if;

r

. . . Kin“!

 

.I
.nii.uu......_mo.. Hun.
11.1...513 nflfidrnoflofflp .
”HPPlt...'.u..u!£ ,5.
If“... . .3. b
.51.... .. 5:17! ”If

mitt”.
.n», I it
I .

.Illl 11.1.1151 If"... 111.39.!

‘. 10' ‘10... A

.11. 38...}... .0132. F .:. ; E . 4.... . . I . . . , , . .a. ;. ..
Kuwait??? ‘ A . , . ,. gfig 8? #1. . . .,.+....L . is. . n. 8.3%.31 . .. M r “fig: 32...,

3119. . , ‘ A. .. I1“ . 3.1.1.1.!(‘1 -

 

S1. . y .a Q 1. Id .7 1.7.7.13

LIBRARY
Michigan State
University

This is to certify that the
dissertation entitled

Examining quantitative and qualitative evidence in perception
changing messages: A study of social norms message
effectiveness

presented by
Thomas A. Fediuk

has been accepted towards fulfillment
of the requirements for the

PhD. degree in Communication

Major Professor’s Signature

Qmém 2/ 200%

Date

MSU is an Affirmative Action/Equal Opportunity Institution

..-.-.-u-n-n-o--.-o--—--c----o-------

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/CIRC/DateDue.p65-p.15

EXAMINING QUANTITATIVE AND QUALITATIVE EVIDENCE IN PERCEPTION
CHANGING MESSAGES: A STUDY OF SOCIAL NORMS MESSAGE
EFFECTIVENESS
By

Thomas A. Fediuk

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Communication

2004

ABSTRACT
EXAMINING QUANTITATIVE AND QUALITATIVE EVIDENCE IN PERCEPTION
CHANGING MESSAGES: A STUDY OF SOCIAL NORMS MESSAGE
EFFECTIVENESS
By
Thomas A. Fediuk
One of the challenges that college campuses face is the harmful consequences of alcohol
consumption by college students. To deal with this problem, university officials have
implemented communication campaigns based on the social norms approach. Social
norm campaigns are designed to provide information about the actual prevalence of
drinking patterns and attitudes about alcohol consumption on campus. Such information
is expected to correct rnisperceptions held by students about alcohol consumption
behavior and attitude. While case studies have documented the impact of social norms
campaigns, very little research has focused on the components of various social norms
message strategies. The primary study examined the effectiveness of different types of
evidence (quantitative statistics or qualitative testimonials) in message believability and
altering student perceptions of peer attitudes about the acceptability of heavy drinking
(injunctive norms). While messages containing both quantitative and qualitative evidence
were found to be more believable than the control and qualitative evidence only
condition, no differences were found across message type for perception change of
injunctive norm beliefs.
Exploratory supplemental studies examined message believability of six different

message components. A message citing evidence based on a moderately large sample

size was found to be more believable than a message citing an extremely large sample
size. A two-sided message explicitly mentioning and refuting widely-held
misperceptions was found to be more believable than a one-sided version simply stating
factual social norms information. No significant differences were obtained in the
believability of three other message variations: presentation of numerical data vs. verbal
labels of normative information, positive vs. negative message flames, and specific vs.
general attributions of testimonial evidence for norm claims. In addition, the quantitative
vs. qualitative induction for injunctive norms in the primary study was replicated with

descriptive norms; the two versions do not differ significantly.

ACKNOLEDGEMENTS

I would like to thank Dr. Atkin for all of his guidance throughout my graduate
studies. Without his assistance, none of this would be possible. Dr. Atkin made himself
available whenever I came seeking his help. Our conversations about the
conceptualization, operationalization, and practical implications of the topic at hand were
invaluable. He also made me aware of the need to focus how I phrased my ideas and the
importance word choice made in the proper understanding and explication of constructs.

I would like to thank Dr. Smith. Our discussions on theoretical issues helped
formalize many ideas in the current study, as well as help lay a foundation for future
studies. She reminded me that the dissertation can be completed. Sometimes I needed to
hear that. Dr. Smith was a significant contributor to the completion of this dissertation.

I would like to thank Dr. Boster. Most importantly, he helped me develop a
framework of understanding social phenomena. This alone has a lasting impact. Dr.
Boster’s insight in the development of the dissertation helped transform many ideas into a
more coherent set of studies. I will forever be grateful.

I am a stronger scholar through my interactions with Dr. Salmon. He helped me
understand the importance of viewing issues from different perspectives. Through classes
and discussions, he helped his students examine ideas through different lenses. I hope
that I never forget this lesson.

Finally, I would like to thank my wife, Isabel Botero. Words can not be written or
expressed to signify the degree of my appreciation.

T.A.F.

iv

TABLE OF CONTENTS

LIST OF TABLES ................................................................................ vii

INTRODUCTION ................................................................................... 1
CHAPTER 1

SOCIAL NORMS .................................................................................. 2

Types of Norms ........................................................................... .3

Message Effectiveness ................................................................... .6

Evidence Type ............................................................................. 8
CHAPTER 2

PRIMARY STUDY: EVIDENCE TYPES ..................................................... 12
CHAPTER 3

SUPPLEMENTAL STUDIES ................................................................... 15

Supplemental Study A: Sample Size ................................................... 15

Supplemental Study B: Numerical Specificity ....................................... 16

Supplemental Study C: Testimonial Attribution ...................................... 17

Supplemental Study D: Disapproval of Unhealthy Behavior ....................... 18

Supplemental Study E: Descriptive Norms and Evidence ........................... 19

Supplemental Study F: Two-sided Refutational Message ........................... 19
CHAPTER 4

METHOD ........................................................................................... 22

Participants ................................................................................. 22

Design ...................................................................................... 23

Message Induction ........................................................................ 23

Primary Study ..................................................................... 23

Supplemental Study A: Sample Size .......................................... 24

Supplemental Study B: Numerical Specificity .............................. 24

Supplemental Study C: Testimonial Attribution ............................ 25

Supplemental Study D: Disapproval of Unhealthy Behavior .............. 25

Supplemental Study E: Descriptive Norms and Evidence ................. 25

Supplemental Study F: Two-sided Refutational Message .................. 26

Procedure .................................................................................. 26

Instrumentation ........................................................................... 27

Perception of Injunctive Norms ................................................ 28

Message Believability ........................................................... 28

Induction Check .................................................................. 29

CHAPTER 5

RESULTS .......................................................................................... 30
Induction Checks ......................................................................... 30
Primary Study ............................................................................. 30
Supplemental Study A: Sample Size ................................................... 34
Supplemental Study B: Numerical Specificity ....................................... 34
Supplemental Study C: Testimonial Attribution ..................................... 35
Supplemental Study D: Disapproval of Unhealthy Behavior ....................... 35
Supplemental Study E: Descriptive Norms and Evidence .......................... 36
Supplemental Study F: Two-sided Refutational Message ........................... 36

CHAPTER 6

DISCUSSION ..................................................................................... 37
Supplemental Studies ..................................................................... 40
Limitations ................................................................................. 44
Future Directions ......................................................................... 45

APPENDICES ..................................................................................... 50

BIBLIOGRAPHY ................................................................................. 74

vi

LIST OF TABLES

TABLE 1:
Means and Standard Deviations for Primary Study .......................................... 47
TABLE 2:
Analysis of Variance for Message Verifiability ............................................. 47
TABLE 3:
Analysis of Variance for Testimonials ........................................................ 48
TABLE 4:
Means and Standard Deviations for Drinking Injunctive Norm Perceptions ............ 48
TABLE 5:
Analysis of Variance for Message Believability ............................................. 49
TABLE 6:
Means and Standard Deviations for Message Believability ............................... 49

vii

INTRODUCTION

A common perception of college life is that students drink alcohol fi'equently and
drink beyond moderation, often getting intoxicated. College students, parents, community
members, and future college students perceive that heavy drinking is a typical behavior
on college campuses (Perkins & Berkowitz, 1986; Perkins, Meihnan, Leichliter, Cashin,
& Presley, 1999; Prentice & Miller, 1993; Sher, Bartholow, & Nanda, 2001). The belief
is that many students drink to extreme levels, which exceed what would be considered
healthy alcohol consumption.

Heavy alcohol consumption is problematic in many ways. Heavy drinking may
lead to harmful consequences, such as alcohol poisoning, vandalism and property
destruction, rioting, fighting, sexual abuse, unprotected sex, and drunk driving. To
combat the adverse consequences of excessive alcohol consumption by the student body,
college campuses have launched numerous persuasive campaigns.

Many public communication campaigns have been designed and implemented targeting
campus-wide alcohol consumption. These communication campaigns are used to inform
or persuade a target audience to ultimately alter their behavior regarding alcohol usage.
The campaigns typically deliver the message with posters, flyers, and newspaper ads. The
message provides information and incentives that would ideally lead the recipient to
accept the proposition of the campaign. Example propositions have included alcohol
abstinence, alcohol consumption moderation, the use of a designated driver, or self and

other protective behaviors such as watching out for a fiiend or pacing drinks of alcohol.

CHAPTER 1

SOCIAL NORMS

The social norms approach is centered on the discovery that people misperceive
the actual normative attitudes and behavior of similar others (Perkins & Berkowitz, 1986;
Perkins et a1., 1999; Prentice & Miller, 1993; Sher, Bartholow, & Nanda, 2001). College
students overestimate the extent to which other students engage in heavy drinking on
college campuses, and that students often overestimate the level of approval for risky
drinking behavior and drunkenness. While the belief that most college students drink
excessively is prevalent, such beliefs are a misperception of actual drinking behaviors
(Perkins & Berkowitz, 1986). When students are asked to estimate the number of drinks
students consume on an average night of partying, the figure is much higher than actual
drinking behavior engaged in by students. These perceptions are proposed to lead to
perceptions of normative expectations (perceptions of how peers wish one to act).
Students are then motivated to act in accordance to the perceived normative expectations
(Perkins, 1997; Perkins, 2002). Students who are moderate in their drinking behavior or
prefer to abstain from drinking are predicted to feel internal pressure to act in accordance
to the perceived expectations of their peer group. Also, those who may be more liberal in
their drinking preferences may act upon consumption desires with relatively little fear of
peer disapproval (Perkins, 2003). These liberal drinkers justify their risky drinking
behavior by pointing out that everyone in college acts in the same way.

Based on the misperception findings, campaigners began to create intervention
strategies that were designed to correct the perceptions of risky alcohol behavior and

attitudes on campuses. Researchers at Northern Illinois University developed a

communication campaign designed to inform students about the accurate amount of
alcohol consumption taking place on campus (Haines, 1996; Haines & Barker, 2003;
Haines & Spear, 1996). Messages informed students about actual drinking norms
regarding moderate drinking and protective behaviors to reduce alcohol related harm
incidents. In a nine-year period, the perceptions of NIU students who engage in risky
drinking fell from an average of 63% to an average of 33%. Self-reported excessive
drinking behavior also fell from 45% to 25%, while self-report alcohol related injury
decreased from 36% to 16%.

Over the years, the social norms approach has increased in popularity, and
currently, 20% of campuses designing campaigns targeting alcohol consumption are
using a social norms approach (Wechsler, Kelly, Weitzman, San Giovanni, & Seibring,
2000). Social norms campaigns and the ensuing literature have generated a body of
research describing the successes of the social norms intervention approach (Haines &
Spear, 1996; Haines, 1996; Johannessen, Collins, Mills-Nova, & Glider, 1999; Perkins,
2002; Perkins & Craig, 2002). While evidence supports a reduction in drinking behavior
and alignment of peer perceptions to actual drinking behaviors after a social norms
campaign, relatively little attention has been given to study systematically the effects of
specific components of a social norms message.

Types of Norms

Misperceptions of campus-wide alcohol norms include both descriptive and
injunctive norms. Descriptive norms serve as a description as to what is normal behavior
(Cialdini, Reno, & Kallgren, 1990). Descriptive drinking norms are defined as “the

perception of other’s quantity and frequency of drinking in discrete drinking situations”

(Borsari & Carey, 2001; pp. 401-402). Such norms represent the behavior of others in
regard to drinking amounts, how frequently they drink, and the normal consumption
levels during atypical night out. Examples of a descriptive norms message could state
that “Most students drink moderately” or “67% of students drink five or fewer.”
Injunctive norms are defined as “the perceptions of others’ approval of drinking. . .and
represent perceived moral rules of the peer group” (Borsari & Carey, 2001; p. 402).
Cialdini, Reno, and Kallgren (1990) refer to injunctive norms as the ‘ought to’ norms, in
that injunctive norms are the rules or beliefs of what is approved or disapproved
behavior. The researchers continue that injunctive norms specify what ought to be done.
Injunctive norms are perceptions of the attitudes that others hold toward the acceptability
of a drinking behavior; whether they approve or disapprove of a specific behavior. An
example of an injunctive norms message may state that “Most students feel that drinking
five or fewer is acceptable.” Both descriptive and injunctive norms are conceptually and
motivationally distinct (Cialdini, Reno, & Kallgren, 1990).

Fediuk (2004) found that injunctive norms perceptions did impact behavior
intentions (r = .50). Descriptive norm perceptions were found to have no significant
impact on behavior intentions (r = .06). While most social norms campaigns generate
descriptive norms messages (Borsari & Carey, 2001), injunctive norm perceptions may
serve as a better target for perception correction since such beliefs were found to have a
strong impact on behavior intentions toward alcohol consumption.

In order to make any causal claims to the effectiveness of the social norms
approach, research must first show evidence that the message corrects misperceptions. In

his experiment, Fediuk (2004) did not find that the presentation of social norms messages

targeting either injunctive or descriptive norms altered normative perception beliefs.
Perceptions toward the message proposition are not altered with the simple presentation
of information (Darley & Gross, 1983; Edwards & Smith, 1996; Kahneman & Tversky,
1973; Lord, Ross, & Lepper, 1979; Munro & Ditto, 1997; Nisbett, Crandall, & Reed,
1976). In fact, a large body of research exists examining the different ways, factors, and
degree of effectiveness of varying persuasive messages and appeals (Chaiken, 1980;
Hovland, Janis, & Kelley, 1953; McCroskey, 1969; McGuire, 2001; Petty & Cacioppo,
1986; Stiff, 1986). Social norms messages are persuasive messages. Social norm
messages are designed to alter existing perceptions from the belief that most students
engage in or approve of a behavior to most do not engage in or approve of such
behaviors. If social norm message effectiveness is defined as the changing of existing
perceptions, such that the perceptions toward the extent of risky drinking behavior or
approval of such behavior on campus reflect reality more accurately, new questions arise
as to what factors will improve message impact on perception change. As called for by
Reinard (1988), research into message effectiveness should isolate the different types of
evidence used and to distinguish the effects of each type of evidence.

The present study attempts to understand how presenting different types of
evidence (qualitative or qualitative) affect student perceptions about peer behavior and
the student’s attitude toward the acceptability of heavy drinking. Injunctive norms were
found to be more impactful on behavior intentions than descriptive norms (F ediuk, 2004).
The primary study was a 2 (no quantitative evidence, quantitative evidence) X 2 (no
qualitative evidence, qualitative evidence) design intended to alter perceptions of

injunctive norms.

A series of exploratory supplemental studies were conducted to examine the
believability of six approaches for presenting quantitative and qualitative evidence. The
first supplemental study compared large versus moderate sample sizes from which the
sample is drawn. The second supplemental study varied the degree of specificity of
quantitative evidence (quantitative statistic compared to a verbal quantitative label). The
third supplemental study used testimonials and examined the specificity of source
attribution and its impact on message believability. The fourth supplemental study
explored the relative believability of disapproval of unhealthy behavior versus approval
of a healthy behavior. The final two supplemental studies focused on descriptive norms.
The fifth supplemental study examined the impact of quantitative versus qualitative
evidence on message believability. The sixth supplemental study compared a one-sided
message with a two-sided refiitational message acknowledging the discrepancy of
expectations about student drinking patterns.

Message E flecti veness

Systematic study into the effectiveness of messages designed to correct
misperceptions has been lacking (F ediuk, 2004). Perception correction messages have
been created across many campuses in the United States, but relatively little has been
done to conduct detailed analysis of the messages presented to correct misperceptions.
The question is whether social norms messages do in fact correct normative perceptions.
Since the message is the key component of the social norms campaign, it is vital to study
the message characteristics and impact on recipients more thoroughly to better advance

social norms campaigns as a legitimate intervention strategy.

To increase the likelihood that a message is accepted, the message presented
needs be believable. Believability of a message grants the message legitimacy
(Rosenthal, 1971). If the audience does not believe the claim or the evidence presented,
perception change is most likely not going to occur. If a message is believable, then it is
more likely that the message will be processed and accepted, and in turn, the recipients
are more likely to be persuaded by the message (Rosenthal, 1971). One way to examine
factors that would bolster beliefs in a claim is to focus on the evidence presented to
support the claim of the message (Baesler & Burgoon, 1994).

Toulrnin (1964) provides a framework for the analysis of persuasive arguments.
Three primary components of the argument are the claim, data, and warrant. The claim is
the persuasive proposal in the argument. In a descriptive social norms message, the claim
can be that “Most students drink moderately or not at all.” Data are the evidence offered
to support the claim of the message. In the case of the social norms message, the evidence
must support the notion that most students drink moderately or not at all. The data may
be in the form of statistics or qualitative testimonials. Finally, the warrant is the statement
that connects the data with the claim. In the social norms message, the warrant is implicit
rather than explicitly expressed.

The current study focused on data presented in a social norms campaign to
support the social norms claim. The basic claim is that most students engage in healthy
drinking practices and approve of others engaging in healthy drinking practices. In other
words, most students do not drink excessively. Evidence is then provided to support the
claim. Recipients of a persuasive message judge the evidence, or data and warrant,

presented in the message (Reynolds, 1987; Reynolds, 2002; Wood, Rhodes, & Biek,

1995). The presentation of evidence has been shown to impact persuasiveness (Allen &
Burrell, 1992; Reinhard, 1988).

Evidence is defined as “data (facts or opinions) presented as proof for an
assertion” (Reynolds, 2002, pg. 429). Two types of evidence are testimonial assertions
and factual information (Reinard, 1988). Testimonials may consist of qualitative
statements or stories which center around the claim. For instance, students may be quoted
supporting the notion that they themselves do not drink excessively. The quotes may also
state that they approve of healthy drinking behaviors and disapprove of excessive
drinking or drinking to intoxication. These testimonials are based on single episodes
rather than summative statistics. Quantitative evidence does consist of summative
statistics, in that evidence is presented in an empirical format.

Evidence Type

For the purpose of the primary study, quantitative and qualitative evidence types
are discussed. Quantitative evidence is information about an object, person, concept or
summary statistical information of a numerical nature (Church & Wilbanks, 1986;
Dickson, 1982; Kahneman & Tversky, 1973; Kazoleas, 1993). Summary statistical
information is “information presented as a rate, frequency, proposition, percentage,
probability, average, median, mode, or some other statistical parameter. Such expressions
often involve the use of mathematical symbols, such as numeric rather than written
figures, the percent sign, and use of the decimal point” (Dickson, 1982, p. 399). Social
norms messages have typically provided claims and evidence of a quantitative nature.

The message may contain a quantitative statistic, such as “67% of students” or a

quantitative label, such as “most” or “majority.” The quantitative label is more
ambiguous than the statistic, but the label still represents a summative quantity.
Qualitative evidence refers to evidence presented in a narrative fashion such as
personal anecdotes, analogies, examples, and testimonies (Kazoleas, 1993). Testimonial
assertions are qualitative in nature. The information presented is not in a summative
form. Rather, the evidence consists of a small sample size, and may include only one case
history or a series of cases (Dickson, 1982). In a social norms campaign, testimonials
could indicate whether peers ‘approve’ or ‘disapprove’ of certain behaviors. A message
may state that most students feel that others ‘ought to’ engage in healthy behavior or
‘should not’ engage in non-healthy behavior. For example, an injunctive norms message
may include students expressing opinions that others ought not to drink excessively, or
that they approve of students drinking moderately. A descriptive norms testimonial may
have students stating that they drink moderately or that they do not drink excessively.
The testimonials can be statements expressing the behavior or attitude of a peer group.
To date, no research has examined the evidence presented supporting a social
norms message claim. Research into whether quantitative or qualitative evidence is more
persuasive in general has produced inconsistent findings. In some studies, qualitative
evidence has been found to be more persuasive than quantitative evidence (Borgida &
Nisbett, 1977; Kahneman & Tversky, 1973; Nisbett & Borgida, 1975). Qualitative
evidence has been found to be more vivid and memorable than quantitative evidence
(Baesler & Burgoon, 1994). On the other hand, some studies have found that quantitative
evidence is more persuasive than qualitative evidence (Allen & Preiss, 1997; Baesler &

Burgoon, 1994; Dickson, 1982; Lindsey & Ah Yun, 2003; Reinard, 1988; Wells &

Harvey, 1977). In a meta-analysis, where the effect size of multiple studies is assessed,
Allen and Preiss (1997) report an average correlation of .10 across 16 studies (N = 1836),
where statistical information is more persuasive than qualitative information. Statistics
are viewed as being obtained fi'om some objective measure or study, as opposed to
personal testimonials, which are statements by individuals who may be biased or not
selected at random.

One explanation as to why quantitative evidence may be more persuasive than
qualitative evidence is that quantitative evidence information is perceived to be
verifiable, and thus more believable (Lindsey & Ah Yun, 2003; Rosenthal, 1971).
Rosenthal (1971) considers a message verifiable if it is precise enough to be empirically
testable. Message recipients who perceive that the evidence is verifiable may perform
further investigation. It is not necessarily the case that the recipients would go out and
verify the information. The key is that the message could be verified if one wishes to do
so. If recipients view that the evidence can be tested in some objective manner, then they
are more likely to believe the message. Such research opportunities may reduce
perceptions that the evidence has been manipulated.

Evidence can vary in its specificity, or how specific the message represents the
quantifiable evidence. Highly specific messages present more detailed information
(Rosenthal, 1971) and are thus more verifiable. The recipient does not need to speculate
or need to provide further cognitive efforts to understand the claim and evidence; the
message provides a sufficient amount of information. For example, “Most students” is
less specific than “67% of students.” The word most can signify anywhere from 50.1% to

100%. Perceptions vary as to what figure would represent “most students.” Such verbal

10

generality and ambiguity reduces the perception that one can accurately verify the
evidence of the claim. Stating that 67% of some population engages in some behavior
would be less exposed to multiple interpretations. The precise figure of 67% would be
expected to be interpreted the same by most recipients.

Another term often used in social norms campaigns is ‘moderately’. The precise
meaning of moderate drinking behavior can be interpreted in many ways. For individuals
who tend not to drink, an evening consuming 2 or fewer may be considered moderate
(whether 2 evenings of alcohol consumption is moderate is open for debate), while a
student who tends to drink a large amount of alcohol, an evening of moderate drinking
may be limiting one’s self to less than eight drinks. A more specific term would be to
define the term moderate as consuming “five or fewer” alcoholic beverages when out
partying-

In brief, social norms messages consist of claims indicating that most students
drink moderately. Such a claim is counter to pre-existing perceptions. In order to present
a persuasive argument, the claim needs backing evidence that supports the main claim.
Evidence may consist of quantitative or qualitative data. The current research examined
the impact evidence type has on message believability and whether perception

differences are impacted by the different evidence types presented in the message.

11

CHAPTER 2

PRIMARY STUDY: EVIDENCE TYPES

F ediuk (2004) found that injunctive norms have a greater impact on behavior
intentions than do descriptive norms perceptions. Inj unctive norms encompass both
perceptions of what behavior is seen to be appropriate, as well as the ‘ought to’ norms.
Therefore, the primary study, as well as most of the supplemental studies, focuses on
inj unctive norms messages.

Messages targeting injunctive norms could provide evidence for the social norms
claim with either qualitative or quantitative evidence. Statistics may report the proportion
of students who hold the attitude or opinion that most students should drink moderately.
For example, “67 % of students report that others ought to drink five or fewer” or “most
students report that other students ought to drink five or fewer when out partying.” Such a
message would be backed by the inclusion of some sample size, which indicates that the
statistics are from a randomized, summary study. Testimonials, on the other hand, could
be created using actual testimonials from students that would indicate students’ attitudes
toward excessive or healthy drinking behavior (approval or disapproval) as well as
statements indicating what students feel others should do in regard to drinking behavior.
Such evidence may or may not include sample size or any indication that the testimonials
come from a larger study.

Messages may also contain no evidence supporting an assertion. Such lack of
evidence has been found to be less impactful than messages that contain supportive

evidence (Kazoleas, 1993; Lindsey & Ah Yun, 2003). Creating messages that claim that

12

most students drink moderately and provide no evidence to support this claim is expected
to be less believable, and less impactful than messages that include supportive evidence.
Students who hold high injunctive norms perceptions feel that most of their peers
approve of drinking to riskier levels, while students who hold low injunctive norms
perceptions feel that most of their peers do not approve of drinking to riskier levels.

H1: Subjects exposed to injunctive norms messages containing supporting

evidence will have lower perceptions of injunctive norms than subjects who are

exposed to messages with no supporting evidence.

H13: Subjects exposed to injunctive norms messages containing supporting

evidence will find the message more believable than subjects exposed to

messages with no supporting evidence.

Allen & Preiss (1997) report that quantitative evidence is more persuasive than
qualitative evidence. Baesler & Burgoon (1994) indicate that recipients of a message that
provides statistics and a large sample size process the information heuristically rather
than systematically. Social norms messages are designed to inform recipients of what the
summative behaviors and attitudes existing on college campuses. Such summative
messages composed of quantitative evidence are expected to be more impactful than
messages containing non-summative qualitative testimonials.

H2: Subjects exposed to injunctive norms messages containing quantitative data

will have lower perceptions of injunctive norms than subjects who are exposed to

messages containing qualitative testimonials.

l3

HZa: Subjects exposed to injunctive norms messages containing quantitative data
will find the message more believable than subjects exposed to messages
containing qualitative testimonials.

Messages containing both quantitative and qualitative evidence may increase
believability and credibility to the social norm claim. The supporting evidence would
consist of both the summative statistic as well as a few testimonials from students.
Research examining the benefits of including both types of evidence is limited (Allen &
Preiss, 1997), but recent research has examined the impact of doing so. For instance,
Allen, et. al., (2000) found that the inclusion of both quantitative and qualitative
information increased message persuasiveness more so than quantitative or qualitative
evidence only messages.

H3: Subjects exposed to injunctive norms messages containing both quantitative

data and qualitative testimonials will have lower perceptions of injunctive norms

than subjects who are exposed to messages containing only one type of evidence
or no evidence.

H33: Subjects exposed to injunctive norms messages containing both quantitative

data and qualitative testimonials will find the message more believable than

subjects exposed to messages containing only one type of evidence or no

evidence.

14

CHAPTER3

SUPPLEMENTAL STUDIES

Supplemental studies were conducted to be exploratory in nature. The purpose is
to identify ideas that are most promising and worth pursuing.

When presenting a message designed to alter pre-existing perceptions, evidence
supporting the claim need to be seen as believable by the target audience. To further
understanding of message features, it is necessary to examine which message features are
seen to be believable. A series of studies to test different message features and the impact
these features have on message believability were conducted. Supplemental studies A, B,
C, and D targeted perceptions of injunctive norms. While targeting injunctive norms may
be more effective (Fediuk, 2004), many current campaigns focus on descriptive norms.
Therefore, the final two supplemental studies (supplemental studies E and F) targeted
descriptive norms.

Supplemental Study A: Sample Size

Message recipients may process a message systematically or heuristically
(Chaiken, 1980). While some recipients of a message will take time to consider the
evidence presented, other recipients will process the evidence fiom some type of message
cue. One such heuristic cue is the sample size of the study. Recipients of a message who
judge the message’s believability based on a one such as sample size are reported to be
processing the message using a sample size heuristic (Baesler & Burgoon, 1994; Lindsey
& Ah Yun, 2003). Baesler & Burgoon (1994) indicate that recipients judge the message

through a law of large numbers. Processing and judging evidence based on the

15

assumption that large numbers are more believable would indicate that recipients are
processing the message heuristically. In brief, if the evidence or statistic used in the
message is from a larger, summative study, then there is a greater likelihood of the
information being viewed as more accurate. Therefore, using the sample size heuristic,
evidence containing a larger sample size would be seen as more believable than evidence
with a smaller or more moderate sample size.

H4: Quantitative social norms messages targeting injunctive norm perceptions

that indicate a large sample size will be seen as more believable than messages

that report a more moderate sample size.
Supplemental Study B: Numerical Specificity

Evidence provided in support of a claim may vary in its specificity. The evidence
may contain very specific numbers and figures or may present ambiguous labels or
qualitative testimonials. Quantitative evidence may be further sub-divided into quantified
statistics or verbal labels. Quantitative statistic evidence would be the most specific
message possible. A quantitative statistic message may indicate that 67% of students
drink 5 or fewer drinks when they party. Verbal labels may be less specific, less precise,
and more ambiguous, but still suggest a summative statistic. The precise number of
occurrences, however, is not presented. An example of a verbal label can be “most
students drink moderately or not at all.” To increase specificity and keep a message
consistent, quantitative statistics contain numerical representations of both percentage
and quantity of drinks consumed. Messages containing more specific evidence would be
seen as more verifiable, and are expected to be more believable than more ambiguous

messages.

16

H5: Quantitative social norms messages targeting injunctive norm perceptions that
use quantitative statistics will be seen as more believable than messages that use
verbal labels.

Supplemental Study C: Testimonial Attribution

While summative statistics are gathered using a large body of a sampled
population, qualitative testimonials represent the opinions of specific students. The
presentation of testimonial evidence does not report results from a specific percentage of
a population. Rather, the message provides specific quotes or testimonials from other
students. For social norms messages to be effective, the recipient must perceive that the
testimonials are representative of a larger body of students. The presentation of one
testimonial may not lead the recipient to feel that the statement is representative of the
student body. Instead, the testimonial could be interpreted as the opinion of one person.
Therefore, multiple testimonials should be presented in the message.

Testimonials can be presented absent of any visually identifiable source. A simple
list of quotations surrounding a group may be presented. The testimonials are not
ascribed to any particular individual in the group, and may in fact come from some
source outside of the pictured group. Testimonials may also accompany pictures of the
source of the testimonial. The testimonial is ascribed to an individual, who may be
accountable for what they say. Since the testimonial is linked to a sender, such messages
may be perceived to be more believable. The message may be even more believable if the

source and testimonial contain the name of the sender.

17

RQI: Do visually identifiable testimonials increase the believability of a social

norms message targeting injunctive norms perceptions more so than testimonials

that are not visually identifiable to a source?
Supplemental Study D: Disapproval of Unhealthy Behavior

Rather than threatening negative consequences due to high levels of alcohol
consumption, social norms messages target healthy behaviors, or alcohol wellness
(Haines & Barker, 2002; Perkins, 1997; Perkins, 2003). One assumption of the social
norms approach is that people are motivated to partake in healthy behaviors. Providing
data that shows that most university students do indeed engage in healthy behaviors or
feel that such behaviors are appropriate are expected to relieve the individual from the
pressures to enact unhealthy practices believed to be the campus norm, and thus allows
students to engage in pre-existing healthy practices, such as abstinence or moderation.

Proponents of the social norms approach insist that a social norms message be
phrased as a positive message (Perkins, 2002). Messages need to stress the healthy
normative attitude or behavior rather than focus on negative consequences of engaging
against the normative healthy behavior. While the belief is that positively framed
messages will be more effective than negatively framed messages, evidence to support
the claim is not evident. It is unknown as to how students would react to a message
stating that most students engage in a healthy behavior or approve of healthy behavior
compared to a message indicating most do not engage in unhealthy behavior and

disapprove of those who do engage in such behavior.

18

RQ2: When presenting a social norms message targeting injunctive norms, are

messages claiming that peers approve of healthy behavior more believable than

messages claiming that peers disapprove?
Supplemental Study E: Descriptive Norms and Evidence Types

While the primary study examined the impact of evidence type on injunctive
norms messages, supplemental Study E examined whether qualitative or quantitative
evidence is more believable when presented with a descriptive norms message.
Descriptive norm messages target perceptions of the behavior of other students rather
than the attitudes and approval of other students. Descriptive norms can be observed by
students. Students do not necessarily need other students to tell them about their
behaviors. One may go to the bar and see how many drinks are being consumed. Such
observability may benefit from summative statistics reporting student drinking behaviors.
Therefore, descriptive norms messages containing summative statistics, which are
verifiable, are expected to be more believable than qualitative testimonials of student
drinking behaviors.

H6: Quantitative social norms messages targeting descriptive norm messages will

be perceived to be more believable than messages providing qualitative

testimonials.
Supplemental Study F: T wo-Sided Refutational Message

College students hold preconceptions as to the amount of drinking and level of
approval toward excessive drinking on campus. Recipients of the social norms message
are provided with information which informs students that most students do not engage in

excessive drinking. Social norms messages tend to offer one-sided messages and

19

assertions without providing reference to pre-existing beliefs. One approach to message
construction is to mention the prevalent beliefs (counterarguments) and then to refute
those beliefs. A social norms message may acknowledge that most people do in fact feel
that most students engage in risky behavior. Following the statement, the message may
present the factual message which provides information countering pre-existing beliefs.

Allen (1991) conducted a meta-analysis comparing the persuasiveness of one-
sided and two-sided messages. One-sided messages only present one side of an argument,
which tends to be favorable toward the message claim. Two-sided messages are messages
that include opposing arguments (Allen, 1998). Results indicate that two-sided messages
are more persuasive than one-sided messages, r = .03. However, two-sided messages can
be separated into two-sided non-refutational and two-sided refutational messages (Allen,
1991). Non-refutational messages only offer counterarguments, but do not refute these
arguments. Refutational messages mention the counterargument as well as refute, or
provide evidence against, the counterargument. Allen (1991) found that two-sided
refutational messages are more persuasive than one-sided messages (r = .07). Two sided
non-refutational messages are less persuasive than a one-sided message (r = -.09).

Most social norms messages can be classified as one-sided messages. However, a
two-sided refutational message may be seen as more believable than a one-sided
message. A two-sided refutational message would include counterarguments and beliefs
held by students. By acknowledging and presenting counterarguments in the social norms
message, students may feel that the message is more believable. Lucas (1989) claims that
communicators who use two-sided refutational messages are seen as more credible than

communicators who do not acknowledge counterarguments. The rationalization for this is

20

that the speaker has taken counterarguments under consideration and is not just
presenting one side of the discussion.
H7: Two-sided refutational messages targeting descriptive norm perceptions will

be rated as more believable than one-sided messages.

21

CHAPTER 4

METHOD

Participants

For the primary study, 230 students were recruited from communication courses
at a large Midwestern university during the spring semester of 2004. Participants were
randomly assigned to experimental conditions. Of the participants, 8% were freshmen,
25% sophomore, 36% juniors, and 31% seniors; 58% were female and 42% were male;
17% reported being involved in a Greek system (fraternity or sorority), 14% were active
in organized athletics, 4% were honor’s college students, and 16% report being members
of religious institutions; 74% reported having seen some type of alcohol related message
on campus, with a median of 5 messages viewed in the last year; 91% reported that they
consume alcoholic beverages. Participants reported partying an average of two times a
week, and consume on average 5.93 drinks when out partying.

For the supplemental studies, 102 students were recruited from communication
courses during the 2004 summer session. Participants were randomly assigned to
experimental conditions. Of the participants, 2% were freshmen, 8% sophomore, 24%
juniors, and 66% seniors; 61% were female and 39% were male; 16% reported being
involved in a Greek system (fraternity or sorority), 7% were active in organized athletics,
4% were honor’s college students, and 18% report being members of a religious
institution; 79% reported having seen some type of alcohol related message on campus,
with a median of 5 messages viewed in the last year; 87% reported that they consume
alcoholic beverages. Participants go out partying an average of three times a week, and

consume on average 5.79 drinks when out partying.

22

Design

A 2 (quantitative evidence, no quantitative evidence) X 2 (qualitative evidence,
no qualitative evidence) between-subj ects factorial design was employed for the primary
study. The dependent variables were message believability and post-message injunctive
norm perceptions.

For the supplemental studies, the dependent variable was message believability.
Supplemental studies A, B, C, and D targeted perceptions of injunctive norms.
Supplemental study A used a l (injunctive norms message) X 2 (moderate one year
sample size, large multi-year sample size) between subjects factorial design.
Supplemental study B used a 1 (injunctive norms message) X 2 (quantitative statistic,
quantitative label) between subjects factorial design. Supplemental study C used a
1(injunctive norms) X 3 (testimonials by aggregate group attribution, testimonials by
individual attribution, testimonials by named individual attribution) between subjects
factorial design. Supplemental study D used a l (injunctive norms) X 2 (approve,
disapprove) between subjects factorial design. Supplemental studies E and F targeted
descriptive norms perceptions. Supplemental study E used a 1 (descriptive norms
message) X 2 (qualitative evidence, quantitative evidence) between subjects factorial
design. Supplemental study F used a l (descriptive norms) X 2 (refutation, no refutation)
between subjects factorial design.

Message Inductions

Primary Study. Participants received one of four messages. The message claim

was that “Most MSU students approve of drinking moderately.” Quantitative and/or

qualitative evidence was provided for support. The no evidence control condition

23

contained the same main claim, but the supporting message simply reiterated the main
claim rather than presenting quantitative or qualitative evidence (i.e., assertion alone).
The quantitative evidence stated that “67%* of MSU students report holding the opinion
that others ought to drink five or fewer when out partying.” The asterisk guided the
viewer to the bottom of the page, which reported the sample size of the study as well as a
web page students may view for further details. The qualitative testimonials had a support
tag line indicating that “MSU students express the opinion that others ought to follow the
five or fewer guideline when out partying.” Four testimonials were provided as support.
The testimonials were created from an in-class exercise using undergraduate students at
the University of Michigan. The message using both qualitative and quantitative evidence
had a tag line stating that “67%* of MSU students express the opinion that others ought
to drink five or fewer when out partying.” Four testimonials were included. The asterisk
guides the viewer to the bottom of the page, which reports the sample size of the study as
well as a web page students may view for further details.

Supplemental Study A: Sample Size. Supplemental study A examined whether
believability of a message varies due to sample size. Both messages contained the same
quantitative evidence. The factor that varied was the sample size reported under the
asterisk at the bottom of the page. One condition reported N = 1,005, while the other
reported N = 5,005. The sample size of 1,005 was based on a one-year study. In order to
attain a plausible sample size of 5,005, the message cites an ongoing study accumulating
a sample over a five year period.

Supplemental Study B: Numerical Specificity. Supplemental study B examined the

impact of message specificity versus verbal generality. Quantitative statistics (more

24

specific evidence) stated that “67% of MSU students“ report holding the opinion that
others ought to drink 5 or fewer when out partying” while the verbally general
quantitative label indicated that “Most MSU students report holding the opinion that
others ought to drink moderately when out partying.”

Supplemental Study C: Testimonial Attributions. Supplemental study C examined
the believability of visual attribution of testimonials to an individual or a group. The
group message showed a group of four students and the four testimonials appearing on
the message, but not visually linked to an individual. The individual message visually
linked the testimonial to an individual. A variation of the message showed the individual
testimonials visually linked to a named source.

Supplemental Study D: Disapproval of Unhealthy Behavior. Supplemental study
D examined whether a message was more believable if framed in a positive approval or
negative disapproval format. The claim was either “Most MSU students approve of
drinking moderately” or “Most MSU students disapprove of drinking heavily.” The
approval message was that “67% of MSU students“ approve of others drinking five or
fewer when out partying.” The disapproval message stated that “67% of MSU students“
disapprove of others drinking more than five drinks when out partying.”

Supplemental Study E: Descriptive Norms and Evidence Type. Supplemental
study B examined the believability of quantitative or quantitative evidence supporting a
descriptive norms message. Descriptive norms messages focus on student behaviors
rather than attitudes. To alter message topics, the descriptive norms message used here
provided a claim about the frequency of college students going out partying. The main

claim indicated that “Most MSU students drink infrequently.” The supporting qualitative

25

evidence indicated that “81% of MSU students“ report drinking twice a week or less” and
the qualitative evidence indicated that “MSU students report drinking twice a week or
less” and contained four testimonials.

Supplemental Study F: Refutation Message. Supplemental Study F examined
whether explicitly referring to student overestimation of college drinking was more
persuasive than not providing such information. The message was a descriptive message
about the frequency of student drinking. The claim was that “Most MSU students drink
infrequently.” The supporting claim read “Many MSU students overestimate how often
other students go out drinking perhaps it’s because some of their fi'iends seem to drink
frequently.” The supporting evidence was quantitative, in that “The fact is that 81% of
MSU students" go out drinking twice a week or less, according to data in a campus-wide
alcohol survey.”

Procedure

For the primary study, participants were randomly given a packet containing the
consent form, message induction, and survey. Participants were asked to estimate the
percentage of peers who approve of excessive alcohol consumption (pre-message
induction injunctive norms). The reported figure served as a baseline for injunctive norm
perceptions. Following the initial questions, participants were exposed to one of four
injunctive norm message evidence types. Participants viewed either a message containing
quantitative statistics, qualitative testimonials, both types of evidence, or no evidence.
After viewing the message, participants answered questions about descriptive and
injunctive norm beliefs, normative expectations, alcohol consumption intentions,

personal approval of excessive alcohol consumption, the believability of the message,

26

and induction check questions. Following these questions, some general questions
pertinent to formative evaluation were asked. Finally, demographic questions were asked.
See Appendix A for messages and Appendix B for the questionnaire.

For the supplemental studies, a separate sample of participants received a packet
containing the consent form, four message inductions, and survey. After each message,
participants answered five items assessing message believability. One of three injunctive
norm messages was received first by the participant: quantitative statistic with a moderate
sample size (N = 1,005), quantitative statistic with a large sample size (N = 5,005), or a
verbal label message (The word “most” rather than a quantitative statistic). The second
message presented was an injunctive norms message providing testimonial evidence. One
of three messages was received by the participant: group-attributed testimonials,
individual-attributed testimonials, or named individual-attributed testimonials. The third
message presented was a descriptive norms message about the frequency of drinking by
students. One of three messages was received by the participant: Quantitative evidence,
qualitative evidence, or a two-sided refutational message. The final message presented an
injunctive norms message that indicated whether other students approved of moderate
drinking or disapproved of excessive drinking. After all messages were viewed,
demographic information was collected, as well as an open-ended question asking “What
were some of the thoughts you had while viewing these messages?” See Appendix A for
messages.

Instrumentation
Perceptions of Injunctive Norms. The primary study assessed perceptions of

norms using two scaling methods developed by Fediuk (2004). To assess perceptions of

27

injunctive norms prior to message exposure, five percentage estimate questions were used
to measure perceived injunctive norms regarding alcohol consumption as well as
frequency of consumption events (See Appendix B for items). Attitudes toward alcohol
consumption behaviors include excessive drinking, students getting drunk, drinking more
than five alcoholic beverages, and frequently getting drunk. The five item scale was
tested using confirmatory factor analysis. Item five (“What percentage of MSU students
approve of their peers drinking 5 or fewer alcohol drinks when out partying?”) was
removed due to lack of internal validity. The final four item scale resulted in a
Cronbach’s alpha of .93. The mean percentage for perceptions of injunctive norms was
that 58.44 % of other students approve of drinking above moderate levels, with a standard
deviation of 22.10 % (See Table 1).

Post-message injunctive norms perceptions were assessed using five 5-point
Likert items developed by Fediuk (2004). The five-item scale was tested using
confirmatory factor analysis, which resulted in a Cronbach’s alpha of .72, with a mean of
3.27 and a standard deviation of .74 (See Table 1).

Message Believability. Message believability was assessed using seven 5-point
Likert items developed by Lindsey & Ah Yun (2003). One item was removed prior to the
study due to lack of face validity: “I believe that the message has integrity.” The 7-item
scale was tested using confirmatory factor analysis. After removing two items: “The
information presented in the message is credible” and “It is difficult to judge if the
description of student opinions is representative of the typical MSU student”), the five
item scale resulted in a Cronbach’s alpha of .86, with a mean of 3.21 and a standard

deviation of .80 (See Table l).

28

For the supplemental studies, five items were used to assess believability. Two
items from the primary study were removed: “I do not believe that the message is
credible” and “The message is not very reliable.” After statistical analysis, an additional
item was removed (“It is difficult to judge if the description of student opinions is
representative of the typical MSU student”). The four item scale across four belief
measures resulted in Cronbach alphas ranging from .89 to .99.

Induction Check. Induction check measures were created for the current study.
Five items were taken from Lindsey & Ah Yun (2004), which were used to measure
verifiability. Quantitative evidence is more verifiable than qualitative evidence. Three
more items were included that ask whether subjects felt that the message presented
statistical representations of data or whether the message contained testimonials by other
students. The 5-item scale was tested using confirmatory factor analysis, which resulted
in a Cronbach’s alpha of .83. The induction check for qualitative testimonials could not
be subjected to a confirmatory factor analysis, since there were only three items. The

three item scale for qualitative testimonials resulted in a Cronbach’s alpha of .79.

29

CHAPTER 5

RESULTS

Induction Checks

Two induction checks were conducted to determine if qualitative or quantitative
data were perceived in the messages. Messages containing quantitative evidence were
perceived to contain more quantitative evidence (M = 3.29, SD = .71) than messages that
did not contain quantitative information (M = 2.72, SD = .89), F (1,229) = 29.05, p <
.001 , r72 = .11 (See Table 2). Messages containing qualitative testimonials were perceived
as presenting more qualitative evidence (M = 3.26, SD = .92) than messages that did not
contain qualitative testimonials (M = 2.47, SD = .92), F (1 ,229) = 42.60, p < .001, If =
.16 (See Table 3). The means are based on a 5-point Likert scale.
Primary Study

Hypotheses one, two and three predicted differences in injunctive norms
perceptions across messages. The higher the injunctive norms perception, the more the
subject believes that others approve of risky drinking, while lower injunctive norms
perceptions reflect perceptions that their peers do not approve of risky drinking. The
hypotheses were tested using a two-way AN OVA omnibus F-test (Quantitative, No
Quantitative evidence X Qualitative, No Qualitative evidence). Similar to the results
obtained by Fediuk (2004), no significant differences on perceptions of injunctive norms
were observed. Perceptions of injunctive norms about drinking did not differ significantly
across all four conditions, F (1,229) = .85, p = .47, r72 = .00. Means are reported in Table

4. Prior perceptions were added to the model as a covariate. Prior perceptions were

30

significant predictors of post-message injunctive norms perceptions, F (1,229) = 50.22, p
< .001, n2 = .18.

Hypothesis one predicted that subjects exposed to messages targeting injunctive
norms perceptions containing evidence would have lower perceptions of injunctive norms
than subjects exposed to messages that do not contain supporting evidence. Hypothesis
one was tested using a one-way AN OVA. Contrast coefficients were employed, where a
contrast coefficient of -3 was assigned to the no evidence control condition, and +1 was
assigned to the three remaining conditions. No significant difference was obtained
between messages that contained supporting evidence and messages that did not provide
supporting evidence, t (226) = -.3 8, p = .70, r = .03. Hypothesis two predicted that
subjects exposed to messages containing quantitative evidence lower perceptions of
injunctive norms than subjects exposed to messages containing no evidence or qualitative
evidence. Hypothesis two was tested using a one-way ANOVA. Contrast coefficients
were employed, where a contrast coefficient of +3 was assigned to the quantitative
evidence condition, and -1 to the control, qualitative evidence condition, and the
condition containing both types of evidence. No significant differences were found
between messages containing quantitative evidence compared to messages with no
quantitative evidence, t (226) = .76, p = .45, r = .05. Messages containing both types of
evidence may confound the results of the analysis of hypothesis two, in that qualitative
evidence is contained in the condition containing both types of evidence. Therefore, a
second analysis was conducted. Contrast coefficients were employed, where a contrast
coefficient of +2 was assigned to the quantitative evidence condition, -1 to the control

and qualitative evidence condition, and a 0 to the condition containing both types of

31

evidence, since qualitative evidence is mixed with the quantitative evidence. No
significant difference was obtained between messages that contained quantitative
evidence and messages that did not contain quantitative evidence, t (226) = .24, p = .81, r
= .02. Hypothesis three predicted that subjects exposed to messages containing both
qualitative and quantitative evidence would have lower perceptions of injunctive norms
than subjects exposed to messages that do not contain both types of supporting evidence.
Hypothesis three was tested using a one-way ANOVA. Contrast coefficients were
employed, where a contrast coefficient of +3 was assigned to the condition containing
both types of evidence, and -l was assigned to the remaining conditions. No significant
difference was obtained between messages that contained both quantitative and
qualitative evidence and messages that did not contain both types of evidence, t (226) = -
1.57, p = .12, r = -.10. The results, while not significant, are contrary to expectations, in
that the trend is that subjects exposed to messages containing both types of evidence held
higher injunctive norms perceptions than subjects exposed to messages that do not
contain both types of evidence.

Hypotheses 1a, 2a, and 3a predicted that evidence type would have an impact on
the believability of a message. The hypotheses were tested using a two-way ANOVA
omnibus F -test (See Table 5). The omnibus F-test resulted in a significant result, F
(1,229) = 3.19, p < .03, rf = .04. Means are reported in Table 6.

Hypothesis 1a predicted that messages targeting injunctive norms perceptions
containing evidence would be more believable than messages that do not contain
supporting evidence. Hypothesis 1a was tested using a one-way AN OVA. Contrast

coefficients were employed, where a contrast coefficient of -3 was assigned to the no

32

evidence control condition, and +1 was assigned to the three remaining conditions. No
significant difference was obtained between messages that contained supporting evidence
and messages that did not provide supporting evidence t (226) = 1.70, p = .09, r = .11.
However, the results are trending toward the predicted direction. Such differences can be
substantial for campaign developers. Hypothesis two predicted that messages containing
quantitative evidence would be more believable than messages containing no evidence or
qualitative evidence. Hypothesis 2a was tested using a one-way AN OVA. Contrast
coefficients were employed, where a contrast coefficient of +3 was assigned to the
quantitative evidence condition, and -1 to the control, qualitative evidence condition, and
the condition containing both types of evidence. A significant differences were obtained,
messages containing quantitative evidence compared to messages with no quantitative
evidence, t (226) = 2.95, p < .01, r = .19. Messages containing quantitative evidence
were found to be more believable than messages that did not contain quantitative
evidence. Messages containing both types of evidence may confound the results of the
analysis of hypothesis two, in that qualitative evidence is contained in the condition
containing both types of evidence. Therefore, a second analysis was conducted. Contrast
coefficients were employed, where a contrast coefficient of +2 was assigned to the
quantitative evidence condition, -1 to the control and qualitative evidence condition, and
a 0 to the condition containing both types of evidence. No significant differences were
found between messages containing quantitative evidence compared to messages with no
quantitative evidence, t (226) = 1.87, p = .06, r = .12. However, the data was trending in
the predicted direction, in that messages containing quantitative evidence were seen to be

more believable than messages that did not contain quantitative evidence. Hypothesis

33

three predicted that messages containing both qualitative and quantitative evidence would
be more believable than messages that do not contain both types of supporting evidence.
Hypothesis 3a was tested using a one-way ANOVA. Contrast coefficients were
employed, where a contrast coefficient of +3 was assigned to the condition containing
both types of evidence, and -l was assigned to the remaining conditions. Messages that
contain both quantitative and qualitative evidence was found to be more believable than
messages that did not contain both types of evidence, t (226) = 2.47, p < .02, r = .16.

Message believability and injunctive norms perceptions were correlated, r = -.17,
p < .01. It remains unclear whether subjects who perceived the message to be believable
were more likely to alter their perceptions, or those with lower injunctive norms
perceptions were more likely to believe the social norms message.
Supplemental Study A: Sample Size

Hypothesis four predicted that messages indicating larger sample sizes would be
perceived as more believable than messages with a smaller sample size (sample size
heuristic). A t-test was conducted to determine any differences. Contrary to expectations,
messages that indicated a moderate sample size (M = 3.20, SD = .97) were perceived to
be more credible than messages with a larger sample size (M = 2.70, SD = .78), t (65) =
-2.29, p < .03, r = -.27. Hypothesis four received no support.
Supplemental Study B: Specificity

Hypothesis five predicted that messages using quantitative statistics would be
seen as more believable than messages using verbal labels. A t-test was conducted to
determine any differences. The quantitative label message was similar to the message

containing a moderate sample size, so the comparison group was the moderate sample

34

size message. No significant difference was found for using a quantitative statistic
message (M = 3.20, SD = .97) or a quantitative label message (M = 2.95, SD = .81), t (69)
= 1.18, p = .24, r = .14. Hypothesis five received no support.

Supplemental Study C: Testimonial Attribution

Research question one asked whether visually identifiable testimonials increased
the believability of a social norms message targeting injunctive norms perceptions more
so than unattributed group testimonials. Three messages were developed varying in
degree of testimonial attribution, where the message could be attributed to either a whole
group, to an individual, or a named individual. A One-Way ANOVA was conducted to
determine any differences. No significant differences were found, F (2, 100) = 1.73, p =
.12. Testimonials attributed to a group (M = 2.70, SD = .78), specific individuals (M =
3.11, SD = .96), or named individuals (M = 3.07, SD = .94) did not alter perceptions of
message believability.

An analysis was conducted combining the two messages attributed to individual
members compared to the message with a group testimonial attribution. Contrast
coefficients were employed, where a contrast coefficient of -2 was assigned to the group
testimonial message, and +1 was assigned to the two individual attribution group
messages. The message with individually attributable testimonials was more believable
than messages without attributable testimonials, t (98) = 2.05, p < . 05, r = .20.
Supplemental Study D: Approval or Disapproval

Research question two tested whether a message indicating approval or
disapproval of an action would be seen as more believable. A t-test was conducted to

determine any differences. A disapproval message (M = 3.04, SD = .91) did not differ

35

significantly from an approval message for message believability (M = 3.09, SD = .91), t
(99) = .29,p = .77, r = .03.
Supplemental Study E: Descriptive Norms and Evidence Types

While the main study examined the impact of evidence type on injunctive norms
perceptions, hypothesis six examined whether quantitative or qualitative messages would
be seen as more believable in a descriptive norms message targeting partying frequency.
A t-test was conducted to determine if any differences were obtained for message
believability. Messages featuring quantitative evidence (M = 2.88, SD = .99) did not
significantly differ in message believability from messages presenting qualitative
testimonials (M = 2.95, SD = 1.10), t (68) = -.306, p = .746, r = .04. When targeting
descriptive norms of partying frequency, the type of evidence did not impact message
believability. Hypothesis six received no support.
Supplemental Study F: T wo-Sided Refutational Message

Hypothesis seven predicted that a two-sided refutational message would be more
believable than a one-sided message. Since the two-sided message includes quantitative
evidence targeting descriptive norms, the two-sided refutational message was compared
to the quantitative descriptive norms message. A t-test was conducted to determine any
differences. The two-sided refirtational message (M = 3.45, SD = .99) was found to be
more believable than the message that does not contain a two-sided refutational message

(M = 2.88, SD = .99), t (62) = 2.36, p < .03, r = .29. Hypothesis six was supported.

36

CHAPTER 6

DISCUSSION

The current series of studies examined how the presentation of different types of
evidence (quantitative or qualitative) affect message believability, perceptions of student
behaviors and perceptions of student attitudes. The primary study examined the effects of
quantitative and qualitative evidence, and supplemental studies focused on message
components such as sample size, testimonial attributions, two-sided refutational
messages, and approval/disapproval terminology.

The primary study tested evidence type on injunctive norm perceptions. Message
believability is crucial to altering student perceptions of injunctive norms. Qualitative
testimonials were not found to differ in believability compared to the control message
containing only the claim with no supporting evidence. Quantitative messages, while not
statistically different, did tend to be more believable than messages containing qualitative
testimonials. Combining both qualitative testimonials and summative statistics was found
to be more believable than the control message or messages containing only testimonials.
These results suggest that quantitative evidence is more believable than no evidence or
qualitative evidence only when presenting social norms claims. Such results are
significant in that persuasive messages must be seen to be believable.

It was expected that subjects exposed to messages that included evidence to
support a claim would have lower perceptions of injunctive norms than subjects exposed
to messages that did not contain evidence. Subjects exposed to messages that contained
quantitative evidence were predicted to have lower injunctive norms perceptions than

subjects exposed to qualitative evidence, and subjects exposed to messages containing

37

both quantitative and qualitative evidence were expected to have the lowest injunctive
norms perceptions. The results did not support the predictions. No significant differences
were found across the four message types in regard to perceptions of injunctive norms.
The majority of subjects did perceive that most students engage in and approve of higher
levels of drinking. On average, students perceived that 55% of students approve of
excessive drinking. The reported figure in the message was that 67% report holding the
opinion that others should drink five or fewer. Thus, the social norms message presented
was counter to pre-existing beliefs held by the subjects.

One explanation for the lack of differences in injunctive norms perceptions due to
message exposure is that people are not impacted by a single message exposure,
especially if the message is counter to existing beliefs. One exposure to a message that
contradicts pre-existing perceptions may have little utility. Often, multiple exposures are
necessary.

Another possible explanation for the lack of results is that subjects do not feel that
the majority of students represent their immediate peer environment. While the majority
may engage in and hold moderate attitudes toward alcohol, subjects may instead interpret
the message through their experiences with their immediate peer groups. Prior research
on evidence suggests that people tend to process messages in a way that disconfirms
presented evidence (Darley & Gross, 1983; Edwards & Smith, 1996; Kahneman &
Tversky, 1973; Lord, Ross, & Lepper, 1979; Munro & Ditto, 1997 ; Nisbett, Crandall, &
Reed, 1976). When presented with evidence that is not consistent with one’s beliefs,
processing by subjects may not necessarily have been an objective examination of the

message evidence. Rather, the subject may have generated counterarguments to

38

disconfirrn the presented evidence. Students may thus feel that the information provided
is inaccurate.

Fediuk (2004) found that many subjects questioned the validity of the claims and
evidence presented in a social norms message. Open-ended responses for the current
study also indicated doubts from a number of subjects. Students may question the
spokespersons in the message, cast doubt to the survey methodology, insist that there is
response bias, or may indicate that the sample does not represent their immediate peer
group. Research needs to continue focus on what students feel is believable and credible
information.

In open-ended responses, some students reported that their responses to the
questionnaire were influenced by their immediate peer group rather than the message
reference group. The implication is that subjects may perceive a difference between
majority beliefs and immediate peer beliefs. Perhaps the majority do hold moderate
attitudes, but their immediate peer group is their primary referent group. Additional
questions in the questionnaire addressing how one’s peer group differs from the majority
of campus students would be beneficial. For example, questions may ask for approval
rates for the entire student body, as well as for one’s immediate peers. Deviations can
then be observed.

Results of the primary study indicate that subjects who felt the messages were
believable also had lower perceptions of injunctive norms. However, it may be the case
that subjects who already have low injunctive norm perceptions may find a message that
is similar to their own perceptions more believable. Subjects who hold high injunctive

norms perceptions may discredit information that counters their own current perceptions.

39

But the argument that subjects who perceive a message to be believable are then
persuaded to lower their perceptions of injunctive norms makes intuitive sense. Future
research should continue focus on message believability and perception change, and how
to best create believable social norm messages.

Supplemental Studies

A series of supplemental studies were conducted examining message variables.
The supplemental studies offered a brief glimpse into message believability. Message
variables were tested using messages targeting both injunctive norms and descriptive
norms.

Contrary to expectations, a moderate sample size was found to be significantly
more believable than a larger sample size and time period. Based on the sample size
heuristic (Baesler & Burgoon, 1994), a larger sample size was predicted to be more
believable than a smaller sample size. It is possible that subjects did not focus on the
sample size message, as no induction check was included. The sample size was placed at
the bottom of the message, and was not highlighted. However, such lack of attention
would not explain the significant contrary difference obtained.

One possible explanation is that the ongoing time period may encompass a time
period that the subject may feel was before their time at the university. The time fi'ame
presented was over a five-year period (1998-2002). In these years, the university has
experienced a number of riots and stories of alcohol abuse in the media. Many
participants have been at the university for four years or less. The thought process may be
that the data may have been accurate prior to their own university attendance, but things

are different now. Perhaps the participant feels that the drinking level is trending toward

40

greater tolerance toward alcohol consumption, and these changes are not captured in a
larger time period. The one-year moderate sample size was conducted in a time period
not far removed from the present year (2002). One possible solution is to have a larger
sample size from a one-year collection period.

Messages that were more specific were not found to be more believable than
messages that were more abstract. High exact percentages were not seen to be more
believable than indicating that verbal labels indicating that most ought to drink
moderately. However, the mean differences were trending in the predicted direction that
more specific messages are seen to be more believable. The lack of power in the
supplemental studies reduces the likelihood of obtaining significant results.

If there genuinely is no benefit to using a quantitative statistic over quantitative
labels, practitioners may use strategic ambiguity to their advantage. Rather than reporting
a specific statistic that requires detailed baseline studies, practitioners may simply state
that most students engage in or approve in healthy behaviors. The word ‘most’ could
encompass any number from 51% to 100%. Also, the lack of specificity removes the
problem of whether students perceive that 57%, 67%, 77%, or 87% is a meaningful
number that would induce behavior or attitude change.

Also unclear is if a believable quantitative statistic is indeed advantageous to
social norms claims. A percentage closer to the student perception would be more
believable than one that is distant from the student perception. For example, a message
indicating that 55% of students drink moderately, which is closer to the perceptions held
by the student, may be more believable than stating that 80% of students drink

moderately. However, the lower percentage, while more believable, may not lead the

41

recipient to feel that the moderate behavior is normative. Further research is warranted to
examine if the pursuit of believable quantitative statistics does not also reduce the
potential for persuasion.

Supplemental studies were designed to test the utility of using testimonial
messages. Testimonial messages that were visually identifiable to an individual were
more believable than testimonial messages attributed to a larger group, with no visual
attribution to an individual. Testimonials may however have little utility, as each
testimonial needs to be thoroughly tested, and meaning may differ between different peer
groups. What may be motivating for one peer group may be humorous to another peer
group. Adding testimonials does not seem to have a strong impact, and the effort in
constructing quality testimonials may be of limited pragmatic usefulness. Further
research to examine if there is any utility in adding student testimonials is warranted.

Social norms proponents advocate the usage of positive message flames rather
than negative message flames toward health or social benefits. The argument would be
that using approval messages will be more beneficial than using disapproval messages.
No significant differences were obtained in flaming a message positively or negatively.
Messages flaming the claim in a way that expresses that other students approve of a
healthy behavior did not differ flom a message stating that other students disapprove of
an unhealthy behavior. One explanation for the lack of findings could be due to a weak
manipulation of the overall message. Another possible explanation is that the
supplemental study had a small sample size, which limits the power to detect significant
differences. Further examination is needed to examine whether a positive incentive such

as being part of the crowd is greater than the incentive of not being a deviant.

42

The benefits of qualitative and quantitative evidence were examined using a
descriptive norms message. The message targeted partying flequency rather than drinking
quantity. No significant differences in believability were found between the two evidence
types. These findings are limited, in that the message was targeting flequency rather than
quantity. It is possible that flequency norms are irrelevant, while quantity norms may be
more impactful. However, F ediuk (2004) found that descriptive norms have limited
impact on behavior intentions.

The message may have also suffered in that the quantitative evidence reported
that 81% of students report drinking twice a week or less. Based on the current study,
only 70% report drinking twice a week or less. The evidence provided in the message
may be overstated. The overestimation may have cast doubts on the believability of the
message.

A two-sided refutational message was found to be more believable than one-sided
messages. Taking into consideration and explicitly stating existing perceptions and
counter-arguments may be more effective than just presenting factual information. Few
social norms messages include counterarguments or acknowledgement of current student
perceptions, even though evidence points to such misperceptions. Explicitly stating and
contesting the misperception appears to be more believable than just stating facts that are
counter to existing beliefs.

The benefits of using two-sided refutational messages for social norms campaigns
need to be further examined. Social norm messages can present different sets of
counterarguments with different sets of refiitations. A series of point-counterpoint ads can

be created. Such features can be presented in multiple messages and appear to be more

43

beneficial than just presenting the same factual information with different visual designs.
Some subjects indicated that 1,000 subjects was not enough to gauge student attitudes on
campus. Two-sided refutational messages may also target the lack of understanding of
statistics and sampling error.

The challenge continues to be to create believable messages to alter injunctive
norms perceptions. Practical challenges are to create believable messages to alter student
perceptions. Based upon the current study, practitioners are encouraged to provide
plausible quantitative evidence to support the main claim. The time period of data
collection should be close to the time period of the statistical evidence presented. Finally,
explicit presentation of counterarguments, followed by factual refutation should produce
greater perceptions of message believability.

Limitations

One main limitation of the current study concerns the supplemental studies. The
supplemental studies were designed to be exploratory in order to assess potential avenues
of investigation. Sample size for these studies was relatively low, which reduces power to
detect any significant differences. Each of the ideas presented in the supplemental studies
warrant greater attention. The current study only focused on message believability.
Future studies could take each supplemental study and further explicate the ideas and
provide more stringent tests while also examining each factor’s effect on message
persuasiveness.

Another limitation is in the graphical presentation of the social norms message.
While the images presented were taken flom existing university campaign material,

further formative development would add to the understanding of the entire message.

44

Many graphical variables and evaluations may have acted as confounds to understanding
the impact of the different types of evidence presented. Perhaps the subjects focused
more on the graphical layout and presentation rather than the text which provided the
message. While the graphical presentation remained fairly consistent across the
conditions, differences in spacing and openness were necessary to differentiate the
statistical evidence and the testimonials. The quantitative messages had much more white
space than did the testimonials, which tended to clutter the overall image.

Message inductions for the primary study were not very strong. While both
induction checks were significant, the overall means suggest that respondents still
perceived quantitative evidence in the qualitative evidence condition and vice versa. The
mean difference obtained was not as strong as one would expect.

Another limitation is that the current study did not include any measure to assess
argument strength. Unknown is whether subjects perceived the argument strength of the
testimonial message the same way subjects perceived the quantitative evidence. Future
studies should include a measure to examine perceptions of argument strength. Without
such measures, comparisons of effect size between evidence type is suspect.

Future Directions

The social norms messages provide evidence that most students drink moderately
or not at all. Future research should examine whether any unintended consequences arise
flom the social norms approach. The results of the current study suggest that little actual
perception change is occurring due to social norms messages. However, researchers and
practitioners have not yet begun focused study on the question if any negative adverse

effects are created by the social norms campaigns.

45

Further efforts should also examine whether implicit or explicit statements of
what actions are desired through the campaign are needed. Social norms messages tend to
implicitly indicate that since most students drink moderately or not at all, so should the
recipient of the message. The effects of explicitly telling students that they should not
drink excessively should be tested. Explicit statements may be even more persuasive if

accompanied by a two-sided refutational message.

46

Table 1

Means and Standard Deviations for Primary Study

 

Variable

SD Range

 

Inj unctive Norms Perceptions

Pre-exposure percentage

58.44% 22.10% 3% - 95%

 

 

 

Post-message perceptions 3.27 .74 1 - 5
Message Believability 3.21 .80 l - 5
Table 2
Analysis of Variance for Message Verifiability
Source Type 111 ss df MS F P 11‘
Quantitative Evidence 18.82 1 18.82 29.05 <.001 .11
Qualitative Evidence .00 1 .00 .08 >05 .00
Quant x Qual .83 l .83 1.28 > .05 .00
Error 146.40 226 .65
Total 166.05 229

 

47

Table 3

Analysis of Variance for Testimonials

 

 

 

Source Type 111 ss df MS F p 11‘
Quantitative Evidence .59 1 .59 .69 >05 .00
Qualitative Evidence 36.18 1 36.18 42.60 <.001 .16
Quant x Qual .76 1 .76 .89 > .05 .00
Error 191.90 226 .85

Total 228.84 229

Table 4

Means and Standard Deviations for Drinking Injunctive Norm Perceptions

 

Qualitative Evidence

 

 

No Yes
Quantitative Evidence M SD M SD
No 3.30 .70 3.30 .74
Yes 3.33 .54 3.15 .72

 

*p<.05

Note: Means are based on a 5-point Likert scale

48

Table 5

Analysis of Variance for Message Believability

 

 

 

 

 

 

 

Source Type III SS df MS F p 11
Quantitative Evidence 5.47 1 5.47 8.70 <.001 .04
Qualitative Evidence .27 l .27 .44 >05 .00
Quant x Qual .27 1 .27 .44 > .05 .00
Error 142.17 226 .63
Total 148.18 229
Table 6
Means and Standard for Perceived Message Believability
Qualitative Evidence
No Yes

Quantitative Evidence M SD M SD

No 3.063 .81 3.06a .77

Yes 3.30 .73 3.43b .86

 

*p<.05

Note: Means are based on a 5-point Likert scale

Differences between superscript a and b are significant at p < .05

49

APPENDICES

50

Appendix A

Messages

Main Study - No Evidence

Most MSU S’Ei—en—ts
Approve of Drinking Moderately

 

 

Based on a survey, MSU students express
the opinion that others ought to follow the
Five or Fewer guideline when out partying

 

51

Main Study - Quantitative Evidence

— Most MSU Students
Approve of Drinking Moderately

‘ .91":
:r: ,e

 

  

67% of MSU students* report holding the
opinion that others ought to drink Five or
Fewer when out partying

 
   
  

 
 

* National College Health Assessment
MSU 2002. N = 1,005, margin of error 1 3%
For more information: vwwvheaithedmsuedulhialhhn

    

52

Main Study — Qualitative Evidence

Most MSU Students

Approve of Drinking Moderately

 

‘I wish people would drink
more mdemte‘ly and watch
how marry they drink'

  
   
   
   

   

,"TIenjoy, :.

'It's great when. people can -
just have fun being around
drinking without ; ' moderate"?
getting drunk" I

1. drinkers" '

 

 

_ ‘People are fun ;;
when theyiren’t 1%7
completely blorlvn 1

0041"

 

MSU students express the opinion that
others ought to follow the Five or Fewer
guideline when out partying

For more information: wwhealthedmsuedu/hialhtm

 

 

 

53

Main Study — Both Quantitative and Qualitative

Most MSU Students

Approve of Drinking Moderately

 

"I wish people would drink
more moderately and watch
how many they drink"

   

“It‘s great when people can
just have fun

drinking without
getting drunk’

 

 

   

‘People arefu'n " .
when they aren't . ,
completely blown

om. ,

 

67% of MSU students* express the opinion
that others ought to follow the Five or Fewer
guideline when out partying

' National College Health Assessment
MSU 2002. N = 1,005, margin of error _+_ 3%
For more information: www.healthed.msu.edu/hial.htm

 

 

 

54

Supplemental Study A — One- Year Moderate Sample Size

i MostMSU Students+
Approve of Drinking Moderately

‘

-.- . ,j

.......

......
D

 
     
 
  

  

67% of MSU students* report holding the
opinion that others ought to drink Five or
Fewer when out partying

 
 
    

" National College Health Assessment
MSU 2002. N = 1,005, margin of error 1 3%
For more intonnation: wwhealthedmsuedulhialhtm

55

Supplemental Study A - F ive- Year Large Sample Size

MostMSU Students
Approve of Drinking Moderately

67% of MSU students* report holding the
opinion that others ought to drink five or fewer
when out partying

' National College Health Assessment
MSU 1998-2002 N = 5,005. margin of error 1 3%
For more information: vwwvhealthedmsuedulhialhtm

 

56

Supplemental Study B — Quantitative Statistics

3 if Most MSUStudents
Approve of Drinking Moderately

 

67% of MSU students* report holding the
opinion that others ought to drink Five or
Fewer when out partying

" National College Health Assessment
MSU 2002. N = 1,005, margin of error 1 3%
For more information: mhealthedmsuedulhialmm

57

 

Supplemental Study B — Quantitative Label

1

Most MSU Students
Approve of Drinking Moderately

 

 

Most MSU students* report holding the
opinion that others ought to drink moderately
when out partying

' National College Health Assessment
MSU 2002. N = 1,005, margin of error i 3%
For more information: vwwv.healthed.msu.edu/hial.htm

 

 

 

58

Supplemental Study C —Testimonia1 Aggregate Group Attribution

Most MSU Students
Approve of Drinking Moderately

"It's great when people can
just have fun A
. drinking‘iwithout
‘ ' getting drunk"

 

 

 

MSU students express the opinion that
others ought to follow the Five or Fewer
guideline when out partying

For more information: www.healthedmsuedu/hialhtm

 

 

 

59

Supplemental Study C — Testimonial Individual Attribution

Most MSU Students
Approve of Drinking Moderately
“I wish people would drink

more moderately and watch
how many they drink"

 

 

   

‘I thinkme‘n‘are . '

    

€““‘"1 enjoy

 

.V ' moreatitrocirive-g
Vbéiné aricund' .Whenthw's‘cirt.
v'r‘moder‘at'e such sloppy drunks";

I ‘ ,, drinkers'

MSU students express the
opinion that others ought to follow the five or
fewer guideline when out partying

 

For more information: www.healthed.msu.edulhial.htm

60

  
    

 

Supplemental Study C — Testimonial Named Individual Attribution

Most MSU Students

Approve of Drinking Moderately

 

"I wish people would drink
more moderately and watch
how many they drink“

 

 

    

- ‘1. €an ,
being around
moderate
drinkers"

   

Kevin Lachlan

MSU students express the
opinion that others ought to follow the five or
fewer guideline when out partying

 

 

For more information: Whealthedmsuedulhiaihtm

 

61

Supplemental Study D — Approval Message

Mme Students “‘ “‘
Approve of Drinking Moderately

 

 

67% of MSU students* approve of others
drinking five or fewer when out partying

' National College Health Assessment
MSU 2002, N = 1,005, margin of error 1 3%
For more information: www.healthed.msu.edu/hial,htm

 

 

 

62

Supplemental Study D — Disapproval Message

Most MSU Students
Disapprove of Drinking Heavily

 

 

67% of MSU students* disapprove of
others drinking more than five drinks when
out partying

' National College Health Assessment
MSU 2002. N = 1,005, margin of error _t 3%
For more information: vmrwhealthedmsuedu/hial.htm

 

 

 

63

Supplemental Study E — Descriptive Norms with Quantitative Evidence

Most moraines“
Drink Infrequently

 

 

81 % of MSU students* report drinking
twice a week or less

' National College Health Assessment
MSU 2002. N = 1,005, margin of error: 3%
For more information: www.healthed.msu.edu/hial.htm

 

 

 

64

Supplemental Study E — Descriptive Norms with Qualitative Testimonials

Most MSU Students

Drink Infrequently

   
   

  
    
 

‘I'm lucky to get out. more
than oncea week"

, ‘ftwhosdrififford going 6‘th
; drinkingmre than a few '
limes/aha“?

 

MSU students report drinking
twice a week or less

For more information: www.healthed.msu.edu/hial.htm

 

 

 

65

Supplemental Study F: Refutational Message

Most MSU Students
Drink Infrequently

 

 

Many MSU students overestimate how often other
students go out drinking perhaps it’s because some of
their friends seem to drink frequently.

The fact is that 81% of MSU students* go out drinking
twice a week or less, according to data in a campus-wide
alcohol survey.

' National College Health Assessment
MSU 2002. N = 1,005, margin of error 1 3%
For more information: www.healthedmsuedu/hialhtm

 

 

 

66

Appendix B
Instrumentation
Instructions

Below are questions about your perceptions and attitudes. Please read the
following questions and answer in a way that best estimates the approximate percentage.

Pre-message Injunctive Norms Perceptions

What percentage of MSU students approve of their
peers drinking excessively when they go out partying?

 

What percentage of MSU students approve of
their peers getting drunk when they go out partying?

What percentage of MSU students approve of their peers
drinking five or more alcoholic beverages when they go out partying?

 

What percentage of MSU students approve of their
peers flequently getting drunk when they go out partying?

What percentage of MSU students approve of their peers drinking 5 or fewer
alcohol drinks when out partying? (R)

Continue to next page

67

The following questions refer to the message you just viewed. Choose a number
flom the scale below that indicates your degree of agreement with each of the following
statements.

1 = Strongly Disagree

2 = Disagree
3 = Neutral
4 = Agree

5 = Strongly Agree
Descriptive Norms Perceptions
Most students on campus drink excessively when they go out partying.

It is a myth that most students get drunk flequently when they go out partying. (R)

 

On average, the majority of students drink at unsafe levels when they go out partying.

 

The campus norm is to drink to get drunk when going out partying.

Overall, only a minority of students drink at unsafe levels when they go out partying. (R)

 

Most students drink at unhealthy levels most of the time when they go out partying.

 

 

Injunctive Norms Perceptions

Most students are favorable toward heavy
drinking by other students when they go out partying.

Most students are favorable toward excessive
drinking by other students when they go out partying.

Most students are favorable toward other students drinking ahnost every day.

Most students disapprove of high levels of alcohol
consumption by other students when they go out partying. (R)

Most students approve of drinking flequently on weekday evenings.

Most students approve of drinking more than 5 drinks when going out partying.

68

Most peers feel that getting drunk is not appropriate behavior when going out partying.
(R)

 

Most students approve of drinking more than twice a week.

 

69

Normative Expectations
Others on campus expect me to drink to extreme levels when I go out partying.

I do not think that others want me to drink to unhealthy levels when I go out partying. (R)

It is expected that college students drink heavily when they go out partying.

I believe that my peers want me to drink as much as they do when we go out partying.

When out partying, I believe that others want me to drink heavily.
Message Believability

The information presented in the message is credible.

 

The information presented in the message is reliable.

I believe that the message is a believable one.

 

It is difficult to judge if the description of student opinions is representative of the typical
MSU student.

The information presented in the message is trustworthy.

I do not believe that the message is credible. (R)

 

The message is not very reliable. (R)

 

Induction Check

The evidence presented is backed by statistical information.

 

The message presents concrete examples of student opinions.

I could confirm the information in the message if I wanted to.

 

The evidence is supported by student testimonials.
The information in the message is verifiable.

The information presented could be validated.

7O

I could check the information presented in the message if I needed to.

The message contains quotes that enhance the message.

71

Media Factors

How effective is this message in influencing the students to consume less alcohol?

VERY FAIRLY SLIGHTLY NOT
EFFECTIVE EFFECTIVE EFFECTIVE EFFECTIVE

What is your personal reaction to the message? Give ratings using a scale flom 0 to 10
(with 10 being the highest score) on each of these factors:

__ INFORMATIVE: tells you something new, increases your knowledge
SENSIBLE: presents wise advice that seems reasonable

_ MEMORABLE: vivid image, fascinating fact, catchy slogan

_ ENJOYABLE: interesting, entertaining, stimulating message

__ USEFUL: valuable information and helpful advice worth remembering

_ IMAGINATIVE: style is refleshing, novel, unique, clever

__ BELIEVABLE: accurate information, sincere/trustworthy characters

__ CONVINCING: presents ideas you agree with

__ PROFESSIONAL: production quality is well done

__ MOTIVATING: presents influential reasons to change behavior

__ ON-TARGET: content is personally meaningful, people/situations you can

identify with

Indicate if the message has any of these negative features:

Yes No PREACHY: Was tone of message too moralistic or righteous?

Yes No DISTURBING: Turned off because it’s too emotional or threatening?
Yes No CONFUSING: Is anything unclear, vague, or hard to understand?
Yes No IRRITATING: Did you find anything offensive or annoying?

Yes No DULL: Was the style boring, stale, or trite?

72

Yes No MISLEADING: Any biased arguments or exaggerated claims?

Overall rating of message: EXCELLENT GOOD FAIR POOR
Please tell us a bit about yourself

Sex: Female Male

Class: Freshman Sophomore Junior Senior Grad

I am a member of: (circle all that apply)

Fraternity / Sorority Athletic team Honor’s College Religious Institution
Do you consume alcoholic beverages? Yes No

On an average week, how often do you go out and party? Days per week

On average, how many alcoholic drinks do you consume when you go out partying?

 

When you examined the message, what specific types of students did you think about
when reading words ‘Most MSU Students’?

 

 

Thank you for your participation. Please turn in this survey and consent form to the
investigator. Make sure that the consent form is separate flom the survey to maximize

privacy.

73

BIBLIOGRAPHY

Ah Yun, K., & Massi, LL. (2000). The diflerential impact of race on the effectiveness of
narrative versus statistical appeals to persuade individuals to sign an organ
donor card. Paper presented at the meeting of the Western States Communication
Association, Sacramento, CA.

Allen, M. (1991) Comparing the persuasiveness of one-sided and two-sided messages
using meta-analysis. Western Journal of Speech Communication, 55, 390-404.

Allen, M. (1998). Comparing the persuasive effectiveness one- and two-sided messages.
In M. Allen & R.W. Preiss (Eds), Persuasion: Advances through meta-analysis
(pp. 87-98). Cresskill, NJ: Hampton Press, Inc.

Allen, M. & Burrell, N. (1982). Evaluating the believability of sequential arguments.
Argumentation and Advocacy, 28, 135-144.

Allen, M., Bruflat, R., Fucilla, R., Kramer, M., McKellips, 8., Ryan, D.J., Spiegelhoff,
M.(2000). Testing the persuasiveness of evidence: Combining narrative and
statistical forms. Communication Research Reports, 1 7, 331-336.

Allen, M., & Preiss, R.W. (1997). Comparing the persuasiveness of narrative and
statistical evidence using meta-analysis. Communication Research Reports, 14,
125- 1 3 1.

Baesler, E.J., & Burgoon, J .K. (1994). The temporal effects of story and statistical
evidence on belief change. Communication Research, 21, 582-602.

Borgida, E., & Nisbett, RE. (1977). The differential impact of abstract vs. concrete
information on decisions. Journal of Applied Social Psychology, 7, 258-271.

Borsari, B., & Carey, K. B. (2001). Peer influences on college drinking: A review of the
research. Journal of Substance Abuse, 13, 391-424.

Brehm, J. W. (1966). A theory of psychological reactance: New York, N.Y., Academic
Press.

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

Cialdini, R. B., Reno, R. R., & Kallgren, C. A. (1990). A focus theory of normative
conduct: Recycling the concept of norms to reduce littering in public places.
Journal of Personality & Social Psychology, 58, 1015-1026.

Church, R.T., & Wilbanks, C. (1986). Values and policies in controversy: An
introduction to argumentation and debate. Scottsdale, AZ: Gorsuch Scarisbrick.

74

Darley, J .M., & Gross, RH. (1983). A hypothesis-confirrning bias in labeling effects.
Journal of Personality and Social Psychology, 44, 20-33.

Dickson, PR. (1982). The impact of enriching case and statistical information on
consumer judgments. Journal of Consumer Research, 8, 398-406.

Edwards, K., & Smith, BE. (1996). A disconfirrnation bias in the evaluation of
arguments. Journal of Personality and Social Psychology, 71, 5-24.

Fediuk, TA. (2004). Social norms: Eflectiveness of descriptive and injunctive norm
messages on students differing in attention to social comparison information.
Unpublished manuscript, Michigan State University, East Lansing, Michigan.

Haines, M. P. (1996). A social norms approach to preventing binge drinking at colleges
and universities. Newton, MA: The Higher Education Center for Alcohol and
Other Drug Prevention.

Haines, M.P., & Barker, GP. (2003). The Northern Illinois University experiment: A
longitudinal case study of the social norms approach. In H.Wesley Perkins (Ed.),

The social norms approach to preventing school and college age substance abuse
(pp. 21-34). San Francisco, CA: Jossey-Bass.

Haines, M., & Spear, S. F. (1996). Changing the perception of the norm: A strategy to
decrease binge drinking among college students. Journal of American College
Health, 45, 134-140.

Hovland, C.I., Janis, I.L., Kelley, H.H. (1953). Communication and persuasion:
Psychological studies of opinion change. New Haven, CT: Yale University Press.

J ohannessen, K. J., Collins, C., Mills-Nova, B., & Glider, P. (1999). A practical guide to
alcohol abuse prevention: A campus case study in implementing social norms and
environmental management approaches. Newton, MA: The Higher Education
Center for Alcohol and Other Drug Prevention.

Kahneman, D., & Tversky, A. (1973). On the psychology of prediction. Psychological
Review, 80, 237-251.

Kazoleas, D. C. (1993). A comparison of the persuasive effectiveness of qualitative
versus quantitative evidence: A test of explanatory hypotheses. Communication
Quarterly, 41, 40-50.

Kopfinan, J.E., Smith, S.W., Ah Yun, J.K., & Hodges, A. (1998). Affective and cognitive
reactions to narrative versus statistical evidence organ donation messages. Journal
of Applied Communication Research, 26, 279-300.

Lindsey, L.L.M., & Ah Yun, K. (2003). Examining the persuasive effect of statistical
messages: A test of mediating relationships. Communication Studies, 54, 306-321.

75

Lord, C.G., Ross, L., & Lepper, MR. (1979). Biased assimilation and attitude
polarization: The effects of prior theories on subsequently considered evidence.
Journal of Personality and Social Psychology, 3 7, 2098-2109.

Lucas, S. (1989). The art of public speaking (3rd ed.). New York: Random House.

McCrosky, J .C. (1969). Summary of experimental research on the effects of evidence in
persuasive communication. Quarterly Journal of Speech, 55 , 169-176.

McGuire, W.J. (2001). Input and output variables currently promising for constructing
persuasive communications. In R.E. Rice & C.K. Atkin (Eds)., Public
communication campaigns (3rd Edition) (pp. 22-48). Thousand Oaks, CA: Sage
Publications, Inc.

Munro, G.D., & Ditto, PH. (1997). Biased assimilation, attitude polarization, and affect
in reactions to stereotype-relevant scientific information. Personality and Social
Psychology Bulletin, 23, 636-653.

Nisbett, R.E., & Borgida, E. (1975). Attribution and the psychology of prediction.
Journal of Applied Social Psychology, 32, 932-943.

Perkins, H. W. (1997). College student misperceptions of alcohol and other drug norms
among peers: Exploring causes, consequences, and implications for prevention
programs. In Designing alcohol and other drug prevention programs in higher
education (pp. 177-206). Newton, MA: The Higher Education Center for Alcohol
and Other Drug Prevention.

Perkins, H. W. (2002). Social norms and the prevention of alcohol misuse in collegiate
contexts. Journal of Studies on Alcohol, supplemental no. 14, 164-172.

Perkins, H.W. (2003). The emergence and evolution of the social norms approach to
substance abuse prevention. In H.Wesley Perkins (Ed.), The social norms
approach to preventing school and college age substance abuse (pp. 3-17). San
Francisco, CA: Jossey—Bass.

Perkins, H. W., & Berkowitz, A. D. (1986). Perceiving the community norms of alcohol
use among students: Some research implications for campus alcohol education
programming. International Journal of the Addictions, 21, 961-976.

Perkins, H. W., & Craig, D. W. (2002). A multifaceted social norms approach to reduce
high-risk drinking. Newton, MA: Higher Education Center for Alcohol and Other
Drug Prevention.

Perkins, H.W., & Craig, D.W. (2003). The Hobart and William Smith Colleges
experiment: A synergistic social norms approach using print, electronic media,
and curriculum infusion to reduce collegiate problem drinking. In H.Wesley
Perkins (Ed.), The social norms approach to preventing school and college age
substance abuse (pp. 35-64). San Francisco, CA: Jossey-Bass.

76

Perkins, H. W., Meihnan, P. W., Leichliter, J. S., Cashin, J. R., & Presley, C. A. (1999).
Misperceptions of the norms for the flequency of alcohol and other drug use on
college campuses. Journal of American College Health, 47, 253-258.

Perkins, H. W., & Wechsler, H. (1996). Variation in perceived college drinking norms
and its impact on alcohol abuse: A nationwide study. Journal of Drug Issues, 26,
961-974.

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

Prentice, D. A., & Miller, D. T. (1993). Pluralistic ignorance and alcohol use on campus:
Some consequences of misperceiving the social norm. Journal of Personality and
Social Psychology, 64, 243-256.

Reinard, J .C. (1988). The empirical study of the persuasive effects of evidence: The
status after fifty years. Human Communication Research, 15, 3-59.

Reynolds, RA. (1987). The eflects of cognitive elaboration, valid arguments, evidence,
and message style in persuasive discourse. Unpublished doctoral dissertation,
Michigan State University, East Lansing, Michigan.

Reynolds, R.A., & Reynolds, J .L. (2002). Evidence. In J .P.Dillard & M. Pfau (Eds) The
persuasion handbook: Developments in theory and practice (pp. 427-444).
Thousand Oaks, CA: Sage Publications.

Rosenthal, RI. (1971). Specificity, verifiability, and message credibility. Quarterly
Journal of Speech, 5 7, 393-401 .

Sher, K. J ., Bartholow, B. D., & Nanda, S. (2001). Short- and long-tenn effects of
flatemity and sorority membership on heavy drinking: A social norms
perspective. Psychology of Addictive Behaviors, 15, 42-51.

Stiff, J .B. (1986). Cognitive processing of persuasive message cues: A meta-analytic
review of supporting information on attitudes. Communication Monographs, 53,
75-89.

Toulrnin, S. The uses of argument. Cambridge, UK: Cambridge Universtiy Press.

Wechsler, H., Kelly, K., Weitzman, E. R., San Giovanni, J. P., & Seibring, M. (2000).
What colleges are doing about student binge drinking. Journal of American
College Health, 48, 219-226.

Wells, G.L., & Harvey, J .H. (1977). Do people use consensus information in making
causal attributions? Journal of Personality and Social Psychology, 35, 279-293.

Wheeler, M. (1976). Lies, damn lies, and statistics: The manipulation of public opinion in
America. New York: Liveright.

77

Wood. W., Rhodes, N., & Biek, M. (1995). Working knowledge and attitude strength: An
information-processing analysis. In R.E. Petty (Ed.) Attitude strength:
Antecedents and consequences. Mahway, NJ: Lawrence Erlbaum.

78

liiiiiiijliiiiiiiiii

31 24