Upload
others
View
7
Download
0
Embed Size (px)
Citation preview
Organizational Genealogies and the Persistence of Gender Hierarchies: Silicon Valley Law Firms, 1970 to 1996.*
Damon J. Phillips Associate Professor of Organizations and Strategy
University of Chicago, GSB Chicago, IL 60637
ABSTRACT
I propose an organizational theory on the genealogical persistence of gender hierarchies that uses insights from organizational imprinting (e.g., Stinchcombe 1965) and gender hierarchies in groups (e.g., Ridgeway 1997) to extend recent research on organizational genealogies (Phillips 2002). I argue that founders replicate the gender hierarchies of their parent firms. Founders from firms that historically lacked women in high-ranked positions are less likely to place women in positions of power within their own firms – independent of the degree of direct contact with high-ranked women. Rather, it is founders from firms that historically have women in prominent positions that are more likely to start firms that in turn hire and promote women into prominent positions. Similarly, founders from firms that historically have women in subordinate positions are less likely to hire or promote women into prominent positions. Also consistent with the theory is the finding that the persistence effect is stronger for founders who were previously lower-ranked employees. Using data on male-founded Silicon Valley law firms, findings support the theory while controlling for several alternative explanations. Supplemental analysis suggests that gender hierarchies may in part be transferred within a set of routines.
* I would like to thank the participants in the Organizations and Markets workshop and the MOB brownbag at the University of Chicago GSB, as well as the participants of the organizations seminar at the Yale School of Management.
INTRODUCTION
At the intersection of research on organizations and gender inequality sits a substantial
body of work on the role that organizations play in creating, sustaining, and altering
differences in attainment between men and women (Baron 1984). Indeed, employment
settings are one of the primary structural arrangements that recreate ordinal status
rankings of individuals by gender – referred to as gender hierarchies (Ridgeway 1997).
Whether the outcome is hiring (Fernandez and Sosa 2003), wages (Madigan and Hoover
1986; Kilbourne, et. al. 1994; Hersch and Viscusi 1996), promotions (Olsen and Becker
1983; Cannings 1988; Lazear and Rosen 1990; Jones and Makepeace 1996), or turnover
(Spurr and Sueyosji 1994), scholars have often found that women are disadvantaged
within organizations. Not surprisingly, attention to gender inequality extends beyond
academia to include members of the popular press (Miller and Friday 1992; Seligman
2001) and business leadership community (Gordon and Whelan 1998).
Despite the vast interest and research in this area, less is understood about the structural
antecedents of gender hierarchies within organizations. Specifically, we know little
about the relationship between founding conditions and subsequent employment and
promotion opportunities for women. Indeed, research within this area is just emerging
(Baron, Hannan, Hsu, and Kocak 2002). At the same time, this new focus is rife with
potential, as it allows us to not only determine the conditions under which gender
inequality emerges in new organizations, but to also improve our understanding of the
persistence of gender inequality across generations of organizations.
2
Drawing upon Stinchcombe’s (1965) attention to organizational imprinting, sociological
research on groups by “microinstitutionalists” (Zucker 1977; Lucas 2003), and a
genealogical approach to organizational populations (Phillips 2002), I propose an
organizational theory of gender inequality that emphasizes the routines (or blueprints)
that founders transfer from parent firms to their new firms. This transfer of routines
reproduces the parent firm’s demographic composition in the new firm. Founders from
firms that historically lacked women in high-ranked positions are less likely to place
women in positions of power within their new firms. In other words, founders from
parent firms that historically had women in subordinate positions (where female
subordination is institutionalized) are less likely to hire or promote women into high-
ranked positions. At the same time, founders from parent firms that historically had
women in high-ranked positions (where female leadership is institutionalized) are more
likely to start firms that hire and promote women into high-ranked positions.
I investigate the persistence of gender inequality using data on 35 years of Silicon Valley
law firm startups that capture the gender composition of law firm progeny and their
parent firms. I hypothesize and find that (ceteris paribus) progeny firms are more likely
to place women in partnership positions to the extent that the progeny’s founders came
from firms that historically had women partners. At the same time, progeny firms are
less likely to place women in partnership positions to the extent that the progeny’s
founders came from firms that historically had women associates (ceteris paribus).
3
I consider these findings alongside two alternative explanations. First, I insure that my
findings are not caused by direct contact with women in parent firms – as Allport’s
(1954) seminal research may suggest (see also Cook 1985; Pettigrew 1998). If my thesis
is correct, founders will replicate the gender hierarchy independent of the degree of direct
contact with women in the parent firm. A parent firm that has had a woman partner for
the first nine of its ten years of existence is more likely to have progeny that will promote
women to partner, even if a progeny’s firm founder worked in the only year in which
there was no woman partner.
A second alternative posits that intergenerational persistence of gender hierarchies occur
when key decision-makers in the parent firm leave to replicate a similar structure in the
progeny firm. In other words, social structures are replicated because the same
individuals that established the opportunity structures in the parent firm also construct the
opportunity structures in the new firm. This alternative suggests that there would be no
replication of the parent’s gender hierarchy by founders who were not decision makers in
the parent firm. On the contrary, I will argue that the lower-ranked, non-decision-making
personnel are more likely to assume a parent’s demographic composition as objectified
fact (Berger and Luckman 1967). This yields an opposite conclusion: lower ranked
employees are more likely to replicate the gender hierarchy of the parent firm than are
higher-ranked employees. Thus, after testing the initial hypotheses, I will examine
whether these effects are greater when the founder was former a lower-ranked employee.
4
The implications of this research present a promising contribution to the literature. In
general, an organizational genealogy framework not only suggests ramifications for the
fate of the parent organization and its offspring (Phillips 2002), but also on the
intergenerational variation and persistence of organizational opportunity structures. As
founders leave their parent organizations, the organizational models they carry with them
reproduce the set of employment opportunities and constraints faced by employees in the
parent firm. In other words, the intergenerational diffusion of employment practices may
contribute to the persistence of sex segregation within organizations.
BACKGROUND AND THEORY
Among the key contributions of Stinchcombe (1965) is that organizational structure is
influenced by conditions that existed at the time of an organization’s founding (see also
Kimberly 1979). Components of organizational environments are “imprinted” on the
structure of new organizations. This insight has influenced many veins of research within
organizational sociology, but few more than population ecology and neo-institutional
theory, whose adherents have shown that not only do exogenous forces influence the
strategy, structure, and vitality of organizations (Meyer and Rowan 1977; Carroll and
Delacroix 1982; DiMaggio and Powell 1983; Swaminathan 1996; Hannan 1998), but so
do those founding conditions which are more endogenous to an organizational population
(Carroll and Hannan 1989; Tucker, Singh and Meinhard 1990). To the extent that we
understand founding conditions, we are better able to understand the emergence and
persistence of an organization’s structure and set of routines. Moreover, since an
organization’s structure at founding substantially influences structure and action later in
the life course of the organization (Stinchcombe 1965; Boeker 1989; Baron, Hannan and
5
Burton 1999), Stinchcombe’s insight becomes an important lens for understanding
organizational structure well beyond the organization’s founding.
While Stinchcombe (1965) draws attention to macro-institutional forces (he uses
examples that contrast organizational forms before and after major societal events such as
the rise of the Soviet Union, the Industrial Revolution, and the Meiji Restoration in
Japan, pp.153-155), I draw attention to the role of micro-institutional forces (Zucker
1977) that lead to imprinting. My motivation is twofold. First, recent research suggests
that to the extent that macro-institutional forces can be isolated, there remains substantial
variation in a cohort’s organizational structure, strategy, and template for the organization
of work (Baron, Hannan and Burton 1999). Second, there is an established literature
demonstrating that given an organization’s structure at founding, there is remarkable
resilience of that structure over time (Meyer and Scott 1983; Hannan and Freeman 1984;
Boeker 1989). My general interest is in examining the factors that constrain founders in
their choice of organizational blueprints.
In understanding the persistence of gender hierarchies across generations of
organizations, I incorporate “micro-institutionalist” research on the transmission and
maintenance of organizational routines (Zucker 1977; also see Levitt and March’s 1988
discussion on organizational memory) with research on organizational genealogies
(Aldrich and Pfeffer 1976; Freeman 1986; Phillips 2002). In particular, I anchor my
model of an intergenerational persistence of gender hierarchies on Phillips’ (2002)
argument that organizations (progeny organizations) often arise as founders leave
6
existing organizations (parent organizations), transferring some of the parent’s routines to
the progeny organization. As a consequence, the structure and the culture of progeny
organizations is in part determined by the structure and culture of their parent
organization. Thus, when examining organizational progeny as proliferators of gender
hierarchies, the parent organization’s routines must be included among the set of initial
conditions that influences a progeny’s tendency to place women in high-ranking
positions.
The parent-progeny transfer of routines that affect gender hierarchies conceptually falls
into two categories. First are routines that are transferred by the founder as formal rules
(Dutton and Starbuck 1978; Nelson and Winter 1982; Baron, Hannan and Burton 1999) –
although they need not be transferred specifically to affect the fate of women employees.
Examples of formal rules are recruitment and evaluation policies, as well as other rules
that dictate the organization of work (e.g., written job descriptions, policies on flexible
work arrangements) (Baron, Hannan and Burton 1999; Fernandez and Sosa 2003). Other
formal, but less direct rules and routines may also include a set of rituals introduced by a
founder to achieve a particular level of employee commitment and cohesion (Schein
1992). All-in-all, this suggests that founders may rewrite the parent’s gender hierarchy
into their new organization through formal rules.
A second category of routines includes “taken-for-granted” rules or interpretations of
legitimate action that founders (perhaps unwittingly) draw from their past organizational
experiences. This category of routines builds upon multiple studies by scholars in the
7
status construction vein (Markovsky, Smith and Berger 1984; Ridgeway and Smith-
Lovin 1999; Ridgeway and Erickson 2000) and recent evidence provided by Lucas
(2003). Using a laboratory setting, Lucas (2003) found that when group members learn
that “successful groups have women as leaders, and when members see that groups
similar to their own have women in leadership positions … female leaders had as much
influence as did male leaders with comparable ability …” In my context, some founders
leave a parent firm in which female leadership is institutionalized. The fact that female
leadership was institutionalized in the parent makes the founder more likely to view
female leadership positively. Ridgeway and Erickson’s (2000) research supports this
contention. Their work demonstrates that beliefs about status hierarchies, once
established, can be spread to others as individuals enact learned beliefs about the relative
status of women and men even as they enter a new context. As a consequence, gender
hierarchy is transferred from the parent to its offspring through the actions of the founder
and those whom that founder “teaches”. Finally, while a founder’s beliefs may be
explicitly stated, it need not be. In fact, Zucker’s (1977) and Lucas’ (2003) work
suggests that founders may never question their views on gender and leadership.
The focus on the transfer of routines, both explicit and taken-for-granted, places this
theory in an agnostic position when it comes to the psychological antecedents of gender
inequality across organizational generations. That is, it may very well be that among the
routines and beliefs transferred is an overt bias against women leaders. However, it may
also be the case that by transferring particular employment models to a new firm (e.g., an
employment model that emphasizes commitment to the employer), the founder may be
8
influencing the present and future demographic composition in ways that are unpredicted
and unintended.
HYPOTHESIZING THE TRANSFER OF GENDER HIERARCHIES
I test my theory by focusing on the degree to which female leadership and subordination
(location in a gender hierarchy) is institutionalized in a parent organization. Ceteris
paribus, founders from parent organizations that have a long history of women in high-
ranked positions are more likely to have high-ranking women in their own organizations
(either by promotion from within or hiring from the external labor market). These
founders transfer routines that facilitate the advancement of women. Capturing the
history of women in high-ranked positions by the proportion of years over a parent’s life
cycle that women have been high-ranked, I hypothesize,
H1 (Institutionalized Female Leadership): Ceteris paribus, the more years over its life-cycle that a parent organization has women in high-ranked positions, the greater the likelihood that its organizational progeny will place a woman in a high-ranked position. A more subtle prediction, but one that is consistent with the framework, is that founders
from parent organizations that have a long history of women in lower-ranked positions
are less likely to have high-ranking women in their own organizations. That is, holding
the representation of women in high-ranked positions constant, the more the parent
organization is characterized as having women in lower-ranked positions, the lower the
likelihood that departing founders will place women in high-ranked positions. Similar to
the argument for H1, having a history of women in subordinate positions increases the
chances that founders carry with them formal and taken-for-granted routines that
replicate women’s subordination.
9
H2 (Institutionalized Female Subordination): Ceteris paribus, the more years over its life-cycle that a parent organization has women in low-ranked positions, the lower the likelihood that its organizational progeny will place a woman in a high-ranked position.
With Hypothesis 2, I am suggesting that having a long-history of women in subordinate
positions has a greater negative effect on the attainment of women in the new firm than if
there were no women at all in the parent firm. In this way, H2 is a powerful test of the
conceptual model, with an interesting implication: women have a higher chance of
making partner in a law firm when the founder came from a firm with no women,
compared to a firm in which women were consistently in subordinate positions.
Finally, there should be differences in the hypothesized effects based upon the former
position of the founder. Lower-ranked members should be more prone to recreate the
gender hierarchy because it is more likely to be seen as an objectified reality (Berger and
Luckman 1967; Zucker 1977). Especially for lower-ranked members with little labor
market experience outside of the parent firm (which is the case in my data), the parent’s
organizational blueprint lacks any competing alternative. This ignorance enhances the
taken-for-granted gender hierarchy of the parent firm.
This prediction is consistent with a long tradition of research shows that individuals that
value membership into in a group yet feel insecure in that membership (e.g., an associate
in a law firm) are more likely to embody the norms of that group (Dittes and Kelley
1956; Blau 1960; Homans 1961). In applying this particular line of reasoning, however,
suggests that the potential founder would be more aware of the transfer of routines from
the parent than an “objectified reality” explanation would emphasize. Nevertheless, both
10
explanations lead to the same conclusion: former lower-ranked employees should be
more likely to replicate the institutionalized gender hierarchy of the parent firm.
H3a: The hypothesized effect in H1 (Institutionalized Female Leadership) should be greater when the founder was formerly a lower-ranked member of the parent firm. H3b: The hypothesized effect in H2 (Institutionalized Female Subordination) should be greater when the founder was formerly a lower-ranked member of the parent firm. While potentially there are several alternative explanations, at least two are critical. First,
it is possible that the parent and progeny are in a market position (or niche) that drives
particular patterns of gender hierarchy across generations, but independent of any
routines that are transferred from the parent firm. For example, in law firms one may
expect that women are disproportionately underrepresented in a particular area of law,
especially at the partnership level. In this case the fact that the gender hierarchy is
replicated from the parent to the progeny is driven more my the economics and social
dynamics associated with the type of service offered than by any parent firm-specific
routines transferred to the progeny law firm. While other empirical studies place doubt
on this alternative explanation (Chambliss 1997), I address this concern in two ways.
First, I control for different areas of practice. Second, I code the proportion of women
associates in the law firm when predicting the likelihood of a woman being placed in as a
partner. This will control for the extent to which the firm is in a market niche or area
over or underrepresented by women.
A second alternative builds upon Allport’s (1954) seminal work on discrimination. His
“social contact theory” suggested that discrimination against lower status minorities
11
could be remedied if: (a) members of the majority could directly interact with the
discriminated member; (b) in a context where the two groups were of equal status; (c)
and were pursing a common goal; (d) through interdependent action. Extending this
logic, it is possible that departing founders who have directly interacted with women in
high-ranking positions are more likely to have women in high-ranking positions in their
new organizations (Lundberg and Startz 1983; Jackman and Crane 1986). Similarly,
founders with direct interactions with women in subordinate positions may be less likely
to place women in high-ranking positions in their own firms. In this second case, the
interaction would amplify any existing negative stereotypes about women.
To test this second alternative, I will control for the number of years each departing
founder worked in which there were women partners and associates, respectively. While
I lack the evidence to confirm actual interaction, it is likely that some form of interaction
would occur given that the organizations in my dataset are relatively small (the average
size of a parent law firm is 9.8 lawyers). Although I expect that Allport’s (1954) thesis
would be confirmed, the transfer of routines that affect women’s attainment in the new
firm should be independent of the direct interaction that the founder had in the parent
firm.
CONTEXT AND DATA Women’s Attainment in Private Practice Law Firms The attainment of women in law firms has received frequent scholarly attention (Epstein
1981; Halliday 1986; Donovan 1990; Spurr, 1990; Hagan and Kay 1995; Dixon and
12
Seron 1995; Kay and Hagan 1998, 1999; Gorman 2004). This topic is among the few
that has engaged not only sociologists, but economists (Spurr 1990), and legal scholars
(Rhode 1988; Babcock et. al. 1996).
Interest in women’s movement into the partnership ranks of law firms is fueled by two
related factors. First, promotion to partner not only involves the greatest increase in
income within a law firm, but partnership includes membership to the professional elite
with access to substantial social and political capital (Nelson 1988). Moreover, women’s
attainment within law firms has larger societal ramifications for access and opportunities
(Hagan and Kay 1995). Accordingly, women’s occupancy of high-ranked positions in
law firms is used as one barometer of generalized gender inequality.
Second, interest remains unabated because the progress of women into the partnership
ranks has been slow (Mossman 1988; Abel 1989; Phillips and Beckman 2004). Although
there has been a substantial rise in the proportion of women law school students and law
firm associates, progress with respect to women promoted to partner has been
comparatively slower (Abel 1989; Abrahamson and Franklin 1986; Hagan and Kay 1995;
Hull and Nelson 2000). Women are less likely to be promoted than men (Spurr 1990;
Kay and Hagan 1998), and more likely to leave before the promotion decision (Spurr and
Sueyoshi 1994), even controlling for experience or human capital-based qualifications
(Nelson 1988; Phillips 2001). Many firms have yet to promote any women to the level of
partner. Differences between men’s and women’s promotion chances seem to remain
after taking into account women’s choice to leave before the promotion decision (Hull
13
and Nelson 2000). In fact, there is some concern that evidence is pointing to a decline of
an already low proportion of women partners (Ziewacz 1996).
The Data: Silicon Valley Law Firms As part of a larger project on the population dynamics of law firms, I collected data from
the annual Martindale-Hubbell Law Directory from 1946 through 1996, for law firms and
attorneys in Silicon Valley, California. The directories list attorney and law firm
characteristics and, when followed across time, provide information on the life chances of
law firms and whether the founder was previously affiliated with another Silicon Valley
law firm. As Suchman (1993, 2000) and Escher and Morze (1998) noted, Silicon Valley
is a relatively self-contained market for legal services in Northern California, with scant
legal activity before World War II. Silicon Valley comprises the following ten cities:
Redwood City, Menlo Park, Palo Alto, Los Altos, Mountain View, Sunnyvale, Santa
Clara, Cupertino, Campbell, and San Jose. In 1946 the area had only six law partnerships.
By 1996, Silicon Valley hosted 209 law partnerships employing 2,375 active attorneys.
For each law firm, I coded its founding date (its first appearance in the directory) to
alleviate any left-censoring. In all, I collected data on 513 law partnerships across the
fifty years, which comprises every firm listed with more than one active attorney -- solo
practitioners were excluded because they cannot be parent firms and lack distinctive
hierarchical positions. Of these firms, 137, or 27 percent, were founded by attorneys who
had a previous affiliation with a Silicon Valley firm (i.e., law firm progeny).
14
I test my hypotheses on a subset of these data, using the last twenty-six years (1970-
1996). The shorter observation period used in this study is driven by the fact that before
1973 there were no women partners in any of the firms in my sample. Accordingly, I
considered my risk set any law firm progeny founded after 1970.1 I also bounded the risk
set by only including law firms with all-male founders. This is to allow the cleanest test
of my theory. Not only would including women founders (who by definition would be
partners in the new firm) confound the measure statistically, but it would add another
consideration to an already complex conceptual framework. The differences by gender in
the response to the institutionalization of women in leadership and subordinate positions
is complex (Ely 1994; Lucas 2003), and thus set aside for future consideration. The
dataset used to test my hypotheses includes 99 progeny law firms (founded by men only)
at risk of placing a woman as a partner.2
Dependent Variable: Having a Woman in a Partner Position. I coded a dichotomous,
variable to capture whether or not a firm had a woman partner. Each firm is followed
annually until one of the firm’s partners is a woman. Sex of the attorney was determined
using first names, with any gender neutral name (e.g., Pat) classified as male. Although
there were no questions of partners with gender neutral names, there were associates for
which sex was unclear. When possible, sex was confirmed with photographs available
on the web or in local bar association membership photographs. The dependent variable
equals one for the firm-year in which there is a woman partner. When this occurs, the
firm exits the risk set. Mirroring the low occurrence of partnering a woman in other
1 In additional analyses, I varied the observation window from 1960-1996 to 1976-1996. None of the results presented here were altered by changes in the observation window. 2 In the discussion I do report supplementary analysis using data that included women founders.
15
samples of law firms, only 16 of the 99 firms listed a woman as partner by 1996.3 Given
that the partnering of a woman was relatively uncommon in my data, and typically of
only one woman at a time, modeling a risk of a firm partnering a woman for the first time
is appropriate.
Independent Variable: Institutionalization of Gender Hierarchies from Parent Firms. I
consider female leadership to be institutionalized the greater the proportion of years that a
progeny firm’s parent had any women partners before an employee left to found a new
firm. This was coded by calculating the proportion of years that there was a woman
partner in the parent firm, beginning with the parent firm’s first year and ending with the
year that the founder departs. Using this scheme, progeny law firms with parents that
never had any women partners are coded as zero. A value of one is assigned to those
progeny of parents that have had women partners in each year of the parent firm’s
existence. Similarly, we coded the proportion of years that a parent firm has women
associates as an indicator of how institutionalized women in subordinate positions were.
Founder’s Former Rank
In general, law firms have two positions: associates and partners. Associates enter the
firm directly from law school or after a one-year judicial clerkship. They generally work
under the firm’s partners, who leverage the work of associates by hiring them at a given
salary and billing them out to the firm’s clients at multiples of that salary. Associates are
considered for promotion to partnership after a period in which they work under the
3 The rate of partnering a woman for all 333 firms in existence from 1970 to 1996 is lower than the percentage partnering a woman by male-founded progeny law firms. Only 14.7% of the 333 firm had made a woman a partner by 1996. In comparing this statistic with data from the ABA, there is an indicator that my sample is representative. 13% of national law firms had women partners in 1995 (A.B.A. Commission on Women in the Profession 1995).
16
supervision of the partners, receive training, and exercise increasing responsibility
(Smigel, 1969; Nelson, 1988; Galanter and Palay, 1991). Those not promoted must leave
the firm (the “Cravath” or “up-or-out” promotion system). Partners own the firm and
share in its profits. The transition from associate to partner is generally accompanied by
increased income, as well as a new functional role within the firm. Partnership
encompasses several new tasks and responsibilities, emphasizing skills in firm
management that transcend the traditional tasks involved with practicing law as an
associate (Nelson, 1988).
To test hypotheses 3a and 3b, I constructed a variable that equals one if the founder was a
former associate, zero if the founder was a former partner. If the intergenerational
persistence of gender hierarchy is more salient for founders who were formerly lower-
ranked members of the parent firm (associates), interacting this term with each of the
institutionalization variables would yield an amplified effect. That is, former associates
should be affected by institutionalized woman leadership and subordination more than
former partners. If this interaction effect is not statistically significant, then former
associates (lower-ranked members) are no less likely to be influenced by the history of
gender hierarchy than former partners (higher-ranked members).
Controlling for Allport’s (1954) Social Contact Theory as an Alternative. As a proxy for
former contact with women at the partner and associate levels, I coded two variables.
The first captures the number of years the founder worked with women partners in the
parent firm. The second captures the number of years the founder worked with women
17
associates in the parent firm. Allport’s (1954) thesis suggests that the greater the number
of years of social contact with women partners, the more likely that the founder will have
a woman partner in his own law firm. Conversely, previous contact with women
associates should make founders less likely to make a woman partner in their own law
firms.
Figures 1a and 1b illustrate the difference between the hypothesized institutionalization
indicator and the Allport social contact alternative. In each figure, a parent firm has a
woman promoted to a high-ranked position in the year five. Also, each example features
an employee that enters the firm in the 15th year and leaves in the 20th year to found a
new firm. The difference between figures 1a and 1b lies in the number of years that the
high-ranked woman remains in the organization. In figure 1a, the high-ranked woman
remains in the parent firm at least fifteen years. When the employee leaves to found a
new firm, the high-ranked woman is still a member of the parent firm. In this case, the
degree to which woman leadership is institutionalized is 0.75, since a woman was high-
ranked in 15 of the 20 years that the parent firm was in existence. Measuring Allport’s
alternative, the new employee had 5 years of direct interaction with the high-ranked
woman. In figure 1b, the high-ranked woman leaves after 5 years – well before the new
employee enters the firm. While the years of direct interaction is 0, the value of
institutionalized woman leadership is greater than zero, or 0.25 (5 years divided by 20
years). This indicates that although the new employee did not personally interact with
18
the woman leader, any organizational routines that were associated with that woman’s
leadership position are transferred by the employee into the new firm that is founded.4
----------------------------------------
Put Figures 1a and 1b About Here
----------------------------------------
Additional Control Variables
I control for the possibility that attorneys (as future founders) seek firms with a particular
gender hierarchy, then leave to found firms with a similar hierarchy. This endogeneity
effect suggests that the informed choices of individual attorneys (e.g., founders who
promote women previously chose to work in firms where women were in leadership
positions) drive any results. Alternatively, I argue that founder choices and action are
framed by the parent firm’s social structure (Blum 1985). I address this alternative by
controlling for the proportion of lower- and higher-ranked women that existed in the year
that the future founder began their career. That is, I follow each founder back to the year
in which they entered the parent firm then record the proportion of women associates and
partners. If it is the case that founder sought particular gender hierarchies earlier in their
careers, then replicated the hierarchies in their newly founded firms, these controls should
explain away my hypothesized effects.
4 Figures 1a and 1b present illustrations to distinguish between the two effects, neither is presented as representative. In fact, situations similar to 1b occur less often in my data. Only 6% of the progeny feature founders that have a zero value of direct contact and a non-zero value of the level of institutionalized woman leadership transferred to the progeny firm.
19
To capture any population-level time trend effects, I generated a count variable ranging
from 1 to 26, for each year of the observation period. I included the log transformation of
the time trend variable since this specification improved the fit of the models. Firm size
was operationalized in two ways: (1) the total number of full-time partners and (2) the
total number of full-time associates. Given that the distribution of firm sizes was log-
normal (skewed to reflect a few relatively large firms), the log of each size variable was
coded.5
In addition to measuring size, I coded a dummy variable for whether a firm was a branch
office, in case the demographic composition of branch offices are influenced by a home
office outside of Silicon Valley. In particular, a woman partner may transfer from
another office of the same firm. I also control for departures at the partnership level,
although the direction of its effect is unclear. While departing partners create vacancies
that can be potentially filled by women, partner turnover may also indicate firm’s poor
health. For each year I coded the number of partners that left the firm at year’s end, then
divided each number by the total number of partners in the beginning of that year. I
similarly calculated the attrition of associates. Each attrition variable varied between 0
(no departures at that rank) and 1 (complete departure at that rank). In addition to the
proportion of departures, I also included two indicators for growth: the proportion of new
associates and the proportion of new partners. For either indicator, the rate of growth
reflects the health of the firm. As noted in Phillips (2001), law firm growth rates drive
5 As elaborated in Phillips (2002), this yields estimates that are easier to interpret than including the leverage ratio, a common measure of firm structure and performance.
20
promotion rates. Thus, we would expect the growth rates to increase the likelihood of
promoting a woman to a firm’s partnership.
I also coded the proportion of women associates. Building off of Kanter (1977), Cohen,
Broschak, and Haveman (1998) note that the proportion of women at lower ranks
influences the likelihood that a woman would be promoted to higher levels of an
organization’s hierarchy. However, research by Ely (1994) on women attorneys
demonstrated that in situations where leadership is male-dominated (as is the case in this
study) junior women may view promotion to partnership as undesirable or unattainable.
Thus, it is difficult to estimate the effect of this variable. I also coded the squared term of
the proportion of women associates. However, the effect of the squared term was not
significant in any model, nor did it improve the model fit. Thus I did not include the term
in the presented results.
Silicon Valley can be roughly distinguished geographically by whether the law firm is
located in the north or south end of Silicon Valley. The north end is represented by the
cities most associated with Silicon Valley by the popular press: Palo Alto, Menlo Park,
and Redwood City. These cities host Stanford University and many of Silicon Valley’s
successful venture capitalists. Not only are these firms considered to be higher status, but
they also serve the larger and more successful corporate clients. To the extent that these
firms have greater market power, I would expect that firms located in the north end of
Silicon Valley to be less likely to promote women to the position to partner. To capture
this effect I coded a one for whether the law firm is located in any one of the three cities.
21
Firm scope was operationalized as a continuous variable from 0 to 1. It captures the
number of law practice areas that a law firm reported in the Martindale-Hubbell Law
Directory in a particular year, divided by the total number of practice areas that were
reported across all firms in that year. This measurement allowed a relative measurement
of firm scope. I also included a quadratic term for firm scope to test for curvilinear
effects. In addition to scope, dummy variables for different areas of law were coded. To
preserve statistical power, I included only those that improved the fit of the model:
corporate securities law, labor/employment law, real estate law, and intellectual property
law.
Not only may opportunities vary by practice area, but through investment in new practice
areas. To capture the role of growth through early-mover expansion, I coded a variable
one if the law firm was the first in Silicon Valley to move into a new area of law before
any other firm in the region, zero otherwise. Not only are these “local innovators”
making a substantial investment for future growth, but they are also seeking new and
different individuals to staff a new area of practice. Moreover, their innovativeness
suggests that these firms are less constrained by the set of legal norms traditionally
practiced.
To minimize concerns on unobserved heterogeneity due to characteristics of the parent
firm, I included the parent firm’s age and size. Larger and older parent firms may have
22
been more likely to have women at the associate and partner levels, confounding the
results.
Method of Estimation
I estimated the likelihood of admitting a woman to a firm’s partnership with piecewise
constant exponential models. In these models, I split the time axis into time periods
according to firm age. Although the models assumed that transition rates were constant
with each of the time periods, base rates varied freely across them. The assumption is that
period-specific baseline rates can vary across time periods, but the covariates have the
same (proportional) effects (Blossfeld and Rohwer 1995; Sørensen 1999).
The resulting models give an age-dependent constant (a “y-intercept”) for each time piece
of the model. There are different strategies for choosing the appropriate time periods.
Some theoretical predictions may require that the time periods take on particular values.
For this paper’s models, I assumed no a priori knowledge of age dependence. The null
model was an exponential model without time periods, in which it was assumed that rates
were time invariant. The y-intercepts included in the model were statistically significant
with respect to a chi-squared model improvement test.
23
RESULTS
Table 1 presents the summary statistics for each of the variables. Table 2 provides
pairwise correlations. The two key independent variables (Institutionalized Woman
Leadership and Institutionalized Woman Subordination) are correlated (r=0.36), but not
to the extent that mutlicollinearity is a concern. Two pairs of variables are correlated
over 0.50. Partner growth is highly correlated with associate growth (r=0.63). In
addition, the correlation between the proportion of years that a parent firm has women
associates (Institutionalized Woman Subordination) and the number of years a founder
formerly worked with women associates (“social contact” with women associates) is
correlated at 0.73. While not cause for alarm, it will be important to insure that the
independent variables are not misestimated when the variable for social contact with
women associates is included. All other correlations are less than 0.50.
----------------------------------------
Put Tables 1 and 2 About Here
----------------------------------------
Testing Hypothesis 1: “Proportion of Years the Parent had Women Partners”
In table 3 are the models that test my hypotheses and the alternative explanations. Model
1 estimates the firm age time pieces (y-intercepts) and the timetrend variable. As
expected, the timetrend variable is positive as a greater proportion of firms made a
woman a partner over time. Model 1 also includes the two female institutionalization
variables. Male founders were more likely to make a woman a partner in their new firms
24
when their parent firms had a history of women partners. This effect remains positive
and significant across the remainder of the models, as a host of control variables are
included and alternative explanations tested. That is, this effect is independent of the new
firm’s size, scope, market position, product/service expansion, location, growth rate,
attrition, the proportion of women attorneys, the founder’s previous position, the presence
of women when the founder first worked in the parent firm, and the size and age of the
parent firm. Additional analyses revealed that this result holds if the two
institutionalization variables are entered separately. Thus, hypothesis 1 is supported.
----------------------------------------
Put Table 3 About Here
----------------------------------------
Testing Hypothesis 2: “Proportion of Years Parent had Women Associates”
While not significant in model 1, there is evidence that hypothesis 2 is supported, albeit
conditionally. Model 2, suggests that male founders were less likely to make women
partners in their new firms when their parent firms had a history of women associates.
The fact that the effect was not significant until the control variables were included
warranted further investigation. Additional analyses revealed two control variables that,
when included, strengthened the statistical significance of the women associate
institutionalization variable: the size of the parent firm and the number of partners in the
progeny firm. Large parent firms are contexts in which the institutionalization of women
associates is more likely (the correlation between the two is 0.20, see table 2). The
25
statistical significance of H2 was also dependent upon the number of partners in the
progeny firm. The rationale here is less clear. The correlation between the number of
partners and the women associate institutionalization variable is -0.05 (not statistically
significant). It is possible that the institutionalization of women associates is more
powerful when there are fewer decision-makers in the progeny firm.
Figure 2 uses model 2 to plot the multiplier of the hazard rate of making a woman
partner. The horizontal axis represents the proportion of years that the parent firm had
women partners (bold line) and associates (dashed line), respectively. The figure
illustrates that as the proportion of years that the parent firm had women partners
increases, so does the likelihood of the progeny firm making a woman partner.
Conversely, the likelihood of the progeny firm making a woman partner decreases as the
proportion of years that the parent firm had women associates increases. Together,
model 2 and figure 2 suggest that the likelihood of a woman being made partner is
greatest when the parent firm had at least one woman partner and no women associates
over the course of its existence. The likelihood is lowest when the parent firm had a long
history of women only in the associate position.
26
Figure 2: A Progeny's Firm Likehood of Making a Woman a Partner as a Function of the Gender Hierarchy in the Parent Firm
-30
-25
-20
-15
-10
-5
0
5
10
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Proportion of Years Parent Firm Had Women Members Before Founder Left
Mul
tiplie
r of t
he H
azar
d R
ate
of M
akin
g a
Wom
an a
Par
tner
(P
roge
ny)
Women Partners in the Parent Firm Women Associates in the Parent Firm
27
Model 2 also suggests that women’s attainment is positively related to the firm’s health.
Firms experiencing high partnership growth were more likely to make a woman partner
(p<0.05). Consistent with this finding, high turnover at the partner level reduced the
chances that a woman would be made partner (p<0.05). Growth by expansion into new
markets positively influenced the likelihood of making a woman partner (p<0.10).
There was little effect for growth and turnover at the associate level. In all, the indicators
for growth were more meaningful than the size, age, and scope of the firm.
Firms located in north Silicon Valley, where the most lucrative and prestigious clients
were located, were less likely to make a woman partner (p<0.01). The practice area
indicators reveal that firms practicing labor/employment law were less likely to make a
woman partner (p<0.10). However, firms practicing in the areas of real estate and
intellectual property were more likely to make a woman partner (p<0.05).
Testing Hypotheses 3a and 3b: Differences by Founder’s Former Rank
Models 3 and 4 test whether there are differences in the hypothesized effects by the
previous rank of the founder (Hypothesis 3a and 3b). This is tested by asking whether
associates (lower-ranked members) are more likely to be influenced by the gender
hierarchy in the parent firm. If so, we should see the two institutionalized variables have
their strongest effect when carried from the parent firm by a former associate, compared
to a former partner (the reference group). Indeed, models 3 and 4 suggest that former
associates are more likely to be influenced by the parent’s gender hierarchy than former
partners. Hypothesis 3a is supported in model 3 (p<0.05). Founders who were former
28
associates are more likely to make a woman a partner when the parent firm had a history
of women leadership. The effect is substantial: a former associate is 158% times as likely
to make a woman a partner as a former partner, given the same level of institutionalized
women leadership. Hypothesis 3b is marginally supported in model 4 (p=0.08).
However, the p value represents the more conservative two-tailed test (it would be
p=0.04 using a one-tailed test), so I do consider the marginally significant result
substantively important. The evidence for H3a and H3b together suggest that a former
associate is more likely to be influenced by institutionalized women leadership and
subordination than former partners.6
For each of the five models in table 3, model fit is determined by the Bayesian
Information Criterion (BIC), which can be applied to survival models (Schwarz 1978;
Burnham and Anderson 1998; Volinsky and Raftery 1999).7 According to Raftery
(1995), BIC approximates the ratio of posterior probabilities for two competing models.
The model with the lowest BIC explains the data with the least expected loss of
information. When comparing two models the difference in the respective BIC values
allows one to determine the strength of the improvement of the better model. A
difference greater than 6 is considered as “strong” evidence that one model is better
fitting; differences greater than 10 are considered “very strong” evidence (Raftery 1995).
6 I also ran analyses that interacted former rank with the social contact control variables. None of the estimates were statistically significant. 7 The formula for BIC is G2 – df*log(N), where G2 is the likelihood-ratio Chi-square, df is the degrees of freedom, and N is the number of cases. This goodness-of-fit indicator is a more conservative means of comparing the relative fit between two models (that need not be nested), especially when the models are complex (e.g., many degrees of freedom). Compared to simply using the log-likelihood, the BIC tends to make it more difficult to reject the null when the alternative is complex.
29
For example, the improvement from model 3 to model 3 is 7.00, suggesting that the
evidence is strong that model 3 has a smaller expected loss of information.
DISCUSSION AND SUPPLEMENTAL ANALYSES
This paper presents an initial effort at understanding the relationship between
organizational genealogies and gender hierarchies. As such, I offer evidence that the
gender hierarchies of the parent firm are important to understanding the attainment of
women in that firm’s offspring. Not only do progeny respond to women leadership in the
parent firm, but to women subordination as well.
Overall, the evidence that institutionalized woman leadership enhances women’s
partnership chances in the next generation of organizations is compelling. Indeed, not
only was this hypothesized effect strong and consistent across the models of table 3, but
the direct contact variable (the Allport hypothesis for women partners) was statistically
significant as well. Men who directly interact with women in high positions, or work in
firms that have historically had women in high positions, are more likely to place women
in high-ranking positions when starting their own firms. These findings suggest that over
time, two intergenerational subpopulations of law firms may form. One subpopulation of
firms may exist in which gender inequality persist across generations. Men leave firms
without women leaders, found new firms in which women are not given leadership
positions, then spawn new firms that recreate the gender inequality. A second
30
subpopulation may consist of the intergenerational transfer of routines, cultures, or
structures that lead to new firms with woman leadership.8
Another finding of this paper is that former associates are more likely to be influenced by
the history of female leadership in the parent firm. This suggests that routines associated
with demographic composition are more likely to be transferred by lower-ranked
employees. Interestingly, Phillips (2002) found that those routines associated with
organizational structure (firm scope and span of control) were more likely to be
transferred when the founder was a former senior member of the parent firm. In fact,
Phillips (2002) argued that the higher the rank, the more resources and routines that the
founder transfers to the progeny firm. These resources and routines enhance the
progeny’s chances of survival.
These findings and arguments can be resolved by recalling Phillips’ (2002) emphasis on
the transfer of resources as well as routines. It is likely that while more resources are
transferred the higher the former rank of the founder, only those routines associated with
those resources are transferred. For example, if a former senior partner leaves a parent
firm, they are more likely to take clients (a key resource) with them. However, taking
clients will also affect the routines transferred to the progeny, especially those routines
directly associated with the client (e.g., a type of law practiced). Here resources dictate
8 I also ran models predicting the rate at which attorneys left firms to found progeny as a function of women partners and associates in the parent firm. The objective was to determine whether founders were more or less likely to have worked in firms with women partners and associates. The models revealed that firms with women partners or associates were no more likely to become parent firms (and thus sources of gender hierarchies for the next generation), although a variable for percentage of women associates was close to significance (p=.103).
31
the routines transferred. Put another way, the routines that are transferred along with key
resources are more likely to be the types of rules and procedures transferred by higher-
ranked attorneys. Founders that are lower-ranked, may be less likely to transfer
resources, but more likely to transfer taken-for-granted sets of routines (e.g., cultural
models).
Supplemental Analyses on Gender Hierarchies as Organizational Routines
I argue that gender hierarchies are transferred along with the same routines that reproduce
the progeny’s structure and organization of work. The imagery is one of founders taking
a bundle of routines with them upon the founding of the progeny. If true, the transfer of
gender hierarchies should be stronger for those progeny that also reproduce other aspects
of the parent’s organization of work. Drawing from Phillips’ (2002) observation that
progeny often had similar leverage ratios (the span of control is calculated as the ratio of
associates to partners and captures the organization of work) as their parent firms, I coded
the absolute value of the difference between the parent’s leverage ratio and the progeny’s.
The greater the difference in leverage ratios, the fewer routines transferred from the
parent to the progeny. To the extent that gender hierarchy is transferred as an
organizational routine, those progeny that are different from their parent firms (with
respect to the leverage ratio) should be less likely to replicate of the parent’s gender
hierarchy. Interacting an “intergenerational difference of leverage ratios” variable with
Institutionalized Female Leadership should result in a negative and significant variable.
Interacting this variable with Institutionalized Female Subordination should produce a
positive and significant variable.
32
Table 4 presents the estimation of the models that examine the relationship between the
proposed institutionalized effects and structural similarity. Model 1 shows that there is
no main effect for the difference in parent and progeny leverage ratios. The coefficient is
negative but the z-statistic is only -0.25. Model 2 includes the covariate for the
interaction between the difference in leverage ratios and Institutionalized Female
Leadership. In support of my thesis, the coefficient is negative and statistically
significant (p<0.05). The finding suggests that routines associated with Institutionalized
Female Leadership are less likely to be transferred when other routines are not
transferred. Put another way, routines that replicate the parent firm’s organization of
work either include or are transferred along side routines that replicate female leadership.
Model 3 represents the interaction between the difference in leverage ratios and the
Institutionalized Female Subordination variable. The coefficient is again in the expected
direction (positive). However, the effect is not statistically significant. Thus, it appears
that the institutionalization of women’s subordination is independent of the replication of
other routines that reflect the organization of work. Overall, Table 4 lends credence to
the argument that gender hierarchies are embedded in the routines and rules that founders
take into their new firms.
----------------------------------------
Put Table 4 About Here
----------------------------------------
33
Other Supplemental Analyses While restricting the sample to progeny that were founded by men was important to
provide the cleanest test of my hypotheses, another critical question involves
understanding the dynamics of women founded firms. While beyond the scope of this
paper, I did rerun the analysis for all progeny since 1970, including a dummy variable for
whether the firm had any women founders. I found that women founded firms were more
likely to partner an addition woman (p<0.001). In addition, I found that Hypotheses 1
and 2 were supported in this model. That is, the institutionalized woman leadership and
subordination effects occur independent of the gender of the founders. The findings
involving the former rank of the founder (H3a and H3b) are contingent upon the founders
being male. Future research should further explore differences by the founder’s gender.
I also reran each of the analyses with different weights for the institutionalization and
social contact variables. Whereas the institutionalization variables were coded as a one if
a woman was an associate or partner for the entire life of the firm, I multiplied the
original variables by the proportion of women at that particular level. For example, if
women partners were an average of 10% of the parent firm’s partners over its life cycle,
the weighted variable would equal 0.10. The weighted variables did not alter the
significance of any of the results however. The only exception is the social contact
control variable for women associates. In these models, the weighted operationalization
of the variable was stronger (negative and statistically significant) than the results
reported in table 3.
34
Conclusion
In sum, my findings add weight to the role of organizations in stratification and gender
inequality by providing evidence that opportunity structures are reproduced across
organizational generations. Within the literature on law firms, there is growing attention
to the persistence of gender inequality (Mossman 1988; Abel 1989). I argue that we
should more often direct our attention to the intergenerational transfer of institutionalized
structures, routines, and values that reproduce gender inequalities. Understanding the
diversity of opportunity structures may not be possible without understanding the
opportunity structures of the previous organizational generation and the characteristics of
organizations that spawn organizational offspring.
While promising, the empirical support of my arguments must come with an
understanding of the context. The organizational structure of law firms and the market
for legal services places a unique set of demands on all attorneys, independent of gender.
Associates are known to labor intensively for several years with little chance of making
partner. Moreover, law firms are human capital-intensive organizations. They are
similar to other professional service organizations where the transfer of routines from a
parent to its offspring may be the easiest to measure. Future research should seek to
verify whether my findings apply to other organizational forms.
My data typically involve small and young organizations. The average sizes of the parent
and progeny firms were 9.8 and 4.6, respectively. The average age of a parent firm was
11.96. For progeny the average age was 6.57. While this may prompt concerns about
35
external validity, there is a substantial benefit. The use of small firms increases the
relevance of group-level research on the emergence and persistence of gender hierarchies
(Berger, Rosenholtz, and Zelditch 1980; Ridgeway 1997; Lucas 2003; Troyer 2004) since
they are closest to the laboratory groups in which “micro-institutionalization” has been
experimentally identified. Moreover, the size and age of my firms allows comparison to
other studies on entrepreneurship, especially those that also examine the attainment of
women (Baron, Hannan, Hsu, and Kocak, 2002, Stuart and Ding 2004)
In addition to exploring other types and sizes of organizations, future work should more
intently document the transfer of routines from the parent to the progeny. Specifically,
what precisely is transferred from the parent to the progeny that causes the persistence of
gender hierarchies? While the findings in table 4 are shed some light on this question,
these data do not allow me to definitively distinguish the transfer of routines as policies
(e.g., recruitment and promotion policies) or as culture (Schein 1992). Future
development should seek to understand whether the transfer need be purposive, and
whether its implications for gender hierarchies intentional.
36
REFERENCES
Abel, Richard L. 1989. American Lawyers. Oxford University Press.
Allport, Gordon W. 1954. The Nature of Prejudice. Cambridge: Addison-Wesley. Babcock, Barbara Allen, Ann E. Freedman, Susan Deller Ross, Wendy Webster
Williams, Rhonda Copelon, Deborah Rhode, Nadine H. Taub, [1976] 1996. Sex Discrimination and the Law: History, Practice and Theory, Little, Brown & Co. (1976, First Ed. 1996, Second Ed.).
Baron, James N. 1984. Organizational Perspectives on Stratification. Annual Review of
Sociology, 10, 37-69. Baron, James N. and William.T. Bielby. 1980. “Bringing the Firms Back in:
Stratification, Segmentation, and the Organization of Work.” American Sociological Review 45:737-765.
Baron, James N., Michael T. Hannan, and M. Diane Burton. 1999. “Building the Iron
Cage: Determinants of Managerial Intensity in the Early Years of Organizations.” American Sociological Review 64:527-547.
Baron, James N., Michael T. Hannan, Greta Hsu, and Ozgecan Kocak, (2002). "Gender
and the Organization-Building Process in Young, High-Tech Firms." In Mauro F. Guillén, Randall Collins, Paula England, and Marshall Meyer (eds.), Economic Sociology at the Millenium. New York: Russell Sage Foundation Press.
Berger, Joseph, Susan J. Rosenholtz, and Morris Zelditch, Jr. (1980). “Status Organizing Processes”, Annual Review of Sociology, 6, 479-508.
Berger, P. and Luckman, T. 1967. The social construction of reality: a treatise in the sociology of knowledge, Penguin Publishers, London.
Blau, Peter M. 1960. “A Theory of Social Integration.” American Journal of Sociology, 65, 6, 545-556.
Blossfeld, H. P., and G. Rohwer 1995. Techniques of Event History Modeling. Mahwah,
NJ: Lawrence Erlbaum. Blum, T. C. 1985. “Structural Contraints on Interpersonal Relations: A Test f Blau’s
Macrosociological Theory,” American Journal of Sociology, 91, 3, 511-521. Boeker, Warren. 1989. "Strategic Change: The Effects of Founding and History",
Academy of Management Journal, 32, 3, 489-515.
37
Burnham, K. P., and D. R. Anderson. 1998. Model selection and inference: a practical information-theoretic approach. Springer-Verlag, New York, New York.
Cannings, Kathy. 1988. "The Earnings of Female and Male Middle Managers: A
Canadian Case Study." Journal of Human Resources 23, 1, 34-56.
Carroll, Glenn and J. Delacroix 1982. “Organizational mortality in the newspaper industries of Argentina and Ireland: an ecological approach”. Administrative Science Quarterly, 27, 167 – 198.
Carroll, Glenn R. and Michael T. Hannan. 1989. “Density Delay in the Evolution of
Organizational Populations: A Model and Five Empirical Tests.” Administrative Science Quarterly 34, 411–30.
Chambliss, Elizabeth. 1997. “Organizational Determinants of Law Firm Integration.”
American University Law Review, 46: 669-744. Cohen, Lisa E., Joseph P. Broschak, and Heather A. Haveman. 1998. “And Then There
Were More? The Effect of Organizational Sex Composition on the Hiring and Promotion of Managers?” American Sociological Review 63:711-727.
Cook, S. W. 1985. “Experimenting on social issues: The case of school desegregation.”
American Psychologist, 40, 452-460. DiMaggio, P. M., and W. W. Powell. 1983. “The Iron Cage Revisited: Institutional
Isomorphism and Collective Rationality in Organizational Fields.” American Sociological Review, 48: 147-160.
Dittes, James E., and Harold H. Kelley. 1956. “Effects of Different Conditions of
Acceptance upon Conformity to Group Norms.” Journal of Abnormal and Social Psychology 53:100–107.
Dixon, Jo and Carroll Seron. 1995. “Stratification in the Legal Profession: Sex, Sector,
and Salary.” Law and Society Review 29:381-412. Donovan, Kathleen. 1990. “Women Associates: Advancement to Partner Status in Private
Firms.” Georgetown Journal of Legal Ethics 4:135-152. Ely, Robin J. 1994. "The Effects of Organizational Demographics and Social Identity on
Relationships among Professional Women." Administrative Science Quarterly, 39, 203-238.
Epstein, Cynthia F. [1981] 1993. Women in Law. 2d ed. Urbana, IL: University of Illinois
Press.
38
Escher, W. J., and S. L. Morze. 1998. “Building a Silicon Valley office.” A Major, Hagen and Africa White Paper. Major, Hagen and Africa, San Jose, California.
Fernandez and Sosa 2003. Gendering the Job: Networks and Recruitment at a Call
Center. Paper presented at the 2003 annual meetings of the American Sociological Association.
Galanter, M., and T. Palay. 1991. Tournament of Lawyers: The Transformation of the Big
Law Firm. Chicago: University of Chicago Press. Gordon, J., & Whelan, K. 1998. Successful professional women in midlife: How
organizations can more effectively understand and respond to the challenges. Academy of Management Executive, 12, 1, 8-27.
Gorman, Elizabeth. 2004. Employers’ Cognitive Biases and Gender Differences in
Hiring: Evidence from Law Firms. Manuscript. University of Virginia. Hagan, John, and Fiona Kay. 1995. Gender in Practice: A Study of Lawyers’ Lives.
Oxford, England/New York: Oxford University Press. Hersch, Joni and Viscusi, W. Kip. 1986.“Gender Differences in Promotions and Wages.”
Industrial Relations, 35, 4, 461-472. Homans, George C. 1961. Social Behavior: Its Elementary Forms. New York: Harcourt,
Brace, and World. Halliday, Terence C. 1986. “Six Score Years and Ten: Demographic Transitions in the
American Legal Profession,” Law and Society Review 20:53-78. Hannan, M.T. 1998. “Rethinking age dependence in organizational mortality: Logical
formalizations.” American Journal of Sociology, 104: 85–123. Hull, Kathleen E. and Robert L. Nelson 2000. “Assimilation, Choice or Constraint?
Testing Theories of Gender Differences in the Careers of Lawyers." Social Forces 79: 229-264.
Jackman, Mary R. and Crane, Marie. 1986. "'Some of My Best Friends Are Black...':
Interracial Friendship and White Racial Attitudes," Public Opinion Quarterly, 50, Winter, 459-486.
Jones, David R. and Gerald H. Makepeace. 1996. “Equal Worth, Equal Opportunities:
Pay and Promotion in an Internal Labour Market.” The Economic Journal 106:410-409.
39
Kay, Fiona M. and John Hagan. 1995. “The Persistent Glass Ceiling: Gendered Inequalities in the Earnings of Lawyers.” The British Journal of Sociology 46:279-310.
Kay, Fiona M. and John Hagan. 1998. “Raising the Bar: The Gender Stratification of
Law-Firm Capitalization.” American Sociological Review 63:728-43. Kilbourne, Barbara Stanek, Paula England, George Farkas, Kurt Beron, and Dorothea
Weir. 1994. “Returns to skill, compensating differentials, and gender bias: effects of occupational characteristics on the wages of white women and men.” American Journal of Sociology 100:689-719.
Lazear, Edward P. and Sherwin Rosen. 1990. “Male-Female Wage Differentials in Job
Ladders.” Journal of Labor Economics 8:S106-S123. Lucas, Jeffrey W. 2003. “Status Processes and the Institutionalization of Women as
Leaders.” American Sociological Review, 68: 464-480. Lundberg, Shelly J. and Richard Startz. 1983. “Private Discrimination and Social
Intervention in Competitive Labor Markets,” American Economic Review, 73, 340-347.
Madigan, R. M., & Hoover, D. J. 1986. “Effects of alternative job evaluation methods on
decisions involving pay equity.” Academy of Management Journal, 29 (1), 84-100.
Markovsky, Barry, Le Roy F. Smith, Joseph Berger (1984). “Do Status Interventions
Persist?” American Sociological Review, 49, 3, 373-382. Meyer, J. W., and B. Rowan. 1977. “Institutionalized organization: Formal structure as
myth and ceremony.” American Journal of Sociology, 83: 340–363. Mossman, Mary Jane. 1988. “"Invisible" constraints on lawyering and leadership: the
case of women lawyers”. Ottawa Law Review, 20, 3, 567-600 Nelson, Robert L. 1988. Partners with Power: The Social Transformation of the Large
Law Firm. Berkeley: University of California Press. Olsen, Craig, and Becker, Brian E. 1983. "Sex Discrimination in the Promotion Process."
Industrial and Labor Relations Review, 36, 624-41 Palmer, Donald A., P. Devereaux Jennings, and Xueguang Zhou. 1993. “Late Adoption
of the Multidivisional Form by Large U.S. Corporations: Institutional, Political, and Economic Accounts.” Administrative Science Quarterly 38:100-131.
40
Pettigrew, T. F. 1998. “Intergroup contact theory.” Annual Review of Psychology, 49, pp. 65-85. Palo Alto, CA: Annual Reviews, Inc.
Phillips, Damon J. 2001. “The Promotion Paradox: Organizational Mortality and
Employee Promotion Chances in Silicon Valley Law Firms, 1946-1996.” American Journal of Sociology, 106, 4, 1058-1098.
Phillips, D. J. 2002. "A Genealogical Approach to Organizational Life Chances : The
Parent-Progeny Transfer and Silicon Valley Law Firms, 1946-1996." Administrative Science Quarterly., 47: 474-506.
Phillips, D. J. and C. Beckman. 2004. Interorganizational Determinants of Promotion:
Client Leadership and the Promotion of Women Attorneys, Manuscript, University of Chicago.
Raftery, A. E. 1995. “Bayesian model selection in social research.” Sociological
Methodology, 25: 111–163. Rhode, Deborah 1988. “PROJECT: Gender, Legal Education, and the Legal Profession:
An Empirical Study of Stanford Law Students and Graduates.” 40 Stanford Law Review. 1209
Ridgeway, Celicia L. (1997). “Interaction and the Conservation of Gendder Inequality:
Considering Employment”, American Sociological Review, 62, 218-235. Ridgeway, Celicia L. and Kristan G. Erickson (2000).”Creating and Spreading Status
Beliefs”, American Journal of Sociology, 106, 3, 579-615. Schein, Edgar H. 1992. Organizational Culture and Leadership, Jossey-Bass. Schwarz, G. 1978. "Estimating the Dimension of a Model", Annals of Statistics, 6, 461-
464. Seligman, Dan 2001. “Equal pay revisited.” Forbes, Jan 22, 80. Smigel, Erwin O. 1969. The Wall Street Lawyer: Professional Organization Man?
Bloomington: Indiana University Press. Sørensen, Jesper B. 1999. Estimation of piecewise-constant hazard rate models. User
written routine for Stata 6.0. www.fmwww.bc.edu/RePEc/bocode/s/stpiece.ado. Spurr, Stephen J. 1990. “Sex Discrimination in the Legal Profession: A Study of
Promotion.” Industrial and Labor Relations Review 43:406-417.
41
Spurr, Stephen J. and Glenn T. Sueyosji. 1994. "Turnover and promotion of lawyers: an inquiry into gender differences." The Journal of Human Resources 29, Sum '94, pp. 813-42.
Stinchcombe, A. 1965. “Organizations and social structure.” In J. G. March (ed.),
Handbook of Organizations, 141–193. Chicago: Rand McNally. Stuart, Toby and Waverly Ding. 2004. “Technology: A New Arena for Gender
Stratification in Scientific Careers?” Four Essays on the Formation and Evolution of U.S. Biotechnology Companies, Ph.D. Dissertation (Ding), University of Chicago.
Suchman, M. C. 1993. On the role of law firms in the structuration of Silicon Valley,
Working Paper, Series 11, presented at the 1993 Annual Meeting of the Law and Society Association, Chicago.
Suchman, M. C. 2000. “Dealmakers and counselors: Law firms as intermediaries in the
development of Silicon Valley.” In M. Kenney (ed.), Understanding Silicon Valley: The Anatomy of an Entrepreneurial Region, 71–97. Stanford, CA: Stanford University Press.
Swaminathan, Anand. 1996. "Environmental conditions at founding and organizational
mortality: A trial-by-fire model." Academy of Management Journal 39(5): 1350-77.
Troyer, Lisa (2004). “The Role of Social Identity Processes in Status Construction”,
Advances in Group Processes: Power and Status, 20, 149-172. Tucker, D J., J. V. Singh, and A. G. Meinhard. 1990 “Organizational form, population
dynamics, and institutional change: The founding patterns of voluntary organizations.” Academy of Management Journal, 33, 151-178.
Volinsky, C. T. and A. E. Raftery 1999. Bayesian information criterion for censored
survival models. Technical Report 349, Department of Statistics, University of Washington. (http://www.stat.washington.edu/tech.reports/tr349.ps). http://citeseer.ist.psu.edu/volinsky98bayesian.html
Ziewacz, Elizabeth K. 2000. “Can The Glass Ceiling Be Shattered?: The Decline of
Women Partners inLarge Law Firms”, Ohio State Law Journal, 57, 971-997.
Zucker, L. G. 1977. “The role of institutionalization in cultural persistence. American Sociological Review, 42: 726-743.
42
43
Table 1. Summary Statistics for Silicon Law Firm Progeny Firms, 1970-1996. N = 623 Firm-Years (99 Firms).
Variable Mean S.D. Min Max DV: Firm has a woman partner 0.03 0.16 0 1 Firm age 6.57 5.66 1 26 Log(Time trend) 2.59 0.64 0 3.26 Prop yrs parent had women partners 0.08 0.26 0 1 Prop yrs parent had women assoc. 0.10 0.23 0 1 Yrs worked w/ women partners 0.35 1.44 0 13 Yrs worked w/ women associates 0.82 2.19 0 13 Log(partners) 1.03 0.55 0 3.04 Log(associates) 0.59 0.66 0 2.77 Firm is a branch office 0.16 0.37 0 1 Relative scope 0.11 0.06 0.01 0.34 New Market Expansion 0.14 0.35 0 1 Firm located in North SV 0.44 0.50 0 1 Associate growth rate 0.05 0.20 0 1 Associate attrition rate 0.10 0.25 0 1 Partner growth rate 0.06 0.15 0 1 Partner attrition rate 0.04 0.16 0 1 Securities law 0.22 0.42 0 1 Real estate law 0.40 0.49 0 1 Labor/Employment law 0.29 0.45 0 1 Intellectual property law 0.11 0.31 0 1 Prop of women associates 0.10 0.24 0 1 Parent firm age when founder left 11.96 11.41 1 60 Parent firm size when founder left 8.04 11.43 1 96 Prop women partners when founder joined parent 0.01 0.07 0 0.41 Prop women assoc when founder joined parent 0.03 0.15 0 1 Founder was a former associate 0.35 0.48 0 1
44
45
Table 2. Pairwise Corrleations for Silicon Valley Law Firm Progeny Firms, 1970-1996 N=623 Firm Years (99 Firms) Firm makes a
wmn partner Firm age Log (partners) Log (assoc.) Rel. Scope New Market
Expansion Located in North SV
Branch Office Log (Time trend)
Partner attr. rate Assoc. attr. rate Prop wmn assoc.
Firm makes a wmn partner
1
Firm age 0.0608 1
Log (partners) 0.1077* 0.2499* 1
Log (assoc.) 0.0658 0.2168* 0.0255 1
Rel. Scope 0.0445 -0.0409 0.2854* 0.0694 1
New Market Expansion -0.0061 0.2280* 0.1872* 0.0192 0.3765* 1
Located in North SV -0.1030* -0.0567 0.0816* -0.0157 0.2313* 0.0673 1
Branch Office 0.1515* 0.0441 0.2657* 0.4387* 0.1529* -0.0976* -0.1021* 1
Log (Time trend) 0.0952* 0.0579 -0.1234* 0.0046 -0.4175* -0.1900* -0.0051 -0.0542 1
Partner attr. rate 0.0143 0.0273 0.2418* -0.0237 0.0794* -0.0512 -0.0653 0.1616* 0.0185 1
Assoc. attr. rate -0.0145 0.0324 -0.1188* 0.3425* -0.0096 -0.0272 -0.0239 0.1306* 0.0319 0.0825* 1
Prop wmn assoc. 0.0478 0.1440* 0.1402* 0.1002* 0.024 0.0933* 0.0799* 0.0386 0.1066* 0.0966* 0.0326 1
Partner growth rate 0.1385* -0.053 0.3186* 0.0194 0.1140* -0.065 0.0108 0.0988* 0.0342 0.1987* 0.0071 0.1456*
Assoc. growth rate 0.029 -0.0134 0.1315* 0.2021* 0.0912* -0.0065 0.0478 0.075 0.0093 0.0004 0.1185* 0.0288
Real estate law 0.0534 -0.1740* 0.1354* -0.1111* 0.3645* -0.0143 0.1982* 0.0293 -0.0929* -0.0488 -0.0037 0.0372
IP law 0.072 -0.0914* 0.0831* 0.0762 0.1970* 0.1257* 0.1613* 0.0145 0.2270* -0.0069 -0.0356 -0.0402
Securities law 0.0111 -0.0828* 0.2406* 0.067 0.4958* 0.1339* 0.2983* 0.0536 0.0204 0.0314 0.0345 0.1310*
Labor/Empl. Law 0.0321 -0.0865* 0.1187* 0.0461 0.2580* 0.1589* -0.045 0.0847* 0.1269* 0.0889* 0.0316 0.0127
Prop yrs parent had women partn.
0.1119* -0.0993* 0.0471 0.2296* 0.1960* -0.1215* 0.1413* 0.3418* 0.1650* 0.1163* 0.0481 0.0185
Prop yrs parent had women asso.
0.0139 -0.1503* -0.047 0.0910* -0.0701 -0.068 -0.1184* 0.1313* 0.2830* 0.0135 -0.0356 -0.0403
Yrs worked w/ women partners
0.045 -0.1277* -0.1022* -0.0459 -0.1059* -0.0713 -0.1500* 0.1016* 0.1794* -0.0379 0.0063 -0.0183
Yrs worked w/ wmn associates
-0.0419 -0.1291* -0.1776* 0.0114 -0.1341* -0.03 -0.1171* 0.0386 0.2487* -0.0675 -0.0421 0.005
Parent firm age -0.0002 0.1040* 0.025 -0.1890* -0.0286 -0.0278 -0.1060* -0.1696* 0.0858* 0.0722 -0.0739 -0.0302
Parent firm size 0.0811* -0.1041* -0.1074* 0.0031 -0.1422* -0.0891* -0.0345 -0.1027* 0.1819* -0.0916* 0.01 -0.0184
Prop wmn partne when founder joined parent
0.0309 0.0460 0.2658* -0.0372 0.1088* -0.0796* -0.0775 0.0381 0.1039* 0.2592* -0.0331 -0.0523
Prop wmn assoc when founder joined parent
0.0413 -0.0701 -0.0836* -0.0294 -0.1025* -0.0596 -0.1381* 0.0117 0.0991* 0.1542* -0.0255 -0.0211
Founder: former assoc. 0.0292 0.1758* 0.3448* 0.0542 0.1524* 0.2999* 0.0114 0.0202 -0.0287 0.0701 -0.012 0.0145
46
Table 2 (cont') Partner
growth rate
Parent Assoc. growth rate
Real estate law
IP law Securities law
Labor/Empl. Law
Prop yrs parent had women partn.
Prop yrs parent had women assoc.
Yrs worked w/ women partners
Yrs worked w/ women associates
Parent firm age
firmsize
Partner growth rate 1
Assoc. growth rate 0.6317* 1
Real estate law 0.021 -0.0221 1
IP law 0.0217 0.0376 -0.0072 1
Securities law 0.1182* 0.1390* 0.3046* 0.3167* 1
Labor/Empl. Law 0.1062* 0.0494 -0.1916* 0.1504* 0.2188* 1
Prop yrs parent had women partn.
0.1103* 0.1276* 0.063 0.0875* 0.2015* 0.2204* 1
Prop yrs parent had women asso.
-0.0026 0.0245 0.002 0.1489* 0.0263 0.2026* 0.3569* 1
Yrs worked w/ women partners
-0.0595 -0.034 -0.0692 0.0137 -0.0758 0.0338 0.0487 0.3853* 1
Yrs worked w/ wmn associates
-0.0552 -0.0096 -0.0715 0.074 -0.0855* 0.1829* 0.1271* 0.7329* 0.3239* 1
Parent firm age 0.0426 -0.0582 -0.0766 -0.0296 -0.1857* 0.2552* -0.0717 0.051 0.1278* 0.1873* 1
Parent firm size -0.0853* -0.031 0.015 0.1003* -0.0643 0.1171* -0.0322 0.2124* 0.1201* 0.2536* 0.4182* 1
Prop wmn partne when founder joined parent
0.0553 0.0457 -0.0771 -0.0256 0.0120 0.3144* 0.1303* 0.0859* -0.0483 0.0155 0.4218* -0.0878* 1
Prop wmn assoc when founder joined parent
-0.0853* -0.0310 0.0028 0.0063 0.0093 -0.0340 0.0755 0.1259* -0.0088 0.0147 -0.1300* -0.0245 -0.0360 1
Founder: former assoc. 0.0084 -0.0026 -0.0403 0.1258* 0.1578* 0.2413* -0.1335* 0.022 0.0172 -0.0515 0.4236* 0.1674* -0.0311 0.3092* 1
Table 3. MLE of a Firm Having a Woman Partner (Testing H1-H3). Silicon Valley Law Firms (1970-1996). N = 623 firm-years; 99 firms. * = p < 0.05 Variable (1) (2) (3) (4) Hypo Firm age: 0 to 3 years
-9.78 (2.62)
-1.46 (12.46)
-21.58 (14.22)
-0.32 (13.67)
Firm age: 3 to 10 years
-9.68 (2.62)
0.25 (12.40)
-18.26 (13.75)
3.81 (13.24)
Firm age: 10+ years
-8.84 (2.61)
5.54 (12.97)
-11.72 (13.01)
10.03 (13.76)
Time trend (logged) Prop. yrs parent had women partners
2.01* (0.90) 1.56*
6.84* (3.16) 20.76*
9.68* (4.13) 18.82*
7.69+ (4.38) 26.88*
(0.67) (7.58) (6.79) (9.73) H1 Prop. yrs parent had women associates
-1.20 (1.10)
-13.57* (5.89)
-20.54+ (11.36)
-15.61 (13.99) H2
Yrs worked w/ women partners
2.07* (0.76)
2.03* (0.86)
3.29* (1.52)
Yrs worked w/ women associates
-0.42 (0.48)
-0.59 (0.57)
-1.37 (2.81)
Log(partners)
0.41 (1.23)
0.41 (1.29)
0.01 (1.77)
Log(associates)
-0.19 (0.87)
-2.11 (3.78)
-0.75 (1.28)
Firm is a branch office
1.45 (2.03)
2.36 (2.40)
1.48 (2.98)
Relative scope
-23.11 (22.12)
-40.51+ (23.36)
-42.15 (33.40)
New Market Expansion
9.28+ (5.03)
9.56* (4.13)
10.84+ (6.13)
Firm located in North Silicon Valley
-13.03* (4.07)
-15.22* (5.49)
-15.80* (6.17)
Partner growth rate
13.52* (5.72)
15.98* (6.73)
16.65* (7.82)
Partner attrition rate
-15.34* (7.65)
-17.82* (8.90)
-18.15+ (9.33)
Associate growth rate
-5.02 (3.05)
-5.65+ (3.15)
-6.35 (4.43)
Associate attrition rate
-2.53 (2.69)
-2.11 (3.78)
-3.83 (4.26)
Prop of women associates
-1.72 (2.36)
-0.97 (2.71)
-1.43 (2.74)
Securities law
-0.94 (2.25)
-1.82 (2.73)
-1.67 (2.81)
Real estate law
6.33* (2.08)
7.68* (2.60)
8.58* (3.05)
Labor law
-6.08+ (3.30)
-4.34* (2.04)
-7.08+ (3.97)
Intellectual property law
6.12* (2.32)
8.93* (3.64)
7.55+ (3.97)
Parent firm age when founder left
-0.18 (0.17)
0.04 (0.15)
-0.30 (0.22)
Parent firm size when founder left
0.11 (0.07)
0.03 (0.06)
0.14+ (0.09)
Prop wmn partners when founder joined parent
28.34 (23.67)
12.10 (15.62)
39.61 (30.72)
Prop wmn assoc when founder joined parent -7.10 (5.88)
-9.98 (8.55)
10.64 (8.11)
Founder was a former associate (Former assoc.)*( Prop yrs parent had women partners)
0.30 (1.90)
-1.88 (2.28) 10.92*
2.07 (2.32)
(4.83) H3a
(Former assoc.)*( Prop yrs parent had women assoc)
-22.26+ (12.83) H3b
N 623 622 622 622 BIC model fit -103.02 -177.22 -184.22 -183.50 BIC improvement versus previous model 4.68 74.21 7.00 0.73 Wald Chi-square 174.04*
(6) 32.36 (28)
26.08* (30)
24.48 (30)
47
Table 4. MLE of a Firm Having a Woman Partner (Testing Structural Similarity). Silicon Valley Law Firms (1970-1996). N = 623 firm-years; 99 firms. * = p < 0.05 Variable (1) (2) (3) Firm age: 0 to 3 years
-0.62 (13.09)
-19.19 (15.52)
-0.23 (13.41)
Firm age: 3 to 10 years
1.20 (13.13)
-15.10 (15.15)
1.85 (13.42)
Firm age: 10+ years
6.51 (13.75)
-8.51 (14.49)
6.74 (14.02)
Time trend (logged) Prop. yrs parent had women partners
6.68* (3.15) 21.06*
9.75* (4.27) 22.72*
6.78* (3.36) 21.10*
(7.77) (7.56) (7.77) Prop. yrs parent had women associates
-13.51* (5.91)
-22.43+ (12.09)
-14.15* (6.66)
Yrs worked w/ women partners
2.12* (0.81)
2.30* (0.93)
2.34* (0.94)
Yrs worked w/ women associates
-0.47 (0.54)
-0.51 (0.60)
-0.61 (0.74)
Log(partners)
0.27 (1.36)
0.28 (1.62)
0.32 (1.45)
Log(associates)
-0.04 (1.04)
0.38 (1.13)
-0.20 (1.16)
Firm is a branch office
1.38 (2.06)
1.57 (2.14)
1.23 (2.27)
Relative scope
-24.64 (23.51)
-29.67 (26.28)
-26.23 (23.95)
New Market Expansion
9.57+ (5.23)
9.32* (4.69)
9.55+ (5.28)
Firm located in North Silicon Valley
-13.09* (4.11)
-18.47* (6.32)
-12.87* (4.15)
Partner growth rate
13.79* (5.91)
17.92* (7.38)
12.80* (6.16)
Partner attrition rate
-15.58* (7.82)
-19.65* (9.28)
-14.54+ (8.08)
Associate growth rate
-5.33 (3.32)
-7.47+ (4.04)
-4.73 (3.47)
Associate attrition rate
-2.55 (2.72)
-4.38 (3.43)
-2.98 (3.11)
Prop of women associates
-1.72 (2.37)
-0.82 (2.61)
-1.37 (2.57)
Securities law
-0.86 (2.30)
2.08 (2.86)
-1.06 (2.47)
Real estate law
6.45* (2.19)
7.43* (2.60)
6.42* (2.17)
Labor law
-6.02+ (3.28)
-4.86* (2.46)
-6.08+ (3.33)
Intellectual property law
6.18* (2.37)
9.88* (3.79)
5.82* (2.55)
Parent firm age when founder left
-0.18 (0.17)
-0.05 (0.16)
-0.20 (0.18)
Parent firm size when founder left
0.11 (0.07)
0.07 (0.07)
0.12 (0.07)
Prop women partners when founder joined parent 29.07 (23.82)
16.74 (21.46)
30.64 (24.35)
Prop women associate when founder joined parent -6.85 (5.89)
-2.70 (8.47)
-7.83 (6.33)
Founder was a former associate
0.02 (2.25)
-1.12 (2.63)
0.29 (2.33)
Logged Absolute Difference in Ratio (Ratio diff)*( Prop yrs parent had women partners)
-0.17 (0.70)
-0.14 (1.09) -3.07*
-0.17 (0.74)
(1.51)
(Ratio diff)*( Prop yrs parent had women assoc) 2.98 (4.96)
N 622 622 622 BIC model fit -177.27 -182.26 -177.88 BIC improvement versus previous model 4.99 0.61 Wald Chi-square
32.36 (30)
26.24* (31)
31.40 (31)
48
49