3 § :13 "WNWJHIIHWHIWNHIHIJHHWWWW”!!!

THE! S

ZLCC

lilii‘iiljillilﬂiiﬂWl '

1 930

 

LIBRARY

Michigan State
University

 

 

 

This is to certify that the

dissertation entitled

The Impact of Varying Argument Quality
and Minority Size on Minority Influence
and Perceptions of the Minority

presented by

M. Sean Limon

- has been accepted towards fulﬁllment
of the requirements for

Ph . D . degree in Communication

 

MJor professor

Date J w a 000

 

 

MS U is an Afﬁrmative Action/Equal Opportunity Institution 0- 12771

 

...___-_— .-

,__.._.—. -——.

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

 

UUL 24200

IV

 

(\Hﬁ

r" n 1 .3
Di -2<. 01‘0-244

 

 

 

 

 

 

 

 

 

 

 

 

 

woo elem-m.“

THE IMPACT OF VARYING ARGUMENT QUALITY AND MINORITY SIZE ON
MINORITY INFLUENCE AND PERCEPTIONS OF THE MINORITY
By

M. Sean Limon

A DISSERTATION

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

DOCTOR OF PHILOSOPHY
Department of Communication

2000

ABSTRACT
THE IMPACT OF VARYING ARGUMENT QUALITY AND MINORITY SIZE
ON MINORITY INFLUENCE AND PERCEPTIONS OF THE MINORITY
By

M. Sean Limon

Past majority/minority inﬂuence research has found that consistency of a
minority subgroup’s position, number of minority subgroup members, and
argument quality, greatly affect the likelihood of that minority subgroup being able
to inﬂuence the majority subgroup. Research regarding minority size and
argument quality and how it affects minority inﬂuence and perceptions of the
minority has been limited and is addressed in this study. Confederates were
used to vary argument quality and the size of the minority subgroup while
keeping consistency of presenting the minority’s side of the issue constant.
Groups discussed a topic and relevant measures were used to assess the
amount of inﬂuence, perceived competence, perceived correctness, and how
much the minority was liked. Results indicated argument quality and minority
size were important when attempting minority inﬂuence and that the perceptions
fostered by the minority were a function of an argument quality by group size

interaction.

Dedicated to my best friend and wife, Sarah,
and to my Mom and Dad

ACKNOWLEDGEMENTS

The completion of a dissertation is rarely, if ever, the sole accomplishment
of one individual. The completion of this dissertation is not any different. The
people who have helped me complete this dissertation have helped me to reach
a place in my life that I thought I would never be. To them, I am very grateful.

I will begin by thanking Sarah, who is my wife, best friend, conﬁdant,
cheerleader, and counselor. It is she who stuck by my side when there was no
light at the end of the tunnel. It is she that picked me up when I was down. It is
she that encouraged me when I needed encouraged. It is because of her love
and support that made it easier on me to ﬁnish this program, and am at this place
in my life that I have always wanted to be.

My parents have always been my most ardent supporters. Their
encouragement has always been a source of support from which I have relied. It
has not been the sage advice they have given to me over the years, nor the
pearls of wisdom they have passed along, but rather the example they have set
by the way they have lived their lives that has guided me in achieving my goals.
To them, I am forever grateful.

I would like to thank a number of other people. I would like to thank
“Shorty" Sparks for teaching me one of the greatest lessons I have ever been
taught: Laugh, and make fun of yourself. Thank you to Betty for being the little
sister I always wanted. Thanks to both Betty and Craig for being two of the best

colleagues and friends a person could have, and for listening to me complain and

iv

telling me exactly what I needed to hear when I needed to hear it while I was
working on my dissertation. I would like to thank Marge Barkman for sitting me
down in “the chair" to counsel and console me a number of times. A gracious
and heartwarming thanks goes to my cohort; there is not another group of people
that I would have rather gone through it with.

l have a lot of gratitude for my advisor and other committee members who
deserve a special thanks for helping me complete my dissertation. I would like to
thank Frank for demonstrating to me what it means to be a scholar, and for all his
wonderful stories. A heartfelt thanks goes out to my other committee members,
Chuck Atkin, Sandi Smith, and Georgia Chao, for being extremely ﬂexible and

understanding throughout the process of my completing this dissertation.

TABLE OF CONTENTS

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

CHAPTER 1

INTRODUCTION ............................................................................. 1
Minority Inﬂuence .................................................................... 2
Consistency ........................................................................... 3
Number of Minority Subgroup Members and Perceptions ................. 5
Argument Quality ..................................................................... 7
Research Hypotheses .............................................................. 10

CHAPTER 2

METHOD ....................................................................................... 14
Participants ............................................................................. 14
Design .................................................................................. 14
Procedure ............................................................................. 1 5
Instruments ............................................................................ 1 7

CHAPTER 3

RESULTS ...................................................................................... 19
Induction Checks .................................................................... 19
Measures .............................................................................. 1 9
Minority Inﬂuence .................................................................... 20
Perceptions of the Minority Subgroup .......................................... 24

CHAPTER 4

DISUCSSION ................................................................................. 33
Future Research and Limitations ........................................... 37
Conclusion ............................................................................ 39

APPENDIX ..................................................................................... 41

REFERENCES ................................................................................ 56

FOOTNOTES ................................................................................. 60

vi

LIST OF TABLES

Table 1
Conﬁrmatory Factor Analysis for the Perception Variables ........................ 45

Table 2
Mean Differences of Inﬂuence for Each Experimental Condition
Compared to the Control Group Using A One-Tailed Dunnett Test .............. 46

Table 3
Cell Means for the Amount of Inﬂuence Based on the Minority Size by
Argument Quality Conditions at the Individual Level of Analysis .................. 47

Table 4
Cell Means for the Amount of Inﬂuence Based on the Minority Size by
Argument Quality Conditions at the Group Level of Analysis ...................... 48

Table 5
Correlations for Inﬂuence and Perceptions at the Individual Level
of Analysis ....................................................................................... 49

Table 6

Cell Means for the Perceived Competence of the Minority Subgroup

Based on the Minority Size by Argument Quality Conditions at the

lndivid ual Level of Analysis ................................................................. 50

Table 7
Cell Means for Perceived Competence Based on the Minority Size by
Argument Quality Conditions at the Group Level of Analysis ...................... 51

Table 8
Cell Means for Perceived Correctness Based on the Minority Size by
Argument Quality Conditions at the Individual Level of Analysis .................. 52

Table 9
Cell Means for Perceived Correctness Based on the Minority Size by
Argument Quality Conditions at the Group Level of Analysis ...................... 53

Table 10
Cell Means for how Liked was the Minority Subgroup Based on the Minority
Size by Argument Quality Conditions at the Individual level of Analysis ........ 54

vii

Table 11
Cell Means for how Liked was the Minority Subgroup Based on the Minority
Size by Argument Quality Conditions at the Group Level of Analysis ........... 55

viii

CHAPTER 1

INTRODUCTION

The purpose of this study is twofold: to identify what affect varying
argument quality and minority subgroup size has on minority inﬂuence and the
perceptions of the minority. The arguments used in this study are either strong
or weak and the perceptions of being competent, correct, and liked are
examined. In the following sections, behavioral consistency and its contribution
to minority inﬂuence research will be discussed, as will past research that
investigated minon‘ty size and argument quality. After the review of pertinent
research, hypotheses that are tested in this study are presented. The method for

testing the hypotheses is explained and the results of this study are discussed.

LITERATURE REVIEW

Minolty' Inﬂuence

Recently, a resurgence in the study of minority inﬂuence in small groups
has occurred (i.e., Gebhardt 8. Meyers, 1994; Meyers, Brashers, & Jerzak, 1998;
Thameling 8. Andrews, 1992, 1999). The reason for renewed interest can be
credited to small group researchers desire to identify the different kinds of
communication that help the minority subgroup to inﬂuence the majority
subgroup as well as the perceptions different kinds of communication foster.1
For example, Gebhardt and Meyers (1995) demonstrated that a minority gives
the perception of being consistent and more inﬂuential when they use statements
that show they support their fellow subgroup member. Thameling and Andrews
(1999) demonstrated that a minority using evidence along with their opinions
were perceived as more interpersonally and task attractive.

Much of the research investigating how a minority inﬂuences a majority
has been conducted using the minority inﬂuence paradigm (Moscovici, 1976).
The minority inﬂuence paradigm is founded on the belief that minorities are not
just targets of inﬂuence, but that they are active agents of inﬂuence capable of
inﬂuencing a group’s opinion (Moscovici, 1976). Whereas once minorities were
seen as receivers of inﬂuence, this paradigm sees them “as potential inﬂuence
emitters and norm originators” (Moscovici, 1976, p. 67). The minority inﬂuence
paradigm also suggests that the behavioral styles of the minority affect how they

are perceived as well as their ability to inﬂuence the majority. The way in which

the minority behaves informs the majority about the minority’s motivation,
certainty, competence, and so on. In sum, the minority inﬂuence paradigm
asserts that the minority is an active agent of inﬂuence and that certain behaviors
aid the minority in inﬂuencing and fostering certain perceptions.

As already noted, the kind of communication used by a minority has been
found to have an impact on how the minority is perceived and how inﬂuential they
are. Nonetheless, much of the minority inﬂuence research has concentrated on
two variables: behavioral consistency and minority subgroup size. One purpose
of this paper is to identify how varying a form of argumentative communication,
speciﬁcally argument quality, as well as varying the minority subgroup’s size,
affects minority inﬂuence. A second purpose of this investigation is to identify the
perceptions the majority has of the minority as a result of the minority’s
arguments and size. In so doing, a better understanding of minority inﬂuence will
be gained.
megs!

Moscovici and colleagues are typically cited as providing the ﬁrst
indications that a minority subgroup is capable of inﬂuencing a majority subgroup
(Moscovici, & Faucheux, 1972; Moscovici & Lage, 1976; Moscovici, Lage, &
Faucheux, 1969). They argued that an important determinant of inﬂuence by a
subgroup was consistency of presentation. Consistency can be deﬁned as “the
maintenance of a given position in interaction by subgroup members” (Gebhardt
& Meyers, 1995, p.151). Of all the behavioral styles that have been investigated,

consistency has received the most attention. Indeed, consistency has evolved

as the cornerstone of minority inﬂuence and is considered to be a necessary, but
not sufﬁcient, component of the inﬂuencing process (Maas & Clark, 1984).
Seminal minority inﬂuence research by Moscovici and colleagues demonstrated
that a minority subgroup could inﬂuence a majority subgroup. to agree with some
of their opinions as long as the minority was consistent in their stance (Moscovici,
& Faucheux, 1972; Moscovici & Lage, 1976; Moscovici, Lage, & Faucheux,
1969). Groups that were inconsistent had little, if any chance at all, of inﬂuencing
the majority.

Subsequent minority inﬂuence research investigating the effect a
consistent minority has upon the majority has continually demonstrated that
when a minority is consistent, they can inﬂuence the majority (e.g. Arbuthnot &
Wayner, 1982; Gebhardt & Meyers, 1995; Nemeth & Wachtler, 1974; Nemeth,
Wachtler, & Endicott, 1977). Meyers and colleagues identiﬁed some speciﬁc
communicative behaviors used for minority inﬂuence (Gebhardt & Meyers, 1994;
Meyers, Brashers, & Jerzak, 1998). Using decision-making groups, they
demonstrated that consistent subgroups, whether a majority or minority, were
more inﬂuential than inconsistent subgroups (Gebhardt & Meyers, 1995).

Further, Meyers et al. (1998) found that winning subgroups, subgroups that had
their ideas adopted as the ﬁnal decision, used more convergence-seeking
statements, messages that are characterized by one’S own subgroup members
as agreement and acknowledgement statements. Hence, members of all
winning subgroups tended to state that they agreed with or acknowledged a point

by their own subgroup member, demonstrating that consistently supporting the

stance of fellow subgroup members helps to inﬂuence other subgroups. Finally,
Meyers, Seibold, and Brashers (1991) found that agreement statements were
used by group members to reinforce one another’s position and to adapt and
adopt a Chosen stance by group members.

The act of consistently supporting other group members is important for
minority inﬂuence. Nonetheless, as noted above, consistency is a necessary, but
not sufﬁcient component for inﬂuencing the majority. The size of the minority
subgroup has been identiﬁed as another mechanism for increasing minority
inﬂuence.

N_L_I_rr_1ber of Minority Subgroup Members and Perceptions

A second ﬁnding from the minority inﬂuence literature is that it typically
takes at least two minority members to be inﬂuential (e.g., Moscovici 8. Lage,
1976; Moscovici, Lage, 8 Faucheux, 1969; Morris & Miller, 1975; Tanford &
Penrod, 1984; Nemeth, Wachtler, & Endicott, 1977). For example, Moscovici
(1976) demonstrated that when trying to inﬂuence the majority, having only one
member as the minority trying to inﬂuence a majority of three did not produce
signiﬁcantly more inﬂuence compared to the control group whereas a minority of
two trying to inﬂuence a majority of four produced signiﬁcantly more inﬂuence
compared to both the control group and the minority of one. A number of
different reviews, both qualitative (Maas & Clark, 1984) and quantitative (Tanford
and Penrod, 1984; Wood, Lundgren, Ouelette, Busceme, & Blackstone, 1994),
concluded that as a minority subgroup increases in size, it tends to be more

inﬂuential. To help explain minority inﬂuence effects Tanford and Penrod (1984)

developed the Social Inﬂuence Model (SIM) using computer simulations. The
SIM asserts that the two most important components in achieving minority
inﬂuence are the size of the minority and consistency of presentation. Although
size is important, the SIM dictates that the minority’s ability to inﬂuence the
majority levels off at three or four members, thus adding minority members
beyond three or four does not increase the likelihood of inﬂuencing the majorityz.
As previously noted, the minority inﬂuence paradigm maintains that the
behavioral style of the minority subgroup affects how the minority is perceived,
and that the perceptions fostered by using a particular behavioral style aid the
minority with inﬂuencing the majority. Surprisingly, past research has
concentrated on identifying the perceptions a minority creates as a result of
employing a particular behavioral style, and not how those perceptions help the
minority to inﬂuence the majority (for a review, see Maas & Clark, 1984). A
notable exception was a study by Nemeth et al. (1977) where they employed the
behavioral style of consistency and varied the size of the minority subgroup,
having anywhere from one to four members trying to inﬂuence a majority
subgroup with majority size held constant at six. Findings indicated that three
members were more inﬂuential than one or two, but not Signiﬁcantly more
inﬂuential than four members in the minority. Additionally, a linear trend was
identiﬁed such that as the minority’s group size increased, so did the perceptions
of them being competent, correct, and liked. These perceptions were also found

to help the minority with inﬂuencing the majority.

In the Nemeth et al. (1977) study, that the minority subgroup fostered
perceptions of competence, correctness and being liked, and that those
perceptions aided the minority with inﬂuencing the majority is consistent with
predictions from the minority inﬂuence paradigm. What is striking about these
ﬁndings is that not only did the behavioral style of consistency contribute to these
perceptions, but the size of the minority greatly affected these perceptions as
well. Hence, a minority subgroup that is consistent in presenting their issue and
larger as opposed to smaller can foster the perceptions of being competent,
correct, and liked, which in turn helps them to inﬂuence the majority subgroup.

In conclusion, the importance of consistency of presentation and the size
of the minority have been well established when the minority subgroup is trying to
inﬂuence the majority subgroup. Further, increasing minority size also helps with
fostering perceptions that can help the minority to inﬂuence the majority.
Nonetheless, to a lesser extent have been investigations examining the different
kinds of communication a minority might use to inﬂuence the majority. Of
concern for this study is argumentative communication, or more speciﬁcally,
argument quality.

Argument Qualig

Argument quality refers to the strength or weakness of an argument. The
strength or weakness of an argument depends on the receiver's perception
(Allen & Burrell, 1992). If a person ﬁnds the arguments compelling, then those
arguments are deﬁned as strong. If not, then the arguments are deﬁned as

weak. Equating strong arguments with those perceived as compelling, and weak

arguments as those perceived as not compelling might be seen as circular.
Nevertheless, deﬁning argument strength in this manner does not imply
circularity of effect. Strong arguments are not always found to be more
persuasive than weak arguments, for example3 (Petty, Cacioppo, & Goldman,
1981)

Boster and Mayer (1984) examined the impact of argument quality on
majority and minority subgroups. They had participants read a ﬁctitious script of
a group discussion pertaining to which of two jobs, one that had high risk
associated with it or one that was more cautious, a recent college graduate
should take. Their results indicated that the majority subgroup inﬂuenced the
perception of argument quality such that both the strong and weak arguments
used by the majority were seen as more inﬂuential compared to any of the
arguments used by the minority subgroup. Put another way, regardless of
whether the majority used strong or weak arguments, they were always seen as
more inﬂuential than the minority. Subsequent research by Garlick and Mongeau
(1993) found similar results.

A different study by Garlick and Mongeau (1992) using the same methods
as the previous two studies reported a different ﬁnding than the two previous
studies. For this study, arguments intended to be strong were perceived as
strong and arguments intended to be weak were perceived as weak, regardless
whether the minority or majority was using them. Consequently, strong
arguments employed by the minority tended to be more inﬂuential than when the

minority used weak arguments.

The inﬂuential effects argument quality had in the Garlick and Mongeau
(1992) study were different than the non-inﬂuential effects obtained in the other
two argument quality studies. One way this study differed from the other two
studies was that rather than have only one member represent the minority
viewpoint, they had an additional condition where there were three members in
the minority subgroup. Conversely, previous studies had only one minority
member. These data demonstrated that increasing the size of the minority
subgroup attenuated the majority's effect concerning argument quality. Further,
their data suggested that as the number of people in the minority subgroup
increased, people began to pay more attention to the minority subgroups’
arguments, diminishing the impact of the majority subgroup. When the minority
subgroup increased in size, participants attended to the minority position and
made it possible for the minority using strong arguments to enact inﬂuence.

It is not surprising that having more than one minority subgroup member
helps with getting others to attend to the minority’s message. When an individual
stands against the majority, the view or arguments of the individual is often
rejected and/or not considered (Moscovici, 1985; Moscovici & Lage, 1976).
Having two or more minority members makes it more difﬁcult for the majority to
reject the minority outright. Further, a minority subgroup induces divergent
thinking, which causes the majority to consider the position of the minority
(Nemeth, 1986; Nemeth & Kwan, 1985; Nemeth, Mayseless, Shaman, & Brown,
1990). When divergent thinking occurs, majority subgroup members are

motivated and willing to discuss with the minority their position and as a result,

strong arguments are more likely to be inﬂuential than weak arguments. This
thinking is consistent with the heuristic-systematic model, which stipulates that
when a person is willing and able to systematically process arguments, strong
arguments will be more persuasive than weak arguments (Chaiken, 1987).
Because divergent thinking motivates the majority to consider the position of the
minority, systematic processing is likely to occur and hence, strong arguments
are likely to be more inﬂuential than weak arguments.

Research Hygotheﬁ

 

Past studies have tended not to investigate the different kinds of
communication that could affect minority inﬂuence. As demonstrated previously,
the kinds of communication used can impact both minority inﬂuence and
perceptions of the minority. Persuasive messages are important in group
decision-making and arguments voiced by group members are an important
inﬂuencing factor for the ﬁnal decision. To understand how argument quality
affects minority inﬂuence, this study investigates argument quality by using actual
group discussions.

For this study both argument quality and group size are varied to identify
the impact they have on minority inﬂuence. This study is different from past
studies in that it investigates argument quality and minority subgroup size holding
constant the behavioral style of consistency. As discussed previously the
importance of consistency in minority inﬂuence has been well established.
Garlick and Mongeau (1992) demonstrated that minority size and argument

quality can impact the minority’s ability to inﬂuence, however, how behavioral

10

consistency works in conjunction with both variables is untested. More
speciﬁcally, allowing groups to discuss a topic while varying argument quality and
minority subgroup size, and allowing a minority subgroup to reinforce their
position (consistency) through interaction, will increase our knowledge in this
area.

A number of speciﬁc predictions are made for this study. A larger
minority, compared to a smaller minority, will be more inﬂuential. Past research
has demonstrated that larger minorities achieve more inﬂuence than smaller
minorities and the same result is expected in this study (e.g., Maas & Clark,
1984; Nemeth et al., 1977; Tanford & Penrod, 1984). Furthermore, Nemeth et al.
(1977) demonstrated that when holding consistency constant, as the size of the
minority increased, the majority’s perception of the minority being competent,
correct and liked increased and these perceptions also helped them to inﬂuence
the majority. Thus, for this study, a larger minority subgroup will be perceived as
more competent, correct and liked compared to a smaller group and these
perceptions will be positively associated with inﬂuence.

It was demonstrated that when the minority subgroup increased in size,
people with a counterattitudinal view started to attend to the minority’s arguments
(Garlick & Mongeau, 1992). When there was only one in the minority subgroup,
people with a counterattitudinal view did not pay much attention to the minority
subgroup’s arguments. Increasing the size of the minority beyond one increases
the likelihood that the majority subgroup will consider their arguments. It follows

then that a minority subgroup larger than one will motivate the majority subgroup

11

to consider the minority subgroups” arguments, and, as demonstrated by Garlick
and Mongeau (1992), in the process of listening to the minority’s arguments,
stronger arguments will be more inﬂuential than weaker arguments.

Overall, strong arguments are expected to be more inﬂuential than weak '
arguments, however, the perceptions that a large consistent minority fosters can
enhance their ability to inﬂuence the majority even when they are using weak
arguments. Past research investigating minority Size indicates that a larger
minority (i.e. four members) subgroup tends to be more inﬂuential than a smaller
minority subgroup (is. two members), relative to majority size. Moreover, the
more widely an opinion is shared, the more likely it is viewed to be correct (Maas
& Clark Ill, 1984). Nemeth et al. (1977) demonstrated that increasing the size of
the minority helped to create the perception of being correct and certain and
these perceptions aided the minority subgroup in inﬂuencing the majority
subgroup. Thus, the larger minority subgroup, as long as they are consistent,
have an advantage over the smaller minority subgroup because the larger
minority is perceived as being more correct due to their larger numbers. In a
discussion it may be easy to dismiss a few people who disagree with the
majority, but when more people share an opinion, it might not be so easy to
dismiss their point of view. Even though a large minority subgroup uses weak
arguments, their size fosters the perception that they are correct. Thus,
argument quality is important when the minority is small, however, when the
minority subgroup is larger, the perception of them being correct will increase

their ability to inﬂuence the majority subgroup, regardless of argument quality.

12

Put another way, the weak arguments from a large minority subgroup will be as
inﬂuential as a large and small minority subgroup’s strong arguments.

The preceding is the rationale for the present study. By designing this
study to allow for a group discussion, it is possible to examine a static variable
like group size in conjunction with a more dynamic variable like argumentative
communication, or more speciﬁcally argument quality, and asceItain how each
affects minority inﬂuence as well as perceptions of the minority subgroup.
Particularly important is to examine how group size and argument quality interact
to impact minority inﬂuence. For this investigation, groups will be comprised of
participants and confederates with the confederates for this study serving as the
minority subgroup and always arguing opposite of what the participants argue.
Confederates will vary argument quality as well as group size.

The speciﬁc predictions for this study are that participants exposed to
strong arguments from the minority will be more inﬂuenced than participants
exposed to weak arguments and participants exposed to a larger minority of four
will be more inﬂuenced than participants exposed to a smaller minority of two. It
is also expected that a large minority’s weak arguments will be as inﬂuential as
minority’s that use strong arguments. Finally, large minorities will foster a greater
perception of competence, correctness, and being liked, compared to the smaller
minority and that these perceptions will be positively associated with minority

inﬂuence.

13

CHAPTER 2
METHOD

Participants

Participants (n=191) were undergraduates from a large Midwestern
university. Participants received either extra-credit or fulﬁlled a course
requirement for participating. The sample was comprised of 37% male and 63%
female participants whohad a mean age of twenty-years old (SD = 1.59). A total
of 24 groups were used for this study with the group sizes ranging from ﬁve to
eleven members.

MD

A 2 x 2 independent groups design with an offset control group was
employed with argument quality (strong vs. weak) and minority size (two, small,
or four, large) varied. The strong and weak arguments were identiﬁed from two
pilot studies. The two and four member minority subgroups were comprised of
trained confederates who were consistent in presenting the minority viewpoint.

An additional variable was created due to the range of group size. Past
research has indicated that as the number of members in the majority increases,
minority inﬂuence might decrease (T anford & Penrod, 1984; Wood et al., 1994).
Thus, minority subgroup size relative to total group size might impact the results
for this study. Taking the number of minority subgroup members and dividing
them by the total group size created a variable labeled percentage. The

percentage variable was treated as a covariate in subsequent analyses.

14

Procedure

Two pilot studies were employed to identify strong and weak arguments.
The ﬁrst pilot study consisted of giving participants (n=6) the stimulus for this
study (Appendix A) and having them list as many arguments as they could for the
topic. The stimulus was an employment item that asked participants to indicate
whether a recent college graduate should take a job with a new computer ﬁrm
(Company A) with an uncertain future and high potential for growth or an older
computer ﬁrm (Company B) with lower potential for growth but a certain future.
Past studies employing a similar (Stoner, 1968) item have demonstrated that
participants tend to prefer the new ﬁrm, and the stimulus was written so as to
encourage participants to prefer the new ﬁrm. Thus, the confederates always
argued for the older ﬁrm, the counterattitudinal position.

A total of 28 non-redundant arguments were generated by the ﬁrst pilot
study and were used for the second pilot study. Participants in the second pilot
study (n=31) were given the stimulus to read, and asked to respond to each
argument created in the ﬁrst pilot study via four, ﬁve-point scales (strong-weak,
convincing-unconvincing, persuasive-unpersuasive, reasonable-unreasonable).
The six highest rated arguments (strong) and six lowest rated arguments (weak)
were identiﬁed and used by the confederates when taking the minority subgroup
stance (Appendix B)‘.

Seven different people served as confederates in this study, and each one
of them participated in both the small and large minority conditions. Each

confederate was given the arguments, told to memorize them, and tested for

15

recall of the arguments. Each confederate was able to recall successfully all of
the arguments. They were then trained to argue consistently for the older, less
risky ﬁrm by using only the arguments they were given. Training consisted of
mock discussions of the employment item used as the stimulus for this study.
The training lasted until all of the confederates could recall all of the arguments,
and only the speciﬁc arguments, given to them for the group discussions.

After the confederates’ training was complete, times were arranged for
participants to report to a lab. Confederates reported to the lab and went through
the same procedures as the participants so as not to arouse suspicion that they
were part of the study, or that they might have status different from the
participants. Once in the lab both the participants and confederates completed a
consent form, read the employment item, and indicated their employment
preference. The experimenter stepped outside of the room to identify
participants that preferred Company B over Company A. Participants preferring
Company A remained in the group with the confederates, and participants
preferring Company B were escorted to another room where they were made
part of another study. The remaining group members were then asked to raise
their hand if they preferred Company A as their choice and the number of
individuals who raised their hand was verbalized for the whole group to hear by
the experimenter. Group members (confederates) who preferred Company B
were then asked to raise their hands. The number of hands raised was also
verbalized for all group members to hear, and participants were able to see who

and how many preferred Company B. The groups were then instructed that they

16

had a maximum of twenty minutes to discuss to consensus their choice of
employers. After answering any questions participants may have had the group
discussion began.

After the groups had reached consensus, or time had elapsed, an
experimenter would enter the room and ask for the ﬁnal decision. Group
members were then asked to raise their hands if they preferred Company A or if
they preferred Company B as their ﬁnal choice. Again, the number of people
choosing either Company A or B was verbalized. Posttest measures were then
distributed, the group members were then debriefed, pledged to silence
regarding the nature of the study, and thanked for their time. The participants in
the control group followed the exact same procedures as the experimental
groups, except there were no confederates in the control groups. The
prediscussion preference for all control group members was Company A.
Instruments

Posttest measures included an induction check to test if participants were
aware of a minority subgroup arguing against them. Gauging how the majority
subgroup perceived the minority subgroup was accomplished by having
participants rate the minority subgroup using ﬁve-point bi-polar response scales
that assessed the correctness of the minority’s position, their competence, and
how much they were liked. Identifying the differences between participants’ initial
and ﬁnal decision for the two companies determined the minority subgroups’
amount of inﬂuence. To indicate which company participants preferred and to

what extent they preferred that company, a 10-point scale was used with a score

17

of 10 reﬂecting the strongest preference for Company A and a score of one

reﬂecting the strongest preference for Company B.

18

CHAPTER 3
RESULTS

Induction checks

Induction checks were employed to test for the minority induction. Two
separate questions asked participants if there was a person or persons who
initially “disagreed” with their position and if there was a person or persons who
initially “agreed” with their position before the group discussion began. Results
indicated that 99.2% of the participants were very aware that there was a
subgroup that initially disagreed with their position. The same result, 99.2%, was
obtained for the participants being aware that there was a subgroup that initially
agreed with their position. Only one person was unsure about a subgroup
initially disagreeing and a subgroup agreeing with their stance.
Measures

The distribution of responses for the change in company preference was
skewed positively with a mean of .95 and standard deviation of 3.4. Conﬁnnatory
factor analysis was used to test the dimensionality of the competence,
correctness, and liked scales. Items that met the requirements of face validity,
internal consistency, and parallelism were retained and summed to create
indices (see Table 1). This procedure resulted in a competence scale consisting
of four items with a distribution of responses that was slightly skewed negatively
with a standardized item alpha of .84 (M = 18.90, SD = 3.95). The correctness

scale consisted of seven items and had a distribution of responses that were

19

slightly skewed negatively with a standardized item alpha of .81 (M = 19.35, SD =
4.30). Finally, the liked scale consisted of ﬁve items and had a distribution of
responses that were skewed negatively with a standardized item alpha of .90 (M
= 17.85, SD = 4.87).

Minority Inﬂuence

Pretest scores were subtracted from posttest scores in order to calculate
the degree to which participants were inﬂuenced. A positive mean indicates that
participants shifted their preference toward Company B, the company the
minority advocated, whereas a negative mean indicates that participants’
preferences shifted toward Company A after group discussion. To identify
differences between the control group and the experimental groups, a one-tailed
Dunnett test was performed (Keppel, 1991). Results are presented in Table 2.
There were statistically signiﬁcant differences (p < .05) between the control group
and both of the strong argument conditions. There were no statistically
signiﬁcant differences (p. > .05) between the control group and the weak
argument conditions. Nonetheless, the probability for the weak arguments, large
minority condition exceeded .05 but it was less than .10.

A number of issues arise concerning the analysis of these data. These
data can be analyzed at both the individual and group level. The individual level
data may be affected by a lack of independence among observations, a problem
that can be addressed by averaging group members’ scores to create a group
level measure. Having only 24 groups with an average cell size of six presents

issues concerning power at the group level while analyzing data from the

20

individual level provides a substantial increase in power with an average size of
30 per cell. Data for this study will be analyzed from both the individual and
group level with careful attention to the pattern of the means. If the pattern of the
means is similar for both the group and individual level analyses, then no
violation of independence has occurred and it is warranted to draw conclusions
based on the more powerful test.

A second issue to be addressed is the analysis of minority size as a raw
number versus minority size relative to total group size. In other words, minority
size as a raw number equals either two or four for this study, whereas minority
size relative to total group size is the percentage of group members who are in
the minority. This issue is pertinent because the percentage of minority
subgroup members compared to total group size varied from group to group and
there is some indication from past studies that this factor could affect the results
(Maas & Clark, 1984; Tanford & Penrod, 1984). Data will ﬁrst be analyzed using
argument quality and minority size as factors from both the group and individual
level followed by analyses treating the percentage of minority group members in
each group as a covariate.

It was predicted that strong arguments would be more inﬂuential than
weak arguments and a larger minority subgroup would be more inﬂuential than a
smaller minority subgroup. Table 3 contains the cell means for the individual
level analysis. Using a 2 X 2, argument quality by minority size, independent
groups analysis of variance, the data were consistent with the two predicted main

effects. There was a main effect for argument quality (F(1,120) = 12.2, p = .001,

21

r = .29) with strong arguments (M = 1.87) being more inﬂuential than weak
arguments (M = -.061). Regarding minority size, there was a main effect
(F(1,120) = 8.30, p = .005, r = .24) such that a larger minority subgroup (M =
1.72) was more inﬂuential compared to a smaller minority subgroup ( M = .091 ).
The interaction was not statistically signiﬁcant (F(1, 120) = <1, p > .05, r = .10).

To identify if minority size relative to total group size also had an effect,
the number of minority subgroup members was divided by the total group size.
This transformation was done for each group and produced percentages for each
minority subgroup. In an analysis of covariance (ANCOVA), argument quality
was used as a ﬁxed factor with the percentage of minority subgroup members as
a covariate. At the individual level of analysis, results indicated that argument
quality still affected inﬂuence such that strong arguments (M = 1.89) were more
inﬂuential than weak arguments (M = -.09) (F(1, 121) = 7.09, p < .05, r = .30).
The effect of the covariate was statistically signiﬁcant as well (F(1, 121) =24.06, p
< .05, r = .45), indicating both argument quality and the percentage of minority
subgroup members relative to total group size affect inﬂuence. Further,
regressing the amount of inﬂuence onto the percentage of minority subgroup
members for both the weak argument condition (b = .19) and the strong
argument condition (b = .17) indicated no differences between the two conditions
and thus, no evidence of an argument by minority size interaction.

Analyzing these data at the group level indicated a similar pattern. Table
4 contains the cell means for this group level analysis. The marginal means for

weak arguments (M = .15) and strong arguments (M = 2.4) followed the same

22

pattern as the individual level marginal means, as did the pattern of the marginal
means for the small minority condition (M = .46) and the large minority condition
(M = 2.1). Although both the effects for argument quality (F(1,20) = 4.00, p = .06,
r = .38) and minority size (F(1,20) = 2.10, p > .05, r = .28) were not statistically
signiﬁcant by conventional standards, this outcome was likely a result of low
power (.47 & .28, respectively). Overall, strong arguments are more inﬂuential
than weak arguments and larger minorities are more inﬂuential than smaller
minorities. Again, the interaction was not statistically signiﬁcant (F(1, 20) = <1, p
>.05, r = .20).

Analyzing the data at the group level indicated that the covariate had a
substantial effect on inﬂuence (F(1, 21) = 8.06, p < .05, r = .57) whereas there
was not a statistically signiﬁcant effect for argument quality (F(1, 21) = 2.03, p >
.05, r = .30), however, this result is explained by the low power (.27) of the group
level analysis. Further, regressing the amount of inﬂuence onto the percentage
of minority subgroup members for both the weak argument condition (b = .23)
and the strong argument condition (b = .21) at the group level indicated no
differences between the two conditions and thus, no interaction.

Expanding on these analyses, one-sample t-tests (two-tailed) were
employed. Individual level analysis indicated that weak arguments (M = -.09)
were not statistically signiﬁcantly different than zero (t(59) = -.24, p > .05) but that
strong arguments (M = 1.89) were statistically signiﬁcantly different than zero
(t(63) = 4.33, p < .05). At the group level of analysis, weak arguments (M = .02)

were not statistically signiﬁcantly different than zero (t(59) = -.03, p > .05) but

23

strong arguments (M = 2.39) were statistically signiﬁcant different from zero (t(63)
= 2.83, p < .05). Thus, regardless of the level of analysis, strong arguments are
more inﬂuential than weak arguments.

Inspection of the standard deviations in each cell of Table 3 identiﬁes a
seriously discrepant standard deviation (SD = 1.63) in the weak arguments, small
minority condition. Indeed, using Levene’s F-test a violation of homogeneity of
variance was detected (F(3, 120) = 10.12, p < .05). This violation is explained by
a basement effect. In this condition change was negative and there was little
room on the scale for participants to move in a negative direction. In all other
conditions change was positive, and there was ample room to change in that
direction. Consistent with this thinking is a substantial positive correlation (r =
.92) between the mean and variance across the four experimental conditions,
indicating that the within condition variances decreased as the condition means
decreased.

Inspection of the correlations in Table 5 among the amount of inﬂuence
and the perceptions of competence, correctness and being liked indicate positive
and substantial correlations, as hypothesized. At both the individual and group
level these ﬁndings are statistically signiﬁcant. Thus, the perceptions fostered by
the minority subgroup help them with inﬂuencing the majority. Moreover, as can
be seen in Table 5, the correlations among the perceptual variables are positive

and relatively strong.

24

Perpeptions of the Minoriy' Subgroup

It was predicted that there would be a main effect for minority size on

 

competence, correctness, and being liked. Table 6 contains the cell means for
the competence measure at the individual level. For the competence measure
there was no main effect for minority size (F(1, 119) = 2.19, p > .05, r = .12),
however, there was a main effect for argument strength (F(1, 119) = 10.9, p <
.05, r = .28). Using strong arguments (M = 16.1) helped the minority subgroup to
be perceived as more competent than did weak arguments (M = 14.1). This
effect was qualiﬁed by a statistically signiﬁcant argument strength by minority
size interaction (F(1, 119) = 13.2, p < .05, r = .32). Inspection of the cell means
indicated that perceptions of the minority subgroup being competent were largest
overall in the small minority, strong arguments cell (M = 21.7), equal in the large
minority condition (M = 18.4 (strong), M = 18.6 (weak)) and smallest in the small
minority, weak argument cell (M = 16.9). Thus, when the minority is small,
making strong arguments produces higher ratings of competence than does
making weak arguments. But, when the minority is relatively large argument
strength has no impact on competence ratings.

ANCOVA was used with argument quality as a ﬁxed factor and with the
percentage of minority subgroup members as a covariate. At the individual level
of analysis using competence as the dependent variable, results indicated that
argument quality still affected perceptions of competence such that strong
arguments (M = 20.0) fostered greater perceptions of competence than did weak

arguments (M = 17.8) (F(1, 121) = 10.4, p < .05, r = .28). The percentage

25

covariate did not have a signiﬁcant impact on perceptions of competence (F(1,
121) = <1, p > .05, r = .03). Thus, argument quality affected perceptions of
inﬂuence whereas the percentage of minority subgroup members relative to total
group size did not. Further, regressing perceptions of competence onto the
percentage of minority subgroup members for both the weak argument condition
(b = .14) and the strong argument condition (b = -.16) indicated differences
between the two conditions. Constructing a conﬁdence interval around b = .14
(95% CI = -.03 to .31) indicates that b = -.16 does not ﬁt, which means that the
two numbers are signiﬁcantly different from one another and thus, that there is an
interaction.

At the group level, similar results were obtained for the competence
measure. Table 7 contains the cell means for this group level analysis. There
was no main effect for minority size (F(1, 20) = 1.74, p > .05, r = .20), however,
there was a main effect for argument strength (F(1, 20) = 9.61, p < .05, r = .46).
Using strong arguments (M = 20.1) helped the minority subgroup to be perceived
as more competent than did weak arguments (M = 17.9). This effect was
qualiﬁed by a statistically signiﬁcant argument strength by minority size
interaction (F(1, 20) = 12.8, p < .05, r = .54) that had the same pattern of cell
means as did the individual level analysis.

Analyzing the data at the group level using ANCOVA with competence as
the dependent variable indicated that there was a statistically signiﬁcant effect for
argument quality (F(1, 21) = 6.90, p < .05, r = .50) with strong arguments (M =

20.2) fostering greater perceptions of competence compared to the weak

26

arguments condition (M = 17.8). The percentage covariate had no effect on
perceptions of competence (F(1, 21) = <1, p > .05, r = .06). Thus, argument
quality affected perceptions of inﬂuence where as the percentage of minority
subgroup members relative to total group size did not. Further, regressing
perceptions of competence onto the percentage of minority subgroup members
for both the weak argument condition (b = .13) and the strong argument condition
(b = -.16) indicated differences between the two conditions. Constructing a
conﬁdence interval around b = .13 (95% CI = -.26 to .52) indicates that b = -.16
does ﬁt within the conﬁdence interval, which means that the two numbers are not
signiﬁcantly different from one another, however, with the small sample size (n =
24), there is a lack of power. Regardless, the two numbers are different from one
another.

Analysis of the correctness measure at the individual level identiﬁed no
main effect regarding minority size (F(1, 119) = <1.00, p > .05.. r = .00). Table 8
contains the cell means for the correctness measure at the individual level.
Analysis of variance identiﬁed a main effect for argument strength regarding
correctness (F(1, 119) = 8.39, p< .05, r = .25) such that minority subgroups using
weak arguments (M = 18.2) were perceived to be less correct in the group
discussion than minority subgroups using strong arguments (M = 20.4). A
statistically signiﬁcant interaction (F(1, 119) = 6.84, p < .05, r = .23) for the
correctness measure was discovered such that perceptions of the minority
subgroup being correct are greatest overall in the small minority, strong

arguments condition (M = 21.5), equal in the large minority condition (M’s = 19.0

27

(weak arguments), 19.2 (strong arguments», and smallest in the small minority,
weak arguments condition (M = 17.4). Thus, when the minority is small, making
strong arguments produces higher ratings of correctness than does making weak
arguments. But, when the minority is relatively large argument strength has no
impact on correctness ratings.

Using analysis of covariance, perceptions of being correct were used as
the dependent variable with argument quality as a ﬁxed factor and the
percentage of minority subgroup members as a covariate. At the individual level
of analysis using correctness as the dependent variable, results indicated that
argument quality still affected perceptions of correctness such that strong
arguments (M = 20.4) fostered greater perceptions of being correct than did weak
arguments (M = 18.3) (F(1, 120) = 6.78, p < .05, r = .24). The percentage
covariate did not have a signiﬁcant impact on perceptions of correctness (F(1,
120) = <1, p > .05, r = .10). Thus, argument quality affected perceptions of being
correct where as the percentage of minority subgroup members relative to total
group size did not. Further, regressing the amount of inﬂuence onto the
percentage of minority subgroup members for both the weak argument condition
(b = .16) and the strong argument condition (b = -.09) indicated differences
between the two conditions. Constructing a conﬁdence interval around b = .16
(95% CI = -.01 to .33) indicates that b = -.09 does not ﬁt, which means that the
two numbers are signiﬁcantly different than one another and that there is an

interaction.

28

At the group level, a similar pattern of means was obtained. Table 9
contains the cell means for this group level analysis. The marginal means for
minority size were alike in both the small minority (M = 19.7) and large minority
(M = 19.0) conditions (F(1, 20) = <1, p > .05, r = .14). The marginal means for
argument quality identiﬁed that strong arguments (M = 20.5) tended to foster
higher perceptions of correctness compared to weak arguments (M = 18.2) (F ( 1 ,
20) = 4.00, p = .06, r = .38). The interaction between argument quality and
minority size was not Signiﬁcant at the group level (F(1, 20) = 2.70, p > .05, r =
.31), however, this was a result of low power (.34). Thus, both levels of analyses
report the same pattern of means with different results being attributable to low
power from the group level analysis.

Using analysis of covariance and analyzing the data at the group level
using correctness as the dependent variable indicated that there was a strong
effect for argument quality (F(1, 21) = 3.62, p = .07, r = .40) with strong
arguments (M = 20.6) fostering greater perceptions of correctness compared to
the weak arguments condition (M = 18.1). The percentage covariate had no
effect on perceptions of correctness (F(1, 21) = <1, p > .05, r = .15). Thus,
argument quality affected perceptions of correctness where as the percentage of
minority subgroup members relative to total group size did not. Further,
regressing perceived correctness onto the percentage of minority subgroup
members for both the weak argument condition (b = .14) and the strong
argument condition (b = -.07) indicated differences between the two conditions.

Constructing a conﬁdence interval around b = .14 (95% CI = -.26 to .54) indicates

29

that b = -.07 does ﬁt within the conﬁdence interval, which means that the two
numbers are not signiﬁcantly different than one another. As previously noted
however, this is a result of low power

Analysis of the liked measure at the individual level identiﬁed no main
effect regarding minority size (F(1, 119) = <1.00, p > .05, r = .00). Table 10
contains the cell means for the liked measure at the individual level. Analysis of
variance identiﬁed a main effect for argument strength regarding how liked was
the minority subgroup (F(1, 119) = 5.50, p < .05, r = .20) such that minority
subgroups using strong arguments (M = 18.9) were more liked than minority
subgroups using weak arguments (M = 16.9). A statistically signiﬁcant
interaction (F(1, 119) = 14.32, p < .05, r = .32) for the liked measure was
discovered such that perceptions of the minority subgroup being liked are
greatest overall in the small minority, strong arguments condition (M = 20.5),
similar in the large minority condition (M’s = 18.3 (weak arguments), 17.2 (strong
arguments», and smallest in the small minority, weak arguments condition (M =
15.5). Thus, when the minority is small, making strong arguments produces
higher ratings of being liked than does making weak arguments. But, when the
minority is relatively large argument strength has no impact on ratings of being
liked.

Using ANCOVA, perceptions of being liked were used as the dependent
variable with argument quality as a ﬁxed factor and percentage as a covariate.
At the individual level of analysis using liked as the dependent variable, results

indicated that argument quality still affected perceptions of being liked such that

30

strong arguments (M = 18.8) fostered greater perceptions of competence than
did weak arguments (M = 16.5) (F(1, 120) = 4.35, p < .05, r = .19). The covariate
did not have a signiﬁcant impact on perceptions of being liked (F(1, 120) = <1, p
> .05, r = .06). Thus, argument quality affected perceptions of being liked where
as the percentage of minority subgroup members relative to total group size did
not. Further, regressing perceptions of being liked onto the percentage of
minority subgroup members for both the weak argument condition (b = .17) and
the strong argument condition (b = -.15) indicated differences between the two
conditions. Constructing a conﬁdence interval around b = .17 (95% CI = .00 to
.34) indicates that b = -.15 does not ﬁt, which means that the two numbers are
signiﬁcantly different than one another and that there is an interaction.

At the group level, similar results were obtained for the liked measure.
Table 11 contains the cell means for this group level analysis. There was no
main effect for minority size (F(1, 20) = <1, p > .05, r = .01), however, there was
a main effect for argument strength (F(1, 20) = 7.52, p < .05, r = .40). Using
strong arguments (M = 19.0) helped the minority subgroup to be perceived as
more liked than did weak arguments (M = 16.5). This effect was qualiﬁed by a
statistically signiﬁcant argument strength by minority size interaction (F(1, 20) =
18.9, p < .05, r = .64) that had the same pattern of cell means as did the
individual level analysis.

Using ANCOVA and analyzing the data at the group level using liked as
the dependent variable indicated that there was a signiﬁcant effect for argument

quality (F(1, 21) = 4.40, p < .05, r = .45) with strong arguments (M = 19.1)

31

fostering greater perceptions of being liked compared to the weak arguments
condition (M = 16.4). The covariate had no effect on perceptions of being liked
(F(1, 21) = <1, p > .05, r = .21). Thus, argument quality affected perceptions of
being liked where as the percentage of minority subgroup members relative to
total group size did not. Further, regressing perceptions of being liked onto the
percentage of minority subgroup members for both the weak argument condition
(b = .25) and the strong argument condition (b = -.14) indicated differences
between the two conditions. Constructing a conﬁdence interval around b = .25
(95% CI = -.14 to .64) indicates that b = -.07 does ﬁt within the conﬁdence
interval, which means that the two numbers are not signiﬁcantly different than
one another. As previously noted however, this is likely a result of low power.

In conclusion, analyzing data from both the individual and group level
demonstrated the same pattern of means. Often the same conclusions were
reached and in cases where they were not, differences were attributable to low
power. Nonetheless, the same pattern of means for both analyses allows for the
conclusion that independence was not violated. Further, treating percentage of
minority subgroup members as a covariate identiﬁed that it affected the
dependent variables in the same manner as minority size. Thus, both minority

size and percentage are important issues to consider for minority inﬂuence.

32

CHAPTER 4
DISCUSSION

The purpose of this study was to extend the minority inﬂuence literature by
investigating the effects of argument quality and group size on inﬂuence and
perceptions fostered by the minority while holding behavioral consistency
constant. Group size for this study varied and the number of minority subgroup
members relative to total group size was examined for any effects it may have
had or caused different from the argument quality and minority size effects.

None were found. Additionally, these data were analyzed at both the individual
and group levels. Regardless of how these data were analyzed, the same
pattern in these data emerged. Because analyzing these data at the group level
produced a less powerful test then analyzing these data at the individual level,
conclusions were drawn from the more powerful test.

It was hypothesized that a minority subgroup using strong arguments
would be more successful at inﬂuencing the majority subgroup than when they
used weak arguments, and that a large minority would be more persuasive than
a small minority. The data were consistent with both hypotheses. Both strong
arguments and a large minority proved beneﬁcial when a minority was attempting
to inﬂuence the majority. Data were not consistent with the hypothesized
interaction that a large minority and a minority using strong arguments are able to

effect the same amount of inﬂuence with the majority.

33

That strong arguments were more inﬂuential than weak arguments was
consistent with the hypothesized effect. Although some studies found that
relative argument quality was more predictive of attitude change than induced
argument quality, it was proposed that in those studies having only one member
as part of the minority subgroup could have contributed to this effect. Garlick and
Mongeau (1992) demonstrated that comprising a minority subgroup of more than
one member attenuated the effect of relative argument quality, and found the
predicted effects for induced argument quality. The present study used minority
subgroups comprised of two and four members and demonstrated that strong
arguments were more persuasive than weak arguments. Because the methods
used for this study differed from the Garlick and Mongeau (1992) study (group
discussion versus ﬁctitious script), it would be interesting to investigate if the
strong arguments used by one lone minority member in a group discussion, with
the help of behavioral consistency, would be perceived as strong and inﬂuential.

The effect of group size proved to be interesting. Whereas the larger
minority subgroup tended to be more inﬂuential than the smaller minority
subgroup, using weak arguments in different size groups produced quite different
results. The expected effect that the large minority using weak arguments would
be as effective as the small and large minority using strong arguments was
obtained. It appears that for a large minority size is advantageous when
attempting to inﬂuence the majority. Although the arguments advanced by a

large minority may be weak, the size of the larger minority compensates.

34

Because a number of other people share a different opinion, the majority may
perceive that a large minority cannot be completely wrong or inaccurate.
Conversely, a small minority using weak arguments proved to be
unproductive when trying to inﬂuence the majority. Majority group members
tended to increase or strengthen their preference for their initial choice in this
condition. If people enter group situations with the expectations that most people
share their belief or perception (Maas & Clark, 1984), then a minority subgroup
has violated the majority’s expectations. In turn, a validation process ensues
whereby the majority tries to understand the minority’s position (Moscovici,
1980). When the minority uses strong arguments, a reasonable attempt at
validating their position has been made. Likewise, when the minority is large, the
larger group size might serve as a heuristic wherein the majority would not
expect so many members of their group to support a bad position, causing them
to think more about the minority’s position. Thus, for large minorities, their size
serves to help them in validating their position. The small minority that
consistently uses weak arguments has neither size, nor reasonable support for
their position. The arguments generated by the majority are countered with weak
arguments by the minority. If the minority has failed to validate their stance and
they consistently support it with weak arguments, the majority may not
understand why these few people do not see their point of view. In the process
of offering their own arguments that were responded to with weak arguments,
majority subgroup members increased their belief in their position. Future

investigations should study this issue.

35

The perceptions fostered by the minority subgroup did not differ based on
minority size as hypothesized. Instead, main effects were found for argument
quality but were qualiﬁed by a Signiﬁcant interaction. In all cases, competence,
correctness and liked, were highest in the small minority, strong argument
condition followed by the large minority condition whereas the small minority,
weak argument condition reported the lowest ratings. Thus, the interaction for
competence, correctness, and liked all took the same form. When the minority is
small but using strong arguments, they produce higher ratings on all three
variables than do minorities that are small using weak arguments. When the
minority is large, argument strength has no impact on how the minority is
perceived with these three variables.

Nemeth et al. (1977) found that as a behaviorally consistent minority
increases in size, the perception of that minority being competent, correct, and
liked increased positively and that these perceptions were positively associated
with minority inﬂuence. The same results were obtained in this study. The
results of this study demonstrate that both minority group size and how well the
minority defends their position (argument quality) are important determinants of
how they are perceived. As evidenced by the ratings received by the different
minorities, larger minorities or minorities using strong arguments foster more
positive perceptions. Groups using strong arguments provide good reasons in
support of their stance, and thus, are perceived more positively, whereas larger
groups, regardless of argument quality, are perceived more favorably because of

their size. Although larger minorities gain favorable attributions because of their

36

size, smaller minorities, when they present their position in a compelling way
(strong arguments) receive favorable attributions as well.

One reason suggested for the interaction effect could be that when two
people comprise the minority it is easier to judge or evaluate them compared to
four people. Having four people in the minority may dilute the majority’s
perceptions of the minority. Having only two people in the minority puts them
more in the spotlight, or, since there are only two people their actions or behavior
are highlighted more than when four people comprise theminority. Thus, if a
minority of two presents their case well, they foster very positive perceptions
whereas if they do not present their case well, favorable perceptions of them are
not as strong. When there are more in the minority, in this case four, it is more
difﬁcult for the majority to focus on the minority because the added number of
minority group members dilutes focus. As a result it is more difﬁcult to judge the
minority which leads to similar ratings regardless of how the minority presents its
case.

Future Reﬁrch & Limitations

Future research should investigate more communicative behavior to
identify how majority group members actually respond to minority inﬂuence
attempts. The long held assertion by Moscovici and colleagues (Moscovici,
Lage, & Faucheux, 1969; Moscovici, & Faucheux, 1972; Moscovici & Lage,
1976) that the number in the minority subgroup and consistency lead to

conformity needs to be amended by adding the quality of the communication by

37

the minority also matters. Thus, simply showing up as a large minority
consistently supporting a position is not always enough.

Future research could be conducted to identify if other factors such as
leadership or credibility affect minority inﬂuence. Leaders of a group may be able
to enact more inﬂuence on a majority subgroup because of their status in the
group as a whole. Depending on the topic, some group members may have
more credibility than other group members because of some unique ability they
possess or information that they may have. The knowledge a group member has
on some topic could make them a credible or expert source in the group and if in
the minority, could use that credibility to inﬂuence the majority.

Another avenue for research could be to investigate if a minority subgroup
is more successful in different situations or contexts. For example, in a situation
where the outcome or ﬁnal decision can be a compromise between the majority
and minority subgroups, it may be easier for the minority to inﬂuence the majority
than it is to inﬂuence them to accept the minority’s opposing categorical decision.
\M'Iatever the case, some situations may be easier or more difﬁcult for the
minority to achieve inﬂuence.

There were limitations to this study. Not everyone was inﬂuenced equally
and regardless of condition, there was not total and complete inﬂuence. An
explanation of this phenomenon Should be examined in the future. Although
differences of opinions existed in the groups, some of the participants may have
not participated or put forth their point of view. They simply may have not taken

the experiment seriously or did not care much for participating. Further, some

38

participants may have reported being inﬂuenced when in actuality they were not.
The purpose for them doing so could have been so the experiment would end.
Finally, the confederates may have exhibited some behavior not measured and it
was this behavior that signiﬁcantly contributed to them obtaining inﬂuence.
Addressing these limitations in future research will help us to better understand
minority inﬂuence in the group context.
Conclusion

Adding to the minority inﬂuence corpus, this study demonstrated that the
way a minority presents its ideas and establishes its position on a topic can
impact their chances of inﬂuencing a majority, as well as how they are perceived
by the majority. Further, this study demonstrated the importance of
argumentative communication when the minority is trying to inﬂuence the
majority. Group members who comprise the minority would best be served by
presenting their ideas with a united front, a large minority subgroup, and via
messages known to be high in quality. Of course, the latter issue is somewhat
mundane. Clearly, any group member would be likely to advocate a favored
position with arguments as strong as could be amassed. Nevertheless, the effect
of argument strength suggests that in order to be effective, members of minority
subgroups might need to accumulate and practice strong arguments prior to
group discussion.

In the absence of having available the information to accumulate and
create persuasive arguments, garnering support from other group members

might prove effective for inﬂuencing the majority. Presenting a uniﬁed front with

39

a large number of minority members can be as effective as having a few
members in the minority subgroup using strong arguments. Of course, having as
many members as possible in the minority subgroup and using strong arguments

increases the minority’s chance of inﬂuencing the majority.

40

APPENDIX

41

 

APPENDIX A

STIMULUS

Directions: Following is a description of a situation a college senior is going to
face in the very near future. Please read the description and then mark what you
think that person should do.

A college senior that is majoring in computer programming is going to be
graduating this semester. Already, this college senior has been recruited by two
organizations to come work for them and the time is coming near to decide
between the two organizations. Following is a description of the differences
between the two companies:

COMPANY A:

Company A has an uncertain future such that it is a new company and whether it
will be around in the future has yet to be determined. However, if the college
senior joins this organization, as the organization grows, the opportunity for
advancement will be great; but there is a lot less job security compared to the
other organization.

COMPANY B:

Company B is a well established company and it is very likely to be around in the
near and distant future. However, if the college senior joins this organization, the
opportunity for advancement is limited; but job security in this organization will be
a lot higher compared to the other organization.

Based on the situation described above, circle the response that best reﬂects
what you think the college senior should do. You can circle any number that you
desire but please note that circling a number closer to “1” indicates that you
believe the college senior should take the job with Company B and by circling a
number Closer to “10” indicates that you believe the college senior should take
the job with Company A.

Take the job Take the job
with Company with Company
B A

 

[112l3l4l5l61718l9l10l

 

42

APPENDIX B

ARGUMENTS
Strong Arguments

1. Since Company B is an established organization, things like vacation time,
work hours and beneﬁts are likely to be more structured than Company A. (M =
16.3)

2. After graduating from college, a person probably does not have much savings
and probably has some bills to pay. Company B will give a person just
graduating a steady, reliable income to pay their bills and begin saving money.
(M = 16.1)

3. Experience at a reputable company could provide opportunities at other
companies. Thus, if a person works for Company B and they decide to leave it,
they might have an easier time getting another job and advancing quicker in that
new job. (M = 14.8)

4. Taking the job with Company B gives a person a job with an established
company that ‘s gonna be around in the years to come, giving them peace of
mind. They have job security with a company that’s gonna last. (M = 14.6)

5. The college senior can always go to an upstaIt like Company A. By starting
with Company B, the senior has a good job and is gaining valuable experience.
(M = 14.5)

6. Taking the job with Company B would afford the college senior the opportunity
to network for future jobs. mm a stable company like B, there a number of
contacts to be made in and out of work. (M = 13.4)

Weak Arguments

1. It seems like a much better decision to take a job with Company B. There will
be less back stabbing if advancement opportunities are not available, so you’ll
have a better chance of developing friendly relationships with your co-workers.
(M = 7.4)

2. Company A is too risky. If you are going to take a risk or gamble, you should
start your own company. That way, you receive all the rewards and beneﬁts. (M
= 7.9)

43

APPENDIX B cont.

3. Working for Company B can afford you the opportunity to diversify yourself in
that you can work at different positions in the company. Vlﬁth Company A, you
might be stuck doing just one thing. (M = 8.1)

4.The great thing about Company B is that it is secure and stable. As a result,
you probably won’t have to work as hard and you would experience a lot less job-
related stress. (M = 8.2)

5. Working for Company B probably means that less dedication and less energy
is required by an employee than if they worked for a new company, like
Company A, that is trying to establish itself. Working for Company B would be a
lot easier. (M = 8.3)

6. Company B offers a lot of job security and job security can be better than how
much you get paid, your rank, or advancement. (M = 8.8)

Table 1

Conﬁrmatom Factor Analysis for the Perception Variables

 

 

 

Measure and Variable M SQ E N
Compgtence Sca_l_e_:_
Competent-Incompetent 4.1 .88 .71 123
Effective-Ineffective 3.5 1 .1 .75 1 23
Proﬁcient-Not proﬁcient 3.5 .99 .85 123
lntelligent-Unintelligent 4.0 .92 .68 123
Correctness Scale:
Correct-Incorrect 2.6 .96 .54 123
Accurate-Inaccurate 2.7 .90 .71 123
Precise-Imprecise 3.1 .89 .56 123
Right-Not Right 3.0 .82 .55 123
Flawless-Flawed 2.7 .99 .71 123
Exact-lnexact 2.5 .83 .59 123
Expert-Not expert 2.8 .82 .68 123
Liked Scale:
Likeable—Unlikable 3.5 1 .1 .71 123
Friendly-Unfriendly 3.6 1 .1 .91 123
Pleasant-Unpleasant 3.6 1 .1 .95 123
Kind-Unkind 3.6 1.1 .88 123
Reasonable-Unreasonable 3.5 1 .2 .58 123

45

Table 2

Mean Differences of Inﬂuence for Each Experimental Condition Compared to the
Control group Using A One-tailed Dunnett Test

 

 

Condition Condition Control Group Difference N
Mean Mean
Weak argument! -1.23 -.37 -.860 30

Small minority

Strong Argument! 1.32 -.37 1.69** 30
Small minority

Weak argument! 1.02 -.37 1.38* 31
Large minority

Strong argument! 2.42 -.37 2.79“ 33
Large minority

 

*p < .10. **p < .05

46

Table 3

Cell Means for the Amount of Inﬂuence Based on the Minong' Size by Ajrgument

Qualﬁy Conditions at the Individual Level of Analysis

 

Argument Quality

 

Minoritv_Si_ze Weak Stronq

Large (4 in subgroup) M = 1.02 M = 2.42
SD= 3.50 SD = 3.55
N = 31 N = 33

Small (2 in subgroup) M = -1.20 M = 1.32
SD = 1.63 SD = 3.39
N = 30 N = 30

 

Control group: M = -.367, SD = .809, N = 30

47

Table 4

C__e__ll Means for the Amount of Inﬂuence Ba_sed on the Minoritv SI_ze bv Argument
Qualily Conditions at the Group Level of Analysis

 

Argument Quality

 

Minoritv Size ng Strong—

Large (4 in subgroup) M = 1.56 M = 2.62
SD= 3.08 SD = 3.24
N=5 N=6

Small (2 in subgroup) M = -1.27 M = 2.19
SD=.834 SD=3.10
N=6 N=7

 

Control group: M = -.367, SD = .463, N = 6

48

Table 5

Correlations for Inﬂuence and Perceptionsiat the Individual Level of Ap_a_lvsis

 

1 2 3 4

 

 

1. Inﬂuence 1.00

2. Competence .22 1.00

3. Correctness .39 .49 1.00

4. Liked .27 .60 .43 1.00
N=123

Correlations for Inﬂuence and Perceptions at the Group Level of Analysis

 

1 L 3 4

 

 

1. Inﬂuence 1.00

2. Competence .56 1.00

3. Correctness .66 .60 1.00
4.Liked .64 .83 .70 1.00
N = 24

49

Table 6

Cell Means for the Perceivep Competenc_s_pf the Minoritv Subgroup Based on
the Minoritv Si_ze bv Argument Quality Conditions at the Individiﬂ Level of

Analysis

 

Argument Qualig

 

Minority Size Wea_k Strong—
Large (4 in subgroup) M = 18.6 M = 18.4
SD= 4.31 SD = 3.85
N = 31 N = 33
Small (2 in subgroup) M = 16.9 M = 21.7
SD = 3.48 SD = 2.61
N = 29 N = 30

 

50

Table 7

Cell Means for Perceived Competence Based on the Minoritv Si_ze bv A_rgument
Qualitv Conditions at the Group Level of Analysis

 

 

Argument Quality

 

 

Minoritv Si_ze Wesk Strong—
Large (4 in subgroup) M = 18.7 M = 18.3
SD= 1.44 SD = 2.00
N=5 N=6
Small (2 in subgroup) M = 17.1 M = 21.8
SD = 2.03 SD = 1.33
N = 6 N = 7

 

51

Table 8

Cell Means for the Percein Correctness Based on the Minoritv Si_ze bv

Argument Quality Conditions at the Individual Level of Analysis

 

Minormg' Size

Argument Quality

 

Large (4 in subgroup)

Small (2 in subgroup)

Wesk Strong

M = 19.0 M = 19.2
SD= 4.85 SD = 4.01
N = 29 N = 30

M = 17.4 M = 21.5
SD = 3.64 SD = 3.88
N = 31 N = 33

 

52

Table 9

Cell Means for Perceived Correctness Based on the Minority Size by A_rgument
Quality Conditions at the Group Level of Analysis

 

 

Minomy' Size

Argument Qualig

 

Large (4 in subgroup)

Small (2 in subgroup)

Wesk Strong—
M = 18.7 M = 19.1
SD= 3.30 SD = 2.40

N = 5 N = 6

M = 17.7 M = 21.8

SD = 2.48 SD = 2.00

N = 6 N = 7

 

53

Table 10

Ce_l____l Means for how Liked was the Minoritv Squoup Based on the Minori_ty Size
by Argument Quality Conditions_ at the Individual Level of Analvsis

 

 

 

Argument Quality
Minom Size Wea_k Strong
Large (4 in subgroup) M = 18.3 M = 17.2
SD== 4.84 SD = 5.47
N = 29 N = 30
Small (2 in subgroup) M = 15.51 M = 20.5
SD = 3.30 SD = 4.14
N = 31 N = 33

 

54

Table 11

Cell Means for how Liked w_a_s the Minoritv Subgroup Based on the Minority Size

by Argument Qual'ﬂ Conditions at the GrOup Level of Analv§i_s

 

ﬂgument Quality

 

Minoritv Size Wea_k Strong

Large (4 in subgroup) M = 18.4 M = 17.0
SD= 2.60 SD = 2.33
N = 5 N = 6

Small (2 in subgroup) M = 14.7 M = 21.0
SD = 2.20 SD = 2.21
N = 6 N = 7

 

55

REFERENCES

56

References

Arbuthnot, J. & Wayner, M. (1982). Minority inﬂuence: Effect of size,
conversion, and sex. The Jourﬂl of chhologv. 111. 285-292.

Asch, S. E. (1951). Effects of Group Pressure upon modiﬁcation and
distortion ofjudgements. In H. Guetzkow (Ed.), Groups. leadershjLand men
(pp. 177-179). Pittsburgh, PA: Carnegie Press.

Boster, F. J. & Mayer, M. (1984). Choice shifts: Argument qualities or
social comparisons? In R. Bostrom (Ed.) Communication Yearbook 8 (pp. 393-
410) Beverly Hills, CA: Sage.

Canary, D. J., Brossman, B. G., 8 Seibold, D. R., (1987). Argument
structures in decision-making groups. Southern Speech Communication Journal,
$18-37. ‘

Garlick, R., & Mongeau, P. (1993). Argument quality in group member
status as determinants of attitudinal minority inﬂuence. Western Journal of
Communication 57 289-308.

 

Garlick, R. 8. Mongeau, P. (1992). Majority/minority size and the
evaluations of persuasive arguments. Communication Research Reports; 43-
52.

Gebhardt, L., & Meyers, R. (1995). Subgroup inﬂuence in decision-
making groups: Examining consistency from a communication perspective.
Small Grmip Research 26 2 147-168.

 

lnfante, D.A., & Rancer, AS. (1982). A conceptualization and measure of

argumentativeness. Journal of Personality Assessment, 46, 72-80.

Jacobs, 8., Allen, M., Jackson, S., & Petrel, D. (1985). Can ordinary
arguers recognize a valid conclusion if it walks up and bites them on the butt? In
J. R. Cox, M. O. Sillars, & G. B. Walker (Eds.), Argument and Social Practice:
Proceedings from the fourth SCA/A_FA conference on _a_rgumentjation (pp. 665-
674). Annandale, VA: SCA.

Latane, B., & Wolf, S. (1981). The social impact of majorities and
Minorities. Psycholpgv Review. 88 438-453.

Moscovici, S., & Faucheux, C. (1972). Social inﬂuence, conforming bias,
and the study of active minorities. In L. Berkowitz (Ed.), Advances in
experimental social psycholpgy (vol. 6. pp.149-202). New York: Academic
Press.

57

Moscovici, S., 8 Lage, E. (1976) Studies in social inﬂuence Ill: Majority
versus minority inﬂuence in a group. European Journal of Social Psychlogy, 6,
149, 149-174.

Moscovici, S., 8 Lage, E., 8 Naffrechoux, M. (1969). Inﬂuence of a
consistent minority on the responses of a majority in a color perception task.
Sociometrv. 32. 365-379.

Maass, A, 8 Clark III, R. D. (1984). Hidden impact of minorities: Fifteen
years of minority inﬂuence research. Psycholpgical Bulletin. 95. 3. 228-450.

Meyers, R., Brashers, D., 8 Jerzak, J. (1998). Majority/minority inﬂuence:
ident'ﬁying argumentative patterns and predicting argument-outcome links.

Paper presented at the National Communication Association in New York.

 

Meyers, R, Seibold, D. R, 8 Brasher, D. E. (1991). Argument in initial
group decision-making discussions: Reﬁnement of a coding scheme and a

descriptive quantitative analysis. Western Journal of Speech Communication,
&, 47-68.

Morris, W., 8 Miller, R. (1975). The effects of conscious-breaking and
conscious-preempting partners on reduction of conformity. Journal of
Experimental Sociaﬂvchologyg 1. 215-223.

Nemeth, C., Mosier, K., 8 Chiles, C. (1992). When convergent thought
improves performance: Majority versus minority inﬂuence. Personally and

Social Psychology Bulletin, 139-144.

Nemeth, C, 8 Wachtler, J. (1983). Creative problem solving as a result of
majority versus minority inﬂuence. European Journal of Social Psychology. 13.
45-55.

Nemeth, C, Wachtler, J., 8 Endicott, J. (1977). Increasing the size of the
minority: Some gains and some losses. Europaan Journal of Social Psycholpgy,
7, 1, 15-27.

Nemeth, C., 8 Wachtler, J. (1974). Creating the perceptions of
consistency and competence: A necessary condition for minority inﬂuence.
Sociometrv. 37. 4. 529-540.

Petty, R. E., Cacioppo, J. T., 8 Goldman, R. (1981). Personal
involvement as a determinant of argument-based persuasion. Journal of

Personalm and Social Psvchologv. 41. 847-855.

Rosenthal, R., 8 Rosnow, R. L., (1985). Contrast analysis: Focused
comparison in the analysis of variance. New York: Cambridge University Press.

58

Socha, T. J. (1997). Group communication across the life span. In L. R.
Frey 8 J. K. Barge (Eds), Managing group life: Communicating in decision-
making groups (pp. 1-28). Boston: Houghton-Mifﬁn.

Tanford, S., 8 Penrod, S. (1984). Social inﬂuence model: A formal

integration of research on majority and minority inﬂuence processes. Psychology
Bulletin. 95. 189-225.

59

Footnotes

‘Throughout this manuscript, anytime the minority or majority is referred
to, it represents a minority or majority subgroup.

2Although it is often reported that at three or four group members inﬂuence
levels off, this number can be controversial.

3Argument strength could be deﬁned employing an alternative criterion,
logical soundness. On the other hand, because persons judge logical soundness
poorly (Jacobs, Allen, Jackson, 8 Petral, 1985), this criterion proves less
effective for varying experimentally argument strength.

4Originally, confederates were armed with only three arguments for both
the strong and weak conditions. After running 4 groups, a decision was made to
increase the number of arguments to 6 for each condition. Before using the

arguments, confederates committed them to memory.

60

"Illllllllllll'lllllllllll