, 74–83.

Tormala, Z. L., Briñol, P., & Petty, R. E. (2006). When credibility attacks: The reverse impact of source credibility on persuasion. Journal of Experimental Social Psychology, 42, 684–691.

Tormala, Z. L., Briñol, P., & Petty, R. E. (2007). Multiple roles for source credibility under high elaboration: It’s all in the timing. Social Cognition, 25, 536–552.

Wegener, D. T., Clark, J. K., & Petty, R. E. (2006). Not all stereotyping is created equal: Differential consequences of thoughtful versus non-thought- ful stereotyping. Journal of Personality and Social Psychology, 90, 42–59.

Wegener, D. T., Downing, J., Krosnick, J. A., & Petty, R. E. (1995). Measures and manipulations of strength-related properties of attitudes: Current practice and future directions. In R. E. Petty, & J. A. Krosnick (Eds.), Attitude strength: Antecedents and consequences (pp. 455–487). Mahwah, NJ: Erlbaum.

Wegener, D. T., & Fabrigar, L. R. (2000). Analysis and design for nonexperi- mental data: Addressing causal and noncausal hypotheses. In H. T. Reis, & C. M. Judd (Eds.), Handbook of research methods in social and personality psychology (pp. 412–450). New York: Cambridge University Press.

Wegener, D. T., Petty, R. E., & Smith, S. M. (1995). Positive mood can increase or decrease message scrutiny: The hedonic contingency view of mood and message processing. Journal of Personality and Social Psychology, 69, 5– 15.

Wood, W., Lundgren, S., Ouellette, J., Busceme, S., & Blackstone, T. (1994). Minority influence: A meta-analytic review of social influence processes. Psychological Bulletin, 115, 323–345.

Source efficacy and persuasion 607

10990992, 2011, 5, D ow

nloaded from https://onlinelibrary.w

iley.com /doi/10.1002/ejsp.787 by Florida International U

niversity, W iley O

nline L ibrary on [08/05/2023]. See the T

erm s and C

onditions (https://onlinelibrary.w iley.com

/term s-and-conditions) on W

iley O nline L

ibrary for rules of use; O A

articles are governed by the applicable C reative C

om m

ons L icense