’ 3‘}

3m ~,,

.. 5.

(in;

 

1:2 .
2.4;)?

 

  

 

.fmmﬂ 5% was

 

. 1, .
., .._ 42w. WE? .

 

“ﬁgmﬁﬁé.

.v ‘ mﬁamﬂ...uﬂwﬁo

11f . .

 

 

7312.13; 8

200».

 

LIBRARY

Em tchigan St“: a.
University

ﬂ"‘ '

 

This is to certify that the
thesis entitled
DIFFERENCES IN SUCCESS AND PARTICIPATION
RATES IN ACADEMIC DEBATE

presented by

Jason Cruce Trice

has been accepted towards fulﬁllment
of the requirements for

M.A. degree in Communications

 

 

memo/K

NMajor prof?

Date 8-24-01

0-7639 MS U is an Afﬁrmative Action/Equal Opportunity Institution

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

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 cJCIRCIDatoDue.p65-p.15

 

 

 

DIFFERENCES IN SUCCESS AND PARTICIPATION RATES BASED ON SEX IN
ACADEMIC DEBATE

By

Jason Cruce Trice

AN ABSTRACT OF A THESIS
Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of
MASTER OF ARTS
Department of Communications

2001

Professor Steven A. McCornack

ABSTRACT

DIFFERENCES IN SUCCESS AND PARTICIPATION RATES BASED ON SEX IN
ACADEMIC DEBATE

By
Jason Cruce Trice

This project begins with an in-depth look at the existing body of literature
pertaining to differences in participation and success in academic debate based on sex.
From the deﬁciencies in existing research, research questions, designed to investigate
changes in participation and success rates attributable to sex since the topic merger of
CEDA and NDT style debate, attempt to discern the effects of discrimination and
determine if the topic being debated impacted the measures of success in any way.

A study analyzing three national circuit college tournaments each year over a
four-year period is analyzed and discussed. Results of this study clearly indicate that
disparate participation and success rates women experience relative to men continue to
plague national circuit tournaments of CEDA and NDT style intercollegiate policy
debate. Men outnumber women competitors by a factor of 3:1. Four-ﬁfths of the judges
placed in the high-high debates analyzed in this project were men. Men are 7 % more
likely than women to win an extra debate, which could make the difference between
advancing in the tournament and being eliminated, and at least part of this variance is
attributable to sex factors rather than the factors incorporated under the rubric of talent.

This study conﬁrms that intercollegiate policy debate is still plagued by inequity
based on sex. Self-reﬂection is encouraged to improve recruiting, scholarship
distribution, hiring practices, judging criteria, and travel decisions to make the debate

community an environment more conducive to participation and success by women.

Copyright by
JASON CRUCE TRICE
2001

TABLE OF CONTENTS

I. INTRODUCTION ................................................................. 1
II. LITERATURE REVIEW ........................................................ 3
Argumentation and Verbal Aggression .................................. 3
Self-Conﬁdence ............................................................. 5
Opinion Leaders ............................................................. 8
Success in Debate ........................................................... 9
Literature Summary ......................................................... 17
III. RESEARCH QUESTIONS .................................................... 21
Topic #1: Participation Disparities ....................................... 21
Topic #2: Disparities in Win-Loss Records ............................. 22
Topc #3: Disparities in Speaker Points .................................. 23
IV. METHODS ....................................................................... 24
Subjects ....................................................................... 24
Procedures .................................................................... 25
Variables ..................................................................... 26
Statistical Design ............................................................ 28
V. RESULTS ......................................................................... 32
Participation Disparities .................................................... 32
Win-Loss Record Model ................................................... 36
Speaker Points ............................................................... 41
VI. DISCUSSION ................................................................... 45
Participation Disparities ................................................... 45
Disparities in Win-Loss Records ......................................... 51
Disparities in Speaker Points ............................................. 53
Limitations ................................................................... 57
VII. CONCLUSIONS ................................................................ 61
APPENDIX A ......................................................................... 64
APPENDIX B ........................................................................ 66
BIBLIOGRAPHY .................................................................... 69

iv

LIST OF TABLES

TABLE 1: DEBATER PARTICIPATION BY TOPIC ......................... 34
TABLE 2: JUDGE PARTICIPATION BY TOPIC ............................. 35
TABLE 3: LINEAR REGRESSION MODEL WITH WIN-LOSS ........... 36

RECORD AS THE DEPENDENT VARIABLE

TABLE 4: LINEAR REGRESSION MODEL OF INTERACTION ......... 39
TERMS WITH WIN-LOSS AS THE DEPENDENT VARIABLE

TABLE 5: LINEAR REGRESSION MODEL WITH WIN-LOSS ........... 40
RECORD AS THE DEPENDENT VARIABLE FOR TEAM SEX
EQUALIZED SAMPLE

TABLE 6: LINEAR REGRESSION MODEL WITH SPEAKER ............ 42
POINTS AS THE DEPENDENT VARIABLE

TABLE 7: LINEAR REGRESSION MODEL WITH SPEAKER ............ 43
POINTS AS THE DEPENDENT VARIABLE THE SAMPLE
EQUALIZED BY TEAM SEX

LIST OF FIGURES

GRAPH I: MEAN WIN-LOSS RECORD BY SPEAKER POINTS......... 37
GRAPH 2: MEAN SPEAKER POINTS BY DEBATER-CRITIC- .......... 44
TOPIC INTERACTION

GRAPH 3: MEAN SPEAKER POINTS BY TOURNAMENT ............... 55

GRAPH 4: MEAN SPEAKER POINTS BY TOPIC ........................... 56

vi

Sex-Based Disparities in Debate

1. Introduction

Women’s participation in academic debate has in many ways mirrored the
changing social role of women in society at-large. A few decades ago, women could not
publicly express opinions without a barrage of criticism from both men and women
(Richmond & McCroskey, 1975). Accordingly, women were excluded from academic
debate altogether until some time this century (Chenoweth & Good, 1972). Even after
women were ﬁnally permitted to participate in debate they were regarded as inferior
debaters. Thirty years ago debate tournaments commonly provided separate divisions
for the two sexes. This practice was grounded in the assumption of male superiority in
forensic endeavors (Hensley & Strother, 1968). While sex-based divisions are a thing of
the past, recent studies reveal that disparities remain between women and men in both
participation and performance in academic debate (Stepp, 1997; Bruschke & Johnson,
1994)

Academic debate provides an ideal context to evaluate the extent to which
women’s ideas and arguments are given equal weight to those of men. Modern policy
debate, as practiced by the National Debate Tournament (NDT) and the Cross
Examination Debate Association (CEDA), pits two teams against one another on
opposing sides of a predetermined topic. Each team, comprised of two individuals,
presents arguments in direct contradiction to those advanced by the opposing team. In
this adversarial environment, countless situations are created in which one’s arguments
are challenged and evaluated by members of both sexes. A judge or panel of judges
evaluates the substance of the arguments in order to render a decision determining the

winning team of each debate. In addition, judges assign speaker points and ranks to each

Sex-Based Disparities in Debate

individual. Thus, in addition to providing a natural context to evaluate differences in the
amount of credibility assigned to arguments based on sex, convenient quantiﬁable
measures exist.

This project seeks to examine differences in levels of success in policy debate
based on sex. That is, the goal is to determine the extent to which the performance
measures used in modern debate evaluate women differently than men based on sex
alone. The study will analyze the effects that speaker sex, partner sex, judge sex and
topic have on win-loss record and the assignment of speaker points after controlling for
the inﬂuence of talent. The outcome will shed light on whether or not male dominance
continues to pose challenges to the debate community. If judges continue to evaluate
women and men differently based on sex alone, only by recognizing the problem can
individuals reevaluate their own conscious and subconscious practices that may
contribute to these disturbing trends.

This thesis is divided into four sections — the literature review, research questions,
research methods, and a discussion of the results. The literature review will begin by
examining several studies relevant to sex differences in social inﬂuence in general before
focusing more acutely on research that has been done on sex differences in academic
debate. The research questions portion describes the deﬁciencies in existing research that
this study seeks to address. The methods section describes the study conducted herein.

Finally, the results, their implications, and suggestions for future studies are discussed.

Sex-Based Disparities in Debate

11. Literature Review

As expected, the social inﬂuence literature on sex differences is quite extensive;
the studies on sex differences discussed herein merely scratch the surface. Many of these
studies establish an empirical basis for the assumption that men tend to employ different
communication tactics than women. Research has shown that men: tend to be rated as
more dynamic speakers than women (Mulac et al., 1986); interrupt women to control
conversations (Tannen, 1994; Spender, 1980); employ less indirection than women
(Tannen, 1994); and control the topic choice in conversations with women (Spender,
1980). One study (Larson & Vreeland, 1985) shows that some of these same
conversational differences exist between men and women during the cross-examination
periods of academic debates. Substantial research during the past three decades indicates
that the majority of societal changes occur through interpersonal communication
(Richmond & McCroskey, 1975). If this is true, the possibility that men still control the
primary vehicle for social change in academic debate warrants considerable concern.

This section will focus on past research that has addressed sex differences in the
areas of social inﬂuence and debate. Topics include: sex differences in
argumentativeness and level of verbal aggression; disparities in self-conﬁdence between
women and men; whether men or women serve as opinion leaders on different topics; and
ﬁnally, different styles and levels of success attained in debate based on sex.
Argumemtion and Verbal Aggression

Infante (1989) conducted a study analyzing sex differences in argumentativeness.
Argumentativeness differs from verbal aggression in the locus of attack. In the former,

criticism of the argument is the focus while the latter involves assassinations of the

Sex-Based Disparities in Debate

personal character of the individual making the argument. Infante cites previous studies
that illustrate the possibility those individuals more inclined toward argumentation were
more likely to be the target of verbal attacks. Further, studies have shown that males tend
to be more verbally aggressive than females. If becoming argumentative entails
increased verbal attacks, students may refrain from using argumentative abilities.
Infante’s study seeks to further clarify the conditions under which argumentativeness
triggers verbal aggression. Preferences for the use of verbal aggression were analyzed as
an effect of an individual's sex, as an effect of an adversary’s sex, and whether or not the
adversary employed verbal aggression.

Subjects included 200 students at a large mid-westem state university divided
evenly between males and females. The subjects were randomly assigned to the four
testing conditions - an equal number of males and females were assigned to each
condition. Each subject was provided a book containing a persuasive situation in which
the following variables were being manipulated: the sex of the adversary and whether
argument or verbal aggression was employed. The situation consisted of three exchanges
in which the subject had a range of message strategies to choose from ranging from
rewarding, punishing, argumentative, to verbally aggressive.

When the adversary responded argumentatively, no difference between male and
female responses occurred. However, when the adversary responded with verbal
aggression male subjects were more likely to select a verbally aggressive strategy while
females were more likely to pursue an argumentative strategy. Whether the adversary

was male or female had no effect on preferences for message strategies. Infante’s results

Sex-Based Disparities in Debate

were consistent with previous research indicating that males are more verbally aggressive
than females, especially when provoked by verbal aggression exhibited by an adversary.

The results of Infante’s study may provide insight into how to avoid provoking
verbally aggressive strategies. Since males are more likely to respond to verbal
aggression in kind, the avoidance of verbal aggression in interactions with males will
likely reduce verbally aggressive responses.

The potential implications for debate are signiﬁcant. It is possible that judges
reward male debaters for verbally aggressive strategies. Forms of verbal aggression, such
as ad homonym attacks, may undermine the perceived credibility of an opponent and may
be perceived as harmless verbal sparring that enhances the enjoyment of the debate round
in the eyes of the judging community. If male debaters are likely to employ more
verbally aggressive strategies than women and are subsequently rewarded for this
behavior by judges, this may provide an explanation for discrepancies in success rates
between males and females, although ﬁirther studies would be required to test this
hypothesis.

Self-conﬁdence

Substantial research has examined success and failure attribution, sex, and self-
conﬁdence; however, few studies have examined these factors in a communication
setting. Moreover, most data focusing on sex differences in the attribution process are
dated given the social changes witnessed over the last decade. Andrews (1987) performed
a study to: determine if a sex difference exists in reported self-conﬁdence in approaching
a persuasive communication task; investigate the possibility of a sex difference in the

self-rating of a performance after the communication task; determine if the sexes differ in

Sex-Based Disparities in Debate

their attribution of success or failure to internal or external factors; and explore
differences in arguments advanced by the opposite sexes.

Andrew’s study (1987) included eighty subjects, evenly split between males and
females, were selected from a large mid-western university. Each subject was provided a
copy of the Kidney Machine Decision case in which the subject must choose one of five
candidates to receive treatment necessary to save their lives due to constraints on medical
resources. The subjects are notiﬁed that they will be asked to justify their decision to a
graduate student. The sex of the listener is a manipulated variable, and an equal number
of males and females are assigned to each condition. Prior to presenting their decision to
the graduate student, the subjects are given a conﬁdence test. The listener then ranks
their performance. After the presentation, the subject is asked to rate his or her own
performance and explain reasons for their success or failure of their persuasive endeavor.
Finally, the transcripts were coded for argument types employed by each subject.

Andrew’s study did not identify any differences between individual confederates,
providing assurance that variability in listener trains is not a source of error. Men were
more conﬁdent prior to presenting in front of a male, while women exuded more
conﬁdence in expectation of appealing to another woman. The ratings from the graduate
student listeners indicated that no success difference in actual performance occurred
between the sexes. In retrospect, men perceived themselves as having argued more
successfully than did women, regardless of the sex of the confederate. Men were more
likely to attribute success to natural ability, while women tended to attribute their success
to hard work. On the other hand, men were more likely to attribute failure to lack of

effort on their part as well as on external factors such as difﬁculty of the task, whereas

Sex-Based Disparities in Debate

females attributed failure to a lack of communication abilities on their part. Arguments
advanced by males tended to be more based in the criteria outlined by the case material
while females tended to advance arguments focusing on familial or social contributions
rather than the preordained criteria. Moreover, both sexes were more likely to use family
or social contribution arguments to appeal to a female audience as compared to a male
audience.

The combined implications of the above ﬁndings are intriguing. Females
approach a task with less conﬁdence and rate themselves lower in retrospect than men
despite the lack of difference in actual performance as measured by the rater. Thus,
negative self-expectations do not appear to impede performance. This also suggests a
perceptual distortion by both men and women on their performance — men evaluated
themselves higher than the rater and women evaluated themselves lower than the rater.
Men take responsibility for their failures, but are also more likely to point to external
environmental factors as explanations for shortcomings.

Future studies are needed to explore the reasons for differences in the way
females and males perceive their communicative abilities and performances, as well as
the effectiveness of different argumentative techniques in various contexts. The
relevance to debate is apparent: if raters found no difference in performance between
males and females, then one would not expect debate judges to evaluate males and
females differently. However, research documenting the relative success of males and
females in debate discussed below contradicts this notion. Andrew’s study is limited to a
single performance by the subjects. Debate involves multiple performances and skills

that are acquired over a period of several years; all of which involves repeated

Sex-Based Disparities in Debate

interactions with the same partners, opponents, and judges. Perhaps if women have lower
self-conﬁdence in their performance both before and after each debate, this has a
cumulative effect that impacts their long-term performance and debate styles. This study
also highlights the possibility that debating in front of a judge of the same or opposite sex
may have an impact on the confidence of the speaker. Andrew’s results also indicate
differences in the type of arguments advanced by males and females that could alter a
judge’s evaluation of which team won or lost, as well as an individual speaker’s
performance in a competitive forum such as debate.
Opinion Leaders

Opinion leaders - people whom others seek out for advice on a particular topic —
play a crucial role in the communication process that is responsible for social change.
These people are perceived as more credible and competent providing them with
substantial inﬂuence. If the vast majority of opinion leaders in our society are men, the
social and political environment will continue to be deﬁned by men. One study
(Richmond & McCroskey, 1975) examines the different issues for which men and
women tend to serve as opinion leaders. This study reveals signiﬁcant progress towards
reducing the sex differences in opinion leaders. Women, who used to suffer criticism
from both sexes any time they offered a public opinion on almost any subject, are now
primary opinion leaders on many topics such as fashion. However, the study concludes
that men are the primary opinion leaders on issues such as public policy and politics. In
these areas, not only were male sources considered more credible overall, but also the
differences were more pronounced for women raters. In short, female deference to male

opinion leaders on political issues was signiﬁcant. If this holds in a debate context,

Sex-Based Disparities in Debate

female judges might contribute more to any disparity in assessing male and female
debaters than male judges on public policy issues. The study proposed below examines
differences between the success of male and female debaters over the course of four years
with a different topic for each year. An interesting comparison should examine any
differences in success rates based on sex that can be attributed to the topic material.
Success in Debate

Research (Stepp, 1997) indicates that women were underrepresented in
intercollegiate debate in the 1980’s and 1990’s. While improvements in female
participation are clear, the proportion of female debate participants still falls short of the
percentage of undergraduate students that are women. Stepp (1997) examines data from
ﬁve CEDA National Tournaments from 1991-1995. This research indicated the ratio of
male to female debate directors and coaches exceeds the proportion of male college
faculty members and graduate students respectively. This under-representation of female
role models might alienate female participants.

Stepp also addresses the relative success of male and female participants at the
CEDA National Tournament over the years. Her analysis concludes that the vast .
majority of ﬁnal round participants, tournament winners, and receivers of top twenty
speaker awards were white males. These numbers were not merely reﬂective of the fact
that males make up a majority of the participants. The proportion of female participants
in the tournaments signiﬁcantly exceeds the proportion of females receiving the
aforementioned accolades. While this study suggests that males, at the expense of
females, attain a disparate level of success in debate, it fails to examine the causes of this

disparity. Perhaps the explanation for this difference is simply the superior skills

Sex-Based Disparities in Debate

exhibited by men in policy debate. To determine whether this is the case, or whether any
part of the difference is attributable to the sex of the participants, requires further study.

Hensley & Strother (1968) published the ﬁrst study that attempted to determine
whether the differential in debate success was attributable to sex. They found that male-
male teams had no signiﬁcant advantage over female-female teams in terms of winning
and losing debates. However, the study concluded that male-female teams performed
better than both male-male teams and female-female teams. These results were generated
by comparing the performance of an equal number of each of the three possible sex
conﬁgurations to what would be expected due to chance alone, 50%. Multiple possible
explanations exist for these results. Perhaps the dynamic of mixed sex teams provided
more versatility than same sex teams. The authors suggested the possibility that coaches
only placed females with the best males, viewed as capable of making up for the
deﬁciencies of their female counterpart. Again, this study is simplistic and does not
attempt to control for any variables such as skill in order to isolate the effect sex has on
success.

Examining the rank order of the four individuals in the round provides an
improvement over the win/loss procedure for comparing males to females. For example,
the win/loss record does not account for the possibility that the majority of the judges
found one debater to be consistently better than her or his partner.

A study (Hayes & McAdoo, 1972) was conducted examining data collected from
several collegiate debate tournaments from 1967 through 1970 ﬁelding a wide range of
experience levels. Debates including only male or only female participants were

excluded from the sample. During each debate, individual speakers were assigned a rank

10

Sex-Based Disparities in Debate

order from 1 through 4. Both males and females should be evenly distributed among the
four ranks according to random chance. This study analyzes whether or not there is a
difference between the actual distribution and the expected distribution for either males
or females.

Results revealed that women received more rankings of “one” and “three” than
would be expected and less rankings of “two” and “four” than would be expected due to
random chance. Conversely, males received a disproportionately high number of “two”
and “four” rankings and fewer rankings of “one” and “three”.

Several possible explanations exist for these results. First, female debaters may
perform better than men. This may not be indicative of any superior natural ability on the
part of females, but merely those women participating in the study exhibited a higher
level of debate performance. Second, the ﬁndings may be a result of some factor on the
part of male judges. Previous studies demonstrate that men in general are more lenient in
their evaluation of women. Thus, it may be the case that the predominantly male judging
pool possesses a leniency factor when evaluating females. This study only addresses the
fact that there is a difference between the speaker rating performance between males and
females without addressing the reasons for this difference. Further research addressing
the reasons for the differences — such as different scholastic achievement between male
and female debaters, differences in high school debate training, and differences in verbal
abilities — is recommended.

Research analyzing the relationship between sex, position, and side to the
outcome of competitive debate rounds is sparse. Previous studies researching these

relationships have several limitations: They are limited to a single forum (i.e. college

Sex-Based Disparities in Debate

debates). Some of the studies are limited to only a few tournaments and all are limited to
a single topic year. The study method did not analyze all ballots and the relationships
between all sex pairs and debate wins (i.e. the winning records for male-male, female-
female, and female-male teams). These studies analyzed the ranking data only (how
individual members performed in direct comparison to the other debaters in the round)
and not quality points which attempt to rate a speaker’s general ability. Finally, many
studies, which attempt to analyze sex issues in debates, do not consider a closed system
of debates. A closed system in the context of debate means every debate is analyzed for
every relevant attribute (i.e. for every win there is a loss, for every person ranked “1”
there is one person ranked “2” etc.).

Subsequent studies have become increasingly complicated in an effort to address
limitation identiﬁed in previous studies. For example, Rosen, Dean, and Willis (1978)
designed a study to: examine a different population, i.e. high school debaters; analyze
debater ranking and quality points in addition to win-loss records; use data from closed
debate systems in which an entire tournament is analyzed; and evaluate results from more
than one debate topic. This study analyzed three high school debate tournaments over the
course of two years in order to reduce the chance the topic might interact with the three
independent variables, which included sex combinations of the two-person debate team,
side (affirmative or negative), and position (i.e. ﬁrst speaker or second speaker). For
each tournament, all ballots were analyzed and the total number of males and females
taking part was recorded, in addition to the number of male-male, female-female and

male-female teams. Victories and losses for the three sex combinations were recorded.

Speaker ranks and quality points were compared to sex of debater, side and position.

12

Sex-Based Disparities in Debate

The proportion of wins and losses for afﬁrmative and negative sides did not differ
for the two topics, and results indicate there is no reliable difference between the numbers
of afﬁrmative or negative wins. The number of male-male, female-female, and male-
female teams did not differ from chance considering the total number of participants and
the fact that a majority of the participants are male (65% in this study). This study did
indicate mixed-sex teams were more likely to have won their debates. The relationship
between win-loss records and rankings of individual teams produced a signiﬁcant effect.
In this study, teams receiving ranks of l and 2 or 1 and 3 always won. When the
distribution of ranks was examined in relation to position, the analysis revealed that the
second position seems to have an advantage over the ﬁrst position for either side of the
debate. The ﬁnal analysis compares quality points, assigned to each of the four debaters,
to the three variables of sex, speaker position, and side. None of these comparisons
yielded interactions approaching signiﬁcance.

The implication for this study is that the perception of male superiority to females
in high school debate is unfounded. Results also indicate that side is not related to
winning the round; however, it appears to be advantageous for the individual debater to

661,,

be afﬁrmative, as the afﬁrmative debaters are more likely to receive the ranking of or
“2.” The conclusions of this study reconcile the inconsistency between the facts that
there is no overall advantage to being afﬁrmative or negative, but that the afﬁrmative
debaters are more likely to receive higher ranks in the round. It is theorized that the
negative debaters are more likely to win in a “split decision” and that this statistic tends

to balance the overall comparison between the afﬁrmative and the negative. Position

(ﬁrst speaker and second speaker) could not be compared to winning since each side

13

Sex-Based Disparities in Debate

contains both positions; however, the second speaker appears to receive high speaker
points than the ﬁrst speaker position.

Brushke and Johnson (1994) collected data from result sheets from several large
intercollegiate NDT tournaments during the 1989 through 1992 debate seasons.
Tournament locations were geographically diverse. To account for the covariance
between talent and success only high-high power matched rounds were used so that the
teams were matched as evenly as possible. In addition, both prior records at the
tournament and victory in the round were treated as covariates to isolate results due
primarily to sex. Four independent variables were coded — speaker sex, partner sex,
judge sex, and topic side. These variables were crossed with one dependent variable,
speaker points.

An AN OVA showed that both previous record and victory in the round had
signiﬁcant effects on speaker points. Again, these two variables were accounted for by
treating them as covariates. Female judges hearing teams on the negative showed
signiﬁcant effects for speaker sex, partner sex, and the interaction between speaker and
partner sex. No signiﬁcant difference existed for female judges hearing teams on the
afﬁrmative. Male judges, when hearing teams on either the afﬁrmative or the negative,
exhibited a signiﬁcant interaction effect, but neither main effect. Female judges gave
females debating on the negative lower points, especially when debating with a female;
no difference exists on the afﬁrmative. Male judges gave same sex teams higher speaker
points than mixed-sex teams but no difference exists between male only and female only

teams, regardless of whether the team was afﬁrmative or negative.

14

Sex-Based Disparities in Debate

Overall Brushke and Johnson (1994) draw three main conclusions. First, females
receive fewer speaker points than their male counterparts due to their sex. That there are
noteworthy examples of successful female debaters or that discrimination is not apparent
from a cursory scan of a results packet from any one tournament does not deny that
females must outperform males to receive equivalent scores. Second, female judges are
the primary contributors to the trend of lower speaker points. Third, teams comprised of
same sex participants are evaluated as better than mixed sex teams, especially by male
judges. Bruschke & Johnson’s study is the most complete study encountered during the
research for this project. By controlling for talent, side of the topic, and looking at cross-
sex effects of the speaker, the partner, and the judge this study addresses some of the
limitations that were apparent in preceding studies. However, this study is not without
limitations. Bruschke & Johnson, by their own admission, make no attempt to analyze
the causes of the sex-based success difference identiﬁed in their study.

Larson and Vreeland (1985) conducted a study of CEDA debate at four California
tournaments that focused on different styles and traits exhibited by women and men in
debates. Exchanges during cross-examinations periods were recorded and subsequently
coded for attempted interruptions, successful interruptions, and question type. Larson
and Vreeland conclude that women asked more open-ended questions than their male
counterparts who relied more on closed or leading questions. In addition men tend to
attempt interruptions more on balance than women, and men were more likely to
interrupt a woman during an exchange than a man. On the other hand, women had a
higher success rate of regaining control and rebuffmg attempts at interruptions than did

males. Perhaps women have simply learned to deal with the tendency of male debaters to

15

Sex-Based Disparities in Debate

interrupt more during the cross-examination periods. Larson and Vreeland theorize that
these differences indicated a superior ability on the part of males to control cross-
exarninations. This could explain discrepancy in participation or success rates, but no
test of this theory is conducted in the Larson and Vreeland study. Further investigation
would be necessary to make the connection between interruptions and question type and
success in debate.

Knutson (1996) is the one attempt surveyed during this project that attempts to
explain the causes of disparate participation and success in academic debate. She
theorizes that the reasons that women are underrepresented and achieve less success in
debate is attributable to the public policy focus of the topics, especially those focused on
the military, and the war metaphors surrounding debates. According to Knutson’s theory,
women are forced to step outside their traditional roles and enter what has traditionally
the male sphere of debates about public policy and military security. In addition, the
terminology of debate relies heavily on the metaphor between arguments and wars:
arguments are classiﬁed as “offensive” or “defensive”, one can “win” or “lose”
arguments, people “attack” their opponents arguments, debaters employee “strategies”,
etc. These metaphors, Knutson argues, force women to engage in rhetoric that less
familiar to them forcing them to speak men’s language, since women have been
historically excluded from military service and military policymaking. Thus, the topics
of the debates and the metaphors utilized to describe arguments create a masculine
environment that discourages female involvement and limits their ability to succeed when
compared to the male standard, according to Knutson. Although the works cited for this

piece offer extensive support for this theory, Knutson did not perform any quantitative

Sex-Based Disparities in Debate

study to test her theory. Perhaps looking at the effect the subject of the debate topic has
on disparate levels of participation and success could provide support for this theory.
Literature Summa_ry

While past research provides some intriguing insights, further investigation is
clearly justiﬁed on several fronts: to update the ﬁndings of past research to determine
whether there continue to be different levels of participation and success based upon sex;
to further determine whether the differences exhibited between men and women are
attributable to discrimination or some other factor such as talent or work ethic; and to
investigate possible causes of the observed disparity such as the topic being debated,
argumentative choices, or stylistic traits.

First, all of the research on debate described above is dated: Brushke and
Johnson’s study (1994) analyzes data from 1989 — 1992; Stepp (1997) analyzes data
ranging from 1991 — 1995; everything else can comfortably be categorized as archaic.
Much has changed in the intervening time period that could have altered the sex
dynamics in academic debate. The older studies focus on individual speaker ranks (one
through four), a measure that has little importance in modern debate given the advent of
the 30-point quality speaker point scale. Hopefully, both the debate community and
society at-large have continued to evolve by embracing a more egalitarian perspective.
The turnover rate for debate is high since one can only compete during the four years of
college; hence, the debaters of today comprise a completely different set of subjects than
those covered in all previous studies. In contrast, the judging pool is more stable since
many people have made careers out of coaching debate. At the same time, however,

there is undoubted a signiﬁcant turnover among judges since many debaters coach and

I7

Sex-Based Disparities in Debate

judge for a year or two once their eligibility expires before moving on to other pursuits.
The ﬂuid composition of the debate community may have ushered in fresh perspectives
that altered the bias indicated by past studies. Most importantly, the formal merger of
CEDA and NDT through the adoption of a joint topic process in 1996 represents a
dramatic cultural shift that changed the composition of both the competitors and the
judges in the world of academic debate. No study examining differing levels in debate
success attributable to sex has been conducted since the merger, and no study done prior
to the merger included subjects from both CEDA and NDT communities. It was widely
held at the beginning of the merger that CEDA judges typically awarded higher speaker
points overall than did their NDT counterparts. Thus, the merger stood to alter both the
social composition of the debate community and the distribution of speaker points. All of
these changes in the intervening period since the studies discussed herein were conducted
justify a more timely study to investigate modern differences in debate success attributed
to sex.

Second, only the Brushke and Johnson study attempt to control for talent when
analyzing the effect of sex on speaker points. While all the previous studies show that
there is a disparity between the results of men and women, without controlling for talent,
these studies do not conclusively prove that sex discrimination exists in debate. It may be
the case that men outperform women because they possess more natural talent, greater
experience, or they work harder to research and prepare for debates. While these ﬁndings
are important, it is also necessary to distinguish between disparate performance based on

talent or some similar factor and differences attributable to sex discrimination. Further

l8

Sex-Based Disparities in Debate

investigation utilizing methods that control for talent is required to better determine the
source of any disparity.

Finally, no attempt has been made to determine the cause of the disparate results
between men and women in debate by any of the studies described up to this point. Most
of the studies cited above focus exclusively on debate success based on sex without
looking at the underlying causes of the differences. Only the Larson and Vreeland (1985)
explores stylistic differences in debate based on sex, but this study does not analyze the
possible effects the stylistic choices may have on success. It is also possible that men and
women make different argumentative choices, rely on different strategies, or employ
different styles to win than men, and that this difference accounts for the discrepancies in
success. The Andrews study (1987) examines the different argumentative choices made
by the sexes, and concludes that women are more likely to advance arguments that appeal
to family or social value while men’s arguments focused more on the pre-ordained
criteria in the study. Both sexes were more likely to make arguments focusing on social
contributions or family when appealing to a female evaluator according to Andrews. It is
possible that men rely more on particular types of arguments in academic debate that tend
to be more appealing to the overwhelmingly male judging pool. Further study is needed
to investigate the possibility that women and men make different argumentative decisions
and that these differences have an effect on success. Richmond and McCroskey’s study
(1975) indicates that women and men are perceived as more or less credible and
competent based on the topic being discussed. This study revealed that women are
treated as opinion leaders when the topic was fashion, but that men were deemed more

credible on topics concerning public policy and politics. Since many CEDA and NDT

l9

Sex-Based Disparities in Debate

debates focus on political subjects that require public policy decisions, the topic of debate
could have a major impact on different levels of participation and success between the
sexes. No study encountered during this project addresses this possibility. It is also
possible that the changes that have occurred in society since the Richmond and
McCroskey study was conducted potentially afforded women more credibility on some
public policy issues than others. Further investigation of these possibilities is clearly

warranted.

20

Sex-Based Disparities in Debate

111. Research Questions

While the present study does not purport to resolve all of the deﬁciencies in the
existing body of research, it is hoped that the study herein makes a positive contribution
in answering some of the remaining uncertainties. Speciﬁcally, this study seeks to
examine whether or not disparate levels of participation and success based on sex
continues to plague academic debate. This project also seeks to discern whether or not
any disparities that emerge are attributable to discrimination. Finally, this research
examines the effect the yearly topic has on varying success levels.

Predicting that there is a difference in success based on sex, or worse that
discrimination is rampant in the debate community, is an uncomfortable proposition.
Beginning with hypotheses, with the goal of determining if the data supports or rejects
the hypotheses, does not seem like a good starting point for investigative research; an
open ended approach that does not seek to prove any suggested theory provides a more
objective starting point. In addition, ﬁnding that there is a difference in participation or
success based on sex is not the only possible signiﬁcant result. Indeed, if no statistically
signiﬁcant differences in success or participation rates based on sex, the results would
indicate a significant and positive result since it illustrates an improvement in the debate
environment for female participants and constitutes a major change from past ﬁndings.
As such, the following research topics and questions are posed for research and
examination:

Topic #1: Participation Disparities
a) Is there a difference in participation based on sex among competitors in

intercollegiate policy debate since the topic merger of CEDA and NDT?

21

Sex-Based Disparities in Debate

b) Is there a difference in actual team composition (all male, all female, and
mixed sex) from what one would expect based on the overall participation rates?

c) Is there a disproportionate representation of the sexes among the judging pool
in intercollegiate policy debate since the topic merger?

(1) Does any tournament exhibit a greater or lesser tendency toward equal
representation of the sexes among either competitors or judges?

e) Does any topic induce greater or lesser equality of representation among
debaters or critics?

Topic #2: Disparities in Win-Loss Records

 

a) Do debaters of both sexes attain equal levels of success in debate as measured
by win-loss record since the topic merger?

b) Does the sex of the speaker’s partner alter the level of success as measured by
win-loss record in the post-merger environment?

c) If disparities are found between female and male debaters, can these ﬁndings
be attributed to sex or are they the result of some other factor such as talent?

d) Does the sex of the judge produce a bias in the way debates are evaluated under
the CEDA-NDT merger?

e) Does the interaction of the speaker sex, partner sex, and judge sex amplify or
mitigate the main effects of each of these variables?

f) Are any observed disparities in success as measured by win-loss record
attributable to observed differences in participation rates between men and women, or are
the differences in win-loss records disproportionate after accounting for differences in

participation rates?

22

Sex-Based Disparities in Debate

Topic #3: Disparities in Speaker Points

a) Do debaters of both sexes attain equal levels of success in debate as measured
by speaker points?

b) Does the sex of the speaker’s partner alter the level of success as measured by
speaker points?

c) If disparities are found between female and male debaters, can these ﬁndings
be attributed to sex or are they the result of some other factor such as talent?

(1) Does the sex of the judge produce a bias in the assigning of speaker points?

e) Are any observed disparities in success as measured by speaker points
attributable to observed differences in participation rates between men and women, or are
the differences in speaker points disproportionate after accounting for differences in
participation rates?

f) Does the tournament have an effect on the assignment of speaker points?

g) Do speaker points vary depending on the topic being debated?

h) Do the interactions of speaker sex, partner sex, judge sex, tournament, and

topic amplify or mitigate any observed main effects?

23

Sex-Based Disparities in Debate

IV. Methods

To address the research questions identiﬁed in the previous section, a study of
three major intercollegiate tournaments held year during the ﬁrst four topic-years after
the CEDA-NDT merger was concluded.
Siam

Three major national caliber tournaments representing geographically diverse
locations were included in the study. These tournaments were hosted by: the University
of Northern Iowa, the University of Kentucky, and Northwestern University. These ’
tournaments were selected because they have consistently been major tournaments
attended by most all, avid college policy debaters. While this sample probably doesn’t
include every single individual that participated in debate during the four years of the
study, these tournaments undoubtedly reﬂect a very substantial sample of the overall
population of collegiate policy debate that this study seeks to examine. Subjects included
all judges and debaters of the aforementioned tournaments from each of the four seasons
between 1996 through 2000. The participants at these three tournaments over four years
yielded 2900 cases for evaluation. Debaters must be currently enrolled undergraduate
college students. Eligible judges must have either already used all four years of debate
eligibility or have completed an undergraduate degree. Most judges are former debaters
that have become graduate assistants, coaches, or administrators employed by the various
debate programs competing at each tournament. Many judges and debaters participate in
more than one tournament over more than one year in this study. Since many debaters
judge debates once their eligibility is exhausted, some subjects may participate as both

competitors and critics; others may participate at only one tournament.

24

Sex-Based Disparities in Debate

Procedures

Each tournament collects data for each debate that occurs. Each judge ﬁlls out a
ballot (see appendix A) in which they record one and only one winner for each debate,
speaker points for each individual ranging from 0-30, and rank order of the four
participants in each debate. It is this data that determines speaker awards, teams qualiﬁng
to compete in elimination rounds, the winners of tournaments, and awards for the schools
and teams with the best records throughout each year. The data for all of the preliminary
debates for each tournament is compiled into results packets.

Data from these results packets will be used to perform this study. Debate
tournaments typically begin with between two and four random preset preliminary
debates. Based on the outcome of these debates, the remaining preliminary competitions
are power matched such that teams with equivalent win-loss records are pitted against
one another to the greatest extent possible. Some rounds are powered high high within
brackets such that the top two teams in each win-loss bracket will compete against one
another; others are powered high-low such that the team with the highest speaker points
will debate the team with the lowest speaker points with the same win-loss record. This
study will only examine high-high rounds of each tournament so as to maximize the
extent to which teams competing against one another are evenly matched ensuring the
results are not skewed by the occurrence of drastically lopsided debates between teams
with drastically different levels of experience, talent, preparation, resources, etc. Some
tournaments may be over-represented in the sample due to years in which they had two
high-high power matched rounds instead of only one. Results listed in cases where a

debate did not actually occur due to a bye or forfeiture were excluded from the analysis

25

Sex-Based Disparities in Debate

since given the lack of a judge and the fact that the speaker points reﬂected an average of
the speaker points in the other seven debates rather than an assessment of performance in
the high-high debate.
Variables
Data for several variables was coded based on the information provided in the
results packets. For each speaker listed in the results packet the following variables were
coded:
0 Tournament Identiﬁcation: University of Northern Iowa, University of Kentucky,
or Northwestern University.
0 Topic Being Debated: United States policies regulating environmental pollutants;
United States security assistance policies toward Southeast Asian nations; United
States civil rights policies protecting against race and/or gender discrimination;
and United States economic sanctions policies toward “rogue” nations.
0 Sex of the Speaker: Male, female, or unknown.
0 Sex of the Speaker ’s Partner: Male, female, or unknown.
0 Sex of the Judge: Male, female or unknown.
0 Judge ’3 Decision: Whether the debate at hand was won or lost by the subject.
0 Overall Win-Loss Record: All tournaments in this study had eight debates, so the
number of wins possible for each subject could range from zero to eight.
0 Speaker Points: The quality points ranging from zero to thirty assigned to each
speaker by the judge.
The sexes of the various participants in the debates were determined by name

recognition. In many cases, ﬁrst hand information was available to determine the sex

26

Sex-Based Disparities in Debate

based upon the familiarity of the author with the debate community. In other cases, the
use of gendered ﬁrst names made the sex of the subject obvious. In situations where the
sex of the individual was not known based upon recognition or gendered naming, efforts
were made to determine the accurate sex by contacting coaches from the school with
which the student was afﬁliated. In cases where the sex of the subject was still
indetemrinate, no data was entered for that variable. The SPSS statistical software
automatically excludes cases as invalid when information is missing for selected
variables.

It should also be noted that there is a 100% overlap between the sex of the speaker
and the sex of the partner variable. For each debate, there are four participants that are
each treated as separate cases. Hence, every speaker has a partner, and in turn functions
as a partner for another speaker. The partner variable was included to allow a
determination of the sex combinations of the teams in the study. Based upon the
aforementioned variables, several additional variables were created to parcel the data in
order to examine interaction effects:

0 Team Sex: Essentially a variable that accounts for the interaction of speaker sex

and partner sex — all female team, all male team, or opposite sex team.

0 Team Sex * Judge Sex: A variable that parcels out the data for the interaction
between sex of the speaker, partner, and judge — female judging all female team,
male judging all female team, female judging all male team, male judging all
male team, female judging mixed sex team, and male judging mixed sex team.

0 Topic * Debater Sex: Females debating environmental regulations; males

debating environmental regulations; females debating security assistance; males

27

Sex-Based Disparities in Debate

debating security assistance; females debating civil rights; males debating civil

rights; females debating economic sanctions, males debating economic sanctions.

0 Topic * Judge Sex: Females judging environmental regulations; males judging
environmental regulations; females judging security assistance; males judging
security assistance; females judging civil rights; males judging civil rights;
females judging economic sanctions, males judging economic sanctions.

0 Topic * Judge Sex * Debater Sex: Females judging females debating
environment, males judging females debating environment, females judging
males debating environment, males judging males debating environment. These
same parcels are made for the data on the security, civil rights, and sanctions
topics.

Statistical Design

Frequency distributions are employed to determine the participation ratios for
both debaters and debate judges. Similar analysis is used to assess the ratio of all female
teams, all male teams, and mixed-sex teams. The interactions of the topic variable and
the tournament variables are explored. A means comparison is conducted to further
illuminate these ﬁndings.

Speaker points and win-loss record are treated as scale variables and function as
the dependent variables. Speaker sex, partner sex, judge sex, team sex, topic, and
tournament are treated as independent variables, as are the interaction variables of team-
judge sex, topic-debater sex, topic-judge sex, and topic-debater-judge sex.

It is widely recognized that there is a substantial relationship between winning or

losing a debate and the quality of individual speaker points received by the participants.

28

Sex-Based Disparities in Debate

In most cases, debaters that successfully persuade judges to vote for them are also
evaluated more favorably when the judge assigns speaker points. Thus, the decision
about who won or lost the debate at hand is treated as a covariate when analyzing the
effects of sex on win-loss record and speaker points in high-high debates.

Because of the large number of variables addressed in this study, a bivariate
correlation analysis (see appendix B) is used as a data reduction technique. This ﬁrrther
conﬁrmed the choice to use the decision in the debate as a covariate; since results
indicated a signiﬁcant relationship between both speaker points (I = .195, p < .001) and
win-loss record (I = .280, p < .001), but had no relationship to any of the independent
variables. This also excluded several variables from the models. For example, neither
tournament nor topic would be expected to have an effect on win-loss record since each
tournament is structured based on power matching schemes in an attempt to produce a
normal distribution of teams at each level of the win-loss spectrum. The correlation
analysis conﬁrmed that no signiﬁcant relationship existed between win-loss record and
either tournament or topic.

The data is analyzed using multivariate linear regression since multiple causes for
each dependent variable are posited. Regression should be preferred over ANOVA if: a
non-experimental design is used, the number of subjects is unequal at different levels of
the independent variable, covariates are used, classiﬁcation variables are used for the
independent variables, and different types of variables are analyzed together. All of the
above stipulations are true of this study; thus, multivariate linear regression was utilized.
In addition, a means comparison analysis in conducted to facilitate a more easily

digestible discussion of the results.

29

Sex-Based Disparities in Debate

In an attempt to discern whether or not discrimination is a factor in any disparity
reﬂected, win-loss record is used as a control variable in a secondary analysis of the
speaker point model. The bivariate correlation analysis indicated a strong relationship
between record and speaker points (1: = .561, p < .001). Ideally debates are decided upon
objective criteria grounded in the substance of the arguments presented by both teams in
each debate. Thus the cumulative win-loss record of each team should represent an
objective assessment of the quality of each team that accounts for experience, talent,
preparedness, resource advantages, etc. Speaker points are a more subjective measure
that attempts to compare each speaker against the backdrop of the entirety of the pool of
the competition. No formal criteria exists to guide the assigning of speaker points, and it
is widely held that judges employ different standards and utilize different point scales. It
is theorized that by using the more objective measure of win-loss record that reﬂects the
quality of a team over the course of eight debates as a covariate, judge bias in assigning
the more subjective measure of speaker points could be ascertained.

One concern about the research design emerged as a result of the disparity in
participation rates among both competitors and judges. It is possible that the
overwhelming number of males could have skewed the results artiﬁcially indicating that
males were more successful than their female counterparts. To discount this possibility,
the cell sizes of team sex were equalized as they would be assuming equal participation
and random partnership pairings. Two hundred cases from each possible team
conﬁguration - all female, female with male partner, male with female partner, and all
male — were randomly selected to form the team sex equalized sample. This also had the

net effect of equalizing cell sizes for speaker sex and partner sex. Linear regression

3O

Sex-Based Disparities in Debate

results were compared to the results from the complete models to conﬁrm that the results

were not skewed by the disproportionate participation.

31

Sex-Based Disparities in Debate

V. Results

This study analyzes participation rates and possible effects on win-loss record and
speaker points from a variety of variables from speaker sex, partner sex, judge sex, and
topic of debate, and the interaction of these variables. For clarity, this section is
organized around the three research topics and corresponding questions identiﬁed in the
previous section. First, participation rates are analyzed to determine if sex-based
disparities exist among the population of debaters attending the selected tournaments, the
conﬁguration of two person teams, or the judging pool. The annual topic being debated
is analyzed to determine if different topics attract a more or less representative group of
participants. Second, the results of the win-loss record model are explored to determine if
disparities exist that can be attributed to the sex of the debater, partner or judge; and to
determine if any disparities can be attributed to bias or discrimination. Finally, the results
of the speaker point model are presented. The variables analyzed in this portion mirror
those in the win-loss model, except that the tournament host and annual debate topic are
included to assess potential contributions to variance in speaker points.
Participation Disparities

At the twelve tournaments analyzed in this study, males represented the
overwhelming majority of competitors, comprising 76.3 % of the sample population of
debaters whose sex was identiﬁed (11 = 2054). Female debaters constitute only 23.7% of
the identiﬁed population (1] = 639). In short, the number of male debaters exceeded the
number of female debaters by a factor of three. Of the total sample, 7.1% of the

participants’ sex could not be identiﬁed (n = 207) and was excluded from the study.

32

Sex-Based Disparities in Debate

Given this distribution of competitors, it becomes possible to compare the actual
sex composition of teams to what would be expected due to chance alone. Since there
were some cases in which the sex of only one of the partners could be determined, there
were slightly less cases available to assess the interaction between speaker and partner
sex than there were in the participation rates cited above for individual speakers (male 1]
= 2004, female 11 = 618). Given this participation ratio, one would expect 5.6% of teams
to be composed of two females; 36% of teams to be mixed sex teams; and 58.4% of the
teams to be all male due to chance alone. Results showed a slightly higher combination
of same sex teams, both male and female, than would be expected: 8% of teams were all
female; 60.9% of teams were all male; and 31.1% were mixed sex teams. The difference
between expected and actual team composition is signiﬁcant at all three levels of the
team sex variable (x2 = 46.481, p < .001).

The female-male ratio among the pool of judges was even more imbalanced than
among competitors. Females represented a mere 15.8% of the sex-identiﬁed judging
pool (n = 436), leaving the remaining 84.2% of identiﬁed ballots to be completed by men
(n = 2326). Roughly four out of every ﬁve judges was male. Judges of unknown sex,
which were excluded from the sample, accounted for only 4.8% of the sample (n = 138).

Both tournament and topic revealed a signiﬁcant relationship to sex of the critic in
the bivariate correlation analysis (p < .005). What follows is a partitioning of the data by
both tournament and then topic to determine if there was a difference in participation
rates.

The tournament under consideration had no signiﬁcant effect on the disparate

participation rates of debaters. Northern Iowa had a slightly greater proportion of women

33

Sex-Based Disparities in Debate

participants (24.8%, n = 297) than did either Kentucky (23.0%, n = 144) or Northwestern

(22.8%, n = 198). A 95% conﬁdence interval around the means of sex composition of

the subjects at each of the three tournaments illustrates the difference between these

ratios as non-signiﬁcant and that the differences could easily be attributed to error.

On the other hand, there is a signiﬁcant relationship between the tournament

being evaluated and the sex composition of the judges of high-high debates. By

constructing 95% conﬁdence intervals around the mean critic sex in each of the three

tournaments, it can be concluded that the Northwestern Tournament had a higher

proportion of female judges (18.7%, n = 158) than either Kentucky (15.4%, n = 110) or

Northern Iowa (14%, n = 168). No signiﬁcant difference between Northern Iowa and

Kentucky tournaments was identiﬁed, and a 95% conﬁdence interval around the mean

proportion of female judges of the entire study (15.5%, n = 436) indicates that both

Northern Iowa and Kentucky were within expected variance due to sampling error.

Table 1: Debater Participation by Topic

 

 

 

 

 

 

 

m % Females 1] Females Mean ratio Standard Error
Sanctions 26.0% 201 .48 .0315
Environment 25.2% 136 .50 .0375
Security 23.1% 145 .54 .0337
Civil Rights 20.9% 157 .58 .0297

 

 

 

 

 

There was a signiﬁcant difference in participation ratios of women based on the

topic being debated (see table 1). By constructing a conﬁdence interval around the

overall mean sex ratio in this study (u = .53), it can be concluded that the economic

34

 

sanctions topic and the civil rights topic both fall outside the allowable difference that can

Sex-Based Disparities in Debate

be attributed to error. Thus the sanctions topic witnessed a proportion of women

participants that was greater than the overall results by a spastically signiﬁcant margin.

On the other hand, during the civil rights topic the ratio of women to men was lower by a

signiﬁcant degree than the overall results, especially those from the environment and

sanctions topic.

Table 2: Judge Participation by Topic

 

 

 

 

 

 

 

EDI—C % Females 1] Females Mean ratio Standard Error
Civil Rights 23.5% 200 .53 .0297
Sanction 1 5.7% 124 .69 .0260
Environment 10.8% 56 .78 .0272
Security 9.3% 56 .81 .0237

 

 

 

 

 

 

Judge participation also varies substantially based on topic (see table 2). A 95%

conﬁdence interval around the mean ratio of the civil rights topic ranges from .47 to .59,

indicating an amount of variation that can be dismissed as error. The mean ratios the

three other topics fall outside this interval, indicating that the proportion of female critic

participation (23.5%) is signiﬁcantly higher on civil rights than any other topic. In turn,

the conﬁdence interval around the mean ratio for the sanctions topic ranges from .63

to .75, indicating that the topic of economic sanctions drew signiﬁcantly more women

than did either the environment or security. No substantial difference exists between the

environment topic and the security topic in terms of sex distribution.

35

13

Sex-Based Disparities in Debate

Win-Loss Record Model

For this analysis, the tournament and topic variables were excluded from
consideration. In a closed tournament environment, the number of wins always equals
the number of losses across all of the topics. The bivariate correlation analysis conﬁrms
that no relationship exists between tournament or topic and win-loss record. This data
reduction technique justiﬁes excluding the tournament and all topic variables from this
analysis.

Win-loss record was modeled using linear regression analysis looking for a
discemable effect that could be attributed to debater sex, colleague sex, critic sex, team
sex (interaction between debater sex and colleague sex), and the interaction of team sex
and judge sex. Decision in the high-high debate and speaker points were treated as

covariates in an attempt to control for talent.

Table 3: Linear Regression Model with Win-Loss Record as the Dependent Variable

 

 

 

Standardi
zed
Unstandardized Coefﬁcien
Coefficients ts
Model B Std. Error Beta t Sig_.__

1 (Constant) -23.757 .890 -26.707 .000
Decision .294 .025 .187 1 1.553 .000
Speaker Points 1.015 .033 .509 31.176 .000
Debater Sex .172 .030 .093 5.781 .000
Colleague Sex .169 .030 .091 5.648 .000
Critic Sex -2.84E-02 .035 -.013 -.816 .414

 

 

 

 

 

 

 

 

Table 3 contains the results of the linear regression analysis of the win-loss record
model with debater sex, colleague sex, and critic sex as the factors under investigation,
while treating decision and speaker points as covariates. As expected both speaker points

([3 = .187, p < .001) and decision in the high-high debate ([5 = .509, p < .001) both had

36

 

Sex-Based Disparities in Debate

signiﬁcant relationships with overall win-loss record. It should not be surprising that
debaters than won their high-high debate won an average of .89 rounds more than those
who lost, since each individual win contributes to the overall win-loss record. Graph 1
shows the linear trend that is evident between speaker points and mean record. The
seeming upward trend in record as speaker points fall below 24.0 can be discounted as
the result of a single score of 23.0 for the entirety of the sample (n = 2862). Taken
together, decision and speaker points account 69% of the variance in win-loss record.
Including these variables, and looking only at high-high debates should parcel the effect
of talent, experience, etc., such that it is possible to test if the remaining variability in

record can be attributed to sex of the participants.

Graph 1: Mean Win-Loss Record by Speaker Points

 

 

Mean Record
N

 

 

I 1 I I I I I

23.0 24.0 25.0 25.5 26.0 26.5 27.0 27.5 28.0 28'.5 29.0 29.5

Speaker Points

The results in table 3 clearly indicate a signiﬁcant effect of both debater sex ([3 =

.093, p < .001) and colleague sex (B = .091, p < .001) on win-loss record. While the

37

 

Sex-Based Disparities in Debate

effects were signiﬁcant, the effects of the control variables dwarfed the magnitudes of the
effects. Still speaker sex and partner sex each account for 9% of the variance in win-loss
record after controlling for the effect of talent, which encompasses experience,
preparedness, resources, etc. Men, on average, win 53.7% of their debates while women
win have a 46.3% winning percentage. Thus, men have a 7% better chance of winning
each debate in which they participate than their female counterparts, and at least some of
this difference is attributable to the sex of the participants. No signiﬁcant effect between
critic sex and win-loss record was observed.

Team sex and the interaction between team sex and critic sex were excluded by
SPSS due to colinearity with other independent variables. This is logical since there is a
direct relationship between these two variables and both debater sex and partner sex.
Critic sex is also colinear with its interaction term with team sex. To rectify this, a
separate linear regression analysis that included only the control variables and the
interaction terms that were excluded by SPSS due to colinearity. The drawback of this
method is that the interaction effect could appear signiﬁcant even though the results are
attributable to variance for which the main effect has already accounted, increasing type I
error (concluding differences exist when none actually does).

Results of the linear regression of the interaction terms are itemized in table 4.
There is a signiﬁcant effect of team sex on success as measured by win-loss records after
controlling for talent factors ([3 = .145, p < .001). Thus, it appears that there is a
compounding effect triggered by the interaction between speaker sex and partner sex,
such that the teams that are evaluated least favorably are comprised of two females. The

interaction effect of team sex and critic sex was non-signiﬁcant (table 4).

38

Sex-Based Disparities in Debate

Table 4: Linear Regression Model of Interaction Terms with Win-Loss as the Dependent

 

 

 

Variable
Standardi
zed
Unstandardized Coefﬁcien
Coefﬁcients ts

Model B Std. Error Beta t Sig.

1 (Constant) -23.757 .889 -26.712 .000
Decision .294 .025 .187 11.555 .000
Speaker Points 1.015 .033 .509 31.182 .000
Team Sex .355 .043 .145 8.166 .000
Team Sex " Critic Sex -1.42E-02 .017 -.014 -.817 .414

 

 

 

 

 

 

 

 

To conﬁrm that the overwhelming majority of both the judges and debaters being
male did not skew the above results, a secondary linear regression analysis was
performed. It seemed preferable to report the results of the entire sample included in this
study since, as previously discussed, this sample is highly representative of the entire
populace of the debate community during the four-year period being examined. To
ensure these results were not caused by the discrepancy in participation rates, another
data reduction technique was employed. The composition of team sex was equalized by
randomly selecting 200 cases of each of the possible team conﬁgurations - all female
team, male debater with female partner, female debater with male partner, and all male
team — as one would expect to ﬁnd if participation rates between men and women were
equal and partnership pairings were random.

This equalizing technique has a couple of drawbacks that should be noted. First,
some teams may be over-represented in the newly constructed sample in all four
categories given the way the data is coded. Since each debate is coded one time for each
speaker, each two-person debate team is represented by two cases in the sample. Thus, it
is possible that some debate teams were included in the reduced sample twice while some

teams may have been excluded all together. Second, by randomly selecting cases to

39

Sex-Based Disparities in Debate

equalize the cell sizes of the team compositions, the cell sizes for variables such as
decision in the high-high debate will not necessarily be equal as would normally occur in
a closed system of tournament debates where for every win there is a corresponding loss.
In the equalized team participation sample, there are 401 wins, 393 losses, and six cases
that were excluded due to missing data. A 95% conﬁdence interval around the mean
decision (0.011, s = .0355) for this equalized sample indicates that there is no signiﬁcant

difference from the mean decision (0.000) given an equal number of wins and losses.

Table 5: Linear Regression Model with Win-Loss Record as the Dependent Variable for
Sex Equalized Sample

 

 

 

 

 

 

Standardi
zed
Unstandardized Coefﬁcien
Coefﬁcients ts
Model B Std. Error Beta t Sig_.__‘

1 (Constant) -25.296 1.565 -16. 168 .000
Decision .306 045 .196 6.759 .000
Speaker Points 1.069 057 .542 18.664 .000
Debaters Sex .127 .044 .081 2.870 .004
Colleague Sex 8.995E—02 044 .058 2.025 .043
Critic Sex -1.45E-02 060 -.007 -.242 .809

 

 

 

 

 

 

 

 

The results in table 5 conﬁrm that the results for the win-loss record regression
model for the equalized team sex sample is consistent with the results obtained by using
the entire sample gathered in this project. The talent control variables, decision ([3 =
.196, p < .001) and speaker points ([3 = .542, p < .001), have a signiﬁcant relationship
with win-loss record. Debater sex ([3 = .081, p < .005) and colleague sex (13 = .058, p <
.05) both still have a signiﬁcant effect on win-loss record. Although it appears that the
effect size of the equalized sample is not as large, indicating a possible exaggeration in
the unequalized sample. In addition, the margin of error for this model is higher for both

the effect of debater sex and colleague sex, but this can largely be attributed to the greater

40

Sex-Based Disparities in Debate

power of the model utilizing the larger sample size. There was no signiﬁcant effect of
critic sex on win-loss record. All of these results are consistent with those found using
the entirety of the sample size, thereby disproving the possibility that the participation
rates skewed the results of the win-loss linear regression model.

Speaker Point Model

The speaker point model provides an opportunity to look at some of the variables
that were excluded in the win-loss record model. Unlike record, which is based on wins
and losses that are always equal across all tournaments and topics, speaker points vary
according the criteria determined by each individual judge. This difference allows the
speaker point model to be investigated for sources of variance attributable to tournament
factors, topic factors, or critic sex.

Speaker points, the dependent measure of success, were analyzed using a linear
regression model that explored the discemable effects of speaker sex, partner sex, critic
sex, tournament, topic, and the interaction terms all of these main variables; including
team sex, team-critic sex, and the interactions of topic with debater sex and judge sex.
As in the previous model, SPSS excluded the interaction terms due to colinearity with the 1
main variable. The same remedy with the associated drawbacks, constructing an
additional linear regression model that excluded the main variables so as to analyze the
interaction effects, was employed. As in the past model, the other measures of success —
decision in the round and win-loss record — are treated as covariates to control for the
factors collectively referred to as talent herein, including experience, preparedness,

natural speaking ability, intelligence, resources, etc., with the goal of parceling these

41

Sex-Based Disparities in Debate

factors out of the data leaving only the variance attributable to other factors such as sex,

topic, tournament and the like to be analyzed.

Table 6: Linear Regression Model with Speaker Points as the Dependent Variable

 

 

 

Standardi
zed
Unstandardized Coefﬁcien
Coefﬁcients ts
Model B Std. Error Beta t Sip.__
1 (Constant) 26.206 .040 647.732 .000
Record .276 .009 .550 31 .561 .000
Decision 2.824E-02 .013 .036 2.104 .035
Tournament .104 .015 .114 6.942 .000
Debaters Sex 2905-02 .021 -.031 -1.369 .171
Colleague Sex 3.742E-03 .015 .004 .242 .809
Critic Sex -1.53E-02 .032 -.014 -.483 .629
Topic -.188 .087 -.379 -2.159 .031
ﬁgzgthnudge-Topuc 5.356E-02 .020 .348 1.924 .054

 

 

 

 

 

 

 

 

Table 6 contains the results of the model just described. As expected, win-loss
record ([3 = .550, p < .001) and decision ([3 = .036, p < .036) both had a signiﬁcant effect
on speaker points. The decision about which teams won the high-high debate reﬂects
which of the two equally talented teams had the better performance in the debate in
which the speaker points were assigned. And the team that goes undefeated during the
eight preliminary debates exhibited superior skills during all of the debates than any of
the other competitors who did not fare as well. Thus, both of these provide good
predictors of the level of speaker points that debaters will receive in high-high debates
that are attributable to talent. After the effects of these two factors are parceled out of the
data, the remaining variance can be analyzed for effects from the independent variables.

The data indicates that no signiﬁcant relationship exists between the dependent

variable, speaker points, and debater sex, colleague sex, or judge sex. To test the

42

Sex-Based Disparities in Debate

Table 7: Linear Regression Model with Speaker Points as the Dependent Variable Examining
the Sample Equalized by Team Sex

 

 

 

 

 

 

 

 

 

Standardi
zed
Unstandardized Coefﬁcien
Coefﬁcients ts
Model B Std. Error Beta t Sig.__
1 (Constant) 26.142 .069 377.458 .000
Record .293 .016 .577 18.584 .000
Decision 1 .979E-02 .024 .025 .813 .417
Tournament 6.256E-02 .027 .069 2.338 .020
Debaters Sex -8.54E-02 .034 -.108 -2.479 .013
Colleague Sex 9.829E-03 .023 .012 .423 .673
Critic Sex -7.37E-02 .058 -.069 -1.277 .202
Topic -.299 .1 57 -.594 -1 .908 .057
ﬁfszzgg’nUdge'mp'c 8.900E-02 .050 .569 1.776 .076

 

 

possibility that these results were skewed by the unequal cell sizes for the variables
representing the sex of the participants, an identical linear regression model was tested
using the sample that was randomly selected to equalize team sex to what would be
expected given equal participation and distribution of team composition described in the
previous section. Results of this model are provided in table 7.

The results from the speaker point model using the team sex equalized data
diverge from the results indicated by the entire sample of this study on the debater sex
factor. According to this analysis, there is a substantial relationship between the success
of debaters as measured by speaker points and the sex of the debater (B = -.108, p < .014)
in so far as male debaters receive less favorable speaker points due to their sex after
accounting for factors dubbed talent.

The tournament variable exhibited a signiﬁcant relationship with the variance in

speaker points in both the entire sample of the study (see table 6, B = .114, p < .001) and

43

Sex-Based Disparities in Debate

the equalized team sex sample (see table 7, B = .069, p < .021). Accounting for this
variance in the model allows for more accurate investigation of other factors.

Graph 2: Mean Speaker Points by Debater-Critic-Topic Interaction

 

27.6

 

 

Mean Speaker Points

 

Debater-Judge-Topic Interaction

F = Females
M = Males
The topic also indicates a signiﬁcant effect on speaker point variance when the

entire sample of this study is examined (see table 6, B = -.379, p < .032). The results do
not support the same conclusion when the equalized team sex sample was analyzed (see
table 7, B = -.594, p > .05), so it is possible that the skewed participation rate could have
contributed to the variance in speaker points across the topics. Finally, the regression
model indicates that the interaction term for debater sex, critic sex, and topic fell just
outside the level of signiﬁcance at the or = .05 level or both the complete sample (table 6)
and the team sex equalized sample (table 7). Graph 2 illustrates the mean speaker points

for this three-way interaction effect.

Sex-Based Disparities in Debate

VI. Discussion

This section raises possible implications and provides possible explanation for the
results outlined for participation disparities, the win-loss record model, and the speaker
point model. The results of this study are contextualized through comparison with
ﬁndings of previous studies described in the literature review. This discussion attempts
to explain the relevance of the ﬁndings of this study to each of the research questions
under the respective research topic in turn. Finally, limitations of the present study will
be addressed.

Participation Disparities

Debate largely remains a male dominated activity, with males comprising 76.3%
of debate competitors and 84.2% of the judging pool. Aﬁer accounting for the overall
participation discrepancies among competitors, analysis revealed that there is a
disproportionate number same sex teams, both all-male and all-female, in comparison to
mixed sex teams. This may reﬂect preferences to have same sex teams to facilitate
teamwork given differing styles or restrictions on males in female dormitories or vice-
versa on some campuses.

Given that past research (Andrews, 1987) indicates that women feel more self-
conﬁdent in anticipation of presenting before a female audience, the lack of female critics
may create an environment that is not conducive to encouraging female participation. On
the other hand, Andrews concludes that conﬁdence does not have an effect on
performance. In addition, Brushke and Johnson (1994) detect results indicating that not

only do women fare less well than men based upon the effect of sex of the competitors,

45

Sex-Based Disparities in Debate

the effect is more pronounced when females are judging females. Thus, the lack of
female judges may actually beneﬁt female debaters in measures of success.

Given the preponderance of males among the pool of debaters and debate judges,
concerns arise regarding those considered opinion leaders (Richmond & McCroskey,
1975). The high proportion of males in the debate community probably indicates that the
balance of female to male opinion leaders, and the corresponding heightened credibility
associated with the status of opinion leaders, is equally skewed. If this is the case, the
styles and arguments advanced by men may be considered the norm or criteria against
which all participants are evaluated; a playing ﬁeld that could scarcely be considered
level for female participants. Women need the space to develop styles and arguments
that are suited to them rather than being forced to conform to the overwhelming male
norm.

It is possible that the disparity observed in judging does not reﬂect an actual
difference in participation rates among the judging pool or among the coaching
population at-large for several reasons. First, while all debaters at a tournament
participate in each round, not all critics at a tournament judge each debate. They may
only be required to evaluate four of the eight rounds at a tournament. Thus, the results
could be a result of a skewed distribution in judging in high-high power matched debates.
Second, all of the tournaments utilize a system of mutual preference judging allowing the
competitors to inﬂuence which judges are assigned to their debates by submitting a
preference ranking for each judge prior to the beginning of the tournament. A bias in the
preferences expressed by the debaters could cause results that diverged from the

participation rates within the judging pool. Also, some debate teams elect to have only a

46

Sex-Based Disparities in Debate

portion of their staff fulﬁll the judging obligations, reserving other coaches to ensure their
competitors are adequately prepared to maximize performance. It is possible that the
skills possessed by women at tournament research and preparation, or some other factor,
encourages teams to utilize their female coaches in capacities other than judging.

Regardless of source, however, it is disconcerting that such drastic imbalances in
participation rates continue to exist among both debaters and judges. Whether the
discrepancy exists due to real participation differences, bias in preference rankings of
judges, skewed judge distribution during high-high debates, or individual teams’
decisions to hold their female critics out of the judging pool; the net effect is that
women’s perspectives were not nearly as prevalent as those of men in the ongoing
dialogue that is academic debate. Numerous studies have shown that participation in
debate teaches students many skills including persuasion, public speaking, proﬁciency in
argument construction, policy evaluation, competitive argumentative speaking, library
research, critical thinking, performance and information processing, integration and
management (Colbert & Biggers, 1985; Colbert, 1987; Cross, 1971; Ehninger &
Brockriede, 1978; Gruner, et al., 1971; Hill, 1987; Jackson, 1961; Whalen, 1991). That
women are underrepresented in an activity that aims to empower participants by teaching
skills necessary for effective advocacy is a tragedy. This conﬁrms that more efforts are
required to evaluate the components of academic debate that discourage female
involvement and to enhance recruiting and retention of female debaters.

A ﬁrst step toward this end is assessing whether some tournaments are more
conducive to heightened female participation than others so the characteristics that may

contribute to any difference can be modeled in future tournament formats or topic

47

Sex-Based Disparities in Debate

choices. Northern Iowa’s exhibited a signiﬁcantly lower proportion of female judges
than the average proportions at other tournaments, while Northwestern attracted a higher
proportion of females than either of the other tournaments examined in this study.
Perhaps Northwestern recruits more female critics to judge at their tournament, or uses
these judges more often during high-high debates. Whatever the source of the difference,
the practices of the Northwestern should be examined to determine if some attributes
could be modeled by other tournaments to encourage greater female participation. At the
same time, it should be noted that Northwestem’s ratio ofwomen (18.4%), while an
improvement when compared to other tournaments in this study, is still a far cry from a
representative ratio of the two sexes. Much more needs to be done to rectify these
imbalances.

The effect the debate topic has on participation rates could also illuminate another
possible tactic to address the participation gap between men and women. Since some
topics - such as sanctions — attract a greater ratio of women, future topics could be
crafted to incorporate the elements of these topics that appeal to female debaters and
critics. Of course, the topic preferences of women would require greater study in an
effort to predict what future topics might encourage more women to participate in debate
since this study merely addresses topics that have already been debated.

The analysis of the effect of debate topic on participation produced some
surprising results. One might expect women to be more active participants on the civil
rights topic that focused to a signiﬁcant degree on gender discrimination in the
workplace. On the contrary, female participation rates were lower on the civil rights

topics than any other topic. Of course, it is possible that the ﬂuctuations in female

48

Sex-Based Disparities in Debate

participation relative to men could be attributable to another cause such as changing
demographics of the overall collegiate population rather than preference for any given
topic. While these results show a signiﬁcant relationship between topic and participation,
this represents a signiﬁcant correlation, not conﬁrmation of causality. To better assess a
causal relationship, one would have to conduct further research examining survey
responses concerning reasons why participants chose to compete during a given year or
topic.

The results on female judge participation seem much more in line with what
might be predicted by past research. Both Richmond & McCroskey (1975) and Knutson
(1996) suggest the potential of topic to have an inﬂuence on the relative levels of female-
male participation and success. Knutson suggests that discussions of military policy, in
particular has historically excluded women’s voices. The low level of participation by
female judges on the security assistance topic is certainly consistent with this theory.
Richmond and McCroskey suggest that, while public policy in general was still largely
male domain at the time the study was conducted, there were some topics on which
women were becoming credible opinion leaders. It could be that the civil rights topic,
which focused largely on gender discrimination in the workplace, was a topic women’s
opinions were held in higher regard than other topics. This may have attracted more
women to the judging pool because of interest and comfort on the topic. Another
possible explanation is supported by the Andrews (1987) study, which concluded that
both men and women tended to rely more on argumentative appeals to family or social
contributions when appealing to female audiences instead of a male audience. Perhaps

the subject matter of civil rights was conducive to the type of arguments debaters deemed

49

Sex-Based Disparities in Debate

more persuasive to females than males. This evaluation could have altered competitors’
preferences for judges in the pre-tournament rankings resulting in a higher proportion of
women judging debates even if the overall judging pool did not change in composition.

It is odd that the results for debater participation interaction with the topic
variable were in such stark contrast to the judge participation contrasts among the topics.
The civil rights topic witnessed both the lowest percentage of female debaters and the
highest percentage of female critics when compared to the other three topics. This
inconsistency obviously raises questions about the link between sex-based participation
and topic. However, it is possible that judge participation is more subject to variance that
is attributable to topic than is the composition of the competitors. Most debaters
probably participate during college regardless of preferences for or against the selected
annual topic because the ability to debate is restricted to those years spent enrolled as an
undergraduate student. Judges, on the other hand, may be more ﬂuid based on topic for
any number of reasons: former women debaters who have chosen to pursue other
interests may return to judge more debates on a topic of interest; women coaches may opt
to judge more on a preferred topic than one they dislike; women judges could be more
preferred on some topics than others increasing the proportion of females judging high-
high debates even if the composition of the judging pool is unaltered by the topic.

In sum, it is important to recognize that even under the most favorable conditions,
female participation in intercollegiate policy debate is far from representative among
either the competitors or the judges. While we should attempt to learn what tournaments
such as Northwestern do differently to attract more female judges, and we should

consider crafting topics that have greater appeal to women, much more needs to be done

50

Sex-Based Disparities in Debate

to rectify these glaring discrepancies in participation rates between women and men in
debate.
Di_sp_arities in Win-Loss Records

Given that substantially greater numbers of males participate as debate
competitors in comparison to females, one would naturally expect aggregate win-loss
records of women and men that corresponds with the disparate participation rates. In
other words, men as a whole should win a greater number of debates simply because
there are more men in the ﬁeld. This explanation, however, would not sufﬁce to explain
disparate levels of success rates attained by the average individual person of either sex. If
the playing ﬁeld in debate is truly level for the sexes, each person should expect to win
roughly half of their competitive debates in a closed tournament environment in which
there are always equal numbers of wins and losses handed down. Of course, tournaments
are paired to induce a normal distribution of win-loss records across the teams competing
in an effort to determine a tournament winner. If one of the sexes outperforms the other
on average after accounting for discrepancies in participation rates, then another factor
must be exerting an inﬂuence on success as measured by overall win-loss record.
Determining if a substantial difference between success rates of the two sexes exists is
the primary goal of this analysis of win-loss record models.

Results of the linear regression model illustrated two main differences in success
based on the sex of the participants. Using win-loss as a measure of success, the analysis
revealed that male debaters are evaluated more favorably than female debaters, and that
speakers with female partners encounter disparate evaluations when compared to

speakers who debate with men. Both of these factors contribute to the variance in

51

Sex-Based Disparities in Debate

success as measured by win-loss record to a statistically signiﬁcant degree after
accounting for variance attributed to talent and all of its variants by treating the decision
in the round and speaker points as control variables. The signiﬁcance effect of the team
sex variable indicates that the sex-conﬁguration of the two-person team can have a
compounding effect on the observed disparity.

Men have a 7% greater chance of winning each debate in which they compete, in
part due to sex factors. Given that the difference of a single victory during preliminary
debates often determines whether a team qualiﬁes for the elimination rounds or is
eliminated from the tournament, the difference in success that is attributable to sex is
discomforting.

An identical analysis was conducted on a sample that had been randomly selected
to equalize the team sex as if participation of women and men was equal and partnership
distributions were assigned randomly. The results of this secondary analysis were
consistent with those gleaned from the unequalized data, illustrating that these results
cannot be attributed to skew resulting from the disparate participation rates described in
the previous section. Since multiple measure were utilized to control for the variants of
talent, these differences indicate that the disparities in part can be attributed to sex bias.

It should not come as much of a surprise that there was no effect of critic sex on
win-loss record. While each debater is involved in all eight of the debates that determine
the win-loss record, each judge contributes only one decision toward the overall win-loss
record of each team. If there were a difference in record between male and female
judges, this would reveal a difference in the quality of teams to which male and female

judges are assigned more than it would reﬂect any difference in the way women and men

52

Sex-Based Disparities in Debate

judge debates differently. This effect could indicate that there is bias toward having men
or women judge the more talented team, which could be an important ﬁnding; but alas,
the ﬁndings of this study indicates no signiﬁcant difference in team record based upon
sex of the judge.

To look at judge discrimination in voting would require a model using decision as
the dependent variable. Since every time judges vote for one team they are voting against
another, there will be no difference between men and women judges in overall voting
record. However, by looking at the interaction effects between sex of the judge and
debater sex, colleague sex, and team sex on decision in the high-high round, it would be
possible to discern any statistical difference between the way women and men judge
based upon sex of the competitors. This analysis was not conducted, however, because
the bivariate correlation (Appendix B) employed as a data reduction technique indicated
no relationship between decision and any of the variables mentioned above nullifying the
need for further analysis.

Disparities in Spea_ker Points

No difference was found in success as measured by speaker points for speaker
sex, partner sex, or judge sex under the speaker point linear regression model. Since the
win-loss regression model indicated a variance in record that could be attributed to sex of
the competitors, it is possible that there is also a disparity in speaker points that is masked
by the inclusion of win-loss record as a covariate. In other words, a discrepancy between
men and women speakers and partners may exist for speaker points, but this difference is
no larger than that already observed in the disparity already accounted for in the win-loss

model. This data does not support the conclusion that speaker points are more likely to

53

Sex-Based Disparities in Debate

exhibit characteristics due to the subjective nature of judging as compared to the more
objective criteria employed to determine winners and losers for debates.

The secondary linear regression analysis of the speaker point model produced an
unexpected result. The sex-equalized data set produced results indicative of speaker
point disparities that detrimentally impacted male debaters. Rather than disadvantaging
women, the data indicates that male debaters are evaluated less favorably as measured by
speaker points due to their sex after accounting for the factors dubbed talent when the
team sex is equalized in the sample. It is possible to conclude, therefore, that the unequal
participation rates among women and men skewed the results in such a way that
increased the risk of type 11 error (data indicated no difference attributable to a factor
when a difference actually exists), which masked the disparity in speaker points
attributable to sex that disadvantages men relative to women. On the other hand, this
ﬁnding should be viewed with skepticism since it is inconsistent with the ﬁndings
described under the win-loss model and previous studies reviewed in this project.

The tournament variable produced signiﬁcant variance in the assignment of
speaker points. Tournament conditions, the judging pool or the quality of the
competition a given tournament attracts, could result in skewed speaker point results.
The speaker point trend based on a means analysis, however, points to another possible
explanation. The tournaments, as represented in graph 3, are in the chronological order
they occur during each year of the study. Northern Iowa is the ﬁrst tournament of the
season on a new topic for most all participants and occurs in mid-September. Kentucky
follows only a couple weeks later. Northwestern does not host their tournament until

February. As the competitors gain familiarity and experience with the arguments and

54

Sex-Based Disparities in Debate

evidence on a given topic, it is likely that they perform better and are evaluated more
favorably by critics as the year progresses.

Graph 3: Mean Speaker Points by Tournament

 

 

 

   

27.4

27.3 I
1”.
.E
o
0.
h
a)
x
i
(I)
c
to
d)
2 27.2

Northern Iowa Touma Northwestern Toumam
Kentucky Tournament
Tournament

The annual debate topic term also yielded a signiﬁcant effect on speaker point
variance. The topics are ordered chronologically in graph 4. The upward trend in
speaker points could be attributed to speaker point inﬂation over the years rather than any
major effect of the topic. This data indicating an average of 27.3 speaker points were
given to debaters in this study between 1996 and 2000, in conjunction with the data from
Brushke indicating an average speaker points ranging from 26.5 to 27.0 during the 1989

to 1992 period, would be consistent with the speaker point inﬂation theory.

55

Sex-Based Disparities in Debate

Graph 4: Mean Speaker Points by Topic

 

 

 

 

27.4
27.3 -
£9.
.E
O
0.
L.
(D
x
as
0)
o.
(I)
r:
(u
a)
E 27.2
Environmental Regula Civil Rights
Security Assistance Economic Sanctions
Topic

In sum, the linear regression model of speaker points as a function of participant
sex yielded few major insights. The covariates, decisions in the high-high debate and
win-loss record, both accounted for substantial variance in speaker points as expected.
The tournament factor demonstrated the most signiﬁcant effect on speaker points. This
effect is consistent with improvement and familiarity with the topic arguments as the
season progresses. The topic variable yielded a signiﬁcant effect, but the trend is
consistent with a chronological increase that could be explained by a speaker point
inﬂation theory every bit as well as a topic variation theory. The topic variable did not
indicate a signiﬁcant response when the equalized team sex sample was modeled, raising
the possibility that the disparate participation results skewed the effect of topic increasing

the likelihood of type I error. Finally, when analyzing the team sex equalized data, a

56

Sex-Based Disparities in Debate

disparate assignment of speaker points to the detriment of male subjects was detected.
This result is the only effect that appeared in the team sex equalized model that did not
appear in the analysis of the entire sample and is inconsistent with every other result
obtained by this study as well as the bulk of the literature on sex-based success in debate.
It seems probable that given the accumulation of error across all of the comparison
performed herein that this result is attributable to type I error.
Limitations

Several limitations of this study have been mentioned throughout the results and
discussions sections: First, unequal participation rates based on sex could skew results in
the regression models of success. Second, the study suffers from a problem of repeated
measures that have not been accounted for in this design. Subjects who competed or
judged at more than one tournament are over-represented. In fact, it is possible that some
subjects participated in all twelve tournaments included in this study while others
participated at only one tournament. Third, the team sex equalized data set used to
mitigate the effect of the unequal participation rates suffers from over representation of
some teams. Fourth, some inconsistent results emerged in the data analysis. Most
notably, the comparisons between the complete data set and the team sex equalized
sample yielded divergent results on the effect of topic and speaker sex on speaker point
variance. Fifth, there are alternative explanations of the trends observed in speaker point
variance attributable to both tournament and topic variables that would require further
testing to clarify. Sixth, it is conceivable that the participation rates reﬂected from this
analysis of high-high rounds is not representative of the sex composition of

intercollegiate debate coaches, or even of the judging pool at the tournaments included in

57

Sex-Based Disparities in Debate

this study. Given the use of mutual preference judging at these tournaments to assist in
determining judge placement, the preferences of the competitors could have resulted in a
skewed distribution given the pairing of the round, such as an effect attributed uniquely
to high-high power matched debates. Analyzing mutual preference requests of teams in
comparison to judge placement and the success rates with preferred or non-preferred
judges offers a host of avenues for further research. It would be interesting to contrast
the sex distribution of judging in high-high debates, which are typically deemed to be the
best and closed debates since the bracket is structured to pit teams of equal talent against
one another, to the distribution in non-power matched rounds or hi gh-low power matched
rounds. Also, preference sheets at some tournaments do not go into effect until the third
round of the tournament. An interesting study might contrast judge distribution based on
sex in rounds employing mutual preference to those using random judging selection.
Seventh, the sex of some of the participants could not be identiﬁed. Further studies
should record the sex of the participants while the tournament is ongoing instead of
attempting to reconstruct the team sex conﬁgurations a few years later. Eighth, the
exclusion of the interaction variables due to colinearity with the primary independent
variables resulted in an analysis of the interaction effects that did not account for the
variance already accounted for by the main effects. Hence, the results reported for the
interaction terms may simply be reiteration of results previously identiﬁed by the primary
variables.

In addition, there are other limitations that have not been addressed. First, the
tournaments selected are only representative of CEDA and NDT national circuit debate.

There are numerous healthy regional circuits that are not represented by this study, and

58

Sex-Based Disparities in Debate

could have differing results. Also, there are other forms of collegiate debate besides
policy debate examined in this project, including such styles as parliamentary debate and
value focused debate. And, this study is not representative of policy debate participants
at the high school level. Further studies on all of these could conﬁrm or deny similarities
to these ﬁndings on intercollegiate policy debate. Second, other variables could be
included as either relevant covariates or independent variables. For example, side of the
topic — afﬁrmative or negative — could produce effects that possibly interact with some of
the results of this study. The Brushke and Johnson (1994) study found some interaction
effects of sex variables and topic. This data is available in the results packets of the
tournaments included in this study. Further work on this project could use topic sides as
contrasts for the results based on topic side. Furthermore, speaker position — whether the
debater is gives the ﬁrst or second speech for the team -— is widely held to inﬂuence
speaker points awarded in the round. The speaker giving the second speech typically
plays a greater role in winning or losing debates for a variety of reasons. Judges often
reward the second speaker with more favorable speaker points for fulﬁlling their greater
responsibilities. The Brushke and Johnson study does not attempt to account for the
speaker position factor either. Again, collecting this data while the tournament being
analyzed is in progress is necessary since it is not provided in many of the results packets.
Finally, the study does not investigate effects attributable to the opponents’ sex. Further
investigation into the interaction effects of speaker sex and opponent sex could provide
insightful results.

In the end, it is difﬁcult to discern the effect the sex the participants has on the

level of success they attain in debate. Too often attempts to examine disparate success

59

Sex-Based Disparities in Debate

rates do nothing to distinguish between effects attributable to sex and those attributable to
different levels of talent, resources, experience, preparedness, etc. While this study
attempts to analyze this distinction by controlling for variables that serve to represent the
non-sex factors that factor into success in debate by looking at high-high rounds, and
treating decision in the debate and overall win-loss record as covariates, ultimately most
of the signiﬁcant effects that were observed based on sex could be attributable to a
variety of non-sex factors. The factors that contribute to success in debate are numerous
and the relationships are complex. That one sex outperforms the other may be a function
of greater experience, more in-depth research, more afﬁnity for the substance or style of
the activity of debate, superior natural speaking abilities, or a number of other
possibilities. In the end, sex clearly matters to the level of success in debate, but it is far
from clear that such disparities amount to discrimination. It is equally possible that the
success disparities result from a mediating factor - such as work habits, argumentative
strategies, or stylistic choices — rather than on any of the sex variables in and of
themselves. Further research should analyze the relationships between some of these

mediating variables and sex-based participation and success differences.

60

Sex-Based Disparities in Debate

VII. Conclusions

Results of this study clearly indicate that disparate participation and success rates
women experience relative to men continue to plague national circuit tournaments of
CEDA and NDT style intercollegiate policy debate. Men outnumber women competitors
by a factor of 3:1. Four-ﬁfths of the judges placed in the high-high debates analyzed in
this project were men. Men are 7 % more likely than women to win each debate, which
could make the difference between advancing in the tournament and being eliminated,
and at least part of this variance is attributable to sex factors rather than the factors
incorporated under the rubric of talent.

Despite the limitations of this and other studies referenced in the literature review,
one conclusion remains consistent: there seems to be a difference between male and
female debaters. The sexes are disproportionately represented in the collegiate debate
community -— both among judges and debaters. Men also win a high proportion of the
awards and honors than even their excessive representation among the pool of
competitors would predict (Stepp, 1997). Males surpass females in measures of success,
even after accounting for the variance in success attributable to talent. Females embrace
different styles that hinder their ability to control cross-examination periods (Larson &
Vreeland, 1985). Females may be or attain mitigated success or experience a lack of
interest in debate due to the public policy focus of the topics or the military metaphors
employed in debate strategy (Knutson, 1996). This study indicates that annual debate
topic has a signiﬁcant impact on female participation rates in the judging pool, which is

consistent with the theories posited by Knutson.

61

Sex-Based Disparities in Debate

Some of the disparities in participation should be readily apparent in casual
observation of tournament demographics. That other disparities are not as easy to detect
does not deny the existence of such difference. Numerous females have attained a
tremendous degree of success by any measure. These women were not just good female
debaters, they compared favorably to their male competition as well. However, these
success stories of great female debaters are the exception rather than the rule.

The disparities should be of tremendous concern to anyone who cares about
debate or the advancement of women. The debate community cannot pretend to be
enlightened if it largely excludes the crucial voices of women in our society. The
ongoing dialogue of debate lacks the perspectives women can offer into the marketplace
of ideas. Debate training enhances critical thinking and hones research and oral
communication skills that can empower women to advance their agenda both in an
outside of the debate community.

The collegiate debate community needs to recognize that the evils of disparate
participation and success based on sex endured the topic merger of CEDA and NDT. Our
activity is not the representative and egalitarian community many have been working so
hard to attain. Current efforts to remedy the gaps in participation and success are
insufﬁcient and must be redoubled and expanded.

Relevant committees within CEDA and NDT, directors of debate programs,
coaches, graduate assistants, judges, and competitors need to use the trends identiﬁed in
this and other studies as a catalyst for self-reﬂection. Perhaps practices we engage as
judges, coaches, or competitors contribute to these problems. An examination of the

topic selection process, recruiting practices, standards for the distribution of scholarships,

62

Sex-Based Disparities in Debate

decisions regarding travel and competitive opportunities for female debaters, choices
about female judging commitments, and criteria employed to evaluate which team won or
lost a debate or to assign speaker points, in an effort to create an environment more
conducive to the participation and success of women in debate. Only through individual
efforts and resistance can the debate community overcome bias and disparities that have

plagued academic debate since its inception.

63

Sex-Based Disparities in Debate

APPENDIX A

Sample Debate Ballot

64

Sex-Based Disparities in Debate

Sample Tournament Debate Ballot

Round: Judge: Room:

 

 

Instructions: Determine who won the debate and vote for one and only one team. Assign
each speaker quality points (0-30) to assess overall performance and contrast them to the
other speakers in this debate by assigning each speaker a rank-order between 1 and 4.

 

 

 

 

 

 

Afﬁrmative: Negative:
1 A: l N:

Rank: Points: Rank: Points:
2A: 2N:

Rank: Points: Rank: Points:

The better debating was done by the (Afﬁrmative or Negative) representing

 

 

School Name Judge’s Signature

 

Reason for Decision:

65

Sex-Based Disparities in Debate

APPENDIX B

Bivariate Correlation Matrix

66

Sex-Based Disparities

€2.94. .25. 3o 55 .5 285:9... 0. 88.2.8 ..
62.2-8 .25. 8... as s .889». 8. c8980 .:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

50« 53 F00« F00« 90« 93 0008 90« 53 $0.... o~0« o~0« 2.3 2
. 8o. o8. 8o. 08. 8... 8o. 23. o8. .8. 80. E. «8. 62.2-8 .90 8.882...
08.. :03. :08. :08. :00«. ooo. :0v«. ooo. :09. ooo.. o—or ooo. .96. 8.8.280 c3800 each-08378880
53 «Sh 50« «EN 0 SN 0 F0« «08 o00« 30.4. «05 oo«« o0~« BR 2
ooo. . ooo. ooo. ooo. 9 P. ooo. 000. 2.0. .00. 2.0. 000. EN 32.28 .90

:80. ooo.. :ooo. :80. :88 «no. :03. 05. h 5. «_.o.- - Po: moo. o«o. 8.8.280 8080a. .80 20:0 . 0.8...
.03 50.0. 80« 83 n E« ««0« 50« ««0« 80« 80« 2.8 2.0« R8 2
ooo. ooo. . ooo. ooo. ooo. ooo. 5o. ooo. one. oi. o««. 03. 32.2-8 0.0

$000. :80. coo; :03. :von :09. :00 F. «no. :vow. 05. o«o.- Koo. o«o. 8.8.280 80800 .80 88an . 08°»
.00« «0R mmo« ooo« 0 ZN ««0« «0.8 83 008 ooo« v03 «00« 2.3. 2
ooo. ooo. ooo. ooo. Kn. ooo. 2.0. won. 00». o«o. So. 000. 32.2-8 .90

:08. :80. :80. ooo.. :o«w. - 8. Low. 08. Koo. ooo. Leo; ooo. Poo. 8.8.0.80 80880 0.8...
n_.0« 98 90« 90« m..0« 90« 90« 06« 93 20« :0« :0« :0« z
ooo. ooo. ooo. ooo. . ooo. ooo. ooo. ooo. .8. Foo. :0. ooo. 32.28 0.0

fog. :3«. row. 83. ooo... :00... :05. :80. :o«0. .03.- :08. ooo.- :Poo. 8.8.280 8082". .80 280 . x00 Eco-r
n .0« 0 F0« ««0« ««0« 0 F0« ««0« m B« ««0« ««0« ««o« ooo« ooo« ooo« z
ooo. 0P _.. ooo. 2.0. ooo. . omo. ooo. ooo. 8F. ooo. 00«. ooo. 32.28 .90

:08. «no. :00". n 5. :09. ooo.. 00o. :02. :13. «no. :09. F«o. f0o«. 8.8.280 8080a. 80 Each
53 «03 $3 «0K 90« 02.8 «0R oom« ..00« «05 85 85 oo«« 2
ooo. ooo. ooo. ooo. ooo. ooo. . on p. «o .. coo. «v«. ««0. ~00. 32.28 .90

:38 :00«. :0. F. :80. :05. 8o. ooo.? o«o. o«o. :39. ««o. woo.- Zo. 8.8.280 8280a. .80 20:0
93 o00« ««0« 008 90« ««0« o00« 00$.- ««0« 003 000« 008 058 z
ooo. 000. Foo. 2.0. ooo. ooo. o9. . ooo. ooo. ooo. 0P0. ooo. 32.28 0.0
00o. 05. «no. ooo. :o«n. :00». o«o. ooo.? :09. 05. .Loo. o—o. fox-3. 88.080 8280 .80 2682.00
room 50.4. 008 80« 93 ««0« F00« ««o« 008 008 Ko« {.8 Row. 2
ooo. 05. ooo. ooh. ooo. ooo. «on ooo. . B«. ooo. ooo. ooo. 32.2-8 .90

:09. to. :19. So. :o«0. v05. o«o. :09. ooo; o«o. :Eo. o«o. :09. 8.8.280 8280.. .80 0.8880
83 «0K 008 ooo« 90w. ««0« «03 no.8 000N- ooo« v08 «00« 0:8 2
50. 80. one. non. Be. not coo. ooo. 5«. . ooo. 000. 000. 32.28 .90
ooo.- «5.- 0 5. ooo. .03.- «no. :36.- 9o. o«o. ooo.. ooo. .8. :9. 88.280 com-80.. EoEoEso»
23 003 EN .08 w —0« ooo« ooR ooo« :8 vow-o. v03 «00« «00« z
ooo. 2.0. oi. o«o.. Foo. ooo. «v«. ooo. ooo. ooo. . ooo. ooo. 32.28 .90
o5- Co.- 0«o.- Leo.- :00o. :n« F. ««o. Loo. Loo. :ooo. ooo.. :03. :50. 8.8.280 80800 2:80 882.0
23 83 2.3 «08 F E« ooo« o9...“ ooo« 2.3. «00« «00« «0.8 ooo« z
E. 000. o«h. «00. 1.0. 08. ««0. 03. won. 000. ooo. . ooo. 32.28 .90
ooo. moo. moo. ooo. moo.- .«o. 3o.- opo. o«o. Foo. :09. ooo.. :oo«. 8.8.280 82800 8.280
2.3 o0..« 2.0« 3.8 .. F.« 88 ooR 028 2.8 $8 «08 83 2.3 2
ovo. 3N 0!. moo. ooo. ooo. ~00. ooo. ooo. 000. ooo. ooo. . 32.2-8 .90

beo. o«o. o«o. Poo. Loo. :oo«. So. :09. .00.. :6.- .50. oo«. ooo... 8.8.280 coﬂmon 230m
8.6225 .80 3.8 x50 5.28 can; .80 2.5 . .80 ES» .80 2.5 x50 80 2228 5558553 250.. 8.280 283.
each-coo . 088 . each .80 Eco-P 3.8280 882.0
2.8880 .

 

 

x5e: 8.8.280 32820

67

Sex-Based Disparities in Debate

BIBLIOGRAPHY

Andrews, RH. (1987). Gender Differences in Persuasive Communication and
Attribution of Success and Failure. Human Communication Research 3 372-
385.

 

Bruschke, J ., & Johnson, A. (1994). An Analysis of Differences in Success Rates of
Male and Female Debaters. Argumentation and Advocacy. 3, 162-173.

Chenoweth, E.C., & Good, U.Z. (1972). The Rise of Women and the Fall of Tradition in
Union Debating at Oxford and Cambridge. Speaker and Gavel. 2. 31-34.

Colbert, Kent and Biggers, Thompson. “Why Should We Support Debate?” Journal of
the American Forensics Association 21 (Spring 1985): 237-240.

Colbert, Kent. “The Effects of CEDA and NDT Debate on Critical Thinking,” Journal of
the American Forensics Association 23 (1987): 194-201.

Cross, Gary P. “The Effects of Belief Systems and the Amount of Debate Experience on
the Acquisition of Critical Thinking,” diss., U of Utah, 1971.

Ehninger, Douglas and Brockriede, Wayne, Decision by Debate (New York: Harper and
Row, 1978).

Freely; C. Gruner, R. Huseman and J. Luck, “Debating Ability, Critical Thinking Ability,
and Authoritarianism,” Speaker and Gavel 8 (1971): 63-64.

Hayes, M. & McAdoo, J. (1972). Debate Performance: Differences Between Male and
Female Rankings. Journal of the American Forensics Association, 3,127-131.

Hensley, W.E., & Strother, DB. (1968). Success in Debate. Saeech Teacher. 3. 235-
237.

Hill, Bill, “The Educational Goals of CEDA: Clariﬁcation and Reforrnulation,” 111;
Forensic of Pi Kappa Delta 73 (October 1987): 17-27.

Infante, DA. (1989). Response to High Argumentatives: Message and Sex Differences.
Southern Communication Journal 2 159-170.

 

Jackson, Teddy R. “The Effects of Intercollegiate Debating on Critical Thinking,” diss.,
U of Wisconsin, 1961.

68

Sex-Based Disparities in Debate

Knutson, Roxanne L. “Metaphorical Construction: Argument is War,” Contemporary
Argumentation and Debate: The Journal of the Cross Examination Debate
Association 17 (1996): 32-42.

 

Larson, Suzanne, & Vreeland, Amy L. (1985). Gender Issues in Cross Examination
Periods in C.E.D.A. Debate. National Forensic Journal, 2, 13-27.

Mulac, A., Lundell, T.L., & Bradac, J .J . (1986). Male/Female Language Differences and
Attributional Consequences in a Public Speaking Situation: Toward an

Explanation of the Gender-Linked Language Effect. Communication
Monographs. 2. 115-129.

Richmond, V.P., & McCroskey, J .C. (1975). Whose Opinion Do You Trust? Journal of
Communication 3 42-50.

 

Rosen, N., Dean, L., & Willis, F. (1978). The Outcome of Debate in Relation to Gender,
Side and Position. Journal of the American Forensic Association. 1, 17-21.

Spender, D. (1980). Man Made Language. Boston, MA: Routledge & Kegan Paul.

Stepp, P. (1997). Can We Make Intercollegiate Debate More Diverse? Argumentation
and Advocacy, 4, 176-188.

Tannen, D. (1994). Gender and Discourse. New York, NY: Oxford University Press.
Whalen, Shawn. “Intercollegiate Debate as a Co-Curricular Activity: Effects on Critical

Thinking,” Arguments in Controversy: Proceedings of the Seventh SCA/AF A
Conference on Argumentation (Annandale: SCA/AF A, 1991) 391-397.

69