Explaining Women’s Careers at a Dutch university: Model
building as a method for knowledge elicitation in gender
analysis
Inge Bleijenbergh', Eelke Blonk', Lodewijk Schulte’, Marloes van Engen”
'Nijmegen School of Management
Radboud University Nijmegen
PO Box 9108
6500 HK Nijmegen
the Netherlands
tel: ++31 (0)24-3611474
ibleijenbergh@fm.ru.nl
*Department of Human Resource Studies
Tilburg University
PO Box 90 153
5000 LE Tilburg
The Netherlands
m.l.vengen@uvt.nl
Paper for the 26" Conference of the System Dynamics Society
July 20-24, 2008
Athens Greece
Abstract
This paper presents model building as a means to study factors explaining the delay in Women’s
Careers at a Dutch university. The causal model is based on analysis of 43 interviews, ten focus
groups and policy documents collected at the five faculties of a Dutch university. The five
researchers involved in this analysis used model building to integrate their knowledge of the
research material and reach a shared definition of relevant variables. Model building proves to
support the qualitative analysis of gender processes by supporting the identification of main
processes and showing the feedback processes at work. The model shows that Women’s careers
are stimulated by the presence of female academics at higher positions, while masculine norms
and the lack of visibility of female academics slow down women’s careers. By focusing on the
role of image shaping about women’s ambitions and performance, it fills in theoretical gaps
identified in earlier research about gender processes in organizations.
Key-words: Model building, Gender equity, Qualitative Research, Women’s Careers, Human
Resource Management, System Dynamics
Introduction
Group Model building is an approved method for supporting managers with strategic decision-
making in organizations (Anderson et al, 2007; Richardson and Andersen, 1995; Vennix, 1996).
In this article we will evaluate whether model building can also be a useful tool for researchers in
structuring, synthesizing and analyzing complex and multifaceted qualitative data, especially
regarding gender processes in organizations. The problem we present in this paper is the under
representation of women in the higher positions at a Dutch university. Data were collected with
the aim to unravel factors influencing the movement of women in, out and trough this particular
university’s hierarchies. As is the case in most countries around the world, Dutch universities
suffer from an under-representation of women in higher positions (Benschop and Brouns 2003;
van den Brink et al 2006). This university in particular performed worse than other Dutch
universities in the advancement of women (Korsten et al, 2006). The research was initiated by
the executive board of the university as it viewed the under representation of women in higher
academic positions as a serious problem. The problem needed tailor-made targeting informed by
research for the university as a whole and its five constituent faculties (Faculty of Economics,
Faculty of Law, Faculty of Social and Behavioral Sciences, Faculty of Humanities and
Theological Faculty). Furthermore the executive board wanted to make haste tackling the
problem and consequently the timeframe for executing the research was relatively short (October
2007-February 2008), as is often the case with practice oriented research.
To be able to gain an in-depth understanding of gender processes within the different
faculties, we used a stepwise qualitative research set-up that was informed by the existing body
of knowledge on the position of women in academia. We held 43 semi-structured interviews,
followed by series of focus groups (ten focus groups in total). The study produced an enormous
amount of qualitative data: 727 pages of interview transcripts, five policy documents and notes
of ten focus group meetings. The five researchers, who all performed parts of the data collection
and processing of the research material,, needed to elucidate their partially implicit knowledge
and interpretation of the abundance of data and integrate it into a coherent analysis of the
problem. Therefore the model building process under study can be characterized as (1) An effort
of a group of researchers which all examined the same problem: the under representation of
women in higher academic positions at a Dutch university, (2) an abundance of data of which the
knowledge was dispersed among the group members. (3) The necessity of integrating this
knowledge into a coherent analysis and (4) a high degree of time pressure.
In this article we first review the existing body of knowledge concerning the position of
women in academia. Herewith we show the difficulties of knowledge elicitation regarding gender
processes. Secondly, we answer the question how we define model building and where we situate
our model building approach in the larger body of literature on System Dynamics. Thirdly we
discuss the content of the model we have build, describing coherent parts of the model separately
and paying special attention to feedback loops in the system. Finally, we reflect on the theoretical
contribution and methodological meaning of our analysis and the role model building played in
it.
Women in Academia; The existing body of knowledge
Women’s scarcity in the academic top has been examined in different academic disciplines such
as psychology, sociology, and recently entered the field of System Dynamics (Campbell and
Hovman 2004, Dudley 2007, Bleijenbergh, Benschop and Vennix 2006). In the following we
briefly review the most common explanations and the empirical evidence for those explanations.
We suggests that dispositional causes (e.g. working time, family responsibilities) may be
overestimated, whereas situational causes (e.g. influence of organizational culture, gender
stereotypes) may be underestimated in the explanation of career progression of women. Our
method of analysis needs to support us in getting insight in the role of culture and gender
stereotypes in this particular organization and to show how these factors relate to more
situational factors.
The most common explanation for the scarcity of women in higher positions is the pipeline
hypothesis. This hypothesis states that it takes a certain period of time, around fifteen to twenty
years, before a growing proportion of female graduate’s moves up the hierarchy into leadership
positions. Research about the Dutch situation shows the pipeline hypothesis doesn’t convincingly
explain the under representation of women in higher academic positions at Dutch universities.
Given the number of female students in the 1980s, a much higher proportion of female managers
and leaders would be expected (AWT 2000; Brouns et al. 2004). A System Dynamics model
about the American situation leads to a comparable conclusion: the pipeline delay hypothesis is
not sufficient to explain the relatively small number of women in American universities (Dudley
2007).
The fact that more women than men work part-time is another situational factor that is often
suggested to be an explanation for women’s career progression. In the Netherlands, in general,
the part-time factor of male and female academics hardly differs (.85 for women, .88 for men).
Analyses of the total amount of work hours of male and female academics within this particular
university also showed hardly any difference between men and women (1.3 hours) (Van Engen,
Bleijenbergh & Paauwe, 2008). A system dynamics model of the American situation suggests the
same. Dudley showed that women’s part-time employment only explains a small part of the delay
(Dudley 2007, 4).
Research is less consistent about the consequences of family responsibilities on career
progression in academia. Some studies show that motherhood negatively influences publication
productivity (e.g. Portegijs & Brugman, 1998), other research shows that women with children
are more successful in their academic careers than women without children (Brouns et al, 2004),
whereas yet other studies conclude that motherhood is unrelated to career development (AWT,
2000; Bordons et. al, 2003; Wessseling, 2001). The most likely scenario Dudley finds is that the
tag of being a potential ‘homemaker’ limits women’s chances of promotion to higher positions at
all levels. Independent of the real behavior of women, employers are reluctant to hire women
because they perceive them as people who may quit the workforce to become homemakers.
Especially when the fraction of homemakers in the population is high, the likelihood of women
moving up the hierarchy is smaller (Dudley 2007, 4).
Herewith the role of gender stereotypes enters the debate. In an American study King (2007)
showed that the way supervisors perceive the ambition and performance of female academics is a
better prediction of their careers than the real ambitions and performance of women. In her study
she shows that supervisors for instance underestimate the number of publications by women and
especially by mothers. We would like to further explore this mechanism and wonder if this
mechanism is also at work in the Dutch University we examine.
Finally, some studies have considered organizational culture as an explanation for the under
representation of women in academia. Some studies suggest women in academia have adjusted to
the organizational culture and are just as happy as men (Portegijs & Brugman, 1998). Yet other
studies have shown women are less satisfied with their employers, especially with the support of
their supervisor (Brouns et al, 2004). We would like to test these opposite claims and would like
to examine the role of organizational culture in this particular university in explaining women’s
career progression.
Summarized, our literature review shows that there are several explanations for the under
representation of women in academia. All the explanations may partially contribute to the delay
of women’s career progress in academia, but further knowledge needs to be developed on the
role of organizational culture and gender stereotyping in particular. We especially need to
examine how situational causes like stereotypes are related to dispositional causes like working
time and motherhood. We need a method of analysis that helps us to unravel gender stereotypes.
Moreover, it should reveal underlying mechanisms that contribute to the reproduction of these
stereotypes. In this paper we will evaluate what the contribution of model building is for
knowledge elicitation on these issues.
Model building in qualitative research
In this study, we use model building as a method to elucidate and integrate knowledge on the
complex issue of gender processes at a Dutch University. We first need to define model building
and situate it in the larger body of literature on System Dynamics. We define Model Building as
representing a real world system into a System Dynamics model. System Dynamics (SD) is a
theory of the structure and behavior of complex systems. The structure has four hierarchical
levels: (1) the closed boundary, (2) the feedback loop as the basic system component, (3) levels
and rates and (4) goals, observed conditions, discrepancy between goals and observed conditions
and desired action” (Forrester 1961, 1987, Vennix 1996, 44-45). The closed boundary does not
mean that the system is not influenced by the external environment, but rather separates the
dynamically significant inner workings of the system from the rather insignificant influences
from the external environment.(Richardson, 1991) SD is used by individual researchers to
combine available research material into a causal loop or stock and flow diagram of the
researched problem. It might also be used by a group of researchers for the same purpose, as we
do in this particular project.
Building SD models with groups of people has become a full sub discipline of System
Dynamics under the name of Group Model Building (Andersen et. Al, 2007, Richardson and
Andersen 1995, Vennix 1996). The main purpose of this method is however to support strategic
decision-making in organizations, rather than to support academic research. Group Model
Building refers to a series of meetings where a professional SD facilitator supports a group of
managers in building a causal diagram of an organizational problem. Vennix (1999) distinguishes
three reasons to involve groups of managers in Group Model Building. One of these is to capture
the required knowledge in the mental models of the client group. Building a model together helps
to explicate the knowledge of individual participants and enhance team learning. The second
reason is to foster consensus on the causes and consequences of an organizational problem. The
causal model is a product of common deliberation and so helps to integrate knowledge. The third
is to create commitment with a resulting decision. Since the model is the product of a group
process, the participants feel connected to the decisions that are enacted from it.
In this research we didn’t perform GMB in the strict sense. We rather used the technique
of model building with a group of researchers to support qualitative analysis in academic
research. The main difference between Group Model Building and model building as applied in
our analysis are the problem characteristics. Group Model Building intends to tackle messy
problems. Messy problems can be defined (characterized?) as problems in which people hold
entirely different views on (a) whether there is a problem, and if they agree there is, and (b) what
the problem is (Vennix 1996; 1999). In the research under study the problem at hand was given
as the research was initiated by the executive board with the aim of identifying factors
influencing the under representation of women in higher academic functions. All researchers
agreed with the problem definition. Therefore the problem cannot be defined as messy. Rather,
the problem should be defined as causally complex, “not because there are too many variables
affecting it, although the number of causal variables is certainly important, but because different
causally relevant conditions can combine in a variety of ways to produce a given outcome”
(Ragin 1989, 26). Moreover, there was no need to create commitment regarding decisions to be
taken from the analysis. The researchers did not have any decision-making power. With their
research report they rather had to deliver input to the executive board of the University, where
subsequent decisions would be taken.
Nevertheless, there are also important similarities between our approach and GMB. The
model building participants had different knowledge regarding the problem, as they had collected
and processed different parts of the huge amount of qualitative material. The model building
needed to support the elucidation and integration of this knowledge. Moreover, they needed to
reach consensus on how to analyze and interpret the material. Aiming for the most broadly
shared interpretation of the empirical material, so called intersubjectivity, is an important quality
criterion in qualitative research (Shank, 2006: 84). So, the model building should support the
validity of the analysis.
Concluding, the model building process discussed in eee eee
this paper differs from regular Group Model Building
applications. Though a group of researchers was involved in
model building, the problem had no messy characteristics. We
had to alter the causal complexity of the problem, but did not —_ avs!
need to incorporate any strategic complexity. It also differed
from individual model building by a single researcher, as
applied by Dudley. Although the model is based on analysis of
empirical material, the analysis had to be supported by the
exchange of knowledge between researchers. Therefore we
situate our model building process in the upper right quadrant of figure 1 and label it collective
structuring of complex causal problems using System Dynamics. Individual researchers such as
Dudley are in the upper left quadrant, and Group Model Building is in the lower right.
‘Strategic complexity
Figure 1: Dimensions of Model Building
Data gathering
Understanding underlying mechanisms that help or hinder careers in each particular
faculty of this university calls for a qualitative research approach. We were interested in the
perspective of different actors (decision-makers, men and women in higher positions, as well as
at the bottom of the university’s career pyramid) within the university on ‘what it takes to get
promoted’ within the universities hierarchies. To be able to collect data from all of these actors
in the limited timeframe granted by the university board, we used a stepwise approach. In the
first phase of the study we held in-depth interviews with 43 academics (see further). The results
of the interviews were discussed in focus groups.
Interviews.
In total we interviewed five deans (all of them male), five chiefs of Human Resources (two
males, three females), five female full professors, five female associate professors, ten female
assistant professors and five male assistant professors, and ten female academics who left the
university during the previous year. The interviews were semi-structured. For each respondent
group a different interview protocol was developed (depending on their level of decision-making
authority). The existing body of knowledge concerning the position of women in academia was
used to develop the interview protocol.
The interviews were transcribed verbatim. After reading a number of interviews the research
team developed a coding format. All interviews were coded according to this format by at least
one researcher and subsequently all coding formats and interviews were read and complemented
by a second researcher. Two researchers then created a draft analysis of a faculty after which the
analysis was discussed and finalized in the research group.
Focusgroups.
The results of the analyses were the input for focus groups with the interviewees to corroborate
our findings and deepen our understanding of the problem. This member check of the interview
analyses made it possible for us to deepen our analyses and correct problems with interpretation.
Next, focus groups were held with all the departmental chairs of each faculty; a total of 35
people, mainly men. First, we collected their personal opinions concerning the causes and
consequences of women’s under representation in higher academic positions on a brief, open-
ended questionnaire at the start of the meeting. We then briefly presented our analysis of the
interview data and asked them to reflect on it and come up with possible intervention strategies.
The insights of these focus groups further improved our understanding of the processes taking
place at the different faculties.
The model building process
After defining the method of model building, we need to specify how we used the method in our
research. After the five researchers had been able to process their part of the material, we came
together in three meetings that were led by a SD facilitator and supported by a SD modeler. From
the start of the model building process there was no doubt on what the central problem was,
namely the under-representation of women in higher academic positions at a Dutch university.
Soon it became clear that this problem was a stock and flow problem, similar to that defined by
Dudley last year (2007). This stock and flow is at the core of the model, and describes the flow
of women in, out and trough the hierarchy of the university. Its stocks representing the
percentage of women in lower and higher academic positions, its flows representing the changes
that occur in them as a result of inflow, outflow and promotions. Discussion within the model
building sessions focussed on the reasons for stagnation in the flow of women through the
university hierarchy. The end result is an explanatory model for this complex causal problem.
The model is based on the perceptions of the academics we have interviewed individually and in
focus groups. We only included explanatory variables in the model that were mentioned by at
least three respondents per faculty, and were found in at least two different faculties. Moreover,
we only included relations that were not falsified by statements of other respondents.
There were a number of cognitive and communicative challenges encountered in the
model building. First of all, the participants had to elucidate, discuss and verify the variables to
be incorporated in the model. Certain issues were dropped from the model because they could
not be verified or were falsified by other research material. An associated problem was keeping
the model free from explanations we knew from literature study and to incorporate solely those
explanations that were actually identified in the empirical material. As discussions became more
vivid, there was a real danger that participants’ theoretical knowledge on gender issues blended
into argumentations. Strict rules on what to incorporate and what not helped to thwart this
problem.
The second challenge was the correct conceptualization of variables and_ their
interrelations. Identifying the relevant issues was relatively easy; incorporating these in the
model in the right terms was much harder. Discussions on conceptualizations and model
structure were quite extensive and sometimes required compromise between participants. The
final model was agreed upon by all participants as an accurate representation of the patterns that
came up from the empirical material.
A total of four meetings of approximately three hours were devoted to the model
building effort. The first meeting was used for the facilitators to get acquainted with the model
building participants, to discuss the research project and establishing a problem variable.
Subsequent meetings were devoted entirely to building the model and refining it.
Model contents
The content of the model will be described in the next paragraphs (Appendix | shows the entire
model). For reasons of intelligibility parts of the model — shown in figures 2 to 7 - will be
described separately. Feedback loops have been identified and will be discussed in order to
create insight into the fundamental behaviour of the system.
Stock and flow of women’s careers
i. Peis
* Seext of ls low of wor
At the core of the model is a stock and eee ee 1 pif renpteorathaletonae
flow diagram. A stock is a unit which +eet
in- or decreases during a specific time ” =
interval, a flow indicates the quantity | Skdlow sfwoman | wget.
with which a stock changes. The values gastos inlowe P97 gooseendersarsy “T v
of the stock and flow diagram were not "=" cee F ‘
quantified. Rather, it served the visual _ vomen trom lower !
iti wdemic postions
and explanatory qualities of the model ae oe
for both the model builders and the —
intended readers of the rapport.
The stocks indicate the percentage of
women currently in higher and lower academic positions, the flows indicate how these values can
change over time. As mentioned before, the stock and flow diagram is the core of the model, and
there are many variables exerting influence on various flows. The stock variable percentage of
woman in higher academic positions influences both masculinity of norms and the percentage of
women in networks.
Figure 2: Stock and flow of women’s careers
Masculinity of norms
a According to the respondents there are
certain norms within the university,
which are determined by the big
aciutment of blicmion — adjystmert of publication proportion of men in higher academic
rien: Speer = pr orme positions. The effect of the masculinity
iY 4 of these norms manifests itself in a
; { ——*tidvtwevor number of variables. Firstly in those
\ 7 variables that constitute the normative
~~ oventine aces ideal of the scientist: (a) The degree to
= eyera—~ Which full time work is set as a
+ ecamilinty standard for being an ideal scientist. (b)
Figure 3: Masculinity of norms The degree to which working overtime
is seen an inevitable part of being the
ideal scientist. (c) The degree to which it is expected that one is available for work every single
day of the week and during the evenings. (d) Status of educational achievements relative to those
in research. Respondents attribute these norms to a culture which is dominated by men and argue
that women in general do subscribe less to these norms than men do. Non adherence to dominant
norms in the faculty environment leads to less congruence with the ideal image of the scientist.
Secondly, masculinity of norms manifests itself in the degree to which management
adjusts productivity output targets when illness or pregnancy occurs, or when one works part-
time. According to respondents, the assignment of research time to individual researchers is
based on the absolute number of publications rather then the number of publications relative to
the time they had actually available for research. This practice bypasses the formal rules for
adjustment and negatively affects those academics that work part-time or took sick or pregnancy
leaf. Respondents agree this disproportionately affects women. When adjustments would be
made on the basis of the relative number of publications, this would make available more
research time for women, enabling them to produce more publications.
Perceived congruence of women with ideal image of the scientist
This variable indicates the perceived a
congruence or discrepancy between Se a " =
perceptions of female academics in \ pee ee
general and scientists in general. This ~—xa se ~ oe
congruence is firstly determined by the ~ a ee \ ctalaing wed
amount to which respondents think gt a
women adhere to the dominant norms 3" cae \ per
mentioned in the previous paragraph. sperience of weenen
Secondly, a number of measurable ~"*#*=%"—__ -| +f 4 cutrepeencuna spat
efforts such as the number of —,. 0. \ | + ais
publications, the amount of ee
international work experience and ——*333%* “once ’
success in acquiring funding were A
found to be of importance. Then there
are some softer variables. namely the Figure 4: Perceived congruence of women with the ideal image of the scientist
perceived household and family responsibilities of women, their entrepreneurial spirit and
visibility that affect women’s congruence with the ideal scientist according to the respondents.
Perceived congruence of women with the ideal image of the scientist has a major influence on
the percentage of women in higher academic positions, as it affects both promotions from lower
to higher academic positions, and outflow of female academics from both the low and high end
of the hierarchy. Respondents agree that managers take their promotion decisions on the basis of
the degree to which a candidate fits in their ideal about what constitutes the good scientist.
Female academics who have left the university expressed their frustration of not to being seen as
serious candidates for promotion. Perceived congruence of women with the ideal scientist is
influenced indirectly by most other parts of the modeled system; networks, support (by
colleagues, spouse and supervisor), the quality of job assessments and the masculinity of norms.
Networks
This part of the model describes the effect of networks on the percentage of women in higher
academic positions. Numerous respondents mentioned an effective network as an important
facilitator for graduates to be accepted into lower academic positions and for academics in lower
positions to be promoted to higher echelons of the organization. Networks are even more
important for entrance into higher academic positions from outside the university. An effective
at ete Reet network improves both the general visibility
} ~ sae ofc of women within the university and the
ms academic world, which offers direct
| } opportunities for career improvement.
According to respondents, the effectiveness
of networks for women increases when there
is a greater overall number of women present
in relevant networks in the academic
ptcfwomenin | environment. The proportion of women
present in relevant networks is partly
determined by the percentage of women that
sis + occupies a higher academic position.
Bectvemess sence nepoet
rework: fe ween (wero)
Figure 5: Networks
Self-confidence
Self-confidence can be seen as the central variable in this Si tn ae
part of the model. Respondents argue that self-confidence
of women influences the visibility, effectiveness of
networks and the entrepreneurial spirit of women. Self- — prerby some sppsrt by
‘ 5 coleagues
confidence is enhanced trough support of a supervisor, co- NZ
workers and spouse, while support of the spouse also Ve
increases the chances of women gaining international work Sg.
experience. :
§ F * \ superstce support
Self-confidence increases the effectiveness of (socio-rmticaat)
networks, as well as increasing the visibility of female Figure 6: Self-confidence
academics in and outside of the organization.
Budget versus Merit principle
x This part of the model refers to the
ll budget principle en ‘ naeraet
_” principles faculties use in their personnel
ee - ks management. The budget principle
# 7 . Fe eeibiec of'sialt: ~ dictates that someone who is applicable
~~ _—— available positions + for a higher function is promoted only
merit principle “. when there are sufficient staff positions
igure 7: Budget verstis Merit principle available within the faculty. The merit
principle dictates that whenever someone
meets the requirements for promotion, he or she is promoted immediately. At the university
examined in this research, both principles were in use. Which principle dominates depends on
the financial resources of the faculty. The principle in use determines the number of free
positions, where the merit principle allows for faster career paths then the budget principle.
The number of available staff positions influences the percentage of external inflow of women in
higher academic positions and promotions of women from lower into higher academic positions.
According to the respondents, when there are fewer positions available, the flow through the
hierarchy is slower for all academics. As the percentage of men is bigger in the higher age
groups, their (partial) replacement by younger women will take place in a slower time frame. The
budget principle effectively slows down the upward mobility of all academics, thereby delaying
the process of reaching gender equity. The number of available positions also affects the outflow
of women from lower academic positions. The frustration of not getting promoted has led to a
number of women leaving the university.
Feedback loops
We distinguish a feedback loop when a change in the value of a variable, via two or more
successive causal relationships in the system, influences the value of the variable itself.
Identifying these feedback loops provides insights on the behavior of the system. There are
positive and negative feedback loops. A positive feedback loops shows that changes in a variable
strengthen each other. A negative feedback loop shows that changes in one variable are
compensated by changes in another variable; a stabilizing process. When the internal structure of
the model is correct, the behavior of the feedback loop corresponds with the characteristics of the
real world system. We discuss the four main feedback loops we distinguish in the model.
Feedback loop 1: Masculinity of norms:
This main feedback loop runs from the percentage
of women in higher academic positions, trough
masculinity of norms, perceived congruence of
women with the ideal scientist back to the —
percentage of women in higher academic
positions. This is a positive feedback loop,
meaning that when there is a higher proportion of
women in higher academic positions, norms
become less masculine, increasing the congruence
of women with the ideal scientist, thereby
decreasing the percentage of women flowing out
of higher and lower academic positions, while
increasing the percentage of women being
promoted from lower into higher academic Feedback loop 1: Masculinity of norms
positions, thus leading to a higher percentage of women in higher academic positions. An
example of this dynamic is found in the Faculty of Law which has a high percentage of women in
high positions. In this faculty there are a relatively greater percentage of women flowing into
high positions and a relatively smaller percentage of women flowing out of high positions.
Feedback loop 2: Publication productivity targets
ee tame Academics agree that performance in their jobs is
mainly measured by the number of publications.
As in the previous feedback loop, a higher
percentage of women in higher academic positions
supports different practices. A higher percentage
of women up the hierarchy leads to better
execution of adjusting publication productivity
targets for part-time work and sick- and pregnancy
leaf, thereby increasing the amount of research
time available for those that have not been able to
work full-time. Via a larger number of
publications, these people come closer to the ideal
Feedback loop 2: Publication productivity targets image of the scientist, thereby increasing their
career chances. As relatively more women work part-time, the practice of adjusting output targets
10
is beneficial for women’s careers. Again we have a positive feedback loop, as a more female
management may decrease the masculinity of norms, thereby supporting the adjustment of
publication productivity targets.
Feedback loop 3: Visibility of women
The percentage of women in higher
academic positions influences the
percentage of women in relevant
academic networks. As the overall
proportion of women in networks
increases, these networks become more
effective for women, thereby increasing
their visibility within and outside of the
organization, increasing the chances of
being promoted and decreasing the
chances women will leave the
university. Again there is a positive
feedback loop, indicating that a Feedback loop 3: Visibility of women
growing percentage of women in academic positions has a self enhancing effect.
Feedback loop 4: Networks and women
This feedback loop shares many similar
variables with the previous one, but
focus here is on the direct opportunities
for career moves offered by networks,
instead of the increased visibility that
comes with having effective networks.
This positive feedback loop shows the
relation between the percentage of
women in higher academic positions
’ and factors that support gender equity.
women a thd When women make up a_ greater
percentage of relevant networks,
networks become more effective for
women, supporting their career paths within the academic world.
Feedback loop 4: Networks and women
Behaviour of the model:
Feedback loops give valuable insights into the behavior of the modeled system. Four major
feedback loops were identified which all had positive polarities. Feedback loops with positive
polarities describe processes which are self reinforcing. Since men are much better represented in
higher academic positions, and thus set the norms and are better represented in relevant
networks, we would expect to find a decreasing number of women in higher positions. With that
respect the model doesn’t fit the real situation at the Dutch university. The amount of women in
higher positions is not diminishing, but has been stable for years and is very slowly increasing.
This is probably due to the compensating force of the stocks-and-flow in the model; the
percentage of women in the younger cohorts has been growing for years. To summarize, the
careers of women are delayed by the processes described in the four feedback loops, but this is
compensated by the fact that the proportion of women amongst graduates en PHD students
grows. Moreover, there have been policies aimed at supporting gender equity. They were aimed
11
at increasing the effectiveness of networks for women and their self-esteem. Finally, also in the
present situation some women do correspond to the image of the ideal scientist. They are
perceived to adhere to the dominant norms, for example full time availability.
The model also offers valuable information on the relative importance of different
variables. All feedback loops contain the variable percentage of women in higher academic
positions, which indicates that this variable is essential for the behavior of the system. The
proportion of women in higher positions at the university has a major impact on the chances
women have to advance their careers. Recognizing the central role of this variable may offer a
starting point for an intervention. A single increase in the proportion of women at higher
positions would increase the perceived congruence of women with the ideal scientist and thereby
increases their chances of promotion.
Evaluation of Model Building
In this paper we show the results of model building in examining the factors influencing the
movement of women in, out and through the university hierarchy. We showed the central role of
four feedback loops in explaining the problem, respectively the dominant norms, publication
productivity targets, the visibility of women and the networks of women. The research question
was: What is the contribution of model building for the qualitative analysis of gender processes
at a Dutch university? We answer this question by reflecting on the theoretical contribution and
methodological meaning of our analysis.
Theoretical contribution
With regard to the theoretical contribution of our analysis, our results confirm the
important role the perceived image of women plays in comparison to their real performance as
identified by King (2007). We show four feedback processes that show how this perceived image
is build up, namely the dominant masculine norms about what it means to be an academic,
publication productivity targets, the visibility of women and the networks of women. The
academics that were interviewed agreed that the dominant norms can be characterized as
masculine. Thus, the ideal scientist behaves like the traditional male breadwinner, working full-
time, prepared to work over-time and is always available. Moreover, the academics perceive
female scientists to fit less to this image than their male colleagues. Both male and female
academics expect women to prefer part-time work, to be less prepared for overwork and
prioritize their family.
Moreover, dominant norms are articulated in terms of productivity output targets. By
describing the feedback loop of publication productivity targets, we show that the (lack) of
adjustment of productivity output targets to actual working time, offers an explanation for
women’s career development. Since publication output is measured in absolute figures rather
than in relation to invested working time, people who work part-time or took pregnancy leave are
put at a disadvantage to their full-time colleagues without career interruption.
Next to the fact that the dominant norms are considered masculine, academics agree that
women are less visible in- and outside of the organization. Although some women are perceived
to adhere to the dominant norms, their visibility within and outside of the organization is less
than the visibility of their male colleagues. Since academics agree that promotion decisions are
taken on the basis of the visibility of potential candidates, a lack of visibility proves a
disadvantage. Therefore, our qualitative analysis shows that the perceived image of women has
to do both with the lens through which people look at them and the spotlight that is directed at
them.
12
Summarized, our model adds to the existing literature an understanding of the particular
mechanisms that help to produce gender stereotypes. Model building helped us to distill the
shared conceptions of what it means to be an academic at a Dutch University, what norms are
dominant and how these norms are reproduced by the lack of women in higher academic
positions. It helps to understand that the careers of women are delayed by the processes described
in the four feedback loops, but that this is compensated by the fact that the proportion of women
amongst graduates en PHD students grows.”
Methodological meaning
What was the methodological meaning of model building to the qualitative analysis of
gender processes? The method of model building first of all enabled processing a huge amount of
research material into a compact Systems Dynamics model. Although the model contains a total
of 37 variables and 61 relations, it remains a very compact summary of all patterns and linkages
identified in the huge amount of research material. It helped us to identify the main variables and
their interrelations and to see the interconnections between the different processes at work.
Model building enabled discussions and sharing of knowledge across the research team.
The researchers participating in the model building sessions had knowledge of the interviews,
focus groups and documents of at least two faculties; none of the participants had a complete
overview of all the research material. Coding the interview material was divided over five
researchers and all analyses were double checked. The interview material, focus group material
and policy document material were thus covered by at least two researchers. Model building
supported a quick exchange of knowledge and identification of the main processes at work, like
for example the relation between the amount of women in higher positions and the visibility of
women. The technique also supported interpreting of the research material. There was agreement
on the general research issue, but intensive discussions were necessary to the bring variables into
operation in the model. The presence of a facilitator and modeler forced us to explicate all our
assumptions. An example of this was the discussion on the variable ‘perceived congruence of
women and the ideal scientist’. By recognizing the fact that model building enabled generalizing
the research findings onto a more abstract level, many interesting details had to be omitted, as
well as relations between variables that were found in only one faculty. Thanks to the presence of
a facilitator these decisions could be made quickly. The decision was made that only relations
which were mentioned by three people in at least two faculties, and could not be falsified, would
be incorporated in the model. This helped us to focus on the relations on which the most
unanimity existed, namely on work and family life, perceived image and networking.
Conclusion
To conclude, what theoretical contribution did model building make to knowledge
elicitation in the analysis of gender processes at a Dutch University? And what is the
methodological meaning of the instrument for such an analysis?
Firstly, the model shows how different, sometimes opposing explanations of women’s careers
progression are interrelated and influence each other. The amount of women in higher positions
at the Dutch University is not declining, but has been stable for years and is very slowly
increasing. While the increase of female graduates would support an increase in women at higher
academic positions (towards a situation of gender equity?) within fifteen or twenty years, the
present low amount of women in these positions slows down this development. The model shows
how this influences dominant norms, productivity output targets, the visibility of women and the
networks of women.
In practical sense, the process of model building helped the research team to identify
intervention strategies. It helped us to identify the essential role of the actual representation of
women in higher positions. Once the amount of women in higher positions increases
13
substantially, this effects the image of the ideal scientist, the amount of women in networks etc.
Since the research team only had an advisory role to the executive board, we formulated a series
of advices linked to the causal model.
The methodological meaning of model building in a qualitative research on this scale, is
that it helped to structure discussions on the interpretation and analysis of the material. Modeling
helped us to elicit the knowledge individual researchers had derived from the qualitative material
and share it within the group. The presence of a facilitator helped us to focus on the main
processes at work and to validate each different relation with all the researchers in the team.
Limitations and opportunities
A limitation refers to the explanatory force of the model. The relations and variables
identified are entirely based on the experiences of academics within the university. The model
represents their beliefs about organizational reality, rather than the reality as it could be
objectively observed. An example is that all of our respondents pointed at part-time work as one
of the most decisive factors in the delay of women’s careers in academia. However, quantitative
analyses of the actual working hours of this university’s academic personnel revealed that
women on average only work 1,3 hours less than men. Another example was that all respondents
believed that parenting would have a negative impact on productivity output. Quantitative
analysis showed that for both men and women, having children was positively related to the
number of publications. The causal loop diagram is a representation of the shared construction of
reality of the 80 academics working at the same Dutch University as reconstructed within the
research team. This doesn’t differ from the results of Group Model Building to support strategic
decision-making amongst managers. The model that results is a social construction. The
constitution of its variables is bases upon the shared beliefs and opinions of the academics which
were interviewed. Constitution is used here as the collective definition of what something is, that
is the intersubjective agreement upon the rules that lay down ‘what counts as’ (Schwaninger
2006, 525). So the model may become less relevant when these constitutions change (van der
Smagt 2006). This means that our explanation is bound to time and context, its universal
applicability has to be further tested.
The theoretical limitations don’t hinder the practical applicability of the model for policy
making. The executive board of the university was looking for a tailor-made analysis of the
actual situation at the five faculties and the university as a whole. The causal model that
supported our analysis was considered as a tool to enable a quick overview of the results of the
research. The fact that the research involved interviews with all the deans and focus groups with
all the heads of departments, gave the conclusions strong legitimacy towards the executive board.
The causal model that summarized the general opinion on the causes and consequences of the
problem was easily accepted as a description of the real situation. To the surprise of the research
team, the executive board took over all recommendations put forth in the final report. The
university policy is based on what people consider as real and so the common construction of
gender processes is also real in its consequences.
14
References
Adviesraad voor het Wetenschaps- en Technologiebeleid (AWT). 2000. Halfslachtige
wetenschap: Onderbenutting van vrouwelijk potentieel als existentieel probleem voor academia.
AWT-advies nr. 43.
Andersen, David, Jac Vennix, George Richardson and Etienne Rouwette, 2007. Group model
building: — problem _ structuring, — policy simulation and decision support.
Http://www.albany.edu/~gpr/JORS06.pdf.
Benschop, Yvonne, and Margo Brouns. 2003. Crumbling Ivory towers: Academic Organizing
and its gender effects. Gender Work and Organization, 10 (2) : 194-212.
Bleijenbergh, Inge, Yvonne Benschop and Jac Vennix. 2006. Group Model Building as a gender
Mainstreaming Instrument. Paper presented at the Conference “Equal is not enough” September
13-15, 2006, Antwerp University, Belgium
Bordons, Maria, Morillo, Fernanda. Fernandeze, Teresa. & Gomes, Isabel. 2003. One step further
in the production of bibliometic indicators at the micro level: Differences by gender and
professinoal category of scientists. Scientometrics, 57 (2): 159-173.
Brouns, Margo, Rie Bosman, and Ilse van Lamoen. 2004. Een kwestie van kwaliteit: loopbanen
van cum laude gepromoveerde vrouwen en mannen. Groningen, Rijksuniversiteit Groningen.
Campbell, Debra and P. Hovman. 2004. /s Your Organization Collegiality-Challenged? A Case
Study Demonstrating the Effect of Unconscious Gender Bias. 22nd Internation Conference of the
System Dynamics Society. The System Dynamics Society: Oxford, England.
Dudley, Richard. 2007. The Equity Supply Chain: Is it the Cause of So Few Women in
Management and Leadership Positions? 25" International Conference of the System Dynamics
Society. The System Dynamics Society: Boston, MA, USA.
http://systemdynamics.org/conferences/2007/proceed/papers/DUDLE171.pdf.
Forrester, Jay. 1961. Industrial Dynamics. MIT Press: Cambridge, MA.
Forrester, Jay. 1987. Lessons from System Dynamics modelling. System Dynamics review 3(2):
136-149
King, Eden. 2007. The effect of bias on the advancement of working mothers: Disentangling
legitimate concerns form inaccurate stereotypes as predictors of career success. Dissertation.
Houston, USA: Rice University.
Korsten, M., Visser, A., Willemsen, T., & van Zwol, W. (2006). Monitor vrouwelijk hoogleraren
2006. Stichting de Beauvoir & Universiteit Leiden, EQUALproject Participatie als Prioriteit/
Universiteit Maastricht, Landelijk Netwerk Vrouwelijke Hoogleraren en Vereniging van
Nederlandse Universiteiten.
Portegeijs, Wil & Brugman, Marga. 1998. Eerdaags evenredig? Belemmeringen en beleid ten
aanzien van de doorstroom van vrouwen naar hogere wetenschappelijke functies. Leiden.
Rijksuniversiteit Leiden.
15
Ragin, Charles. 1989. The comparative method; moving beyond qualitative and quantitative
strategies. University of California press, Los Angeles.
Richardson, George, and David Andersen. 1995. Teamwork in group model building. System
Dynamics review 11(2): 113-137.
Richardson, George. 1991. Feedback Thought in Social Sciences en Systems Theory.
Philadelphia. University of Pennsylvania Press.
Schwaninger, Markus. 2006. Comment on the paper by van der Smagt. Systems Research and
Behavioral Science 23: 525-527.
Shank, Gary 92006) Qualitative Research: a personal skills Approach: New Yersy: Pearson
Prentice Hall
Van den Brink, Marieke, Margo Brouns, and Sietske Waslander. 2006. Does Excellence have a
gender? A national research study on recruitment and selection procedures for professorial
appointments in the Netherlands. Employee Relations. 28 (6) 523-539.
Van Engen, Marloes, Bleijenbergh, Inge & Paauwe, Jaap. 2008. Vrouwen in hogere
wetenschappelijke functies aan de Universiteit van Tilburg; Processen van instroom, doorstroom
en uitstroom nader bekeken, Tilburg: Universiteit van Tilburg
http:/Awww.uvt.nl/medewerkers/vrouwen.html
Van der Smagt, Ton. 2006. Causation and Constituition in System Dynamics: Modelling a
Socially Constitutied World. Systems Research and Behavioral Science 23: 513-524
Vennix, Jac. 1996. Group model building — Facilitating team learning using System Dynamics.
John wiley & Sons Ltd, Chichester.
Vennix, Jac. 1999. Group model-building: Tackling messy problems. System Dynamics review.
15(4): 379-401.
16
Appendix 1: The total model
17
