Professional Documents
Culture Documents
experience), women had lower probabilities of being selected than men. Further,
applicants' sex interacted with other job-irrelevant variables (age and race) to affect
such decisions.
Proponents of the sex-based-bias perspective hold that the highest cadres of
corporations function as old boys' clubs with glass ceilings limiting the ascension
of women to the topmost leadership ranks (Morrison, White, Van Velsor, & the
Center for Creative Leadership, 1987; Solomon, 1990). The good old boys barrier
enforced by stereotyping, excluding women, and causing them social discomfort
holds women down (Haskell, 1991), so much so that, of 400 female executives
surveyed in two studies, 70 percent of one group and 56 percent of the other
reported a male-dominated corporate culture and the existence of a glass ceiling as
obstacles to their success (Business Week, 1992).
In support of the view that women face enormous barriers in the form of sex-based
bias, Kanter (1977) presented evidence that regardless of their qualifications, when
placed in groups in which they are significantly outnumbered by men, women
become tokens and are faced with predictable treatment from others that forces
them into roles that limit their probabilities of success. However qualified they are,
token women become subject to excessive scrutiny, their differences from men are
highlighted and exaggerated, and their attributes are distorted so that they become
trapped in stereotypical roles. These phenomena are so prevalent at the highest
levels that in one study as many as 79 percent of the chief executive officers (CEOs)
and personnel officers at 1,000 leading companies surveyed acknowledged that
women faced the hurdles of stereotyping and of corporate unwillingness to risk
promoting them (Business Week, 1991). In a replication and extension of a 1973
study, Heilman, Block, Martell, and Simon found little change over 15 years in
respondents' perceptions of men, women, and managers: men in general are still
described as more similar to successful managers than are women in general
(1989: 935).
Again, although much of this research pertains to women who are managers rather
than directors, the sex-related-bias argument can be extended into the boardroom
to address issues of committee staffing. Since these subgroups represent the
innermost circles of corporate decision making and power, even greater barriers
against the entry of women are likely to arise in the forms of stereotyping,
exclusionary practices, and social discomfort. Hypothesis 1b: After experiencebased characteristics are controlled, the odds of female directors' membership on a
given board committee will differ significantly from the odds of male directors'
membership.
Additionally, there is some evidence that women's characteristics and behaviors are
not evaluated in the same way as men's (Case, 1993). For example, Jago and Vroom
(1982) documented the existence of a double standard in the performance
evaluation of individuals in leadership roles: men behaving autocratically received
modestly positive evaluations by managers, but women behaving autocratically
were evaluated negatively. Nieva and Gutek summarized previous research findings
as follows: Studies focusing on the evaluation of qualifications in selection and
promotion situations and research on the perceived causes of performance show
fairly consistent bias in favor of men. In sum, sex-related evaluation bias presents
the greatest problem for successful or competent women, in situations where there
is considerable ambiguity, and that involve sex-inappropriate situations or require
sex-incongruent behaviors (1980: 273274).
Since the sex-based-bias argument points to the relevance of director sex for
committee membership decisions, it is likely that a particular experience
characteristic will have a different impact on women's and men's odds of committee
membership. Hypothesis 2b: The influence of directors' experience-based
characteristics on the odds of membership on a given board committee will differ
significantly for male and female directors.
In summary, the experience-based-bias argument and the sex-based-bias argument
provide competing accounts for the representation of male and female directors on
corporate board committees. This study tested competing hypotheses drawn from
these two perspectives in terms of the direct and interactive effects of director sex
on the odds of committee membership.
Previous SectionNext Section
METHODS
Kesner (1988) explored the differences between members and nonmembers of key
board committees. Our study differs from Kesner's in four important ways. First,
Kesner found that women were not disproportionately represented on audit and
compensation committees, but that there were disproportionately few women on
executive and nominating committees. In ancillary analyses, she found that female
directors were generally less experienced than male directors. Kesner argued that
the observed gender differences in committee membership may have been due to
women's generally lower level of experience. In the present study, we directly
tested Kesner's argument that differences in men's and women's experience
account for the representation of women on corporate board committees.
Second, we expanded the range of director characteristics Kesner (1988)
investigated. In addition to the director type, tenure, and occupation variables she
examined, we also examined directors' external corporate and noncorporate
connections since these characteristics are powerful signals of people's experience,
visibility, prestige, and centrality in the business and general community (e.g.,
Lorsch, 1989; Pfeffer & Salancik, 1978; Useem, 1980, 1984; Worthy & Neuschel,
1984). In previous research (Bilimoria & Piderit, 1994), we found that directors'
external linkages distinguished among new members, continuing members, and
nonmembers of key corporate committees.
Specifically, the experience-based director characteristics investigated in this study
were director type (insider or outsider), board tenure, occupation (business or
nonbusiness), business interlocks (directorships of other corporate boards), and
nonbusiness interlocks (directorships of noncorporate institutions such as social,
cultural, educational, and religious organizations). Inside directors have day-to-day
familiarity with company performance
P
Data Set
Data collected for a previous study on the boards of directors of the first 300
companies of the 1984 Fortune 500 list (Bilimoria, 1990) and additional data were
used in this study. The first 300 companies alone were examined because of time
and resource constraints on data collection. We obtained the names and titles of all
directors of the 300 corporations from the 1984 edition of Standard & Poor's
Register of Corporations. Subsequently, information for each director was collected
from corporate proxy statements. Data were collected on 3,924 male and 175
female directors from 270 companies.
To minimize bias, we included only firms with at least one woman on their board of
directors. The final set of individuals consisted of 1,940 male and 175 female
directors from 133 companies; 85.2 percent of the men and 91.1 percent of the
women served on at least one committee. No significant differences existed
between female and male directors on the mean number of committees on which
they served (t = .67, df = 202, p = .505).
Director Characteristics
Director type was measured as a dichotomous variable with inside directors (0)
defined as all current or former employees of a company or its subsidiaries and all
relatives of such persons, and all other directors defined as outsiders (1). Tenure on
a board was measured in years. Occupation was measured as a dichotomous
variable, with directors whose primary employer was a corporation, financial
institution, or law firm classified as holding business occupations (1) and all other
directors categorized as holding nonbusiness occupations (0). Business interlocks
were defined as the total number of concurrent directorships of other corporations
or financial institutions a director held. Nonbusiness interlocks were defined as the
total number of concurrent directorships or trusteeships of political, cultural,
educational, religious, or other nonprofit institutions a director held.
Other variables.
Director sex was measured as a dichotomous variable, with 0 representing women
and 1 representing men. The committee membership variables for the audit,
compensation, executive, finance, nominating, and public affairs committees were
measured dichotomously, with 0 representing nonmembership and 1 representing
membership.
Data Analysis
Logistic regression analysis was used because this method is particularly
appropriate for predicting states of dichotomous dependent variables and permits
use of a mix of continuous (tenure, business interlocks, and nonbusiness interlocks)
and categorical (type, occupation, and gender) independent variables. We
estimated the influence of each variable on the odds of an individual's membership
on a given committee, with odds defined as the probability of serving on the
committee divided by the probability of not serving on the committee.
Each of the director characteristics variables was entered into the regression
equation iteratively so that we could examine the incremental contribution of each
separately. Then the gender variable was entered, followed by the interactions
between gender and each director characteristic. We performed a separate logistic
regression analysis in this way to examine the impact of these variables on the odds
of membership in each committee. The variables were entered in the same order for
each equation, and we attempted no rescaling of variables so that we could
compare results across the different committees. However, in a few cases involving
director type, the first-order interaction could not be entered into the equation
because of extreme anomalies in the distributions resulting from the lack of female
inside directors on certain committees in the data set.
The significance of each variable in the equation was evaluated by examining the
improvement chi square (SPSS Inc., 1990) after the variable was initially entered
into the regression equation.1 The logistic regression coefficients (betas) indicate
the direction and magnitude of each variable's influence on the odds of committee
membership. If the beta for a variable is significant and greater than zero, an
increase of one unit in the variable increases the odds of an individual's serving on
the focal committee. If the beta for a variable is significant and less than zero, then
the odds of committee membership decrease for each one-unit increase in the
variable.
Previous SectionNext Section
RESULTS
This study directly addressed the question of whether, after relevant experiencebased characteristics are considered, director sex influences the odds of committee
membership. We present a systematic interpretation of the results of the logistic
regression analyses designed to predict the odds of membership on each
committee, first for the main effects of the experience-based characteristics, then
for the main effect of sex, and finally for the first-order interactions of sex with each
experience variable. Next, in post hoc analyses, we examined the differences in
director characteristics between men and women. Table 1 presents summary
statistics and correlations.
View this table:
In this window
In a new window
Table 1
Summary Statistics and Correlations
Table 2 presents the results of the analyses predicting the odds of committee
membership. Three sets of coefficients are reported: experience-based director
characteristics, director sex, and interaction variables. The model chi-square for five
of the six committees examined is significant, indicating that the models fit
satisfactorily and that the results can be meaningfully interpreted. Since the chi-
square for the audit committee is not significant, the results for this committee
should be interpreted with caution. The likelihood ratio index, or pseudo-R2, which
provides a measure of fit analogous to the R2 in regression models (Judge, Griffiths,
Hill, Lutkepohl, & Lee, 1985), is also reported. We computed the pseudo-R2 as 1
(La/L), where La is the log likelihood of the hypothesized model and L is the log
likelihood of the null model, which sets all parameters to zero. The pseudo-R2
statistics reported in Table 2 indicate that the improvements over the null models
for the different committees ranged from .05 to .18.
View this table:
In this window
In a new window
Table 2
Results of Logistic Regression Analysisa
Main Effect of Each Characteristics Variable
The first set of coefficients reported in Table 2 indicates how each experience-based
characteristic influences committee membership odds. Outsider status is a favored
qualification for committee membership for four committees (audit, compensation,
nominating, public affairs), and insider status is valued for executive committee
membership. High board tenure is important in gaining membership on four
committees: compensation, executive, finance, and nominating. A high number of
corporate directorships is a significant qualification for three committees
(compensation, executive, and nominating), and fewer corporate directorships are
favored for the public affairs committee. A high number of nonbusiness interlocks is
a favorable characteristic for membership on the audit and finance committees. A
business occupation is important for three committees (executive, finance, and
public affairs), and a nonbusiness occupation is favored for compensation
committee membership and marginally favored for nominating committee
membership.
Main Effect of Sex
Because we entered the sex variable in each logistic regression equation after the
set of experience-based director characteristics, any significant main effect of sex
indicates that it still influences membership odds after other measured differences
in directors' characteristics have been accounted for. The results in Table 2 indicate
that male directors are significantly favored for membership on three of the six
committees examinedcompensation, executive, and financeand female
directors are favored for membership on one committee, public affairs. For the audit
and nominating committees, no significant main effect of sex was detected. Since
the sex variable was incrementally influential for four of the six committees
examined, these findings support Hypothesis 1b rather than Hypothesis 1a, in that
corporate board committee membership is subject to sex-based bias.
The results of the analyses of the main effects of the sex variable indicate an
interesting pattern of sex-typing of committees. Although two committees, audit
and nominating, show no evidence of sex-based bias, certain other committees
appear to be biased for or against women's membership. Men have a marked
membership advantage on the compensation, executive, and finance committees. It
appears that these committees truly operate under the influence of the BOGSAT
phenomenon: the idea that the most important decisions in organizations are
made by a bunch of guys sitting at a table (Willis, 1989: 20, citing Hardesty &
Jacobs [1986], emphasis added).
Female directors have higher odds of serving on a public affairs committee than
equivalent men. The committee's functions may explain this finding. The public
affairs committee is charged with directing and evaluating a corporation's social
performance. This mandate clearly appears to be in the stereotypically perceived
soft area at the periphery of a board's governance work rather than part of the
hard-core work deemed more central to governance like that undertaken by the
compensation, executive, and finance committees. This finding of sex-typing by the
nature of the work undertaken by committees is analogous to Harrigan's (1981)
results that indicated, among Fortune 500 firms, a relatively large number (between
5 and 14 percent) of female directors on boards in the banking, insurance, retailing,
food and agriculture, utilities, and drugs, soaps, and toiletries industries, all of which
are mostly service-oriented, labor-intensive, or women's products industries, but no
female directors in manufacturing and diversified industries such as conglomerates,
electronics, measurement equipment, metals fabrication, paper, rubber, specialized
machinery, steel, and wholesaling.
The absence of sex-based bias in audit committees may be explained by the fact
that this committee is an extremely important board committee, as evidenced by
the extensive scrutiny and regulation to which it is subjected by external agencies
such as the New York Stock Exchange. Therefore, a high degree of clarity about this
committee's purpose, tasks, and evaluation is likely. Previous experimental studies
have shown that information about task requirements is associated with decreased
bias against women (cf. Tosi & Einbender, 1985).
In summary, the results of the main effects of the experience characteristics and
the sex variable on membership odds indicate the pervasive presence of sex-based
bias in the selection of committee members. Only two committees, one of which is
extensively regulated and scrutinized, showed no sex-based bias in membership.
For the other committees, the pattern of bias appears to follow traditional sextyping: committees stereotypically perceived as attending to hard governance
issues favored male membership, and the committee stereotypically perceived to
deal with soft governance issues favored female membership.
Interactions
The third set of coefficients reported in Table 2 reveals the interactions between
experience characteristics and the sex variable. Interactions indicate whether male
and female directors are evaluated differently; they compare whether the
relationship between the experience variable and the odds of membership on a
committee varies with a director's sex. Significant, positive interactions indicate that
a unit change in the experience variable is associated with a larger change in the
membership odds for men than for women. Such interaction terms indicate an
advantage for male directors, since the relationship between the experience
variable and membership odds is stronger for them than it is for female directors.
That is, changes in the experience variable have a greater impact on membership
odds for men than for women, regardless of the sign of the main effect of the
characteristic variable. For example, if directors with low tenure are more likely to
serve on a particular committee (a main effect), a positive sex-by-tenure interaction
would indicate that the difference in odds between a man with low tenure and a
man with high tenure will be greater than the difference for women with similar
tenures. Conversely, significant negative interaction terms signal an advantage for
women; these indicate that the relationship between the experience variable and
membership odds is stronger for women than for men.
The results reported in Table 2 reveal four significant interactions. The sex-by-type
interaction is positive and marginally significant for the audit committee, indicating
that men gain more from outsider status than do women. The interaction of sex with
nonbusiness directorships is negative and significant for the audit committee,
indicating that a unit increase in the number of nonbusiness interlocks is associated
with a bigger increase in membership odds for female directors than for male
directors. None of the interaction coefficients are significant for the compensation,
executive, or public affairs committee, indicating that the director experience
characteristics had similar impacts on membership odds for men and women for
these committees. For the finance committee, the sex-by-occupation interaction is
negative and significant, indicating that a business occupation is more of an
advantage for female than for male directors for gaining membership on this
committee. In contrast, for the nominating committee, the sex-by-occupation
interaction is positive and significant, indicating that men holding business
occupations are favored over similar women for membership. The results also
indicate that the sex-by-type interaction could not be entered into the equations of
the finance and public affairs committees since there were no female inside
directors on them.
In summary, the analyses of the first-order interactions revealed little evidence of
differential treatment based on male and female directors' characteristics: only 4 of
the 28 interaction terms (14.3%) revealed a significant or marginally significant
impact on committee membership odds. In two of the four cases, an identical
change in director characteristics resulted in disproportionately higher increases in
membership odds for men, and in the other half of the cases, the reverse was
observed. Thus, although Hypothesis 2a clearly prevailed in most cases, Hypothesis
2b received some support from the data.
Overall, however, the main and interaction effects presented in Table 2 suggest
evidence of certain sex-based biases against female directors. For all the
committees examined except the public affairs committee, some or all of the male
directors were significantly or marginally favored in ways that the female directors
were not. Men had higher membership odds than equally experienced women for
the compensation, executive, and finance committees, male outside directors were
preferred over female outside directors for membership on audit committees, and
male directors holding business occupations were favored for nominating
committee memberships over similar female directors.
However, sex-based bias also favored women in certain cases: women had a higher
likelihood of public affairs committee membership than similarly qualified men,
female directors with several nonbusiness interlocks were preferred for audit
committee memberships, and female directors with business occupations were
favored for finance committee memberships over similarly qualified men.
Sex Differences in Relevant Experience-Based Characteristics
Before going on to discuss the implications of the results obtained in the logistic
regression analysis, we report a post hoc analysis of the data examining whether
male and female directors differ on the experience-based characteristics important
for committee membership. Table 2 indicates the importance of characteristics, and
Table 3 shows how men and women differ on each of the five experience variables.
View this table:
In this window
In a new window
Table 3
Sex Differences in Director Characteristics
The significant chi-square for the type variable reported in Table 3 indicates that sex
and director type were associated; a larger proportion of female than male directors
were outsiders. Since the main effects of director type reported in Table 2 suggest
that outsider status is an advantage in gaining membership on four of the six
committees studied, this finding does not indicate a general lack of qualifications on
this characteristic for female directors. In fact, in most cases, using outsider status
as a membership criterion would favor women.
Examination of the occupation and sex variables indicates that these characteristics
are not independent: more men than women have business occupations. Since
business occupations are significantly important for membership in the executive,
finance, and public affairs committees, nonbusiness occupations are significantly
and marginally important for membership in the compensation and nominating
committees, respectively, and occupation is not important for membership on the
audit committee (Table 2), the overall lower proportion of female directors with
business occupations appears to be prima facie disadvantageous for women on
three of the committees studied. However, even though a business occupation is
important for membership on the public affairs committee, women are preferred
over men there, suggesting that the distribution of business occupations among
female directors is detrimental to their membership on only two of the six
committees examined, executive and finance.
Like Kesner's (1988) results, the Table 3 findings indicate that overall, the female
directors studied had significantly lower tenures on their boards (5.48 years) than
male directors (8.80 years). Since board tenure is important for four out of the six
committees examined (from Table 2), this finding suggests that women in general
may be less qualified than men on this variable for membership on the
compensation, executive, finance, and nominating committees of boards. This
finding appears to support the notion that women have just not been at the top as
long as men have.
Table 3 also indicates that male and female directors do not differ on the number of
corporate boards on which they serve, indicating