A MODEL OF NORMATIVE INFLUENCE:
MEMBERSHIP AND REFERENCE CONCEPTS
By
David M. Keating

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirement
for the degree of
Communication—Doctor of Philosophy
2014

ABSTRACT
A MODEL OF NORMATIVE INFLUENCE:
MEMBERSHIP AND REFERENCE CONCEPTS
By
David M. Keating
A model of interpersonally communicated normative influence is put forth proposing that
conformity to normative influence messages is a function of a membership concept and a
reference concept. The major constructs outlined are (interpersonally communicated) normative
influence, conformity, membership, and reference; propositions are put forth regarding the
relationships among them. The proposed model generally is consistent with a number of
explanations for the influence of social groups and normative influence on behavior (e.g., social
identity theories, theory on social norms, process of identification). Some preliminary evidence
for the model is first reported across two studies; following this, an experiment is reported that
did not give further evidence for the model. Implications of the observed data and limitations of
the studies are discussed.

ACKNOWLEDGMENTS

I am indebted to a number of amazing people. First and foremost, thanks must be given to
my adviser, Sandi Smith. Sandi has been an endless source of knowledge and encouragement
throughout this process and I could not have asked for a better scholar or person to guide me
through my degree. Second, I would like to thank the rest of my committee—Frank Boster,
Amanda Holmstrom, and Brent Donnellan—for their insightful questions and comments across
both my preliminary paper and my dissertation. Without these people, and the opportunity to
learn from them, I would not be the scholar that I am today. Third, I would like to thank other
professors and colleagues that I have been fortunate to learn from and to know, including Steve
Rains, Chuck Atkin, Joe Bonito, Tim Levine, Dana Mastro, Hee Sun Park, and Joe Walther.
Fourth and finally, I would like to thank the friends and family who have supported me over the
last four years, whether the support was sent from Utah, Arizona, Michigan, or elsewhere. I
would like to especially thank Diane, Michael, and Melissa Keating for being my rock; Manata
and Evan Perrault for their humorous and often wry takes on our shared experience; and Eric
Leckman and Tyler Green for keeping me sane.

iii

TABLE OF CONTENTS

LIST OF TABLES…………………………………………………….….….….… vi
LIST OF FIGURES………………………………………………………………... vii
INTRODUCTION…………………………………………………………………. 1
NORMATIVE INFLUENCE ……………………………………………...….…..
Interpersonally Communicated Normative Influence …….…………..…..
Conformity to Normative Influence………………….…….…………..….
Membership and Reference……….………………….…….…………..….
Assumptions of the Model……….………………….…….…………..…...
Specific Predictions of the Present Investigation…….…….…………..….

2
3
4
6
8
11

PRELIMINARY STUDY 1……………………………….…………………….…
Method…………………………………...…………………………….…..
Respondents………………………………………………………..
Measures………..…………………………………………….……
Results…………………..………………………………………….………
Discussion…………………………………………………………….……

13
13
13
14
14
15

PRELIMINARY STUDY 2………………………………………………………..
Method……………………...……………………………………….….….
Respondents………………………………………………………..
Scenarios……………………...……………………………………
Measures……………………..…………………………………….
Results…………………………………………………….…….…….……
Discussion……………………………………………..………….………..

17
17
17
17
18
18
19

PRIMARY LAB EXPERIMENT………………………………………….….…… 20
Method………………………………………………………. …………… 20
Pilot Study……...…………………………………………………. 20
Procedure………………………………………………….. 21
Results…………………………………………………….. 24
Primary Study……………………………………………………... 25
Participants………………………………………………... 27
Results…………………………………………………………………….. 27
DISCUSSION……………………………………………………………………… 30
CONCLUSION………………………………………………………………….…. 34
NOTES…………………………………………………………………….…….….35

iv

APPENDICES…………………………………………………………….…….…. 38
Appendix A…………………………………………………………..…….. 39
Appendix B………………………………………………………………… 40
Appendix C…………………………………………………………..…….. 41
Appendix D………………………………………………………………… 42
Appendix E…………………………………………………………..…….. 43
Appendix F………………………………………………………………… 44
Appendix G…………………………………………………………..…….. 45
Appendix H………………………………………………………………… 46
Appendix I…………………………………………………………..…….. 47
Appendix J………………………………………………………………… 48
Appendix K…………………………………………………………..…….. 51
Appendix L………………………………………………………………… 52
Appendix M……………………………………………………………..…. 53
BIBLIOGRAPHY…………………………………………………………….……. 54

v

LIST OF TABLES

Table A.1

Examples of Groups that are Positive Reference or Negative
Reference and in which People Hold Membership or Not………… 39

Table B.1

Examples of Normative Influence Messages Reported by
Respondents in Study 1……………………………………………. 40

Table C.1

Results of the Moderated Multiple Regression Analysis………..

Table M.1

Means and Standard Deviations for each Outcome Variable
in each Experimental Condition …………………………………. 53

vi

41

LIST OF FIGURES

Figure D.1

The Membership-Reference Interaction for Conformity
in Preliminary Study 1……………………………………….……..42

Figure E.1

Predicted Conformity Following Normative Influence in
Preliminary Study 2……………………………………….……….. 43

vii

INTRODUCTION
Group influence has remained a central focus of research for over 50 years. From
foundational work on conformity (Asch, 1956) to more recent empirical and conceptual work on
social norms (Cialdini, Reno, & Kallgren, 1990; Lapinski & Rimal, 2005), there is no dearth of
research on the impact that groups have on behavior. However, in spite of the abundance of
research in this realm, there is still room to expand knowledge about the role that groups play in
the influence process. Within work on normative influence, for instance, various types of norms
have been identified (see Lapinski & Rimal, 2005; Park & Smith, 2007). Though explication of
these types of norms is invaluable, furthering knowledge about the normative influence process
is warranted.
The following sections present a model of normative influence that occurs during
interpersonal communication episodes and that focuses on the roles of group membership and
desired affiliation. First, the primary constructs of the model are detailed (i.e., normative
influence, conformity, membership, reference). Second, propositions are presented for the nature
of the relationships among the major concepts. Third, the results of two preliminary studies and
the primary lab experiment are described which provide mixed evidence for the predictions.
Along with this, the implications of the results and limitations are discussed.

1

NORMATIVE INFLUENCE
Normative influence has been treated in several ways. Some have considered normative
influence in terms of descriptive and injunctive norms, discussed further below; subjective
norms, via the theories of reasoned action and planned behavior, which are considered as the
expectations of significant others regarding the enactment of some behavior (e.g., Ajzen, 1991);
and personal and societal norms, which refer to norms within important groups and within
society at a larger level (Park & Smith, 2007).
A classic definition defines normative influence “as an influence to conform with the
positive expectations of another” (Deutsch & Gerard, 1955, p. 629). A slight revision of this has
argued that normative influence is “based on the goal of obtaining social approval from others”
(Cialdini & Goldstein, 2004, p. 606). In the model put forth here, an extension of these
definitions is adopted: Normative influence is information about normative behavior within a
group that is perceived, on the part of the message recipient, as signaling behavior that group
1

members—perhaps including the message recipient—are expected to adopt.

Much of the literature on norms has focused exclusively or partly on the differential
impact that descriptive norms and injunctive norms have on people’s behaviors. It is generally
agreed that descriptive norms identify “what is typical or normal. It is what most people do…”
(Cialdini et al., 1990, p. 1015), and that injunctive norms refer “to rules or beliefs as to what
constitutes morally approved and disapproved conduct…[and] specify what ought to be done”
(Cialdini et al., 1990, p. 1015). An array of empirical studies have presented evidence that
descriptive and injunctive normative information are meaningful factors influencing a range of
behaviors, such as littering in various environments (Cialdini et al., 1990), household energy

2

consumption (Schultz, Nolan, Cialdini, Goldstein, & Griskevicius, 2007), and alcohol
consumption intentions (Rimal & Real, 2005).
In many cases, normative influence may be interpersonally communicated. For instance,
a member of a given ethnic group may receive messages about typical behavior for members of
that ethnic group; a father might tell his child about how to act when around extended family
members; or a work group might tell a newcomer that laptops are discouraged during meetings,
and that legal pads typically are used. Whatever the specific instantiation, it is plausible that
normative influence often comes from interpersonal sources.
Interpersonally Communicated Normative Influence
Interpersonal communication also has been defined in several ways. The definition
adopted here, labeled as the message-centered approach (Burleson, 2009), puts forth that
interpersonal communication “is a complex, situated social process in which people…exchange
messages in an effort to generate shared meanings and accomplish social goals” (p. 151). In
other words, this approach proposes that interpersonal communication involves a transaction of
messages among interactants in a concrete episode where at least one party is seeking to share
some internal state (e.g., beliefs, feelings) and/or to accomplish some sort of goal (e.g., influence,
comfort).
It is useful to consider interpersonally communicated normative influence (ICNI) in this
way because it suggests that it occurs during concrete situations (i.e., it is situated within a
specific episode and within the larger normative influence process) and revolves around the
sharing of information about normative behavior and the attempt to move a target toward that
behavior. Given this and the definition of normative influence above, ICNI is defined broadly as

3

communication of information about normative behavior during an interpersonal interaction that
is perceived as signaling behavior that group members are expected to adopt.
Though the focus here is on how such communication impacts recipients’ behaviors and
not on the specific features of the messages, some speculation is warranted as to whether
different message characteristics would meaningfully impact upon the ICNI process. For
instance, beyond the distinction between descriptive and injunctive messages, one potentially
important dimension might be the explicitness of the normative influence. Some normative
messages may be very direct (e.g., in our group, we do a certain behavior, so you should, too);
however, others may be more indirect (e.g., simply talking about, even vaguely, the general
behaviors that the group tends to enact without an outright request that the target does the same).
Both cases would constitute ICNI if the recipient believed the message to be intentionally geared
toward changing the recipient’s behavior, but it is likely that the former case—more direct
messages—are more frequently perceived as having such intent and thus are more common
instantiations of ICNI.
Conformity to Normative Influence
When faced with normative influence, including ICNI, recipients are faced with at least
two clear options: Conform to the normative behavior, or do not conform to it. Conformity is
defined here as “the act of changing one’s behavior to match the responses of others” (Cialdini &
Goldstein, 2004, p. 606). In addition, this investigation proposes that conformity is continuous in
nature: Targets of influence may conform none of the time or all of the time, but they may also
conform rarely, often, or nearly always.
This focus on behavioral conformity does not include an attitudinal component. Attitudes
are relatively enduring positive or negative evaluations of some person or thing (Stiff &

4

Mongeau, 2003) and tend be strong predictors of behavior (Kim & Hunter, 1993). In spite of
this, the present investigation considers public conformity, regardless of whether the normative
behavior is associated with private acceptance (Festinger, 1953).
Beyond the roles of descriptive and injunctive norms lie additional factors that arguably
impact conformity to normative influence. Among the concepts that might be particularly
interesting from a theoretical standpoint are group similarity, historical norms, source
prototypicality, membership, and reference status.
Group similarity refers generally to how much group members are perceived to be alike;
especially in ambiguous situations in which the “correct” behavior to enact is unknown, as group
similarity increases, so too should the motivation to conform to normative influence (Rimal &
Real, 2005). Historical norms refer to beliefs about past norms; perceptions about such prior
normative behaviors might result in an increase in the adoption of this type of behavior
(Carcioppolo & Jensen, 2012). Source prototypicality refers to the extent to which a member of a
group typifies or exemplifies the “ideal” member of the group (e.g., news personalities on a
conservative or liberal news network); prototypicality might be positively related to adoption of
normative behavior, such that normative influence from more prototypical members is more
likely to be followed (Hogg & Reid, 2006). Membership refers to whether a recipient of
normative influence is or is not a member of the relevant group and reference status refers to the
extent to which membership within a group is desirable to members and non-members (Siegel &
Siegel, 1957); these last two concepts of membership and reference status are the primary focus
of this investigation.
Now that some basic distinctions and definitions have been given, a central question can
be asked: Why do some people conform to the behaviors presented by normative influence,

5

while others do not conform, and some might even adopt opposing behaviors? In considering the
answer, the two concepts are discussed next: Membership and reference. These are defined
before presenting the model’s core assumptions and predictions.
Membership and Reference
A higher-order assumption of the model is that any given person belongs to, and believes
that he or she belongs to, a number of social groups. A social group is a collection of people that
share some sort of external trait (e.g., biological sex) or internalized affiliation (e.g., group of
friends), and references some social category, such as the middle class or a political party
affiliation (Tajfel, 1982). Though on its face this might appear to preclude task-oriented groups
(e.g., a marketing team within an organization), this is not necessarily the case: An
organizational team or a group of students working on a project share some sort of social tie, in
that they have interdependent goals, often are affiliated to some extent, and must communicate
with one another. In any case, there also exist some groups to which a person does not belong,
and it is assumed that people understand that they lack membership to such groups.
In the existing literature, membership groups are denoted as those to which a person
belongs (e.g., family, work group, political party), whereas reference groups are denoted as those
to which a person “aspires to attain or maintain membership” (Siegel & Siegel, 1957, p. 360).
Though some have used the term “reference group” to refer broadly to any group in which an
individual is a member, it is considered here as a group one desires to be a part of, regardless of
actual group membership. Many membership groups also are reference groups (e.g., a close-knit
family), but this is not an inevitability (Sherif & Sherif, 1953). Indeed, there may be groups that
a person wants to be a member of but is not (e.g., a high school clique) as well as groups to
which a person belongs without having any real affiliation (e.g., employee of a fast-food chain).

6

Further, a given group may be a negative reference group—one in which someone actively
opposes or does not wish to be a member (Sherif & Sherif, 1953). Regarding the impact on
outcomes of influence, conceptual arguments and empirical findings suggest that people are most
influenced by groups that are both membership and reference groups and least influenced by
those that are neither (Sherif & Sherif, 1964; Siegel & Siegel, 1957).
This model adopts tailored views of these concepts such that they are slightly different
from yet remain conceptually consistent with the foundational work. For the membership
concept, the following view is adopted: You are a member of a given group or you are not a
2

member of that group. For the reference concept, the following view is adopted: Desired
affiliation falls along a continuum, such that there may be positive reference groups in which
membership is desired or negative reference groups in which a person desires to not be a
member, and everything in between. (See Appendix A for some examples, broken down by
whether or not there is membership to the group, and whether the group is a positive reference or
negative reference group.)
So what do membership and reference have to do with ICNI? Some examples may help
to illustrate how ICNI can be understood in terms of these concepts. For instance, consider a
family where the parents expect strict child compliance to their requests. When a teenaged son
stays out too late, his mother might offer a message such as, “So long as you are living under our
house, you will be home when we tell you to be home.” In this scenario, a group member (the
mother) is attempting to establish a norm (conformity to a parentally defined rule about when to
be home at night) to another group member (the son). This could be considered as a case of ICNI
within an important membership group. However, the son’s affiliation toward the family as a
group might be open to interpretation.

7

Not all normative influence concerns a membership group. For example, consider a
graduate student who one day aspires to go into the academy—a goal she believes she can
reasonably attain. When thinking about this motivation, she may recall a time when a tenured
faculty member said, “I don’t worry too much about getting conference papers out there—
publications are more important.” The graduate student might subsequently not fret about
conference deadlines, and in her attempts to focus on journal submissions, may ultimately have
few conference papers listed on her curriculum vitae. In this scenario, the graduate student is not
yet a member of the group (i.e., professors) about which she perceived there to be ICNI, though
she is working toward becoming, and most likely desires to be, a member of this group.
Now that the major constructs of the model have been defined (i.e., normative influence,
conformity, membership group, reference group), a series of propositions can be advanced. The
following sections propose some broad assumptions of the model and predictions that are drawn
from them. While these assumptions are not limited to ICNI per se, their generality should allow
them to be applied to interpersonal instantiations of normative influence.
Assumptions of the Model
As stated, this conceptual model is framed by the assumption that individuals both belong
to and do not belong to different social groups, and they perceive themselves as members or nonmembers of those groups. This is consistent with social identity theories (SITs), which generally
argue that people place themselves into different social categories (i.e., groups) and further
delineate between people who also are within their groups (in-group members) and those who
are not (out-group members) (Ellemers & Haslam, 2011; Tajfel & Turner, 1986).
Group distinctions may be made based on shared biological characteristics, such as
ethnicity, biological sex, and age cohort, as well as on internal affiliation with similar others,

8

such as sharing the same religion or political party (Harwood, Giles, & Palomares, 2005).
Conceptual work on intergroup communication has argued that there are times when individuals
interact based on their group identities, and this tends to occur when group membership
distinctions are made salient to them (Harwood et al., 2005). Such salience is likely to result
when a group identity is important to a person and/or when that identity is particularly relevant to
the communicative context (Hogg & Reid, 2006).
Following the definition of normative influence presented previously, messages that are
perceived as being about normative behavior within a given group should make that group
relatively salient to the message recipient. Two broad and related assumptions are now offered.
First, people belong to, and perceive that they belong to, different social groups; they also do not
belong to, and perceive that they do not belong to, certain social groups. Second, normative
messages concerning a particular group draw message receivers’ attention to their current
affiliation regarding that group.
Membership within social groups additionally should be associated with motivations to
adopt group behavior. SITs propose that individuals derive at least some of their self-image from
their group identities (Tajfel & Turner, 1986). As a result, individuals evaluate in-group
members more positively (Ellemers & Haslam, 2011) and are more highly motivated to behave
similarly to in-group members compared to out-group members (McGarty, Haslam, Hutchinson,
& Turner, 1994). In the realm of social norms, the theory of normative social behavior (TNSB)
additionally argues that people are highly likely to follow the behavior of similar others (Rimal
& Real, 2005). This leads to a third assumption: People may be motivated to engage in the same
behaviors as other group members.

9

Beyond simply being a member of a group, the desire to be a member of some group—
regardless of actual membership—should play a role in the adoption of normative behavior. The
process of identification, for instance, argues that people conform to influence because of its
association with a desired network or relationship; in other words, people conform because they
“want to establish or maintain a satisfying self-defining relationship to another person or a
group” (Kelman, 1958, p. 53). The TNSB proposes that individuals are particularly motivated to
emulate the behaviors of other group members when they have strong affinity toward those
members (Rimal, 2008). Put another way, people should be more likely to conform to normative
behavior when they hold a particularly positive attitude toward members of that group (Lott &
Lott, 1961). Thus, a fourth assumption is that people are most highly motivated to enact
normative behavior within a membership group when it is a positive reference group. This is not
to claim that membership groups that are negative reference groups are never influential; indeed,
undesirable groups to which people belong often are associated with factors that keep members
from leaving the group or that imply negative social sanctions for distancing oneself from the
group (Festinger, 1953). Rather, this assumption argues that, on average, the most influential
groups are those that can be classified as membership and positive reference groups.
Finally, the reference concept should affect conformity to normative influence when
message receivers are not members of the group that the behaviors concern. Consider the earlier
example of the graduate student. She does not yet have her degree and a job at a university, and
thus is not yet a member of her desired group (i.e., she is not yet a professor). However, she
probably is more likely to engage in behaviors that are characteristic of that group (e.g., working
on a research project late into Thursday night) than are not characteristic of that group (e.g., bar
hopping on a Thursday night). Indeed, empirical work (Siegel & Siegel, 1957) supports the idea

10

that people are likely to conform to positive reference groups even when they are not
membership groups.
When normative influence concerns a negative reference group that is not a membership
group, however, an opposite pattern of behavior should result. This type of group clearly is an
out-group—a social category to which an individual does not belong (Harwood, 2010), that is
comprised of members that are judged less positively than in-group members (Ellemers &
Haslam, 2011), and that may be perceived in a particularly negative or competitive way (Tajfel,
1982). In sum, out-groups are those to which a person is not a member and to which membership
often is not desired.
Because group identities are tied to self-concept and people typically strive to maintain
positive self-concepts (Tajfel & Turner, 1986), individuals tend to differentiate between in-group
and out-group members such that the out-group is judged more harshly (Ellemers & Haslam,
2011). It is no surprise, then, that out-group members usually are unsuccessful influence sources
compared to in-group sources (Mackie & Queller, 2000). Thus, when normative influence
involves a negative reference group that is not a membership group, targets should have little
motivation to adopt the specified normative behavior. The final assumption of the model, then, is
as follows: People are motivated to engage in the normative behavior of a non-membership
group when it is a positive reference group; they are motivated not to engage in the normative
behavior of a non-membership group when it is a negative reference group.
Now that the major assumptions have been laid out, specific predictions can be advanced.
The following section argues for two major hypotheses that are consistent with these general
assumptions.
Specific Predictions of the Present Investigation

11

If it is the case that individuals are cognizant of their affiliation within different groups
and that normative influence makes group membership distinctions salient, then conformity to
normative behavior is likely to be affected by the membership and reference concepts. Regarding
the former, if it is the case that individuals are motivated to engage in behavior that is normative
within their membership groups, then conformity to normative influence should be more likely
when it involves a membership group than when it involves a non-membership group.
The impact of membership on conformity, however, ultimately should be affected by the
extent to which normative influence involves a positive or negative reference group. It was
proposed that individuals are motivated to conform to positive reference groups, regardless of
whether they are membership groups. When normative influence involves a more negative
reference group, however, some substantial differences should be observed.
If it is the case that individuals are somewhat motivated to conform to negative reference
groups when they are membership groups but are motivated to not conform to negative reference
groups that are non-membership groups, then the reference concept should modify the
relationship between membership and conformity. In other words, there should be a monotonic
interaction of the membership and reference concepts for conformity to normative influence,
such that people tend to conform to positive reference groups regardless of whether they are
membership groups; tend to conform to negative reference groups that are membership groups;
and do not tend to conform to negative reference groups that are non-membership groups.
3

The proposed predictions initially were examined in three investigations. The first study
had two purposes: To assess whether people actually receive ICNI and to statistically test the
predictions. The primary purpose of the second and third studies was also to test these
predictions, but in more controlled contexts. Each of these is reported next.

12

PRELIMINARY STUDY 1
Method
Research participants were asked to retrospectively report on their experiences in order to
assess the extent to which ICNI actually is received and to test the predictions regarding the
model’s constructs (i.e., membership, reference). Respondents answered a number of questions
via an online survey. They were first presented with the following instructions:
Sometimes, a person who is a member of some group will tell you about how people in
that group do act or should act, and sometimes their intention is to get you to behave in
that way. For example, a co-worker might tell you how people in your work group act,
your mother might talk to you about appropriate behavior in your family, a friend might
try to get you to behave in a way that is typical of your group of friends, or someone in
your political party might tell you about appropriate action to take. If you can, please
think about a time when someone, in a face-to-face setting, talked to you about normal
behavior in some group. The person may or may not have been trying to persuade you,
and you may or may not have been an actual member of this group.
Following this, they were asked additional questions about the recalled experience and about
demographic information.
Respondents. A total of 357 students at a large Midwestern university responded to the
online survey. Of these, only 34 (9.5%) refrained from reporting on a normative message. The
remaining respondents who did report on a normative message they received (N = 323) were
between 18 and 41 years of age (M = 20.03, SD = 1.97), and primarily were women (64.1%) and
Caucasian (81.1%). They reported receiving the normative influence an average of 13.17 months

13

(SD = 29.69) prior to completing the survey. Normative influence concerned a number of groups,
including friendship networks, work groups, family, political parties, and social clubs.
Measures. The membership variable was assessed with a single yes/no question asking,
“At the time you heard this message, were you a member of the group?” The reference variable
was assessed with one 5-point semantic differential item asking, “At the time you heard this
message, to what extend did you prefer to be a member of this group?” The response scale
ranged from “Really did NOT want to be a member of this group” to “Really DID want to be a
member of this group.” Conformity to normative influence also was assessed with one 5-point
item that asked, “Since you heard this message, to what extent have you behaved in the way that
the person was talking about?” Responses ranged from “I have never behaved in that way” to “I
have always behaved in that way.”
Results
The normative influence received by respondents (see Appendix B for some examples)
covered a vast array of groups (e.g., family, political parties, sororities/fraternities, ethnic groups)
and behaviors (e.g., personal conduct, voting decisions, dress attire, treatment of others). In
general, most respondents who received ICNI reported on a group in which they had
membership (n = 241, 74.6%), with many fewer reporting on one in which they did not (n = 82,
25.4%). Reference scores were somewhat skewed (M = 3.57, SD = 1.25, Mdn = 4.00, range =
1.00 – 5.00), as were conformity scores (M = 3.38, SD = 1.27, Mdn = 4.00, range = 1.00 – 5.00).
The membership variable (0 = no membership, 1 = membership) was positively correlated with
both the reference variable, rpb (321) = .32, p < .05, and conformity variable, rpb (321) = .30, p <
.05. Reference and conformity also were positively correlated, r (321) = .49, p < .05.

14

To test the predictions, a moderated multiple regression analysis was conducted with
conformity set as the criterion variable. In the regression model, the membership variable was
dummy-coded (non-member = 0) and the reference scores were mean-centered (Cohen, Cohen,
West, & Aiken, 2003); the membership-reference interaction term entered was the product of
these two former scores.
2

The results of the overall model, F (3, 319) = 39.55, p < .05, adjusted R = .26 (also
displayed in Appendix C), illustrate that the set of variables accounted for a statistically
significant amount of variance in conformity scores. In addition, as can be seen, all three
predictor variables were statistically significant. The predicted non-additive effect was further
analyzed by computing the average scores at one standard deviation above, at the mean, and at
one standard deviation below for the reference scores separately for those who reported on a
membership group and those who reported on a non-membership group. The following equation
was used to do so: Y = 3.14 + 0.37(X1) + 0.61(X2) + (-0.23)(X3), where X3 = (X1)(X2). As can
be seen in Appendix D, when normative influence involved conformity to a positive reference
group (i.e., at one standard deviation above the mean of reference scores), there was virtually no
effect of membership on conformity. On the other hand, when normative influence involved a
more neutral reference group (i.e., at the mean of reference group scores) and a negative
reference group (i.e., at one standard deviation below the mean of reference group scores),
conformity was higher when it concerned a membership group. Thus, the data were generally
consistent with the predictions.
Discussion
This study sought to explore the frequency with which normative influence is
communicated interpersonally as well as to test the predictions regarding the membership and

15

reference constructs. The majority of respondents indeed recalled receiving such normative
influence (90.5%), though it is possible that the example given impacted the types of groups that
were reported on by participants. In addition, the results were consistent with the predictions.
However, there was extreme variability in the groups, sources, and behaviors reported—in fact,
it is prudent to assume that no two respondents reported on the same thing. To further test the
predictions in a more controlled way that eliminated the effect of relevant extraneous variables
(e.g., group similarity, relationship to message source), a second preliminary study was
conducted using hypothetical scenarios.

16

PRELIMINARY STUDY 2
Method
Research participants were randomly presented with one of four hypothetical scenarios in
which they were asked to gauge the behavior of a fictional person following their reception of
ICNI.
Respondents. A total of 271 students from the same university as Preliminary Study 1
read one of four scenarios and responded to questions in an online survey format. The age of
these respondents ranged from 18 to 28 years (M = 20.34, SD = 1.61) and the majority of them
were female (56.1%) and Caucasian (75.4%).
Scenarios. Each scenario was about the same group (an unnamed fraternity) and the
same behavior (dress attire on a specific day of the week). However, each scenario varied in
terms of whether the fictional receiver (Todd) was a member of the group or not as well as
whether it was a positive reference or negative reference group. In the membership/positive
reference condition, for example, respondents were given the following scenario:
While walking on campus, Todd bumps into an acquaintance—Bob. Todd knows Bob
because they are both members of the same Fraternity. Even though Todd *loves* being
a part of this group, he does not yet know everyone well (including Bob). Todd and Bob
strike up a brief conversation. At the end of their conversation, Bob says, “You’re not
wearing a solid black shirt today? On Fridays, members of our Fraternity wear solid
black shirts.”
After reading this, respondents were asked a number of questions about how they thought Todd
would behave following this situation. Groups were approximately equal (ns = 67 – 68).

17

Measures. Expectations about Todd’s behavior were measured in two ways. First,
respondents were asked to indicate, from 0% to 100%, “After this interaction, for the rest of the
semester, what percentage of the time do you think Todd will wear a solid black shirt on
Fridays?” (M = 59.41%, Mdn = 75.00%, SD = 39.44%). Second, they were asked a 5-point
semantic differential item about publicly observable behavior: “If Todd expected to *always* see
members of this group on Fridays, how often do you think he would wear a solid black shirt on
Fridays?” (M = 3.75, SD = 1.51). Responses ranged from “none of the time” to “all of the time.”
Results
The percentage item was statistically significantly correlated with the publicly observable
behavior item, r (267) = .77, p < .05. Thus, they were combined to form a single measure; this
was done by transforming each measure into a standardized score and then averaging them. This
combined measure was then subjected to a 2 (membership/no membership) by 2
4

(positive/negative reference group) analysis of variance (ANOVA). There were statistically
significant main effects for both the membership factor, F (1, 267) = 61.67, p < .05, r = .31, and
the reference factor, F (1, 267) = 233.74, p < .05, r = .61, and there was also a statistically
significant interaction effect, F (1, 267) = 62.90, p < .05, r = .32. The means, which are visually
represented in Appendix E, were consistent with the predictions.
To further probe the predictions (i.e., that conformity would be expected to occur most
frequently when normative influence involved membership/positive reference and nonmembership/positive reference groups; that conformity would be expected, though less
frequently, when normative influence involved a membership/negative reference group; and
conformity would not be expected when normative influence involved a nonmembership/negative reference group), a contrast ANOVA was conducted. In particular, the two

18

positive reference conditions were given a contrast coefficient of +1, the membership/negative
reference condition was given a contrast coefficient of 0, and the non-membership/negative
reference condition was given a contrast coefficient of -2. This contrast model additionally was
statistically significant, F (1, 267) = 358.15, p < .05, r = .76, and explained the remaining
residual variance.
Discussion
The results again were consistent with the predictions put forth. However, given that the
majority of respondents were communication majors at a large Midwestern university, it is
plausible that existing perceptions of the group type, including historical norms (Carcioppolo &
Jensen, 2012), impacted upon the expectations of how the fictional recipient of normative
influence would behave. In other words, this demographic likely is familiar with the importance
of tradition in the Greek system, and many might have been in a sorority or fraternity
themselves. In addition, the respondents were not subjected to normative influence and given a
chance to adopt (or not adopt) a normative behavior themselves. Thus, a lab experiment was
conducted that sought to further control for extraneous variables as well as to assess actual
behavioral choices. Participants were presented with a normative influence message, rather than
a hypothetical episode, involving novel groups and behaviors with which they had no prior
experience.

19

PRIMARY LAB EXPERIMENT
Method
Following a pilot study, an experiment was conducted in which research participants
were randomly assigned to one of eight experimental conditions designed to represent the
different membership/reference configurations (membership/positive reference;
membership/negative reference; no membership/positive reference; no membership/negative
reference) across two different “personality characteristics” that were part of the experimental
procedures. In each condition, participants received a normative message (a descriptive norm)
from an experimenter. The scripting of these messages was designed with the intent of
containing explicit conclusions (Cruz, 1998) without arousing suspicion; explicit conclusions are
those that make clear how the target of influence is expected to believe or behave following an
influence message. All participants received course or extra credit for their participation.
Pilot study. Prior to the primary experiment, a pilot study was conducted. The chief
goals of the pilot study were to assess the believability of the protocol (i.e., the extent to which
the true nature of the experiment was successfully concealed from participants) and to gauge the
success of the experimental induction. The primary method by which the first goal was evaluated
was through a thorough debriefing process in which participants were asked a series of questions
about any suspicions regarding the claimed nature of the study. The second goal was evaluated
via an induction check, which is specified below.
Research participants (N = 19) were randomly assigned to one of the four primary
conditions (membership/positive reference, n = 5; membership/negative reference, n = 4; no
membership/positive reference, n = 5; no membership/negative reference, n = 5). These
participants were primarily female (n = 12, 63.2%) and Caucasian (n = 13, 68.4%) and ranged in

20

age from 18 to 25 years (M = 20.68, SD = 1.86). Only one experimenter—the author—conducted
the pilot study.
Procedure. Research participants were brought into the lab under the guise of helping
researchers to enhance a measurement tool. Once the participants arrived, the experimenter
administered a brief questionnaire that asked a series of bogus personality questions (see
Appendix F). Although this questionnaire was made up of irrelevant items, after it was
administered, the experimenter claimed that the questions measured a personality characteristic
(i.e., “discerning”; “circumspect”) and that subsequent placement into a “high” or “low”
5

category would be based on the participants’ responses ; they were additionally told that they
may later be asked to interact with other members who fall into their category. Of the 19
participants, about half were randomly assigned to each trait (ns = 9 – 10).
At this point, participants were asked to complete a set of demographic items as well as
some basic attitude items about three types of programs (i.e., food bank, animal shelter, job
placement) that also played into the cover story; during this time, the experimenter “tallied up”
the results of the questionnaire. Following this, the experimenter claimed that research tends to
find that people who are high in the characteristic “tend to be highly valued at a societal level,
and people who are high in the characteristic actually tend to be evaluated by others as fairly
6

creative, likeable, and charismatic.” At this point, the experimenter casually said, “So, like, for
example, your scores put you as [high/low] on this dimension.” In reality, participants were
randomly assigned to different groups. These procedures were chosen in order to prime beliefs
that membership in the “high” group was more desirable than membership in the “low” group; a
similar type of priming has been used successfully in previous research (Spears, Doosje, &
Ellemers, 1997).

21

At this point, consistent with the cover story that the goal of the study was to enhance a
new measure of this dimension, the experimenter stated that the measure has received some
criticism and has not yet been validated. Participants were then told that their performance on
some decision-making tasks would be better indicators of their true scores, and that their
completion of these would be compared to their responses on the measure they just completed in
order to assess its validity. This was meant to give participants the opportunity to essentially
transition to a different category. In addition, the first message about normative behavior was
given at this time. Specifically, the experimenter “let slip” the following:
So, like, even though your scores on that placed you as [high/low], what you do on these
tasks will either give more evidence that you are [high/low], or suggest that you’re the
opposite. Personality researchers use these exercises a ton because people who are high
and low on this characteristic tend to make the same decisions. It’s really interesting,
actually—like on this first exercise, people who are [high/low] tend to choose the
parachute as the #1 most important item.
Following this, participants were given the first exercise. Presumably, participants placed
in the “high” group who received normative information about other “high” members should
constitute the membership/positive reference condition; people in the “low” group who received
normative information about other “low” members should constitute the membership/negative
reference condition; people in the “low” group who received normative information about the
“high” group should constitute the non-membership/positive reference condition; and people in
the “high” group who received normative information about the “low” group should constitute
the non-membership/negative reference condition.

22

The first task was a desert survival exercise (see Appendix G). In this exercise,
participants are asked to imagine themselves as having crash-landed in the Sonoran Desert
during a plane ride; after salvaging the wreckage, they are left with 15 items (e.g., sunglasses, a
mirror, a compass). They are asked to rank, from 1 to 15, the importance of each item for their
survival. Participants were given about five minutes to make these rankings. Participants’
rankings of the parachute item served as one measure of conformity. Participants knocked on the
door after they completed this first exercise, and the experimenter entered the room to take the
form and to administer the next exercise.
Prior to administering the second task, it was explained that the attitude items from
earlier were going to be used to help statistically control for whether people’s existing attitudes
have an impact on their responses. The experimenter then again, seemingly in an impromptu
manner, gave a normative message:
So, like, in other words, like—people who are [high/low] tend to give the most money to
this job placement program, and we basically need to make sure that your existing
attitudes don’t end up affecting the results.
Following this, the second exercise—a monetary donation task—was administered (see
Appendix H). In this exercise, participants are given a fictional amount of money ($1,000) and
asked to divvy it up, in whatever amount they see fit, among three different “causes.” They are
told that they must donate at least $100 to each organization and that they must donate all of the
money, but beyond this, there are no restrictions on the nature of donations. Participants were
again given about five minutes to complete the exercise. The amount of money that participants
chose to donate to the job placement organization (range: $100 – $800) served as another
7

measure of conformity to normative behavior.

23

After this exercise was completed, the experimenter administered the induction checks
(see Appendix I). Experimental participants were asked several questions about the groups into
which they were placed prior to completing the first exercise. First, participants were asked to
note the group into which they had been placed. Second, they were given items intended to
assess the desirability of both the “high” and the “low” groups (i.e., the extent to which each
group is a desirable one in which to be a member). This gauged the relative success of the
reference manipulation.
Finally, research participants were debriefed. Before explaining the true nature of the
study, pilot study participants were asked (1) what they believed were the hypotheses,
predictions, or questions of the study and (2) what about the study, if anything, made them
suspicious of the goals that were explained to them. Of the 19 participants, only one (5.3%)
correctly saw through the cover story. Following these questions, the experimenter explained the
true purpose of the study, assured participants that the personality characteristic questionnaire
was bogus and that they are not actually “high” or “low” in the dimension, and asked participants
to refrain from speaking about the study with other students until the study’s completion.
Results. Every participant correctly noted that they were in the group into which they
were placed. The items intended to measure the desirability of membership in the “high” group
(M = 3.69, SD = 0.58, αSI = .81, 95% CI [3.42, 3.96]) were statistically significantly above the
scale’s midpoint, t (17) = 5.09, p < .05; however, the items intended to measure the desirability
of membership in the “low” group (M = 2.79, SD = 0.73, αSI = .79, 95% CI [2.46, 3.12]) were
not statistically significantly below the midpoint, t (18) = -1.26, p > .05. The results of a pairedsamples t-test gave evidence that participants tended to hold a more favorable attitude toward the
high-trait group than toward the low-trait group, t (17) = -3.94, p < .05. Together, these findings

24

suggested that the experimental induction was indeed successful in placing people into different
categories and in creating beliefs that the “high” group was more desirable.
The mean score for the importance of the parachute item in the first task was 9.16 (SD =
3.95, range: 2 – 15) and the mean amount of money donated to the job placement program in the
second task was $356.12 (SD = 108.62, range: 100.00 – 500.00); neither distribution of scores
was skewed. There were no statistically significant differences among the outcome measures
depending on the characteristic (discerning versus circumspect), ts (17) = 0.39 – 1.00, ps > .05.
There were no statistically significant main effects for the membership, Fs (1, 15) = 0.00 – 0.06,
rs = .00 – .02, or the reference variables, Fs (1, 15) = 0.05 – 0.51, rs = .02 – .05; there also were
no statistically significant interaction effects, Fs (1, 15) = 0.65 – 3.52, rs = .08 – .12.
Primary study. The procedures in the main experiment were the same as those described
in the pilot study with some exceptions. First, two additional experimenters were recruited in
order to help ensure that the experimenter of the pilot study (i.e., the author) did not bias the
results. These experimenters were undergraduate students who received course credit for their
assistance. Second, instead of a thorough debriefing as in the pilot study, participants were asked
at the end of the experiment to write down their guesses as to the “hypotheses, predictions, or
questions of the study”; they also were asked to write down whether they were suspicious of the
goals that were explained to them and what, if anything, made them suspicious. This helped to
gauge whether participants were “correctly suspicious” about the primary experimental
induction, and to help assess whether such suspicions impacted the findings in any meaningful
way.
Two additional changes were made in order to enhance the believability of the study’s
process and, accordingly, resulted from the debriefing from the pilot study. First, parts of the

25

script were revised in order to make the normative messages more colloquial and to account for
changes to the exercises (the final script for the experimenters is shown in Appendix J). Such
changes also helped to further establish the experimenter as someone who “let’s things slip,”
including the normative messages. Second, the decision-making exercises were revised in order
to be more believably related to one’s personality. This decision was made in light of the
remarks by some participants during the debriefing in the pilot study; in particular, a few
wondered why the exercises would be related to their personality. It was reasoned that such a
disconnect between the exercises and the personality characteristic in the cover story of the
experiment might help to explain why the results of the pilot study were not consistent with the
predictions.
In the first exercise (see Appendix K), participants were given a list of four people with
different attributes and asked to rank, from first to fourth, the people with whom they would
most and least prefer to be friends. In reality, all four of the potential friends have extremely
similar attributes. The normative message from the experimenter was that “people who are
[high/low] tend to rank Taylor the best.” This message explicitly pointed participants to the
normative behavior.
In the second exercise (see Appendix L), participants were put in a hypothetical situation
wherein they are playing a hand of poker against one of the best players in the world and must
make a wager between $0 and $500. They are told that the other player will most likely bet the
same exact amount and that the winner will take all of the money, but also that they are only
50% sure that they will win the hand. The normative message was that “people who are
[high/low] tend to be very careful and not willing to take big risks.” This message was less
explicit than for the previous exercise; however, although the communicated behavior was more

26

generally about risk-taking, the exercise and script were designed in such a way as to clearly
suggest that members of the mentioned group would be expected to bet less money.
Participants. The experimental participants (N = 114) were undergraduate students from
the same university as the respondents in the two preliminary studies. These participants were
between the ages of 18 and 26 (M = 20.33 years, SD = 1.55) and were primarily Caucasian (n =
79, 69.3%) and female (n = 74, 64.9%). They were about equally assigned to each condition
(membership/positive reference = 29; membership/negative reference = 28; nonmembership/positive reference = 29; non-membership/negative reference = 28); within this,
about half were randomly assigned to each of the two traits (discerning = 56; circumspect = 58).
Of the 114 participants, 20 (17.5%) reported being suspicious about the normative
messages (i.e., suspicious about the primary aspect of the experimental induction). When these
participants were excluded, the results did not change; thus, the analyses reported below include
these people. It appears that the majority of this suspicion was associated with one of the
experimenters in particular: Of these 20 cases, 14 (70.0%) occurred with one experimenter; a 2
(suspicious/not suspicious) by 3 (experimenter) chi-square test gave further evidence that
2

suspicions were more likely to arise for this experimenter than would be expected, χ (2, n =
114) = 7.12, p < .05, Cramer’s V = .25.
Results
Every participant but two correctly noted that they were in the group into which they
were placed; these two participants did not complete the item, were both in the nonmembership/negative reference condition, and were excluded from the analyses reported. The
items intended to measure the desirability of membership in the different groups initially were
subjected to an eight-item, two-factor confirmatory factor analysis. The centroid method was

27

used to estimate factor loadings, and the internal consistency and parallelism theorems (Hunter &
Gerbing, 1982) were analyzed; the factor loadings were used to calculate predicted correlations,
which were compared to the observed inter-item correlations in order to estimate residual error.
Overall, the error was small, root mean square error (RMSE) = .05. In addition, none of the
residual errors were statistically significant. Thus, all items were retained.
The items were averaged to form an index for desirability of membership in the low-trait
group, M = 2.81, SD = 0.78, αSI = .86, inter-item correlations = .59 – .69, 95% CI [2.67, 2.96],
and an index for desirability of membership in the high-trait group, M = 3.79, SD = 0.63, αSI =
.81, inter-item correlations = .36 – .60, 95% CI [3.67, 3.90]. A paired-samples t-test indicated
that the participants rated membership in the high-trait group as more desirable than membership
in the low-trait group, t (110) = -8.71, p < .05. In addition, on average, membership in the lowtrait group was rated as statistically significantly below the midpoint of the scale, t (111) = -2.57,
p < .05, and membership in the high-trait group was rated as statistically significantly above the
midpoint, t (110) = 13.06, p < .05.
After testing for any statistically significant differences among the outcome measures
depending on the characteristic (discerning versus circumspect), ts (110) = -0.10 – 0.95, ps > .05,
each of the measures of adoption of normative behavior (i.e., rated preference of being friends
with Taylor, money bet) was subjected to a 2 (membership/no membership) by 2
(positive/negative reference) ANOVA test. The scores for the preference rating of Taylor were
approximately normally distributed and had a mean of 2.12 (SD = 0.92, Mdn = 2, range: 1 – 4).
There were no statistically significant main effects for the membership, F (1, 110) = 0.59, p >
.05, r = .03, or reference factors, F (1, 110) = 2.57, p > .05, r = .06; there also was no statistically
significant interaction effect, F (1, 110) = 0.05, p > .05, r = .01. The scores for the amount of

28

money bet in the second task were approximately normally distributed and had a mean of
$269.23 (SD = 99.30, Mdn = 250.00, range: 0.00 – 500.00). There were no statistically
significant main effects for the membership, F (1, 110) = 0.00, p > .05, r = .00, or the reference
factors, F (1, 110) = 0.65, p > .05 r = .03; there also was no statistically significant interaction
effect, F (1, 110) = 1.61, p > .05, r = .04. The means and standard deviations for each outcome
measure in each condition are reported in Appendix M.

29

DISCUSSION
Unlike the initial studies, the lab experiment did not give evidence for the model
presented. Although the experimental induction appeared to be effective—all but two
participants correctly identified the group into which they were placed and participants tended to
rate membership to the high-trait group as more desirable than membership to the low-trait
group—the predicted effects ultimately were not observed. Some aspects of the experimental
design potentially could have led to this.
First, it is possible that participants were unconvinced by the information presented about
the supposed personality characteristic. More specifically, when told that people who are high on
the characteristic tend to be evaluated in various ways (i.e., as creative, likable, and charismatic),
at least some participants, regardless of the experimental condition, may have disagreed with the
information. In fact, on the debriefing form, some participants noted being taken aback at being
“told” how they are. If this is the case, the subsequent normative messages about how different
types of people tend to behave might have been overridden by participants’ existing notions
about their own personality traits. In other words, participants’ disbelief about what the bogus
personality measure had to say about them might have worked against the effectiveness of the
normative messages.
Second, it is possible that the goal of controlling for extraneous variables other than the
membership and reference factors actually worked against this test of the model. Although the
experimental inductions appeared to have worked, it is plausible that the checks assessed
participants’ perceptions of the desirability of membership within the groups, but not their
individual desire to be affiliated with the high-trait group per se. If this is the case, then on

30

average, motivations to conform to the normative messages would have been relatively low for
all participants regardless of the condition into which they were placed.
Third, it might be the case that the experiment artificially created some form of
conflicting roles (Goode, 1960). Role strain can occur when the normative expectations of an
individual person within one group differ from that person’s expectations within some other
group. In the lab experiment presented here, many participants are faced with messages about
how they are expected or not expected to be (i.e., the “high” group is evaluated in a number of
positive ways) and how they would be expected or not expected to act (in particular, to be low on
risk-taking). However, it is possible that such normative expectations differ from those
experienced by these people in their day-to-day lives. This might help point to why the
participants ultimately did not act in the predicted manner.
Some noticeable limitations should be noted about this experiment. First, though the
model should robustly apply to many types of normative influence episodes if it is valid, the
primary focus here was on ICNI. As it stands, given the definition of ICNI presented (i.e.,
communication of information about normative behavior during an interpersonal interaction that
is perceived as signaling behavior that group members are expected to adopt), the experimental
procedures did not map neatly onto this focus. What is more, the normative messages did not
come from a member of the group that the messages concerned; rather, they came from a neutral
authority (i.e., the experimenter). This suggests a boundary condition: The model likely applies
solely when normative influence comes from group members themselves. It is reasonable to
argue that ICNI messages from someone who is involved with the group—who is within the
group—carry more force than such messages from neutral or non-member sources. Second, the
sample size of this study was not as large as desired (for the analyses, N = 112). However, in

31

spite of this, these results suggest that the predicted effects would not have been found even if
the sample had contained a larger number of participants.
The series of studies reported here ultimately are mixed with regards to the validity of the
presented conceptual model. The first preliminary study asked respondents to retrospectively
report on a recent time when they experienced ICNI; in this study, the results were consistent,
albeit weakly, with the predictions of the model. The second preliminary study asked
respondents to report on their expectations of how someone who received ICNI (under different
membership/reference configurations) would behave; the results again were consistent with the
predictions, and the effects were stronger than in the first preliminary study. The primary study
was a lab experiment in which participants were experimentally placed into either a desirable or
undesirable group and then given normative messages about either the desirable or undesirable
group, thus simulating the different membership/reference configurations; however, unlike the
preliminary studies, the results were not consistent with the model’s predictions.
A follow-up lab experiment might look markedly different from the one undertaken here.
For instance, as noted, the message source should be a member of the group itself, rather than a
third-party such as an experimenter. Perhaps a confederate who is from one of the group types
(membership/positive reference, membership/negative reference, etc.) would serve as a more
suitable source of an ICNI message. Or perhaps the cover story and primary method could be
altered; rather than begin with the false feedback technique, participants could be asked to selfselect into a group and then the desirability of the group could be experimentally controlled in
order to prime the desired group types.
Although it cannot be firmly concluded that the predictions are valid, the results do
suggest that they deserve further consideration. Such consideration might question the boundary

32

conditions of the model as well as the extent to which a true experiment allows for an ideal test
of the predictions. One potential explanation for the mixed findings is that the predictions apply
primarily to naturally existing groups, or perhaps to groups with which one is familiar. Both
preliminary studies (where the predicted effects were observed) involved groups with which
respondents directly or indirectly had experience; however, the primary study (where the
predicted effects were not observed) involved artificially created groups with which participants
likely had no experience. Given this rationale, a field experiment might be a more apt method by
which to test the predictions. If comparable naturally occurring groups could be found (i.e.,
existing groups that are similar in composition) that fell into the varying membership/reference
configurations, normative messages about public behavior could be implemented and (non)group
members’ behaviors subsequently observed; this would yield a less sterile environment in which
to investigate the roles of the membership and reference concepts during the normative influence
process.

33

CONCLUSION
The primary predictions of this investigation were that conformity to normative
influence—and to interpersonally communicated normative influence in particular—would be a
function of two major factors: Membership within the group (i.e., membership versus nonmembership) and the reference status of the group (i.e., positive reference versus negative
reference). Although two preliminary studies gave evidence for these predictions, the primary
study did not. Further tests are warranted using a wide range of methods, one of which (i.e., field
experiments) could prove especially promising for attempts to replicate the findings of the two
preliminary studies.

34

NOTES

35

NOTES

1

It has been argued persuasively that descriptive and injunctive norms often work
together—either complementing or contradicting each other—and thus that it is important to
understand both types (Cialdini et al., 1990). However, for the sake of this consideration of
normative influence, focus is placed primarily on descriptive norms.
2

The membership concept is treated as a dichotomous dimension here. However, an
interesting question arises as to whether it could be treated in a continuous fashion: In addition to
being a full-time group member or being in a group none of the time, can people be a member of
a group some of the time or most of the time? This question is not answered by this investigation
but warrants further consideration.
3

Across these studies, one often-considered concept in social influence research—source
credibility—is ignored. In the hypothetical scenarios study and the lab experiment, this is held
constant. Also in these latter two studies, the potential role of injunctive norms also is explicitly
controlled.
4

Given that the hypothetical scenario involved a man and a male-only social group, the
potential for biological sex of the respondents to alter the nature of the results also was
considered. A 2 (biological sex) by 2 (membership/non-membership) by 2 (positive/negative
reference) ANOVA was conducted on each of the outcome measures. For all tests, there were no
statistically significant main effects or two-way interaction effects regarding biological sex of the
respondent.
5

The two characteristics were chosen following a pretest in which online survey
respondents at the same university (N = 283) were asked to report, on a 7-point scale, whether
they would characterize themselves as “very high” or “very low” on a large set of personality
traits. The characteristics of “discerning” and “circumspect” were chosen because their
distributions were clustered around the midpoint of the response scale, Ms = 4.33 – 4.42, SDs =
1.14 – 1.29. Given these responses, it seemed probable that respondents could be believably,
albeit experimentally, designated as “high” or “low” in these characteristics. This pretest was
completed two months prior to the pilot study.
6

Although participants were told that, in a broad social sense, people who are in the
high-trait group are more highly valued and evaluated compared to low-trait groups, one major
limitation of this design is that the specific target groups (i.e., high-trait vs. low-trait) themselves
do not hold sanctioning power. In other words, a major aspect of group influence is the potential
for negative social sanctions to be levied for noncompliance (Festinger, 1953), and this
experimental design does not completely account for this.

36

7

The protocol of this experiment has no dynamic component because it precludes
normative change. How normative behaviors and influence shift over time is an important
question; however, for the purposes of this investigation, where the goal is to assess the validity
of a novel model, the dynamic nature of normative influence is simplified.

37

APPENDICES

38

Appendix A
Table A.1
Examples of Groups that are Positive Reference or Negative Reference and in which People
Hold Membership or Not
Positive Reference

Negative Reference

Membership

A campus sorority or fraternity
that one has dreamed of joining
since becoming a student.

A work group that has been put
together by an outside party, in
which a person must work with
unproductive others.

No Membership

An “Honors Society” group that a
person has just discovered and
wants to join.

An opposing political party.

39

Appendix B
Table B.1
Examples of Normative Influence Messages Reported by Respondents in Study 1
Group Type

Normative Influence Message

Political Party

[He] told me the ins and outs of the health care system and how [it] will
destroy the quality of care we receive today and…it would be very
appropriate to vote against Obama this next election.

Friendship
Group

Hey man, everyone that hangs out with our group drinks and smokes.
We don't want to pressure you and all but we are being real with you by
telling you this.

Sorority

Make sure to wear your letters around campus now. Also, you need to
always look presentable and put together while wearing your letters or
anything that has our house name on it.

40

Appendix C
Table C.1
Results of the Moderated Multiple Regression Analysis
b

SE

β

t

sr

Membership

0.37

0.15

.13

2.40*

.12

Reference

0.61

0.09

.60

6.50*

.31

Membership X Reference

-0.23

0.11

-.18

-2.06*

-.10

2

2

Overall model: F (3, 319) = 39.55*, R = .27, adj. R = .26
Note. b = unstandardized regression weight; SE = standard error of the unstandardized regression
weight; β = standardized regression weight; t = t-value for the significant test of the standardized
weight; sr = semipartial correlation. Reference group scores were mean-centered. Membership
group scores were dummy coded (0 = non-membership group). Outcome variable = Conformity
scores. *p < .05. N = 323.

41

Appendix D
Figure D.1
The Membership-Reference Interaction for Conformity in Preliminary Study 1
5.00
4.50
4.00
3.50
Member

3.00

Non-Member

2.50
2.00
1.50
1.00
1SD below

Mean

1SD above

42

Appendix E
Figure E.1
Predicted Conformity Following Normative Influence in Preliminary Study 2
0.80
0.60
0.40
0.20
0.00
-0.20

Membership

-0.40

Non-Membership

-0.60
-0.80
-1.00
-1.20
-1.40
Negative Reference

Positive Reference

43

Appendix F
Please respond “yes” or “no” to the following statements about yourself.
1. My decisions are based more on opportunity than on following set plans.
a. Yes
b. No
2.

My behavior can be easily affected by my mood.
a. Yes
b. No

3. I try to think three moves ahead, rather than “living in the moment.”
a. Yes
b. No
4. I prefer individual work in class, rather than group work.
a. Yes
b. No
5. I find it difficult to sympathize with complete strangers.
a. Yes
b. No
6. I am rarely affected in a negative way by news stories.
a. Yes
b. No
7. I try to be vocal when I think someone is making a bad decision.
a. Yes
b. No

44

Appendix G
It is approximately 10 a.m. in mid-August. Returning from a recent trip in a twin engine plane,
you are flying over the Sonora Desert in the southwestern United States. You notice that the
immediate area is quite flat and, except for occasional barrel and saguaro cacti, appears to be
barren. The last weather report indicated the temperature would reach 110 degrees that day.
The pilot just indicated that you were approximately 70 miles from a mining camp, which is the
nearest known habitation, when the plane begins to experience difficulty. The plane crash lands
in the Sonora Desert, burning up. Before the plane caught fire your group was able to salvage 15
items. While salvaging the items, you notice that the impact killed both the pilot and the co-pilot,
but your group survived. You survived the crash, now you must survive the desert.
The following list describes the 15 items you salvaged. Rank the items according to their level of
importance. Specifically, enter a number between 1 and 15 next to each item, where "1" indicates
the most importance to your team's survival, and "15" indicates the least importance.
ITEM

Your Ranking

Flash light (4 battery size)
Jackknife
Sectional Air Map of Crash Area
Plastic Raincoat (large size)
Magnetic Compass
Bandage Kit with Gauze
.45 Caliber Pistol (loaded)
Parachute (red and white)
Bottle of Salt Tablets (1000)
1 liter of water per person
Book Entitled: ‘Edible Animals of the Desert’
2 Pairs of Sunglasses per Person
2 liters of 180 Proof Vodka
1 top coat per Person
Cosmetic Mirror

45

Appendix H
You currently have a sum of $1,000 that you want to donate. You have searched and you have
found three non-profit organizations that you can donate to.
The first is a Food Bank that provides food to homeless individuals.
The second is an Animal Shelter that finds homes for abandoned cats, dogs, and other animals.
The third is a Job Placement program that helps unemployed workers find new jobs.

There are only two restrictions on how you donate: Your donations must add up to $1,000 and
you must donate at least $100 to each. Beyond this, you can donate your money in any way you
want.

Food Bank: $____________

Animal Shelter: $____________

Job Placement: $_____________

46

Appendix I
1. Which describes how the initial measure placed you?
a. Relatively LOW in the characteristic
b. Relatively HIGH in the characteristic
2. Consider people who are LOW in the characteristic. In your opinion, to what extent is
being a member of that group…
(semantic differential items with five-point scaling)
a. desirable – undesirable
b. attractive – not attractive
c. enjoyable to be a member of – not enjoyable to be a member of
d. preferable – not preferable
3. Consider people who are HIGH in the characteristic. In your opinion, to what extent is
being a member of that group…
(semantic differential items with five-point scaling)
a. desirable – undesirable
b. attractive – not attractive
c. enjoyable to be a member of – not enjoyable to be a member of
d. preferable – not preferable

47

Appendix J
() denotes conversational/informal dialogue; [[]] denotes administering of materials
(I have this list of stuff I’m supposed to tell you, so I’m just going to read straight from this.)
Thank you for agreeing to participate in this study. First, I need you to read through and
complete this consent form.
[[ADMINISTER CONSENT FORM]]
Now that you have formally agreed to participate, we will begin.
The purpose of this study is to help us assess a new measure of a personality characteristic,
which I will tell you more about after you have completed this first set of items.
(So we’re doing this at a bunch of schools, and some of them can’t do online surveys.)
I’m going to step out of the room while you complete this. When you are done, please open the
door and I will come in to explain the next part.
[[ADMINISTER PERSONALITY ITEMS and leave the room]]
This questionnaire that you completed is actually a new measure of a personality characteristic
known as [“discerning”/“circumspect”]. The purpose of this study is to try to validate this
measure—in other words, we want to see if it is an accurate measure or not. We may also ask
you to interact with other people who score the same way on this measure at a later time.
(So right now, I’m going to tally your scores and see whether you are high or low on this
dimension. While I do that, could you do this next set of questions?)
[[ADMINISTER TASK CHECKS AND DEMOGRAPHIC ITEMS and “tally” scores]]
(OK…oh, now I’m supposed to tell you about the goals of the study.)
In general, research on this personality characteristic has shown that people who are high on this
characteristic tend to be highly valued at a societal level, and people who are high in the
characteristic actually tend to be evaluated by others as fairly creative, likeable, and charismatic.
(So like from your scores, it looks like you are [high/low] on the dimension. Umm…)

48

This measure you filled out has received some criticism. Some researchers do not think it is a
very good measurement tool. The goal of this study is essentially to assess how valid of a
measure it is. What we are going to have you do now is complete a few decision-making
exercises on paper. We are going to use your choices on these exercises to assess whether the
questionnaire you completed matches them.
Although your scores on the initial set of items might suggest you are high or low, what you do
on these tasks will either give more evidence that you are that way, or show that you may
actually be the opposite. Personality researchers use these exercises frequently because those
who are high and low on this characteristic tend to act in relatively consistent ways.
(So it’s kinda weird, like, I guess on this first one, people who are [high/low] tend to rank
TAYLOR the best, but I guess it’s a good thing for those researchers, or something? Umm…)
First, we are going to have you complete a “friend preference” exercise. You will find the
instructions on the paper. I’m going to step out of the room while you complete this. You will
have about 5 minutes. Once you are finished, please open the door.
[[ADMINISTER FIRST EXERCISE and step out of the room]]
Next, we are going to ask you to complete a risk-taking exercise. The reason we asked you some
questions earlier was to help us statistically control for your sensation-seeking tendencies.
(So, like, in other words, like—people who are [high/low] tend to be very careful and not willing
to take big risks, but for a person like that, we’d need to make sure their existing tendencies don’t
affect the results. Anyway…)
I am going to step out of the room while you complete this. You will have about 5 minutes.
Again, once you are finished, please open the door.
[[ADMINISTER SECOND EXERCISE and step out of the room]]
Next, we would like to ask you to report on your opinions about these two different types of
groups: People who are high in the characteristic and those who are low in the characteristic,
although you may not have a firm opinion at this time. This will help us to further control for
extraneous variables statistically. (So it’s like another analysis thing.)
I am going to again step out of the room. Once you are finished, please open the door.

49

[[ADMINISTER INDUCTION CHECK and step out of the room]]
There is one final thing we would like you to do. I will remain in the room while you respond to
these questions.
[[ADMINISTER DEBRIEFING ITEMS]]
The study is now over. Just so you know, there is no such measure of this personality
characteristic, so you are not actually high or low on this characteristic. Please do not talk to any
students about this until the semester is over.

50

Appendix K
st

th

Please read the attributes of the four people below and rank, from 1 to 4 , who you would
*most prefer* to be friends with. In other words, rank “1st” the person who you would most like
to be friends with, and rank “4th” the person who you would least like to be friends with.

SAM:
•
•
•
•

Magnetic
Cautious
Intelligent
Outgoing

TAYLOR:
• Fun-loving
• Hard worker
• Charming
• Careful

JORDAN:
• Meticulous
• Sociable
• Charismatic
• Clever

ADRIAN:
• Witty
• Thorough
• Extraverted
• Polite

Please make your rankings below:

SAM: ________

TAYLOR: ________

JORDAN: ________

51

ADRIAN: ________

Appendix L
Consider the following situation:
You are playing poker against three of the best players in the world. The game is nearing its end,
and in the current hand, you are only going up against one of the players. You are about 50%
confident that you are going to win the next hand. You have a total of 500 dollars left that you
can bet, and you are very sure that the other player will also bet the same amount as you. The
winner of this hand will take all of the money that is on the line.
Below, please indicate the amount of money that you are most comfortable with betting. You
may bet any amount between $0 and $500.

I would bet $_____________

52

Appendix M
Table M.1
Means and Standard Deviations for each Outcome Variable in each Experimental Condition
Friend Preference Rating

Money Bet

Positive
Reference

Negative
Reference

Excluding
Reference

Positive
Reference

Negative
Reference

Excluding
Reference

Membership

1.90
(0.94)

2.21
(0.96)

2.05
(0.95)

249.31
(110.74)

288.36
(107.40)

268.49
(109.91)

NonMembership

2.07
(0.80)

2.31
(0.97)

2.18
(0.88)

274.14
(75.44)

265.38
(101.51)

270.00
(87.98)

Excluding
Membership

1.98
(0.87)

2.26
(0.96)

261.72
(94.75)

277.30
(104.26)

Note: N = 112; Membership/Positive Reference n = 29; Membership/Negative Reference n = 28;
Non-Membership/Positive Reference n = 29; Non-Membership/Negative Reference n = 26.
Means are presented above the standard deviations, which are noted parenthetically. “Excluding
Reference” = the means for the membership and non-membership groups, ignoring the reference
factor; “Excluding Membership” = the means for the positive reference and negative reference
groups, ignoring the membership factor.

53

BIBLIOGRAPHY

54

BIBLIOGRAPHY

Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision
Processes, 50, 179-211.
Asch, S. E. (1956). Studies of independence and conformity: A minority of one against a
unanimous majority. Psychological Monographs: General and Applied, 70(9), 1-70.
Burleson, B. (2009). The nature of interpersonal communication: A message-centered approach.
In C. R. Berger, M. E. Roloff, & D. R. Roskos-Ewoldson (Eds.), The handbook of
communication science (2nd ed.) (pp. 145-163). Thousand Oaks, CA: Sage
Carcioppolo, N., & Jensen, J. D. (2012). Perceived historical drinking norms and current
drinking behavior: Using the theory of normative social behavior as a framework for
assessment. Health Communication, 27, 766-775.
Cialdini, R. B., & Goldstein, N. J. (2004). Social influence: Compliance and conformity. Annual
Review of Psychology, 55, 591-621.
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 and Social Psychology, 58, 1015-1026.
Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. (2003). Applied multiple regression/correlation
analysis for the behavioral sciences (3rd ed.). New York: Routledge.
Cruz, M. G. (1998). Explicit and implicit conclusions in persuasive messages. In M. Allen & R.
W. Preiss (Eds.), Persuasion: Advances through meta-analysis (pp. 217-230). Cresskill,
NJ: Hampton Press.
Deutsch, M., & Gerard, H. B. (1955). A study of normative and informational social influences
upon individual judgment. The Journal of Abnormal and Social Psychology, 51, 629-636.
Ellemers, N., & Haslam, S. A. (2011). Social identity theory. In P. A. M. Van Lange, A. W.
Kruglanski, & E. T. Higgins (Eds.), Handbook of theories of social psychology, volume II
(pp. 379-398). Thousand Oaks, CA: Sage.
Festinger, L. (1953). An analysis of compliant behavior. In M. Sherif & M. O. Wilson (Eds.),
Group relations at the crossroads (pp. 232-256). Oxford, England: Harper.

55

Goode, W. J. (1960). A theory of role strain. American Sociological Review, 25, 483-496.
Harwood, J. (2010). The contact space: A novel framework for intergroup contact research.
Journal of Language and Social Psychology, 29, 147-177.
Harwood, J., Giles, H., & Palomares, N. A. (2005). Intergroup theory and communication
processes. In J. Harwood & H. Giles (Eds.), Intergroup communication: Multiple
perspectives (pp. 1-17). New York: Peter Lang Publishing.
Hogg, M. A., & Reid, S. A. (2006). Social identity. Self-categorization, and the communication
of group norms. Communication Theory, 16, 7-30.
Hunter, J. E., & Gerbing, D. W. (1982). Unidimensional measurement, second order factor
analysis, and causal models. Research in Organizational Behavior, 4, 267–320.
Kelman, H. C. (1958). Compliance, identification, and internalization: Three processes of
attitude change. The Journal of Conflict Resolution, 2, 51-60.
Kim, M.-S., & Hunter, J. E. (1993). Attitude-behavior relations: A meta-analysis of attitudinal
relevance and topic. Journal of Communication, 43, 101-142.
Lapinski, M. K., & Rimal, R. N. (2005). An explication of social norms. Communication Theory,
15, 127-147.
Lott, A. J., & Lott, B. E. (1961). Group cohesiveness, communication level, and conformity. The
Journal of Abnormal and Social Psychology, 62, 408-412.
Mackie, D. M., & Queller, S. (2000). The impact of group membership on persuasion: Revisiting
“Who says what to whom with what effect?” In D. J. Terry & M. A. Hogg (Eds.),
Attitudes, behavior, and social context: The role of norms and group membership (pp.
135-156). Mahwah, NJ: Lawrence Erlbaum Associates.
McGarty, C., Haslam, S. A., Hutchinson, K. J., & Turner, J. C. (1994). The effects of salient
group memberships on persuasion. Small Group Research, 25, 267-293.
Park, H. S., & Smith, S. W. (2007). Distinctiveness and influence of subjective norms, personal
descriptive and injunctive norms, and societal descriptive and injunctive norms on
behavioral intent: A case of two behaviors critical to organ donation. Human
Communication Research, 33, 194-218.

56

Rimal, R. N. (2008). Modeling the relationship between descriptive norms and behaviors: A test
and extension of the theory of normative social behavior (TNSB). Health
Communication, 23, 103-116.
Rimal, R. N., & Real, K. (2005). How behaviors are influenced by perceived norms: A test of the
theory of normative social behavior. Communication Research, 32, 389-414.
Schultz, P. W., Nolan, J. M., Cialdini, R. B., Goldstein, N. J., & Griskevicius, V. (2007). The
constructive, destructive, and reconstructive power of social norms. Pscyhological
Science, 18, 429-434.
Sherif, M., & Sherif, C. W. (1953). Groups in harmony and tension. New York: Harper.
Sherif, M., & Sherif, C. W. (1964). Reference groups: Exploration into conformity and deviation
of adolescents. New York: Harper.
Shulman, H. C., & Levine, T. R. (2012). Exploring social norms as a group-level phenomenon:
Do political participation norms exist and influence political participation on college
campuses? Journal of Communication, 62, 532-552.
Siegel, A. E., & Siegel, S. (1957). Reference groups, membership groups, and attitude change.
The Journal of Abnormal and Social Psychology, 55, 360-364.
Spears, R., Doosje, B., & Ellemers, N. (1997). Self-stereotyping in the face of threats to group
status and distinctiveness: The role of group identification. Personality and Social
Psychology Bulletin, 23, 538-553.
Stiff, J. B., & Mongeau, P. A. (2003). Persuasive communication (2nd ed.). New York: The
Guilford Press.
Tajfel, H. (1982). Social psychology of intergroup relations. Annual Review of Psychology, 33,
1-39.
Tajfel, H., & Turner, J. C. (1986). The social identity theory of intergroup behavior. In S.
Worchel & W. G. Austin (Eds.), Psychology of intergroup relations (2nd ed.) (pp. 7-24).
Chicago: Nelson-Hall Publishers.

57