AMBITION IS NOT ENOUGH: EXPLAINING CANDIDATE EMERGENCE IN STATE

LEVEL POLITICS

By

Jamil S. Scott

A DISSERTATION

Michigan State University

in partial fulﬁllment of the requirements

Submitted to

for the degree of

Political Science – Doctor of Philosophy

2018

ABSTRACT

AMBITION IS NOT ENOUGH: EXPLAINING CANDIDATE EMERGENCE IN STATE

LEVEL POLITICS

By

Jamil S. Scott

While the literature on political ambition has progressed, there is still much to be learned about
the path to elected o ce for current o ceholders and the potential o ceholders in the eligibility
pool. The most recent research has moved toward looking at candidates in the pre-emergence
stage , but the study of individuals in candidate and leadership training programs has largely not
been considered. In the dissertation, I argue that training programs account for all three aspects
necessary to spur political participation - capacity, motivation, and recruitment. Thus, candidate and
leadership training programs should be considered in the our rethinking of the candidate eligibility
pool as these programs not only contain individuals who have expressed interest in being politically
engaged, but also provide individuals with the tools and resources to be politically engaged.
Throughout the dissertation, I examine the relationship between political training programs and
representational outcomes. Speciﬁcally, I address the role that training programs play in fostering
political ambition, particularly for women and minorities, and how completing a training program
inﬂuences electoral chances of participants. I provide a comprehensive analysis of candidate and
leadership training programs in American politics using observational data, ﬁeld experiments, and
interviews. I derive much of the data from a partnership with the Michigan Political Leadership
Program. I situate this work in the candidate emergence literature as well as the literature about
participation as these programs have assumed, but largely untested beneﬁts for individuals with
political ambition.

To my mother, for believing that this was possible and for all of her support.

iii

ACKNOWLEDGEMENTS

There are a number of people who have made this journey through graduate school possible. I
am thankful that I had people not only believe in my ideas, but also believe in me. My academic
work has beneﬁted from the ﬁnancial support of the Alliance for the Graduate Education and the
Professoriate (AGEP), the Graduate School, and the Department of Political Science. I also have
great appreciation for the Summer Research Opportunities Program (SROP), which started my
journey at Michigan State University.

My committee has been an important part of this journey of developing my research ideas and
my identity as a scholar. I am so grateful that Nadia Brown agreed to join my committee. I not
only admire her work, but also have the privilege of calling her a co-author. Melinda Gann Hall is
state politics and I have beneﬁtted greatly from taking her class and getting her input. Cory Smidt
is largely responsible for my graduate education in American Politics. This was a good thing. I
learned so much from him and I would like to think that I gave him things to think about too. Matt
Grossmann was an instrumental part of my dissertation process, from the data collection to pushing
my ideas further than I thought they could go. My advisor, Eric, showed me what it meant to be
a scholar, was always there to provide advice along the way, and gave me the space to ﬂourish.
That made all the di erence. I could not have dreamed of a better committee and I am extremely
grateful for them all.

My family and friends, back on the east coast, indulged my long absence from home, but are no
less important to my journey. I received so much encouragement and love from them. Sometimes
it was from a far and other times it was a warm welcome home when I needed it. I’m blessed to be
able to join them again as I start a new adventure as a professor.

I can truly say that I was supported by an amazing group of women graduate students, both
in and outside of my department. You all were so important to my success as a graduate student.
You all are more than friends. You are my Michigan family. To my cohort, I must say that
Michigan State did a good thing when they admitted ﬁve women at the same time. Erica Maylee,

iv

Elizabeth Lane, Emma Slonina, and Erika Rosebrook I will truly miss you. To Lora DiBlasi, Jessica
Schoenherr, and Kesicia Dickinson, I appreciate you all so much and I will miss you all as well.
The co-authorship won’t end and WHIPS (Women Hustling in Political Science) will live forever.
To Stefanie Marshall, my accountability partner and friend, thank you. You read my work and
made sure I worked. I hope that we continue to support each other and hold each other accountable
for achieving great things. Graduate school would not have been the same without any of you.

v

TABLE OF CONTENTS

LIST OF TABLES .
LIST OF FIGURES .
CHAPTER 1

.
.
.
.
INTRODUCTION .

.
.

.
.

.
.

.
.

.
.

.
.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1

CHAPTER 2 UNDERSTANDING CANDIDATE SUPPLY: THE ROLE OF THE MICHI-

GAN POLITICAL LEADERSHIP PROGRAM IN MICHIGAN POLITICS . 16

CHAPTER 3 EXAMINING THE CANDIDATE ELIGIBILITY POOL USING FIELD

EXPERIMENTS .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

CHAPTER 4 TRAINING CANDIDATES: AN EXAMINATION OF HOW POLITI-
CAL TRAINING PROGRAMS INFLUENCE STATE LEVEL REPRE-
SENTATION .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

.

.

.

CHAPTER 5 WAITING TO RUN: CONSIDERING THE EXPERIENCES OF WOMEN

CHAPTER 6 CONCLUSION .
.
.
APPENDICES .

OF COLOR AS LATENT CANDIDATES . . . . . . . . . . . . . . . . . . 70
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
APPENDIX A FIRST APPENDIX . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
APPENDIX B SECOND APPENDIX . . . . . . . . . . . . . . . . . . . . . . . . . . 93
APPENDIX C THIRD APPENDIX . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
APPENDIX D FOURTH APPENDIX . . . . . . . . . . . . . . . . . . . . . . . . . . 100
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102

BIBLIOGRAPHY .

.
.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

vi

LIST OF TABLES

Table 1.1 Candidate and Leadership Training Program Curriculum Coding Scheme . . . .

9

Table 2.1 Comparison of MPLP Racial Breakdown to Michigan Census Data

. . . . . . . 23

Table 2.2 The Impact of Political Training Programs on Running for O ce . . . . . . . . . 28

Table 2.3 The Impact of Political Training Programs on Running for O ce: Matched

Interaction Models .

.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

Table 2.4 Bootstrapped Full Model (Matched) . . . . . . . . . . . . . . . . . . . . . . . . 35

Table 3.1

(Experiment 1) Response Rates by Condition . . . . . . . . . . . . . . . . . . . 46

Table 3.2

(Experiment 2) Response Rates by Condition . . . . . . . . . . . . . . . . . . . 46

Table 4.1 The Impact of Training Programs on Candidacy at the State Level

. . . . . . . . 62

Table 4.2 The Impact of Training Programs on Strategic Entry . . . . . . . . . . . . . . . 65

Table A.1 Balance Statistics for Matched Sample . . . . . . . . . . . . . . . . . . . . . . . 89

Table A.2 Percent Balance Improvement for Matched Sample . . . . . . . . . . . . . . . . 89

Table A.3 Sample Size of the Control and Treated Groups . . . . . . . . . . . . . . . . . . 89

Table A.4 The Impact of Political Training Programs on Running for O ce . . . . . . . . . 90

Table A.5 The Impact of Political Training Programs on Running for O ce . . . . . . . . . 91

Table A.6 The Impact of Political Trainings Programs on Running for O ce: Matched

Interaction Models .

.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92

Table C.1 Cronbach’s Alpha of Items Used to Create Strategy Measure . . . . . . . . . . . 96

Table C.2 Factor Analysis of Items Used to Create Strategy Measure . . . . . . . . . . . . 97

Table C.3 The Impact of Training Programs on Candidacy at the State Level: Model with

Time E ects Displayed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98

Table C.4 The Impact of Training Programs on Strategic Entry: Ordered Logit Models . . . 99

vii

LIST OF FIGURES

Figure 1.1 Proportion of Women and Minority Candidates Across States in 2012 and 2014

12

Figure 2.1 Gender and Racial Composition of MPLP Applicant Pool

. . . . . . . . . . . . 25

Figure 2.2 Desire to Run for O ce Across Racial Groups . . . . . . . . . . . . . . . . . . 26

Figure 2.3 Discrete Change in Predicted Probabilities . . . . . . . . . . . . . . . . . . . . 31

Figure 2.4

Interaction E ects for Race and Gender . . . . . . . . . . . . . . . . . . . . . . 33

Figure 3.1 Professions of Applicants to the Michigan Political Leadership Program . . . . 38

Figure 3.2 Facebook Ad .

.

.

.

.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

Figure 3.3 Recommendation Page

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

Figure 4.1 Number of Programs That Fit Training Program Coding Scheme . . . . . . . . 56

Figure 4.2 Emergence of Training Programs Over Time . . . . . . . . . . . . . . . . . . . 57

Figure 4.3 Training Program Participation . . . . . . . . . . . . . . . . . . . . . . . . . . 60

Figure 4.4 Predicted Count of Candidates Given the Number of Women Focused Programs

63

Figure 4.5 Optimal Scale of the Dependent Variable (Strategic Behavior) . . . . . . . . . . 67

Figure 5.1 Percent of Women in State Legislatures . . . . . . . . . . . . . . . . . . . . . . 70

Figure B.1 Text of Postcard 1- Control

. . . . . . . . . . . . . . . . . . . . . . . . . . . . 93

Figure B.2 Text of Postcard 2 - Treatment 1 . . . . . . . . . . . . . . . . . . . . . . . . . . 93

Figure B.3 Text of Postcard 3 -Treatment 2 . . . . . . . . . . . . . . . . . . . . . . . . . . 94

Figure B.4 Text of Email - Control

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94

Figure B.5 Text of Email - Treatment

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95

viii

CHAPTER 1

INTRODUCTION

In the wake of Hillary Clinton’s loss to Donald Trump in the 2016 presidential election, many have
taken to the streets in protest. As estimated by the Crowd Counting Consortium, from January 21,
2017 to December 31, 2017, between 5.9 and 9 million people have engaged in protest activity
in opposition to President Trump’s political stances and policies (Crowd Counting Consortium,
2017). Notable examples of this protest activity include the wave of Women’s Marches across the
country and the Day Without Women. While this wave of protest is noteworthy and certainly worth
analysis in its own right, what is most interesting and pertinent here is that the wave of protest
was accompanied by a wave of interest in running for o ce. Journalists have dubbed this surge in
interest in political candidacy the “Pink Wave" (Alter, 2018) because a record number of women
are running for o ce and even more are ﬂocking to candidate training organizations to gain the
tools to run for o ce (Landsbaum, 2017). However, this wave of political interest is not exclusive
to women, the surge in political candidacies has included men as well as people of color (Summers,
2018; Dittmar, 2018a). Though one election cycle will not change the gender and racial disparities
in political o ce, this is still an exciting and critical time for the candidate emergence literature.

While the candidate emergence literature has provided some explanations of why women and
minorities do not run for o ce (Fox and Lawless, 2011b; Kanthak and Woon, 2015; Sigelman
et al., 1995; Johnson, Oppenheimer and Selin, 2012; Highton, 2004), we still do not know why
these groups do run for o ce. Certainly, political ambition - the desire to run for o ce - matters.
Schlesinger (1966) was the ﬁrst to recognize the importance of political ambition as a motivation
for o ce seeking and the idea that o ce seekers are driven by ambition has served as the premise of
the candidate emergence literature (Fowler, 2010, see). However, I seek to advance the literature by
arguing that ambition is not enough. Although Fox and Lawless’s (Fox and Lawless, 2011a) Citizen
Political Ambition Panel study shows that there are di erences in political ambition by gender and
race, what is often the less discussed implication of this work is that there are individuals who have

1

an interest in running for political o ce but for one reason or another are not acting on this interest.
If we know that political ambition matters and that there are individuals who do not act on this
interest, then this leads to the fundamental question of why some ambitious people run for o ce
while others do not?

In arguing that ambition is not enough to explain o ce seeking behavior, I seek to advance
the literature by considering the act of running for o ce for what it truly is - a form of political
participation. It is one of the rarest forms of participation, but participation nonetheless. Because
the political participation literature does provide a framework to understand why people decide to
participate in political activities, I frame this work using Verba, Schlozman and Brady’s model
of civic voluntarism (Verba et al., 1995). Verba, Scholzman and Brady suggest that there are
three factors that explain political participation: resources (the civic skills, time and money to
participate), motivation (actual interest in being involved, self conﬁdence that you can be involved),
and recruitment (encouragement). Using the civic - voluntarism model, I contend that candidate and
leadership training programs are both a tool to identify latent candidates and a means through which
these latent candidates can obtain the capacity and encouragement to run for o ce. Candidate and
leadership training programs are an ideal means to understand who runs and how capacity matters
because if we can start to get answers as to why some ambitious people run for o ce and others
do not, as well as how elites matter in this process, we may start to understand why we see gaps in
candidate emergence.

The current moment presents an important opportunity for candidate emergence scholars to
test our standing theories about interest in political candidacy and explore new questions that need
answers. Despite the likelihood of electoral success for women and minority candidates (Fox and
Carroll, 2006; Highton, 2004; Juenke, 2014; Shah, 2014), prior work suggests that the barrier to
political entry remains the decision to run for o ce (Juenke and Shah, 2016; Fox and Lawless, 2004;
Shah, 2014). The early focus on who wins o ce (Bond, Covington and Fleisher, 1985; Canon,
1990; Maisel and Stone, 1997) has shifted to understand who runs and where (Sanbonmatsu, 2010;
Wilcox and Norrander, 2005; Juenke and Shah, 2015). This current wave of political empowerment

2

and political interest has produced the election of the ﬁrst women and minorities in state legislatures,
mayoral seats and stat level o ces across the country (Nilsen, 2017). While these historic electoral
victories bode well for what we already know about candidate emergence, the literature is still
incomplete.

To understand the behavior of politically ambitious people, I address four questions throughout
the dissertation. 1) What role do candidate and leadership training programs play in the candidate
emergence process for women and minorities? 2) How and when is political ambition primed? 3)
What impact do candidate and leadership training programs have on the candidate emergence of
women and minorities across states? 4) What role do candidate and leadership training programs
play in the candidacy of women of color and can these programs mitigate potential barriers they
may face in seeking to run for o ce?

Although candidate and leadership training programs are not necessarily a new phenomenon,
the proliferation and use of these programs have increased over time. The National Women’s
Political Caucus, founded in 1971, is one such organization that has been instrumental in training
female candidates for o ce throughout its existence. To date, the Center for American Women
and Politics (CAWP) names over 350 training programs across 38 states.1 A number of these
programs cater speciﬁcally to women, but many of these programs facilitate the political ambition
of both men and women. The fact that women are more likely to have attended a training program
(Sanbonmatsu, Carroll and Walsh, 2009), coupled with the evidence that training programs act as a
source of information and skill development (Sanbonmatsu, 2015; Hennings, 2011), speaks to the
important role these programs may play in American politics.

This chapter frames my argument for the dissertation and also provides a road map for what I
hope will be a clear and cohesive story about how training programs matter in state level politics
as well as how we should think of the act of running for o ce. First, I present a discussion of the
perspectives on political candidacy. The examination of political candidacy has a long history in
the congressional and state legislative literature. Particularly important to my work here is how the

1http://www.cawp.rutgers.edu/education/leadership-resources

3

more recent approach, candidate emergence, has combined the process and rational actor schools
of thought to further ambition theory. Next, I present my theoretical approach to understanding
political candidacy using political training programs. Finally, I describe the multi-method approach
I use to understanding the role of training programs at the state level and provide an outline of the
chapters to come.

Perspectives on Political Candidacy

Over time, scholars have grappled with the questions of who runs for o ce, why, and how this
matters for democratic outcomes. The aforementioned questions fall into three main theoretical ap-
proaches, sociological, process, and rational actor. Because of the di culty in identifying potential
candidates for o ce, much of the literature in these three streams rely on observed o ceholders
to express interest in running for higher o ce or to identify their motivations for seeking their
current o ce. The earliest approach to examining political candidacy was sociological in nature.
It originated with European sociologists like Weber, Mosca, and Pareto (Fowler, 2010). After the
behavioral revolution, the modern approach to this literature focused on understanding the social
backgrounds of o ceholders, placing emphasis on educational attainment, age, occupation, as well
as other factors. However, much of this work was descriptive in nature (Prewitt, 1970; Prewitt and
Eulau, 1971) and tried to draw links between background characteristics and legislative behavior.
Because there was mixed evidence that one’s background and identity inﬂuenced legislators’ career
experiences and policy decision making (Moncrief, 1999; Matthews, 1984), this approach is largely
not used today. Although, it can be argued that there is still reason to study the backgrounds of
o ceholders for the purpose of understanding representational outcomes - expressly for descriptive
and symbolic representation (Canon, 1994).

A second stream of literature on political candidacy is concerned with the recruitment process
for elected positions. This literature emphasizes the role of gatekeepers, like political parties
and interest groups, as well as the election and nomination process (Moncrief, 1999). Most of

4

the studies falling under this approach examine the ways in which o ceholders were recruited
(Matthews, 1984). However, two studies from the literature stand out - Wahlke et al. (1962) and
Seligman et al. (1974). Wahlke et. al ties social factors, institutional factors, and legislative
behavior together to examine recruitment patterns across states. Because of the design of the
study, Wahlke and his associates were able to observe and discuss the systematic di erences across
states that inﬂuence recruitment patterns. The other study of note in this literature is the work by
Seligman et. al. due to the focus on candidates for the Oregon state legislature. These authors
ﬁnd various patterns of recruitment including friends, families, local organizations, and political
parties. Despite the fact that recruitment does a play role in political candidacy, it does not fully
account for the personal decision making involved in the decision to run.

The idea that o ce seekers are driven by ambition is the premise of the rational actor approach to
studying political candidacy. Later work extends the theory by suggesting that potential candidates
for elected o ce emerge based on strategic decision making (Black, 1972; Jacobson and Kernell,
1983). For instance, quality, potential candidates are less likely to enter an electoral race when
an incumbent runs for o ce, but are more likely to do so in the absence of an incumbent (Bond,
Covington and Fleisher, 1985; Canon, 1990) or when they perceive that they are likely to win
(Maisel and Stone, 1997). There is also evidence to suggest that higher quality incumbents deter
challengers. More speciﬁcally, challengers tend not to be well funded in races with high quality
incumbents (Mondak, 1995) and losing a race is costly for strong challengers (Zaller, 1998).

Because of the perceived limitations of the rational actor approach, scholars have sought to
extend the study of candidacy to include both the recruitment process and political ambition
(Fowler, 1993; Fowler and McClure, 1990; Maisel and Stone, 1997; Maisel et al., 1990) as the
candidate emergence approach. Not only does this stream of work consider political candidacy,
but also the pre-candidacy phase. Thus, the candidate supply or the candidate eligibility pool is
important to identify (Kazee, 1994) as well as the personal and institutional factors that inﬂuence
the decision to run (Fowler and McClure, 1990). More recent work has extended the candidate
emergence literature to explain women’s under-representation in o ﬁce (Lawless and Fox, 2005;

5

Fox and Lawless, 2011b, 2004). For instance, the concept of nascent ambition comes from the
work of Fox and Lawless (2005) and it characterizes interest in running for o ce before making
the decision to actually run. These authors ultimately ﬁnd that women are less likely to express
interest, which has paved the way to pinpointing the barriers to greater political representation
(Dolan, 2006). However, this same e ort has largely not been extended to the study of emergence
amongst minority groups (Reny and Shah, N.d.), despite the fact that both minority o ceholders
and potential o ceholders express desire to run for some elected position (Jensen and Martinek,
2009; Lawless, 2012).

Theoretical Framework

The body of work on political participation is continuously growing and expanding our knowledge
of who participates and why. Scholars have studied how one’s identity matters for participation
(Harris, Sinclair-Chapman and McKenzie, 2005; Harris, 1994; Calhoun-Brown, 1996; Chong and
Rogers, 2005; Leighley and Vedlitz, 1999), how one’s resources (Leighley, 1990; Verba and Nie,
1972; Verba, Nie and Kim, 1987; Verba et al., 1993; Kenny, 1992), skills (Verba et al., 1995,
1993), and even political environment matter (Matsubayashi, 2010; Rosenstone and Hansen, 1993;
Caldeira and Patterson, 1982; Boyd, 1989; Cox and Munger, 1989) as well. In addition, scholars
have examined the political activities in which individuals choose to engage. Most notably, there are
a variety of explanations for voter turnout on election data (Bedolla and Michelson, 2012; Gerber,
Green and Larimer, 2008; Gerber and Green, 2000; Nickerson, 2008; Fraga, 2016). Whether it be
through the use of experiments or by leveraging observational data, scholars continue to ﬁne tune
what we know about the the factors that lead to increased voter turnout (Green and Schwam-Baird,
2016).

This same attention to explaining voter participation has not been paid to explaining participation
as a candidate (who runs for o ce). This is despite the fact that who sits in o ce has been linked to
increased participation for both women (Campbell and Wolbrecht, 2006; Stokes-Brown and Dolan,

6

2010; Atkeson, 2003) and minority groups (Platt, 2008; Tate, 1991; Gay, 2001; Whitby, 2007). In
fact, only two studies include running for o ce in the possible realm of participatory acts - the
work by Shingles in 1981 and the work by Hennings in 2011. In trying to clarify the scope of
Black political engagement in the United States, Shingles (1981) lays out a typology of political
activities that suggests running for o ce (along with a number of other activities like lobbying,
protesting, and being involved in community organizations focused on public issues) both require
high initiative and an intent to inﬂuence government outcomes. Certainly Shingles is correct that
all of those activities require great e ort and are often done in pursuit of a goal, but running for
o ce is a unique activity and I do not take for granted that political ambition does matter. What I
do take from Shingles’ typology is that if we can imagine a realm in which running for o ce is on
a continuum of political activities - all be it at the extreme end, then we can apply the same models
that explain who votes to who runs. In applying a model of political participation to candidate
emergence, I take a cue from Hennings (2011) who similarly seeks to explain political ambition
expression using a model of political participation.

I focus on Verba, Schlozman and Brady’s model of political participation as a framework. In
their 1995 work, Verba et al. (1995) put forth the civic - voluntarism model in which resources,
motivation and recruitment are used to explain acts of political participation in the American public.
In their model, resources refer to time, money and civic skills. Political motivation refers to political
interest, political e cacy, and political information. Finally, recruitment refers to the act of being
encouraged to participate.

The Verba et.al participation model ﬁts nicely here as it provides a means of theorizing what
beneﬁts candidate training programs can have for its participants. By completing a training
program, participants have already demonstrated their interest or motivation to seek political o ce.
Stock’s 2012 work adds to this as training program alumni are more likely to report increased
political activity, e cacy, and conﬁdence. Furthermore, there is an indication that candidate-
training programs can serve in the capacity of encouraging potential candidates to run (Hennings,
2011) as well as identifying potential candidates for o ce and providing these individuals with

7

resources(Sanbonmatsu, 2015; Burrell, 2010; Rozell, 2000). Therefore, it would appear that
candidate training programs account for all three aspects necessary to spur political participation -
capacity, motivation, and recruitment.

What are candidate and leadership training programs?

Candidate and leadership training programs are organizations that emphasize instruction in leader-
ship development and/or how to run a political campaign. While there are some organizations that
have a national proﬁle in my dataset, I focus speciﬁcally on programs that are based in a state. This
does include national organizations with state based chapters.

In identifying candidate and leadership training organizations across the country, I classify the
organizations on a set of curriculum objectives as identiﬁed by each organization’s website. This is
an important task because it represents one of the ﬁrst attempts to distinguish what these programs
o er to its participants in the political learning experience as well as how these o erings di er by
program type and state. I detail each curriculum item in the classiﬁcation in Table 1.1.

I include both candidate and leadership training programs in the analysis because some organi-
zations self-identify as leadership training organizations, but are similar in proﬁle to self-identiﬁed
candidate training programs. An example of this would be the Michigan Political Leadership Pro-
gram, which ﬁgures prominently in the analysis throughout the dissertation, and Camp Wellstone, a
well known progressive political training and advocacy organization. While the two organizations
di er in partisanship, the Michigan Political Leadership Program being non-partisan and Camp
Wellstone being a Democratic leaning organization, both organizations attest to o ering the same
learning outcomes in the description of the curriculum. That is, both organizations describe their
curriculum objectives as encompassing instruction on campaign planning, campaign strategy, voter
outreach, fundraising, networking, public speaking, governance and ethical behavior and leader-
ship. If I only included the programs that self-identity as candidate training programs, then I would
miss programs like the Michigan Political Leadership Program that take on traditional campaign
focused curriculum topics.

8

Table 1.1 Candidate and Leadership Training Program Curriculum Coding Scheme

Campaign
Planning

Campaign
Strategy

Voter
Outreach

Fundraising
and
Budgeting

Networking

Public Speak-
ing

How to get on the ballot; understanding what personnel are necessary to run a
campaign and the role these personnel play; campaign timeline and scheduling

How to develop e ective messaging for the campaign; how candidates should
present themselves to voters

How to coordinate get out the vote e orts; how to engage with voters and
develop e ective ﬁeld organization; how to utilize social media when engaging
voters

Strategies to raise money for the campaign; how campaign money should be
used during the campaign

Developing and using contacts for purposes beyond an initial meeting

Developing the skills to e ectively deliver a message in front of a group

Governance

How to be an elected/appointed o cial

Ethical
Behavior and
Leadership

Any curriculum based material having to do with the importance of following
the law, outlining what the laws are, and/or emphasizing appropriate behavior
for elected o cials and candidates

In addition to the curriculum, candidate and leadership training programs di er on a number of
other factors, which include cost, length of the experience, number of people admitted, and focus
on an identity group. Program participation costs vary widely. The range of program participation
costs extend from $0 to almost $6000. Similarly, the length of the program experience varies
widely as well. For example, the programs in the Ready to Run Network hold 1 to 2 day seminars
for its participants, while the New Leaders Council has chapters that facilitate a 5 month fellowship
experience. The longest noted program experience in the data is 2 years. Furthermore, there
are programs geared toward speciﬁc identity groups. There are 87 programs that are focused
speciﬁcally on developing women’s leadership and/or preparing them to run for o ce. Similarly,

9

there are 32 programs that focus on racial minorities and 1 organization that focuses on advancing
the political candidacy and leadership of individuals who identify as LGBTQ.

Deﬁning the Candidate Pool

Within the literature, there are assumptions about who runs for o ce. The accepted view is that
candidates (both men and women) are likely to be lawyers, businessmen or women, educators and
political activists (Moncrief, Squire and Jewell, 2001; Fox and Lawless, 2004), yet this view is largely
based on the observed winners of electoral races - members of the legislature. It does not account
for the professional and educational backgrounds of the losers. What’s more, this assumption
about who runs for o ce underlies many of our ﬁndings in the candidate emergence literature.
Despite that, Carroll and Sanbonmatsu (2013) suggest that the candidate pool has expanded for
women. Their qualitative study indicates that the common experience among the interviewed
female legislators is strong community involvement and civic engagement. In addition, the study
conducted by Sanbonmatsu, Carroll and Walsh (2009) implies that women do not necessarily follow
the same career paths as men on the road to political candidacy. Thus, this would suggest that
rethinking the candidate pool is necessary to better understand who runs for o ce. This is where
candidate training programs play an important role.

Given the growth in the number of candidate training programs over time, it is somewhat
surprising that the literature remains relatively silent on what these programs mean for legislative
politics and representation in the United States. However, of note are the works by Hennings
(2011) and Stock (2012) as they both examine how training programs inﬂuence the behavior of
its participants. Hennings draws on the Verba et al. (1995) participation framework as well as
social identity theory to argue that training programs provide an opportunity for its participants to
view themselves as political actors and develop a civic identity. Using a survey and interviews,
she ﬁnds that amongst program alumni, women are more likely to participate in political activities
(like attending meetings, making speeches, volunteering for political campaigns, etc.), the majority
of program alumni increase political activity after program completion, and participants credit

10

completing the program as their motivation for seeking o ce. Similarly, Stock examines the
psychological beneﬁts of training program participation. In particular, she argues that exposure to
the program and its curriculum is a means of increasing political interest and e cacy. Stock uses
a quasi-experimental approach and interviews to evaluate the e ectiveness of the program. She
ﬁnds that participants and non-participants (individuals who applied, but were not selected for the
program) signiﬁcantly di er in knowledge about women in politics and methods of participation, as
well as interest in running for o ce, with participants being advantaged on all three factors. These
ﬁndings are further bolstered by the qualitative data in which the participants emphasize positive
personal outcomes (like increased e cacy and conﬁdence).

While both Hennings and Stock provide an important discussion of training programs as a
causal mechanism, I seek to di erentiate this work by focusing on the inputs and outputs of training
programs. That is, Hennings and Stock address how these programs matter, while I seek to
address why these programs matter, particularly for electoral outcomes at the state level. The really
interesting aspect of studying training programs is that these programs identify a set of individuals
with nascent ambition. However, these programs have not been conceptualized as such in the
literature. Speciﬁcally, individuals who seek to complete, and do complete, a training program
are to some degree motivated or interested in running for o ce. This might especially be the case
when there is a cost to participate and a screening process. It would follow that the expectation
would be that those individuals who actually participate in the program would be more likely to
engage in o ce seeking behavior.

Through my focus on the "why" question, I advance the work of Hennings and Stock by going
beyond self report measures of political ambition expression and measure actual ambition expression
by examining electoral outcomes. Additionally, I acknowledge the selection bias problem in the
candidate emergence literature. The literature has begun to examine how both the winner and losers
of electoral races matter (see Juenke, 2014) and the literature has made inroads in identifying latent
candidates or candidates with nascent ambition (Fox and Lawless, 2004, 2005), but much of what
we know about candidate emergence derives from the observed cases of individuals running for

11

o ce. Scholars still do not fully understand the structural or institutional factors, as well as the
individual level factors, that contribute to latent candidates emerging. More simply, we have not
fully examined all the factors that might explain why candidates get elected in the places that they
do (Broockman, 2014a).

Figure 1.1 Proportion of Women and Minority Candidates Across States in 2012 and 2014

Women Candidates

Minority Candidates

 
s
e
t
a
t
S

AZ
CO
MT
NM
WA
NH
ID
IL
ME
MD
MN
KS
FL
CT
NV
WI
OR
ND
OH
UT
DE
SD
GA
NE
MO
IA
OK
NC
RI
CA
MI
AR
NY
WY
PA
TN
TX
WV
IN
KY
AL
SC

 
s
e
t
a
t
S

NM
CA
GA
TX
FL
AL
NY
SC
MD
IL
TN
AZ
NC
DE
CT
NV
MI
OK
OH
CO
MO
AR
RI
NE
PA
IN
KS
KY
UT
WV
WI
WY
WA
MN
OR
IA
NH
MT
ID
ND
ME
SD

0.2

0.4

0.6

0.8

Proportion

0.2

0.4

0.6

Proportion

0.8

Why States?

This work focuses on potential candidates at the state level. The most practical reason for this
decision is that the subnational level a ords more opportunities to see women and minorities
candidates in electoral races. In Figure 1.1, I show the proportion of women and minority state
legislative candidates, respectively across states. The data shows that women candidates are
emerging in some states more than others. The highest proportion of women running for o ce in

12

2012 and 2014 is in Arizona at 35%, while the lowest is in South Carolina at 16%. Although there
are no states that are at gender parity in the proportion of women running for o ce, states like
Arizona, Colorado, and Montana are getting close. In addition, we would likely not see this level
of candidate gender representation in national level races.

On a similar note, the proportion of minority candidates in state level races also varies across
states. The proportion of minority candidates is as high as 41% in New Mexico and as low as
1% in South Dakota. These patterns for minority candidates not only speak to the distribution of
minority populations in the United States, but also the opportunity structures to run for o ce at
the state level. In addition, the 2010 Census2 puts the growing minority population at about 38%
in relation to non-Hispanic whites. States like New Mexico and California are at or are reaching
racial parity in candidate representation.

Given the patterns in where women and minorities tend to run (Sanbonmatsu, 2010; Branton,
2009) and the variation across states (Morehouse and Jewell, 2004), it would make sense to ensure
that I am able to observe as many cases of these candidates emerging as possible. Also, the cross state
variation in where minorities and women tend to emerge is interesting and worthy of explanation.
Moreover, there is certainly a need to explain to what extent candidates are self-motivated to run
for o ce and to what extent motivation comes from recruitment.

Why focus on Women and Minorities?

Despite the record number of women and people of color who are seeking o ce during this
election cycle, as alluded to earlier, women and minorities are still under-represented in American
politics. There is no state in which women’s representation in the legislature is on par with their
representation in the total population. A similar story can be told for minority o ceholders. Despite
their underrepresentation, Women and minorities’ representation in legislatures is meaningful.
While Mansbridge provides theoretical justiﬁcation for the importance of diverse voices in decision

2This estimate is based on Census Data Quickfacts provided at

https://www.census.gov/quickfacts/fact/table/US/PST045216

the following link:

13

making (Mansbridge, 1999), others show the presence of diverse voices means that womens’ and
minorities interest are represented in the legislative agenda (Minta and Brown, 2014; Bratton and
Haynie, 1999; Swers, 2005; Preuhs and Juenke, 2011; Juenke and Preuhs, 2012).

Though we have seen electoral success for some women and minorities during the most recent
election cycle, this will not mean success for all (see Dittmar, 2018a). Beyond this wave of political
excitement and interest, we must understand women and minorities pathways to political o ce
as well as what encourages them to enter politics. While the political excitement may not last,
the underlying political ambition of potential o ceholders will. If we can bring those politically
ambitious people into politics, then we may see some meaningful change in representation.

A Look Ahead

Throughout the following chapters, I lay out the implications of candidate and leadership training
programs for representation as well as political participation - particularly for women and minorities.
I provide a comprehensive analysis of political training programs in American politics using
observational data, ﬁeld experiments, and interviews. I derive much of my data from a partnership
with the Michigan Political Leadership Program.3

The ﬁrst part of the dissertation uses an original dataset that I created from the Michigan
Political Leadership Program. It encompasses the total applicant pool of the program from 2003
to 2015. The nature of the data provides leverage to examine the electoral trajectory of individuals
who were accepted in the program and those who were not. Furthermore, I use matching to be
able to speak to the causal implications of the results. I ﬁnd that individuals who were accepted to
the program were more likely to run for o ce, which suggests that training programs motivate its
participants to run for o ce.

3The Michigan Political Leadership Program is a training program associated with the Institute
for Public Policy and Social Research at Michigan State University.
It is a bipartisan training
program. Each program class includes 12 men and 12 women from across the state and from
diverse backgrounds.

14

The second part of my dissertation primes political ambition through two ﬁeld experiments.
Field experiments have long been used as a means to mobilize voters, but the literature has not fully
explored whether or not mobilization can extend to other forms of political participation. Further-
more, the literature has not fully addressed if participation in one activity leads to participation in
others. Focusing on individuals who show a strong voting record, the ﬁrst experiment examines if
presenting encouraging appeals to active voters will increase interest in political candidacy. The
second experiment hones in on the importance of being asked to run for o ce. Using the Facebook
platform, I examine who gets referred to events about running for o ce and whether or not a
personal or impersonal appeal matters for that individual to attend an event about running for o ce.
Next, I examine the impact of candidate training programs across states. After identifying
over 350 candidate and leadership training programs, I classify these programs to determine what
they o er participants as well as the states in which they have a presence. This e ort is unique in
that no other work classiﬁes training programs, at this scale across states, and available work has
focused on a few programs to understand program impact on self-reported political ambition and
self-e cacy. Here, I focus on how presence of a training program in the state inﬂuences women’s
candidate emergence. I ﬁnd that the presence of a training program in a state signiﬁcantly increases
the likelihood that a woman candidate runs for o ce in a state.

Finally, I use a qualitative approach to contend that the study of women of color in identity
politics is largely missing in the candidate emergence literature.
I focus on women who have
completed the Michigan Political Leadership Program as a means of identifying women who have
shown interest in running for o ce. I conduct interviews in which participants are asked questions
about their political and social networks, what considerations they have made in running for o ce
(or not), the role that the training program has played in their political candidacy, the factors that
have sparked their interest in politics, as well as the barriers to running for o ce.

15

CHAPTER 2

UNDERSTANDING CANDIDATE SUPPLY: THE ROLE OF THE MICHIGAN

POLITICAL LEADERSHIP PROGRAM IN MICHIGAN POLITICS

Introduction

One question of interest in the identity politics literature is whether descriptive representation mat-
ters for American Politics.1 Beyond the normative arguments in favor of descriptive representation
as a means of improving deliberative democracy (Mansbridge, 1999), there is evidence both at the
state and national level that the presence of minority and women legislators does have an impact on a
variety of political outcomes. In particular, minority legislators are more likely to sponsor and sup-
port legislation that speak to the interests of minority constituencies (Canon, 1999; Whitby, 2000;
Hero and Preuhs, 2010; Juenke and Preuhs, 2012). Similarly, women legislators are more likely
to initiate legislative action on women’s issues (Bratton and Haynie, 1999) and prioritize gendered
issues in their legislative agenda (Thomas and Welch, 1991; Swers, 2005). Although minority and
women legislators are uniquely responsive to the needs of women and minority constituencies, the
reality is that these groups are still underrepresented in American politics.

The fact that women and minority legislators can inﬂuence policy outcomes suggests that they
can access the ranks of elected o cials, but there is a still a candidate supply problem. Women and
minority candidates are less likely to run for o ce. That is, prior work on the electoral success of
women indicates that they are just as likely to win as men (Fox and Carroll, 2006), but the patterns
of women’s electoral success varies across states (Sanbonmatsu, 2010; Wilcox and Norrander,
2005) and the relative growth in women o ceholders overtime has plateaued (Hennings, 2011;
Sanbonmatsu, 2010). Similarly, there is evidence that minority candidates can and do win in
non-majority minority districts (Highton, 2004; Juenke, 2014; Shah, 2014), but electoral success
is conditioned by the presence of minority candidates willing to run for o ce (Juenke and Shah,
1Descriptive representation is achieved when a constituency is represented by someone who

shares their identity (Pitkin, 1967).

16

2016, 2015). All together, this suggests that there is still much to be learned about the paths to
elected o ce for both current and potential o ceholders. If voter - demand side theories cannot
explain the representation of women and minorities in state legislatures, then we must learn more
about how ambitious women and minorities become candidates.

In the classic book, Why Parties?, Aldrich 1995 argues that one of the fundamental purposes
of parties is to serve as gatekeepers in order to provide a check on political ambition and regulate
access to political o ce. The gender and politics literature, in particular, suggests that this check
on political ambition has been detrimental to whose candidacy gets supported and who gets asked
to run(Niven, 1998; Sanbonmatsu, 2006). Furthermore, the strength of the recruitment e ort for
diverse candidates depends on the message (Karpowitz, Monson and Preece, 2017) as well as the
means through which party leaders choose to recruit (Crowder-Meyer, 2013). With that being
said, the party does not appear to be the main means through which diverse candidates are being
recruited and encouraged to run for o ce. In fact, it can be argued that both, on the political right
and left, independent political and activist groups are stepping into the role of determining who
runs for o ce and are reducing the parties’ ability to gatekeep (Rauch and Raja, 2017). These
organizations are not only identifying talented individuals, but gaining access to those whom the
party may have traditionally missed or may not have the networks to recruit.

In this study, I argue that candidate and leadership training programs, are a mechanism for
identifying potential candidates. Program participants present a clear example of nascent ambition
(Fox and Lawless, 2005) precisely because applicants are expressing initial interest in running for
o ce or, at a minimum, interest in being politically engaged. To this end, I focus speciﬁcally on
the Michigan Political Leadership Program to develop an original dataset of applicant information
spanning from 2003 to 2015.
I further expand on the data contained in the application ﬁles
by including a measure of whether or not an applicant ran for o ce and if the o ce seeking
attempt was successful. This presents a unique opportunity to not only examine political ambition
expression across race and gender, but also to focus on the shift from nascent political ambition to
ambition expression (e.g running for o ce). Using coarsened exact matching, I show how program

17

participation plays a distinct role in candidate’s decision to run for o ce and their subsequent
electoral outcomes.

Who Runs For O ce

Voter preferences (demand) for minority and women candidates are often presented as a factor to
explain the representation of these groups in o ce. Indeed a candidate might take these preferences
into account, but the demand side story does not fully account for the emergence and electoral
success of women and minority candidates. While there is experimental and survey evidence that
voters prefer representatives from their own racial group (Canon, 1999; Hutchings and Valentino,
2004), demand for a representative of a speciﬁc race does not hinder candidates’ electoral chances
in the real-world (Citrin, Green and Sears, 1990; Highton, 2004; Voss and Lublin, 2001). Similarly,
there is evidence that both men and women prefer a greater proportion of men to women as
representatives (Dolan and Sanbonmatsu, 2009). Nevertheless, co-partisanship overrides gender
and racial preferences for representation (Kam, 2007; Sigelman et al., 1995; Dolan and Lynch,
2015). This provides some indication that demand alone is not dictating when women and minority
candidates decide to seek political o ce.

In the candidate emergence literature, scholars have grappled with the questions of who runs
for o ce, why and how this matters for democratic outcomes (Fowler, 2010). The most recent
research has moved toward looking at individuals in the pre-candidacy stage to better answer these
questions. More speciﬁcally, this stream of literature not only seeks to understand the process of
recruitment, but also the motivational and rational considerations that factor into o ce seekers’
decision making (Moncrief, 1999; Fox and Lawless, 2004; Kanthak and Woon, 2015). While this
literature has made great strides in understanding the factors that hinder o ce seeking behavior,
especially as it pertains to women and minorities, the reasons why individuals decide to run for
o ce are less clear (Dolan, 2006).

What is relatively well deﬁned in the literature are the assumptions about who runs for o ce

18

- the traditional candidacy pool. The accepted view is that candidates (both men and women) are
likely to be lawyers, businessmen or women, educators and political activists (Moncrief, Squire and
Jewell, 2001; Fox and Lawless, 2004), which is largely based on the observed winners of electoral
races, members of the legislature. What’s more, this assumption about who runs for o ce underlies
many of our ﬁndings in the candidate emergence literature. However, there is reason to believe
that the candidate pool has expanded, particularly for women. In a qualitative study, Carroll and
Sanbonmatsu (2013) ﬁnd that the common experience among the interviewed female legislators
is strong community involvement and civic engagement.
In addition, the study conducted by
Sanbonmatsu, Carroll and Walsh (2009) implies that women do not necessarily follow the same
career paths as men on the road to political candidacy. The literature is not so clearly deﬁned as
to paths to political o ce for minority o ceholders. Inherently, there is a selection bias problem.
We tend to know a lot about the winners, but not very much about the losers, (see Juenke, 2014).
Similarly, the literature has made inroads in identifying latent candidates or candidates with
nascent ambition (Fox and Lawless, 2004, 2005), but much of what we know about candidate
emergence derives from the observed cases of individuals running for o ce. Scholars still do not
fully understand the individual level factors, that contribute to latent candidates emerging. More
simply, we have not fully examined all the factors that might explain why candidates get elected
in the places that they do (Broockman, 2014a). A related concern is that the desire to run for
o ce may not translate to actual outcomes because of the costs involved in running for o ce, both
personal and monetary. In addition to the ways in which parties can act as gatekeepers (Broockman
et al., 2014; Sanbonmatsu, 2006) to keep particular individuals out of o ce, political ambition can
be dynamic (Fox and Lawless, 2011a). The implication of this is that individuals may be willing to
express interest in running for o ce, but may never take the actual steps to run. This is a problem
for the literature if its goal is to answer why and when people decide to run for o ce. All together,
this would suggest that rethinking the candidacy pool is necessary to better understand who runs.
Identifying latent candidacy is very di cult though. Given the above critiques, a latent candidate
must both desire to run and demonstrate interest in running in some meaningful way. For this

19

reason, I turn to political training programs as a means of identifying latent candidates.

Redeﬁning the Candidacy Pool

One factor that has played an important, but understudied role, in political candidacy is the
political training program. As it pertains to women, these organizations have been instrumental
in advancing women’s political ambition by undertaking recruitment, training and developing
resources for candidates (Burrell, 2010). To date, the Center for American Women and Politics
(CAWP) names over 350 training programs across 38 states.2 A number of these programs cater
speciﬁcally to women as well as other minority groups, but many of these programs facilitate the
political ambition of both men and women across racial groups.

Despite the proliferation of these programs, there is very little research that examines the
outcomes of these organizations, as it pertains to training and recruitment, and how they matter
for the candidacy of disadvantaged groups in the political realm. Furthermore, there is little to
no research about the individuals who complete these programs and if they ﬁt the traditional
conception of the candidacy pool. However, there is limited evidence that women are more likely
to have attended a training program (Sanbonmatsu, Carroll and Walsh, 2009), and some evidence
that training programs act as a source of information and skill development (Sanbonmatsu, 2015).
Given the growth in the number of candidate training programs over time, it is somewhat surprising
that the literature remains relatively silent on what these programs mean for legislative politics and
representation in the United States. However, of note are the works by Hennings (2011) and Stock
(2012) as they both examine how training programs inﬂuence the behavior of its participants. I
di erentiate this current work from Henning and Stock by examining the step prior to program
participation - the application process. In focusing on program applicants, I am able to address
why these programs matter for identifying candidates as well as for electoral outcomes in state
and local politics in Michigan through an in - depth analysis of the Michigan Political Leadership

2http://www.cawp.rutgers.edu/education/leadership-resources

20

Program (MPLP). I seek to situate programs like MPLP into the candidate emergence literature by
examining if MPLP applicants are latent candidates.

There are a number of ways in which an individual can be politically engaged in the American
political system, running for o ce and serving in o ce are some of the more extreme and rare
forms of participation. Since running for o ce is a form of political participation, I draw on the
seminal work of Verba, Scholzman and Brady 1995. Verba, Scholzman and Brady’s framework
suggests that people who are active in political activities are those who have capacity (time, money,
and civic skills) and motivation (political interest, information and e cacy). According to the third
prong of the Verba et. al framework, these politically active individuals just need to be recruited or
encouraged to participate. The Verba et. al participation model ﬁts nicely here as it provides a means
of theorizing what beneﬁts training programs can have for its participants. There is an indication
that candidate-training programs can serve in the capacity of encouraging potential candidates to
run (Hennings, 2011) as well as identifying potential candidates for o ce and providing these
individuals with resources (Sanbonmatsu, 2015; Burrell, 2010). I take these prior arguments a step
further by examining actual electoral outcomes as a result of program participation.

Assuming that political training programs account for all three aspects necessary to spur political
participation, individuals who seek to complete, and do complete, a training program will be
motivated, have the capacity, and have been encouraged to participate. It would follow that the
expectation would be that those individuals who actually participate in the program would be more
likely to engage in o ce seeking behavior. Thus, I hypothesize that training program participants
will be more likely to run for o ce than non-participants and being accepted into MPLP will be a
greater predictor of running for o ce than the mere desire to run. In addition, conditional on being
accepted into the program, participants with civic skills will be more likely to run for o ce.

21

Data

In order to test the hypotheses, I create an original dataset that consists of applicants to the Michigan
Political Leadership Program. To provide some background about the program, the Michigan
Political Leadership Program (MPLP) was established in 1992. MPLP is a non-partisan program
that emphasizes partisan and geographic parity in each class. Each year, the program chooses 12
men and 12 women through an application and interview process. The application and interview
process are important features of the program because not every political training program has a
gatekeeping function. That is, some political training programs are open to all who pay to attend
a training. If accepted in the program, participants, or fellows as described by the MPLP program,
are required to attend one weekend session per month for about a year. In total, the fellows attend
10 training sessions, which cover topics like public policy issues across the state, public speaking,
campaign planning, and leadership development. The program is explicitly advertised as a space
for individuals interested in running for o ce and participating in the political process.

The dataset includes information about each applicant from two time points - at the point in
which they apply to the program and any electoral or appointed political activity after the application
process. Data from the ﬁrst time point was obtained from individual application ﬁles from 2003
to 2015. The MPLP application ﬁles contain information about prior political o ce and other
political experience, beliefs about important issues in Michigan, organizations that the applicant
participates in, demographic information (gender, race, education, and occupation), and social ties
(each applicant has to provide 2 reference letters). Each ﬁle also contains an aggregate interview
score. At the interview, each applicant is given scores from a panel consisting of the program
directors and former program participants. The interview score is a unique feature of this dataset
and is a rating of an individual’s likelihood of running for o ce and program ﬁt.
In addition,
applicants are speciﬁcally asked their expected outcomes for participating in the program and
whether or not they plan on running for o ce.Data from the second time point includes a measure
of whether or not a program applicant ran for o ce or was appointed to a position, the year, and if

22

the o ce seeking attempt was successful.

The data collected here is unique because it includes self-report data on demographic character-
istics, activities and desire to run for o ce as well as observed o ce seeking attempts. In essence,
I am able to capture applicant quality before they complete (or do not complete) the program.
Furthermore, the dataset includes all applicants to the program. I not only can track outcomes for
individuals who have participated in the program, but also the outcomes for individuals who were
not selected. The nature of the data provides leverage in examining latent candidacy across appli-
cants over time as well as the impact of program participation. Descriptive statistics for the dataset
are provided in Table 2.1. From 2003 to 2015, 530 individuals have identiﬁed as white (69.6%
of applicants), 174 individuals have identiﬁed as Black (22.9% of the applicants), 13 individuals
have identiﬁed as Latino (1.7% of applicants), 15 have identiﬁed as Asian (2% of applicants), 17
have identiﬁed as other (2.2%), 5 have identiﬁed as Arab American and 7 have identiﬁed as Native
American (both less than 1% respectively). Table 1 also shows that there are no stark di erences
in the racial demographics of individuals who were accepted in the program and those who were
not. In fact, the racial demographics of all program applicants is about on par with Census data for
Michigan - also depicted in the table.

Table 2.1 Comparison of MPLP Racial Breakdown to Michigan Census Data

Racial Group
(Percentages)

Census Data
for Michigan

V. 2015 (Estimated) V. 2010 (Actual)

MPLP Racial
Demographics

Not Accepted

White
Black or African American
American Indian and Alaska Native
Asian/Paciﬁc Islander
Latino
Arab American *
Other

75.6
14.2
0.7
3.0
4.9

76.6
14.2
0.6
2.4
4.4

Accepted
71.67
21.16
0.68
2.39
2.05
0.34
1.71

68.38
23.93
1.07
1.71
1.5
0.85
2.56

*I include Arab American as a racial category here because in Michigan it is a salient identity. The largest Arab
American population in the United States resides in Michigan.

23

While the demographic proﬁles of MPLP applicants do not present real di erences from other
members of the public, there are some strong di erences in racial and gender demographics between
MPLP applicants and the composition of the Michigan legislature as depicted in Figure 2.1.3 In
comparing the composition of the Michigan Legislature to MPLP applicants, there is certainly
more representation of minority groups and women among the MPLP applicants. Although this is
a descriptive representation of the data, Figure 2.1 speaks to what scholars are missing when only
drawing inferences about candidate emergence from o ceholders. Furthermore, the fact that there
is greater representation of minority groups in the MPLP applicant pool is encouraging and speaks
to the evidence from (Lawless, 2012) about the political ambition of minority groups, Blacks in
particular.

Although the process of completing a political training program is certainly meaningful, the
process of applying to one of these program provides some suggestion that an individual is interested
in politics - motivated to some degree. I show this descriptively with the MPLP participant sample
in Figure 2.2. Program applicants show interest in running for o ce across racial groups. More
speciﬁcally, over 50% of applicants across groups say that they want to run when asked on MPLP
application materials. The same can be said across genders. Both men and women suggest that they
are interested in running for o ce more often than not. This suggests that not only do applicants
understand the purpose of the program, but also are seeking the program to accomplish the outcome
of running for o ce. This further suggests that political training programs may be used as a means
to identify potential candidates for o ce.

Moving forward, I assume that the process of applying to a political training program accounts
for at least one aspect of motivation to participate - political interest. Previous work has similarly
examined political training outcomes for participants in comparison to non-participants and suggests
that political training program participants are more likely to have knowledge about politics and
methods of political participation as well as feel that they can run for o ce (Stock, 2012). This
3This comparison is based on demographic characteristics of the 2015 Michigan Legislature

according to the National Conference of State Legislatures

24

Figure 2.1 Gender and Racial Composition of MPLP Applicant Pool

MPLP Applicants

MI Legislature

  White 

 Black

 Multiracial

Am Ind/Nat Alaskan

Asian/Pac Islander

Latino

Ethnicity

MPLP Applicants

MI Legislature

80

60

40

20

0

80

60

40

20

0

t

n
e
c
r
e
P

t

n
e
c
r
e
P

80

60

40

20

0

80

60

40

20

0

Men

Women

Gender

would suggest that political training programs inﬂuence levels of information and political e cacy,
but an individual must ﬁrst be interested in running for o ce to glean those beneﬁts.

Empirical Analysis

I turn to the empirical analysis to examine the relationship between program acceptance and
indicators associated with political participation (motivation, capacity and encouragement) in Table

25

Figure 2.2 Desire to Run for O ce Across Racial Groups

Other

Black

Hispanic

Native American

White

Arab American

Asian

No

Yes

Unsure

No

Yes

Unsure

No

Yes

Unsure

Desire to Run

Women

Men

0.8
0.6
0.4
0.2
0.0

0.6

0.4

0.2

0.0

t

n
e
c
r
e
P

t

n
e
c
r
e
P

0.8
0.6
0.4
0.2
0.0

0.8
0.6
0.4
0.2
0.0

0.6

0.4

0.2

0.0

No

Yes

Desire to Run

Unsure

2.2. In analyzing the data, I use a matching approach.4 In particular, I want to reduce any potential
bias or variance that may simply be a product of the sample (Ho et al., 2007), especially given
my interest in examining outcomes for women and minorities. Speciﬁcally, I use coarsened exact
matching because of the straightforward nature of the approach and the evidence in favor of this
method reducing imbalance in the data (Iacus, King and Porro, 2012).

I match applicants on gender, race, education, and the applicants’ aggregate interview score.
4I provide a discussion of the matched sample and balance statistics in the Appendix as Table

A.1 and Table A.2.

26

Matching on the interview score is particularly important because applicants with the highest
interview score are not necessarily admitted into the program. MPLP emphasizes balance on
factors like gender, partisanship, race and geography. Time and commitment as well as the rule
against running while in the program, are also considered in the decision to accept an applicant. This
means that an individual with a stellar application and high interview score might not be admitted.
The person admitted might have a low or midrange interview score and average recommendations.
The advantage that the less than stellar candidate has might be location or partisanship. The
program seeks to have applicants from all over the state and requires gender as well as partisan
parity. In addition, this focus on balance means that training and encouragement might matter even
more in this context. MPLP is not accepting the participants who will necessarily win an electoral
contest, but the ones who ﬁt their balance criteria.

The dependent variable, or the “treatment", in the matching model is program acceptance. That
is, I am mitigating the di erences between accepted applicants and non-accepted applicants to make
a causal claim about participation in MPLP. I ﬁrst present results using all of the observations in
the dataset and then presents results on the matched subset of the data. While using the full sample
will allow for more generalizable claims about what the presence of a training program means for
representative outcomes in the Michigan context, using the matched subset of the data will allow
for a quasi-causal test of whether participation in MPLP matters or provides its participants with
some set of skills that makes them more likely to seek o ce.

I next use logistic regression to model the likelihood of running for o ce. The baseline model
for both the unmatched and matched data (shown in Table 2.2 as Model 1 and 2 respectively)
include the interview score, race,5 gender, education, age, and desire to run for o ce.6 The
baseline models account for how demographic characteristics and sheer political interest might
5I dichotomized race to include all persons of color as 1 and whites as the comparison case. A
model with race including categories for Whites, Blacks, Asians, Latinos, Native Americans, and
Others is included in the Appendix as Table A.5. I chose to use a dichotomous measure here to
maintain a reasonable sample size in the matched sample. Additionally, the racial group di erences
in running for o ce, in comparison to whites, remain non-signiﬁcant.

6Desire to run for o ce is dichotomized here and in subsequent models

27

Table 2.2 The Impact of Political Training Programs on Running for O ce

Accepted

Prior O ce

College

Higher Degree

POC

Involved in Community Group

Involved with Party

Female

Interview Score

Desire to Run

Age

Year Fixed E ects

Constant

Observations
Log Likelihood
Akaike Inf. Crit.
Note:

Dependent variable: Ran for O ce

Unmatched (1) Matched (2)
1.047⇤⇤⇤
(0.226)

1.000⇤⇤⇤
(0.218)

Unmatched (3) Matched (4)
1.451⇤⇤⇤
(0.346)

1.214⇤⇤⇤
(0.318)

1.695⇤⇤⇤
(0.294)

1.790⇤⇤⇤
(0.321)

0.378
(0.212)
 0.362
(0.301)
 0.296
(0.214)

 0.181
(0.196)

0.007
(0.005)

0.582
(0.323)

0.044⇤⇤⇤
(0.009)

0.420
(0.224)
 0.431
(0.335)
 0.272
(0.239)

 0.100
(0.212)

0.004
(0.006)

0.691⇤⇤
(0.347)

0.046⇤⇤⇤
(0.009)

0.371
(0.234)
 0.236
(0.328)
 0.127
(0.235)
 0.499
(0.347)

0.900⇤⇤⇤
(0.274)
 0.059
(0.218)

0.006
(0.005)

0.888⇤⇤
(0.403)

0.032⇤⇤⇤
(0.010)

0.364
(0.250)
 0.369
(0.370)
 0.134
(0.265)
 0.639
(0.386)

0.970⇤⇤⇤
(0.294)

0.032
(0.238)

0.0001
(0.007)

1.057⇤⇤
(0.444)

0.034⇤⇤⇤
(0.011)

X

X

 4.041⇤⇤⇤
(0.646)

 3.981⇤⇤⇤
(0.767)

 4.314⇤⇤⇤
(0.795)

 4.411⇤⇤⇤
(0.955)

586

501

 285.696
619.392

 247.287
542.573
⇤⇤p<0.05; ⇤⇤⇤p<0.01

586

 324.701
667.403

501

 285.402
588.804

28

inﬂuence the decision to run for o ce given that the literature suggests that some groups are more
likely to express political ambition than others (Lawless and Fox, 2005). In the case of the baseline
model for the unmatched data, the signiﬁcant indicators for the observed behavior of running for
o ce are age and being accepted into the program. Gender, race, education, interview score, and
desire to run for o ce are not signiﬁcant here. For the matched model, age and being accepted into
the program is also signiﬁcant. The matched model di ers from the unmatched model in that the
desire to run for o ce is signiﬁcant.

Given that there is some evidence for MPLP providing its participants with motivation to run
for o ce, I now turn to Models 3 and 4 to assess capacity and encouragement in the context of
the full model. Models 3 and 4 include the baseline variables along with an indicator for having
previously served in o ce, political party involvement, community group involvement, and time
ﬁxed e ects.7 The inclusion of the prior o ce variable gets at the impact that having previously
served in o ce has on the decision to run. This is especially pertinent given that candidates
are more likely to enter an electoral race when they perceive they are likely to win (Maisel and
Stone, 1997). Including that variable in the model with program acceptance might speak to the
e ect that political training program participation is actually having on encouraging participants
to run. To account for capacity, I include dichotomous measures of political party involvement
and community group involvement. Party involvement and community group involvement fall into
classic deﬁnitions of civic engagement. Verba et. al Verba et al. (1995) suggest that it is through
this type of participation that individuals gain civic skills.

The signiﬁcance and direction of the coe cients in model 3 and 4 are consistent despite the
use of matching.8 Once again, being accepted into the MPLP program is a positive predictor
7The ﬁxed e ects are not substantively interesting for this analysis, but are included because
the political climate in a given year may encourage or discourage potential candidates to run for
o ce. In Table 2.2, ﬁxed e ects are denoted with a checkmark. The full results, with ﬁxed e ects
denoted in the table, are available in the Appendix as Table A.4

8The likelihood ratio test for the baseline models in comparison to the full models (both
matched and unmatched respectively) suggest that the additional variables signiﬁcantly contribute
to the model.

29

of running for o ce as desire to run for o ce. Education, race, interview score, and gender
remain non-signiﬁcant predictors of whether an applicant runs for o ce. This seems to indicate
that perceptions of likelihood to run and assessment of program ﬁt in the interview process is
not a good indicator of who actually runs. In terms of capacity, community group involvement
is negatively related to running for o ce, but is not signiﬁcant in either of the models. Party
involvement, on the other hand, is positively related to running for o ce and is signiﬁcant in both
models. This makes sense given the discussion of the role of the party in the state politics literature.
Although that discussion largely relates to how the party can make or break the candidacy of women
(Sanbonmatsu, 2010; Niven, 1998; Carroll and Sanbonmatsu, 2013), being in the party network
appears to be an important factor in o ce seeking behavior (Crowder-Meyer, 2013; Karpowitz,
Monson and Preece, 2017; Preece, Stoddard and Fisher, 2016). This indicates that all of forms of
civic engagement are not created equal in encouraging candidate emergence.

I turn to the predicted probabilities from the matched model (model 4) to further tease out the
e ects of the signiﬁcant variables on running for o ce. In particular, I examine the discrete change
in predicted probabilities in Figure 2.3.9 When considering the e ect of participation in MPLP,
there is a signiﬁcant di erence in the probability of running for o ce between accepted and non-
accepted applicants. More speciﬁcally, a person who completes a program like MPLP has about
a 0.16 higher probability of running for o ce. The prior o ce variable also shows a signiﬁcant
di erence. Individuals who have previously held o ce have about a 0.22 higher probability of
running for o ce than those who have not. On other hand, there is no signiﬁcant di erence in the
probability of running for o ce as it pertains to individuals who do and do not express a desire
to run. Similarly, political party involvement does not increase the probability that an individual
will run for o ce in comparison to those who do not participate in that activity. All together, this
9The variables for being accepted in the program, prior o ce, desire to run, race, gender,
education, being involved in the community and being involved in the party are categorical variables.
Thus, I estimate the discrete change in the predicted probabilities when each of these variables take
on the value of 0 and 1, respectively. The variables for interview score and age of applicants are
continuous. Thus, I estimate the discrete change for a centered standard deviation increase (plus
and minus one-half standard deviation around the median).

30

Figure 2.3 Discrete Change in Predicted Probabilities

y
t
i
l
i

b
a
b
o
r
P
n

 

i
 

e
g
n
a
h
C
d
e

 

i

t
c
d
e
r
P

0

.

1

5

.

0

0

.

0

.

5
0
−

0

.

1
−

Accepted Prior Office

Gender

POC

Community

Party

Education

IW Score

Desire

Age

appears to suggest that training program participation is providing some beneﬁt to its participants
such that they feel prepared to run and do indeed run for o ce. The fact that individuals who
have held prior o ce are likely to run is not surprising. There is a long literature on progressive
political ambition (see Fowler, 2010) that would suggest that ambitious politicians use lower o ce
as a stepping to stone to higher o ce. The fact that individuals who have held prior o ce are
interested in completing a training program may speak to potential network and resource beneﬁts
of training programs that are not explicitly captured here.

Finally, I contend with potential interaction e ects with gender and race in the Table
Despite the fact that the product term is not signiﬁcant in either model in Table 2.3, this does
not rule out the potential that the second di erence is signiﬁcant (Ai and Norton, 2003; Norton

31

Table 2.3 The Impact of Political Training Programs on Running for O ce: Matched Interaction Models

Dependent variable: Ran for O ce

Gender Interaction Model

Race Interaction Model

1.394⇤⇤⇤
(0.368)

1.789⇤⇤⇤
(0.320)

0.030
(0.239)
 0.234
(0.348)
 0.653
(0.387)

0.969⇤⇤⇤
(0.294)

0.362
(0.251)
 0.366
(0.369)

0.0002
(0.007)

1.040⇤⇤
(0.446)

0.034⇤⇤⇤
(0.011)

0.237
(0.528)

X

 4.358⇤⇤⇤
(0.961)

501

 247.186
544.371

⇤⇤p<0.05; ⇤⇤⇤p<0.01

Accepted

Prior O ce

Women

POC

Involved in Community Group

Involved in Party

College

Higher Degree

Interview Score

Desire to Run

Age

Accepted x Women

Accepted x POC

Year Fixed E ects

Constant

Observations
Log Likelihood
Akaike Inf. Crit.
Note:

1.311⇤⇤⇤
(0.407)

1.788⇤⇤⇤
(0.321)
 0.096
(0.310)
 0.134
(0.266)
 0.646
(0.386)

0.972⇤⇤⇤
(0.294)

0.362
(0.251)
 0.347
(0.371)
 0.0002
(0.007)

1.051⇤⇤
(0.444)

0.034⇤⇤⇤
(0.011)

0.302
(0.466)

X

 4.327⇤⇤⇤
(0.966)

501

 247.076
544.152

32

et al., 2004). In Figure 2.4, I compute the marginal e ect of the change in the product term for
race and being accepted into the program as well as gender and being accepted into the program.
The graphs on the left show the magnitude of the interaction e ect for race and gender. The e ect
appears to be quite small for both race and gender. In looking at the signiﬁcance of the interaction
e ects, both appear to be non-signiﬁcant. Thus, this would suggest that the product term is not
meaningful for the model. Substantively, this ﬁnding indicates that race and gender are not not
barriers for potential o ceholders who complete programs like MPLP.

Figure 2.4 Interaction E ects for Race and Gender

e
c
a
R

 
r
o
f
 
t
c
e
f
f

E
 
n
o
i
t
c
a
r
e
t
n
I

r
e
d
n
e
G

 
r
o

f
 
t
c
e

f
f

 

E
n
o

i
t
c
a
r
e

t

n

I

0
.
1

5
.
0

0
.
0

5
.
0
−

0
.
1
−

0
.
1

5

.

0

0

.

0

5

.

0
−

0

.

1
−

0.0

0.0

c
i
t
s
i
t
a
t
S
Z

 

3

2

1

0

1
−

2
−

3
−

1.0

0.0

0.4

0.2
0.8
Predicted Pr(Running for Office)

0.6

0.4

0.2
0.8
Predicted Pr(Running for Office)

0.6

c
i
t
s
i
t

t

a
S
Z

 

3

2

1

0

1
−

2
−

3
−

0.4

0.2
0.8
Predicted Pr(Running for Office)

0.6

0.4

0.2
0.8
Predicted Pr(Running for Office)

0.6

1.0

0.0

33

1.0

1.0

Robustness Check

While using matching can be instructive in making causal claims about relationships in observa-
tional data, the use of the method can reduce the sample size for further analysis because some
observations do not have a match and are subsequently not included in the matched data. For this
reason, I bootstrap the matched model in order to ensure that the inferences I make here are not bi-
ased by the size of the sample. In particular, I use random-x resampling, which selects bootstrapped
samples of the observations, ﬁts a model, and then determines the standard errors from the bootstrap
distribution. I present the model as Figure 2.4. I resample data from the matched model 1000 times
and use bias corrected and accelerated conﬁdence intervals to determine coe cient signiﬁcance.10
The bootstrapped model appears to approximate the full matched model. The coe cients that were
signiﬁcant in the matched model, being accepted into the program, having held prior o ce, and
desire to run for o ce, maintain their signiﬁcance in the bootstrapped model. All other variables
are non-signiﬁcant. This would suggest that the speciﬁcation in the original model is capturing an
e ect that is not an artifact of sample size and that training program participation is meaningful.

Discussion

While the literature on political ambition has progressed, there is still much to be learned about the
path from the pool of eligible candidates to elected o ce for current and potential o ceholders.
The most recent research has moved toward looking at individuals in the pre-emergence stage,
but the study of individuals that participate in candidate training programs has not been fully
considered. Political training programs present a meaningful way to look at latent candidacy
particularly because of the di culty in identifying all potential candidates for electoral races. We
can certainly identify incumbents and their likelihood of running for the same position again, but
challengers are a di erent story because their decision process to enter the race is unknown and
10Bias corrected and accelerated conﬁdence intervals correct for skewness in the bootstrap

distribution and gives the conﬁdence interval on the original scale (DiCiccio and Efron, 1996).

34

Table 2.4 Bootstrapped Full Model (Matched)

Dependent variable: Ran for O ce
Bias
-0.478
0.108
0.154
0.006
-0.017
-0.054
0.064
0.034
-0.028
0.004
0.152
-0.645

Intercept
Accepted
Prior O ce
Gender
Race
Involved in Community Group
Involved with Party
Education: College
Education: Higher Degree
Interview Score
Desire to Run
Age
Year Fixed E ects
Note:

Original
-4.654⇤⇤
1.477⇤⇤
1.838⇤⇤
2.465
-1.456
-6.464
9.829⇤⇤
3.635
-3.846
-1.793
1.309⇤⇤
8.894

X

X

⇤⇤p<0.05

Std. Error

1.240
0.382
0.381
0.266
0.298
0.435
0.269
0.278
0.442
0.016
0.572
3.371

X

more simply we do not know who these individuals are. The study of political training programs
could provide further insight into candidates’ decision to enter the electoral arena as well as better
specify what types of people are likely to run for o ce.

This work makes a couple of important contributions to the current literature. First, I theoret-
ically situate the discussion of political training programs in the political participation literature
using Verba, Scholzman and Brady’s theory of participation as a framework. Running for political
o ce is most certainly a form of political participation and should be treated as such. By my
account and that of previous work, political training programs account for the main tenants of the
theory - capacity, motivation, and recruitment - that lead to political engagement. However, just
what activities, above and beyond the contribution of political training programs, lend to o ce
seeking behavior still requires further examination.

Second, I show that participation in a political training program is positively associated with the
observed behavior of running for o ce. This result stands with matched data (a quasi-experimental
Indeed, the examination of MPLP applicants presents a unique
context) and unmatched data.

35

sample of individuals, who for the most part, express interest in running for o ce. I present these
individuals as latent candidates because of their political aspirations and the fact that the desire to
run is positively associated with running for o ce. Although desiring to run does not signiﬁcantly
predict o ce seeking behavior, Figure 3.1 shows that it is the individuals who have participated in
the program that are more likely to enter an electoral race.

Finally, I present an interesting puzzle in my ﬁndings. While I was able to show a relationship
between participating in the Michigan Political Leadership Program and the expression of political
ambition (running for o ce), how participation in a program like MPLP inﬂuences the decision to
run for o ce remains a question. More speciﬁcally, my analysis here presents MPLP as a black
box and future work should focus how the inner workings of political training programs inﬂuence
outcomes like running for o ce. Further research is certainly necessary to understand what factors
are meaningful for members of minority communities to seek o ce. Sanbonmatsu 2015 provides
some qualitative evidence that the decision to seek o ce, at least among minority women from
candidate training programs, is inﬂuenced by the ability to build support and resources to run.
Understanding how these factors may be external to or even associated with program participation
however might be useful particularly because many candidate training programs target women
and minorities for public o ce. The success rate of these programs might speak to where these
candidates in particular emergence and what o ces the choose to run. Further examination of
training programs might also prove useful in better understanding the paths to candidacy for
women and minorities.

36

EXAMINING THE CANDIDATE ELIGIBILITY POOL USING FIELD EXPERIMENTS

CHAPTER 3

Introduction

Within the literature, there are assumptions about who runs for o ce. Traditionally, individuals
who belong to professions involving law, business, or education have been considered part of this
eligibility pool (Dolan and Ford, 1997; Moncrief, 1999). However, this view is largely based
on the observed winners of electoral races, members of legislatures. It does not account for the
professional and educational background of the electoral losers. What’s more, this assumption
about who runs for o ce underlies many of our ﬁndings in the candidate emergence literature.

Some scholars have sought to move beyond this traditional candidacy pool. For instance,
Fox and Lawless (2004) have indicated political activists as members of the pool. Carroll and
Sanbonmatsu (2013) suggest that the candidate pool has expanded for women. Their qualitative
study indicates that the common experience among the interviewed women in legislators is strong
community involvement and civic engagement. In addition, the study conducted by Sanbonmatsu,
Carroll and Walsh (2009) implies that women do not necessarily follow the same career paths as
men on the road to political candidacy. Thus, this would suggest that rethinking the candidate
pool is necessary to better understand who runs for o ce. This is where candidate and leadership
training programs might play an important role.

I focus on the role of candidate and leadership training programs as a means of identifying
latent candidates (individuals who are interested in running for o ce). I argue that by seeking to
participate in one of these programs, individuals are showing interest in political o ce and therefore
exhibiting political ambition. In chapter 2, I used the Michigan Political Leadership Program as
an illustration of how applicants to the program, across racial groups, by and large show interest
in running for o ce. Continuing with the Michigan Political Leadership Program as an example,
in Figure 3.1, I show the professions of the applicants. This graph adds credence to the idea that

37

latent candidates are not necessarily who we would traditionally think of as potential candidates.
The majority of the applicants have careers that are not related to business1, law or education.
This other category encompasses a number of positions like local and county level o ceholders,
government o cials, non-proﬁt workers, etc. All together, this suggests that we should look beyond
the traditional candidacy pool in the conversation about political candidacy.

Figure 3.1 Professions of Applicants to the Michigan Political Leadership Program

s
t

n
a
c

i
l

p
p
A

 
f

o

 
t

n
e
c
r
e
P

0.6

0.4

0.2

0.0

Business

Education

Law

Legislative Staff Other

Profession

I designed two ﬁeld experiments that look beyond the traditional candidacy pool in targeting
potential latent candidates. These ﬁeld experiments seek to answer two questions: Who counts as
a latent candidate and does it matter who asks potential candidates to take a step toward political
1I determined who belonged in this profession category by counting the number of applicants
who described themselves as a business owner, president, chief executive o ce, chief ﬁnancial
o cer, or vice president of an organization

38

o ce? In partnership with the Michigan Political Leadership Program, I invited active voters from
the state of Michigan to apply to the program. I randomly varied the content of the message voters
received to examine how the message might matter for recruitment. In the second experiment, I
advertised the Michigan Political Leadership Program to Facebook users and asked them to invite
a friend to apply. I randomly varied the message that the invited friend received about the program
and why they should apply. Although there were di erences in the response rate across groups in
the study, I cannot make any substantive conclusions about the impact of the treatments due to low
response rates.

Theoretical Background

The traditional eligibility pool still stands as a benchmark for who might be considered a latent
candidate.2 The accepted view is that candidates (both men and women) are likely to be lawyers,
businessmen or women, and educators (Moncrief, Squire and Jewell, 2001; Fox and Lawless, 2004),
yet this view is largely based on the observed winners of electoral races. The fact that scholars base
their assumptions about latent candidates on individuals who tend to run for o ce may mean that
the candidate emergence literature is under identifying the segment of the population who has an
interest in public o ce.

In the ﬁrst experiment, I target active voters using the State of Michigan Qualiﬁed Voter
Files (QVF). I target these individuals because Verba, Scholzman and Brady’s 1995 framework of
political participation suggests that people who are active in political activities are those who have
capacity (time, money, and civic skills) and motivation (political interest, information and e cacy).
According to the third prong of the Verba et. al framework, these politically active individuals just
need to be recruited or encouraged to participate. By identifying the individuals who are active
2Although, there has been some indication that the pool should be expanded. Sanbonmatsu
2010, in particular, suggests that convention delegates, sta  members for politicians, leaders in civic
and community groups, and members of appointed boards and commissions should be included in
the eligibility pool as well.

39

voters in both state and local elections, I am expanding the universe of potential latent candidates
given that these individuals are likely to be motivated as deﬁned by the Verba et. al framework.

One factor that has emerged as a clear indicator for women’s decision to seek political o ce is
being asked or encouraged to run(Fox and Lawless, 2010). Indeed, much of the literature has placed
emphasis on the role of political actors in candidate recruitment. Previous studies have shown how
the party, in particular, plays an important role in the candidate recruitment process whether it to
be the beneﬁt or detriment of potential candidates(Crowder-Meyer, 2013; Niven, 1998; Carroll and
Sanbonmatsu, 2013; Karpowitz, Monson and Preece, 2017; Preece, Stoddard and Fisher, 2016).
However, even when women are recruited by the party, they tend to be skeptical of how much the
party will actually do to support their candidacy (Butler and Preece, 2016).

There is similar evidence that other political actors can also have an impact on encouraging
potential candidates (Broockman, 2014b); however, it remains unclear to what extent the identity of
the "recruiter" matters. That is, is a woman just as likely to respond to recruitment or encouragement
appeals to run for o ce from non-political actors? This question not only has theoretical importance
to the gender and politics literature, but also practical importance as there are a number of non-proﬁt
organizations that are operating with the message that the gender imbalance in government can
start to be remedied when we all "ask a woman to run". There are even organizations that create
scripts and venues through which women can recommend other women to run for o ce.

Although the literature would suggest that we should be asking political actors to encourage
women to run (Karpowitz, Monson and Preece, 2017, see), there is reason to believe that leveraging
the networks of potential candidates might be meaningful. Work in the political networks arena
suggests that networks play an important role in political behavior. Not only does the network
inﬂuence ordinary citizens in terms of information ﬂow and vote choice (Huckfeldt and Sprague,
1995; McClurg, 2006; Nickerson, 2008), but also inﬂuences the choices of political actors (Fowler,
2006a,b). There is even evidence that one’s network can change one’s political behavior (Sinclair,
2012). While political networks have been examined in the context of basic political acts (like
voting and donating to political organizations), political networks have not been examined in the

40

context of running for o ce.

In the second experiment, I examine how who acts as a recruiter matters in motivating potential
candidates to express political ambition. This experiment relies on individuals who show interest
in attending a candidate training event. It is my contention that political training programs are a
mechanism for identifying latent candidates because these individuals are demonstrating interest
in public o ce. Instead of directly targeting latent candidates, I attempted to target the friends of
and family of latent candidates. By clicking on a Facebook ad, the Facebook user was directed to
a page to provide information about the person they would like to recommend. The recommended
individual then receive an email about a candidate training event and how to register for the event.
Field experiments have long been used as a means to mobilize voters (Gerber, Green and
Shachar, 2003; Gerber, Green and Larimer, 2010; Nickerson, Friedrichs and King, 2006; Green,
2004). Overall, the literature suggests that when voters receive an appeal to vote (whether it be by
phone or direct mail) they tend to do so. Voting is just one form of political participation though.
While there is evidence that factors such as socioeconomic status and education matter in whether
one participates or not, the literature has not fully addressed if participation in one activity leads to
participation in others (Leighley, 1995). That is, it could be the case that an active voter might also
attend local organization meetings, and/or be interested in running or o ce. Given the fact that
being asked to run for o ce does factor into the decision to do so (Huckshorn and Spencer, 1971;
Seligman et al., 1974; Moncrief, Squire and Jewell, 2001; Maestas et al., 2006; Broockman et al.,
2014), it could be the case that potential candidates can be mobilized in a similar way as mobilized
voters. There is some indication that this logic is right. Broockman2014b ﬁnds that presenting
encouraging appeals to activists can increase interest in running for o ce. Similarly, Preece,
Stoddard, and Fisher 2016 demonstrate that individuals active in the political party do respond to
elite recruitment, albeit in gendered ways. In an attempt to rethink the political candidacy pool, I
follow the lead of recent work (see Broockman, 2014b; Preece, Stoddard and Fisher, 2016; Preece
and Stoddard, 2015) that has used ﬁeld experiments to mobilize potential candidates.

41

Experiment 1

In order to better understand the boundaries of latent candidacy, who counts as a potential candidate,
the ﬁrst experiment involves making contact with active voters in the state of Michigan using the
Qualiﬁed Voter File (QVF). By using the QVF, I have access to the widest potential eligibility
pool for interest in political o ce, which includes 381,362 voters across the state of Michigan.3
Active voters were randomly assigned to one of three experimental groups (the Control condition,
the Encouragement condition, and the Gender Appeal condition).

Individuals assigned to the groups were sent one mailing during the recruitment period for
the Michigan Political Leadership Program. For the Control and Encouragement conditions, men
and women were equally sampled to receive a message and included a sample of 1400 and 800,
respectively. For the Gender Appeal Treatment, participants in the treatment condition were
identiﬁed using the gender designation in the QVF and included 800 participants. Active voters
were sampled for no more than one condition. Active voters, randomly assigned to one of the
groups, were mailed a postcard and only postcard was sent per household. There were multiple
outcome measures to capture potential interest in running for o ce. These outcomes measures
include: completing the program application, calling to learn about the program4, visiting the
designated website associated with a given experimental group, or sending an email about the
program using one of the designated email addresses.

Treatments and Hypotheses

The mailed postcards included three main parts: the headline (which varied across experimental
groups), the body of the message (which was consistent across groups), and modes of contact (a
website, and email address, and a phone number - which all varied slightly based on the condition).5
3Active voters are deﬁned as individuals who voted in both state and local elections in 2015
and 2016. This excludes the 2016 Presidential election as the experiment was conducted prior to
November.

4Participants were asked about the postcard they received when they called.
5See Appendix B.1, B.2, and B.3 for the full text of the postcards.

42

The headline of each postcard varied in the following way for each experimental group:

Control (C): Do you want to have an impact on Michigan Politics? Learn the skills to run for
o ce.

Treatment 1 (T1): We have identiﬁed you out of 7 million Michigan voters. You’re in the top
5% most politically active and engaged.

Treatment 2 (T2): Women are half the state, but only 20% of the legislature. Isn’t it time for
women to take the lead? We need you!

In terms of expectations for the outcomes of the experiment, based on the experimental groups,
it might be expected that active voters will be likely to respond to leadership recruitment appeals
given the literature and that recruitment messages suggesting that these participants are needed in
politics will be more appealing. Following this logic, I hypothesize that presenting encouraging
appeals to active voters will increase in interest in political candidacy in comparison to the control
condition. Relatedly, I hypothesize that priming gender identity will signiﬁcantly inﬂuence interest
in an elected position among women in relation to the control condition.

Experiment 2

In the second experiment, I seek to capture how networks matter for latent candidates. Understand-
ing the networks of potential candidates is particularly important because of the power that social
networks have been shown to have on political decision making and participation among ordinary
citizens (Sinclair, 2012). More speciﬁcally, if your social circle can encourage your vote choice,
then it may also be able to encourage your decision to run for o ce. In that spirit, the purpose of
this experiment is examine how “who” acts as recruiter matters in motivating potential candidates
to express political ambition.

43

Participants were recruited via Facebook using the ad in Figure 3.2. Facebook allows for
targeted messages based on demographic characteristics and locations. I targeted Facebook users
in the state of Michigan. The ad reached 40,584 people (16,925 women and 23,0003 men).

Figure 3.2 Facebook Ad

Upon clicking on the ad, Facebook users were directed to a page in which users were prompted
to provide their name and email address as well as the name and email address of someone they
would like to recommend and send an email to about running for o ce. The recommender was also
given the opportunity to provide a description of why they think the recommended person should
run. The screen appeared as seen in Figure 3.3. While the ad reached 40,584 people, 670 people
clicked on the ad (299 women and 363 men). Of those Facebook users who clicked the ad, only 33
users completed the recommendation page to have a message sent to a friend.

44

Figure 3.3 Recommendation Page

The recommended persons served as the sample group for the experiment. The sample was
randomly placed in two groups a control group and a treatment group. The control group received
an email about the Michigan Political Leadership Program addressed from the program. The
treatment group received a personalized invitation addressed from their recommender.

Treatment and Hypotheses

the email subject line (which varied between
The emailed message included four main parts:
groups), the personalized section of the email (varied between groups), the body of the email
(remained consistent across groups), and mode of contact to learn morn about the program (the
email address varied).6 The emails varied in the following ways for the experimental groups:

Control (C):

6See Appendix B.4 and B.5 for the full text of the emails.

45

Subject Line: Your Friend thinks you have what it takes to be a Michigan political
leader

Personalized Section: Your friend, (Insert name here), thinks you should run for o ce.
If your friend, (Insert name here), thinks you have what it takes, then you should too.

Treatment (T):

Subject Line: You have what it takes to be a Michigan political leader

Personalized Section: I think you should run for o ce.
(Insert recommender’s reason why recommended person should run for o ce.) If I think
you have what it takes, then you should too.

The outcome measure here is whether or not the recommended individual expresses interest in
the Michigan Political Leadership Program by sending an email to learn more about the program.
I expect that people who receive an initial personalized message from a friend will be more likely
to show interest in the program.

Results

Table 3.1 (Experiment 1) Response Rates by Condition
Group
Control
Treatment 1
Treatment 2

Responded Received Response Rate

0.504
1.26 ⇤
0.761

1389
794
788

7
10
6

* Statistical signiﬁcance relative to the control at the .10 level, two sample z-test for proportions.

Table 3.2 (Experiment 2) Response Rates by Condition
Group
Received Response Rate
Control
Treatment

Responded

45.5
36.4

11
22

5
8

In Table 3.1 and Table 4.1, I report the response rates for the ﬁrst and second experiments.
The “responded" category captures the individuals who responded to the experimental stimuli.

46

The “received" category captures the actual number of postcards (and emails) that were sent and
received by participants. In Table 1, the “received" category does not reﬂect the initial number
of postcards that were sent out because 29 postcards were returned to sender. The “response rate‘
category reﬂects the number of people who responded to the experimental stimuli divided by the
total number of participants in the experimental group. Because of the low response rate, I do not
distinguish between types of responses here. However, I will note that the majority of the responses
were garnered via email. The low response rate for both experiments is not surprising. A number of
previous studies have shown the di culty in mobilizing citizens to engage in political behaviors like
voting (Gerber and Green, 2000; Gerber, Green and Shachar, 2003), making political contributions
(Miller and Krosnick, 2004), and showing interest in political o ce (Preece and Stoddard, 2015).
Although there are di erences in the response rate in both Table 3.1 and Table 3.2, the general
low number of responses is not amenable to drawing strong conclusions about causal e ects.
Nevertheless, I performed two sample z-tests for proportions comparing the control group and
treatment groups, respectively, from the ﬁrst experiment as well as the control group and treatment
group from the second experiment.7 The only signiﬁcant di erence in the response rate is between
the Control Group and Treatment 1 from the ﬁrst experiment. This provides some evidence that
the encouragement condition may have an impact; though, more evidence is certainly necessary
here. Moving forward, it would be worthwhile to test how the "encouragement" condition might
work in another experiment setting. That is, while active voters did respond, to some degree, to the
experimental stimuli, it appears that the active group is too big a group to consider as part of the
candidate eligibility pool.

In regards to the second experiment, it may be worth exploring this in another setting as well.
Facebook may not be the best space to conduct an experiment of this nature given the that the
experimental sample is dependent on an initial response to a prompt. Although, it should be noted
that the response rate in Table 3.2 is encouraging. The networks of potential candidates are an
7This method of analysis is ideal here because it allows me to consider the response rate as
a variable of interest and deals well with categorical data. T-tests, on the other hand, require
continuous data for estimation as well as multiple observations.

47

untapped resource in understanding who runs for o ce. For example, if your friend circle can
encourage your vote choice, then it can surely encourage or discourage your decision to run. The
same may be said for family given that women often cite familial considerations as a reason why
they choose not to run (Fox and Lawless, 2004).

Discussion

Although both of the ﬁeld experiments were not successful, non-signiﬁcant results can still tell
an important story here about what we can say about the candidate eligibility pool. It can likely
be said that the political ambition requires more than just being an active voter. Broockman
2014b performs a similar experiment with political activists and ﬁnds a signiﬁcant e ect for his
encouragement condition. Though he does not prime gender in any of his conditions, his ﬁndings
are instructive for how the Verba, Scholzman and Brady participation model matters for political
ambition. Broockman’s ﬁndings are encouraging for the idea that recruitment matters, but he also
provides something to thing about in terms of the motivation aspect. It is likely that individuals
who identify themselves as activists engage in politics more or in di erent ways than active voters.
This gives me something to think about moving forward in terms of how assumptions about the
candidate eligibility pool can matter for this type of endeavor.

This work does represent an important step in the right direction though. One of the most
interesting features of this current political climate is the surge in political candidates, both men
and women (see Dittmar, 2018b). In addition, the resounding narrative of many of these emerging
candidates is that they do not necessarily look like the candidates we have seen in the past (Holder,
2017). Whether this increase in political candidacy is on a wave of empowerment after the 2016
election or a response to the outcome of the 2016 election, there are ﬁrst time candidates running
at all levels of political o ce. Although the literature has considered the impact of non-traditional
o ceholders (see Canon, 1990), we do not know why these individuals choose to run for o ce
or why they choose to run when they do. Given the growing number of candidate and leadership

48

training programs as well as how much attention these organization are receiving, we may see more
candidates like this in the future.

Because these experiments were conducted before the 2016 election, I can only speculate about
the impact of the election on the emerging candidates that the literature has traditionally excluded
from the candidate eligibility pool. However, this does provide conceptual clarity about the impact
of the treatments in the experiment as opposed to the impact of the political environment. With
that being said, future work should model the inﬂuence of the election and the subsequent political
climate on latent candidates’ desire to run for o ce.

49

CHAPTER 4

TRAINING CANDIDATES: AN EXAMINATION OF HOW POLITICAL TRAINING

PROGRAMS INFLUENCE STATE LEVEL REPRESENTATION

Introduction

Each election cycle, candidates vie for the opportunity to represent their respective electorates in
state legislatures and municipalities. Over time, scholars have sought to understand who decides
to run for o ce and what motivates them to do so. Certainly political ambition matters for o ce
seeking behavior (Schlesinger, 1966), but a successful campaign for public o ce requires more
than just the desire to run. A potential candidate must have some information about important tasks
like getting on the ballot and running a campaign. However, even more important than knowing
how to get in the race is knowing when to get in the race.

The political candidacy literature has long suggested that quality challengers know when to
run. These o ce seekers are assumed to be rational and to recognize the risks and costs associated
with o ce seeking (Black, 1972; Rohde, 1979). For instance, quality challengers are less likely
to enter a race when an incumbent holds the seat and are more likely to run when they perceive a
high probability of winning (Bond, Covington and Fleisher, 1985; Canon, 1990; Jacobson, 2001;
Maisel and Stone, 1997; Jacobson, 1981). The quality of the incumbent matters as well given that
higher quality incumbents deter strong challengers (Mondak, 1995; Zaller, 1998).

Looking speciﬁcally to women’s political candidacy, the literature suggests that women engage
in strategic behavior as well. Like Black (1972) ﬁrst suggested for o ce seekers more generally,
women tend to consider the costs and the probability of winning in their decision to run for o ce
(Palmer and Simon, 2003). Factors like open seat elections and a history of women o ceholders
in the state matter for women’s candidate emergence (Ondercin and Welch, 2009; Ondercin et al.,
2005; Palmer and Simon, 2001). Furthermore, women tend to emerge in what Palmer and Simon
(2010) call women friendly districts. These districts tend to be democratic, urban, and home to a

50

higher educated and higher income population.

Taken all together, if o ce seekers are rational and strategically determine when and where they
enter the electoral arena, then the question remains where do candidates receive their information
and learn the rules of the electoral game. For incumbents, the answer to the question posed is
relatively easy. As political insiders, they have access to information and resources that the average
challenger would not. However, for challengers, the question posed is more complex, especially as
it pertains to women candidates.

Women are under represented at all levels of o ce. Though prior work on the electoral
success of women indicates that they are just as likely to win as men (Fox and Carroll, 2006),
the patterns of women’s electoral success varies across states (Sanbonmatsu, 2010; Wilcox and
Norrander, 2005) and the relative growth in women o ceholders overtime has plateaued (Hennings,
2011; Sanbonmatsu, 2010).
In addition, women’s journey to political candidacy involves more
considerations, ranging from personal factors (Fox and Lawless, 2004; Preece, 2016; Kanthak and
Woon, 2015; Fox and Lawless, 2011b) to how they are perceived by their party and the public (Niven,
1998; Sanbonmatsu, 2010; Crowder-Meyer, 2013; Carroll and Sanbonmatsu, 2013; Sanbonmatsu,
2002; Dolan and Sanbonmatsu, 2009).

Though it can be the case that ambitious career politicians at the local level are strategically
using lower o ces as a stepping-stone to higher o ce (Palmer and Simon, 2001; Maestas et al.,
2006), in this work, I focus on the role of candidate training programs in the candidate emergence
process.
I argue that candidate and leadership training programs can be a mechanism through
which women receive the information and resources to learn how to achieve electoral success. In
particular, I focus on how the presence of state level candidate and leadership training programs
inﬂuence the number of women who seek o ce in a state as well as how these programs prepare
candidates to be "strategic." I ﬁnd that women are more likely to run for o ce in states that have
training programs speciﬁcally geared toward women. In addition, individuals who participate in
training programs are more likely to act, as what the literature as traditionally described, as quality
challengers.

51

Structural and Personal Factors in the Decision to Run

As it pertains to women’s political candidacy, Fox and Lawless Fox and Lawless (2004) suggest
that potential candidates, or individuals with nascent ambition, base their strategic electoral entry
decision on such factors as being encouraged to run for o ce and feeling qualiﬁed to run for
political o ce. Similarly, self-e cacy also plays a strong role in women’s interest in political o ce
(Preece, 2016; Kanthak and Woon, 2015). In addition, their emergence as candidates is hindered
by personal and professional factors, like considerations of family and children, as well as whether
or not to leave their careers for public service (Fox and Lawless, 2011b).

In addition to the personal and professional factors, there are also external factors that can either
constrain or enable their ability to seek o ce. Beyond the presence of an incumbent in an election,
factors such as prestige of o ce and constituency composition have an inﬂuence on o ce seeking
behavior (Goodli e, 2001; Moncrief, Squire and Jewell, 2001; Maestas et al., 2006). In addition,
studies suggest that ﬁnances can also inﬂuence candidate entry (Gimpel, Lee and Kaminski, 2006;
Hamm and Hogan, 2008). Moreover, Hogan (2008) ﬁnds that the probability of a challenger
entering the race decreases when the district is larger, the incumbent’s margin of victory in the last
election is large, and in non-presidential election years.

The party is another factor that can make or break female political candidacy. Given that women
are more likely to run when they are encouraged to do so, the evidence that suggests male party
elites are likely to be biased against potential women candidates is problematic (Niven, 1998).
In addition, Sanbonmatsu (2010) indicates that state party structures in which party elites have
more control over the nomination process disadvantages women, because candidate recruitment
practices in these states largely favor male networks. This work also suggests that women may be
more strongly recruited in states in which parties have di culty in ﬁnding candidates for o ce.
Carroll and Sanbonmatsu (2013) further express the importance of party recruitment for female
candidate emergence in that they ﬁnd women were more likely to run for o ce if they were recruited
by party elites and more likely to cite being asked to run for o ce as their reason for o ce seeking.

52

The discussion of structural and individual factors that inﬂuence candidate emergence for
women lends itself to an analysis of what role candidate training programs play in state level
politics across states. The Verba et al. (1995) participation model ﬁts nicely here as it provides
a means of theorizing what beneﬁts candidate training programs can have for its participants. By
completing a training program, participants have already demonstrated their interest or motivation
to seek political o ce. Stock’s 2012 work adds to this as training program alumni are more likely
to report increased political activity, e cacy, and conﬁdence. Furthermore, there is an indication
that candidate-training programs can serve in the capacity of encouraging potential candidates to
run (Hennings, 2011) as well as identifying potential candidates for o ce and providing these
individuals with resources(Sanbonmatsu, 2015; Burrell, 2010). Therefore, it would appear that
candidate training programs account for all three aspects necessary to spur political participation -
capacity, motivation, and recruitment.

The study of candidate training programs is not only crucial because of the resources that
they provide to potential candidates and the fact that the number of these programs continue to
grow across the country, but also because the study of these candidate training programs may
inform the candidate emergence literature and candidate emergence process for women. While
the candidate emergence literature has made inroads in identifying latent candidates or candidates
with nascent political ambition (Fox and Lawless, 2004; Lawless and Fox, 2005), the literature has
not fully addressed the crucial step that changes nascent political ambition to expressed political
ambition. This point is critical due to the evidence that suggests that recruitment matters for women
candidates and that there are barriers to women political candidacy. In order to determine what
beneﬁts candidate training programs have for identiﬁed latent candidates, it must ﬁrst be understood
if candidate-training programs give their participants’ information and resource that set them apart
electorally.

53

Data

In this study, I examine two questions involving how candidate-training programs inﬂuence women’s
candidacy at the state level. First, I seek to examine whether women are more likely to be elected
in states with candidate-training programs. Second, I seek to examine how candidate-training
programs inﬂuence strategic behavior among former participants.
I rely on an original data
collection endeavor as well as data from the following sources: the Center for American Women
and Politics (CAWP), Party Competition and Conﬂict in State Legislatures Dataset1,the State
Partisan Balance Dataset2, data from The Correlates of State Policy Project3, and a measure of
state level acceptance of women’s political candidacy (Scott and Smidt, 2017).4.

The original data collection e ort here involved identifying and classifying candidate and
leadership training programs across states. I have identiﬁed 450 candidate and leadership training
programs. I have classiﬁed these programs based on the following criteria 5:

• Campaign Planning

• Campaign Strategy

• Voter Outreach

• Fundraising and Budgeting

• Networking
1Hinchli e, Kelsey L. ; Lee, Frances E. , "Replication data for: Party Competition and Conﬂict in
State Legislatures", http://dx.doi.org/10.15139/S3/12128 UNF:5:z2VoaixuEdmJZHDE8Qet0A==
State Politics Policy Quarterly [Distributor] V1 [Version]

2Klarner,

Carl,

2011’,

-
http://hdl.handle.net/1902.1/20403 IQSS Dataverse Network [Distributor] V1 [Version]

Balance Data,

Partisan

’State

2013,

1937

3Jordan, Marty P. and Matt Grossmann. 2017. The Correlates of State Policy Project v.1.14.

East Lansing, MI: Institute for Public Policy and Social Research (IPPSR)

4The is variable is worth noting for inclusion in the dissertation because it captures state level
public opinion about women as o ceholders. Using data from the General Social Survey, Scott and
Smidt advance a measure of women candidate support at the state level using multilevel regression
and post-stratiﬁcation

5Please see Table 1.1 in Chapter 1 for further explanation of the coding scheme.

54

• Public Speaking

• Governance

• Ethical Behavior and Leadership

This e ort is unique in that no other work classiﬁes training programs, at this scale, across
states based on what they o er to participants. Previous work has focused on a few programs to
understand program impact on self-reported political ambition and self-e cacy. In addition, the
only available quantitative example measures training program presence dichotomously with null
results (Hennings, 2011).

Figure 4.1 is a graphical depiction of how many of the classiﬁed programs fall into the coding
scheme previously described.6 A program feature score of 1 suggests that a program only addresses
one of the coded categories above, while a score of 8 suggests that the program addresses information
about all of the coded categories in its curriculum. As shown in the graph, there are a number of
programs that are self-deﬁned as leadership training programs and have a score of 8. This suggests
that these programs take on what would be considered campaign based curriculum (coded categories
1 -4) as well as what might be considered leadership focused curriculum (coded categories 5-8).
The number of campaign training programs that have a score of 8 is much lower.

The programs included in this analysis are deﬁned as serving either a campaign training purpose
or are focused speciﬁcally on women as candidates and leaders, respectively. This is not to say that
these two types of programs are mutually exclusive. The programs deﬁned as serving in a campaign
training capacity in this study are so deﬁned because they engage campaign planning, campaign
strategy, voter outreach and fundraising as part of the program curriculum (coded categories 1-
4). This campaign training variable is measured dichotomously. The programs that are exclusively
geared toward women are included in order to understand how women focused programs might

6This graph does not show all of the programs I identiﬁed.

I coded these programs based
on information from each program’s respective website. For programs that did not describe the
extend of it’s curriculum, the relevant category was marked "not applicable" or "not stated". Those
responses were excluded from the analysis.

55

Figure 4.1 Number of Programs That Fit Training Program Coding Scheme

Candidate Training

Leadership Training

20

40
Number of Programs

60

80

s
e
r
u
t
a
e
F
m
a
r
g
o
r
P

 

8

7

6

5

4

3

2

1

0

8

7

6

5

4

3

2

1

0

matter for candidate emergence among women. This variable is measured as a count of programs
in each state by year. I present a graphical display of both variables as Figure 4.2. The graph in
Figure 4.2 shows the growth of these campaign based curriculum programs and women focused
programs over time. Although my analysis does not extend into 2018, I show these additional years
in order to present a clear picture of how the training program landscape has changed and continues
to change. With that being said, there is noticeably more programs that are geared toward women
that have developed over time.

56

Figure 4.2 Emergence of Training Programs Over Time

57

The dependent variable in the model consists of a count of candidates by year and state. CAWP
has collected data on the count of women in state level elections, by state, from 1994 to 2016.7 The
data focuses on elections on even years, which excludes election year data for states like Louisiana,
Mississippi, New Jersey and Virginia.8 To account for those states, the data includes a count of
current women legislators that are not up for re-election for each state during the period of 1994 to
2016. The data also includes a count of women who ran and won during each election year. I use
variables from the Party Competition and Conﬂict in State Legislatures Dataset, the State Partisan
Balance Data and the Correlates of State Politics in order to capture state legislative competition,
legislative professionalism, partisan balance in state level governments, state party structure, and
other state level political demographics.9 My analysis focuses on the election period from 1994 to
2012 due to the available of the covariates in my model.

In regards to the second question under examination here, I focus on whether former participants
in candidate training programs are more likely to be strategic in their entry into the electoral arena.
I use data from the CAWP Recruitment Study (Sanbonmatsu, Carroll and Walsh, 2015) as it focuses
on individual level factors and choices of current legislators and city level executives. The main
independent variable is a dichotomous measure of whether or not the survey respondent participated
in a candidate-training program. The central piece in answering the question I pose is deﬁning
what counts as strategic behavior. The dependent variable is an additive index including measures
of strategic legislative behavior as deﬁned by the literature. The variables included in the index are
measures of factors that have played a role in the survey respondent entering the electoral race for
his or her ﬁrst political position.These factors include the presence of an open seat election, having
the ﬁnancial resources to run for o ce, whether the partisan composition of the district matches
the participant, and whether the previous o ceholder is from the same party as the participant, the

7 http://www.cawp.rutgers.edu/sites/default/ﬁles/resources/canwinleghistst.pdf
8 These states tend to hold elections on odd numbered years.
9In addition to measuring the percentage of a state that is considered urban, I also include
a measure of the percent of individuals in a state who identify as democrat, the public opinion
measure of acceptance of women’s candidacy, and a measure of state level policy liberalism as
constructed by Caughey and Warshaw (2018).

58

role of party support in the decision to run, and having prior political experience.

All of the items are scaled for the analysis such that the highest value is 1 the lowest value is 0.
When summed, the additive measure for "strategy" takes on values from 0 to 6. Individuals with a
higher score are more strategic and individuals with a lower score are less strategic. The six item
measure has a Cronbach’s alpha value of 0.37. I conducted factor analysis on the items as well and
the model explains 58% of the variance.

To be clear, although there are a number of training programs that are geared toward women,
men participate in training programs as well. Figure 4.3 displays the number of men in comparison
to women who have participated in a training program. Women certainly utilize these programs
more than men do; however, this graph suggests that conclusions drawn from the relationship
between strategic behavior and training program participation may not just be gender speciﬁc. That
is, women and men might be using these programs to learn the rules of the electoral game.

Empirical Analysis

I turn to empirical analysis to examine the relationship between training programs and candidate
emergence across states as shown in Table 4.1. Because the dependent variable is a count of
candidates who have run for o ce, I use a negative binomial model to perform the analysis.10 I
display four models in Table 4.1, the baseline model when the independent variable of interest
is campaign focused programs (Model 1), the baseline model when the independent variable of
interest is programs geared toward women (Model 2), and the respective full models as Model 3
and Model 4.11

The baseline models account for legislative professionalism, the percent of the state considered
10I also performed the analysis with the poisson model and checked for over dispersion. The
equidispersion assumption was violated here, which suggests that using the poisson model would
result in ine cient model estimates. I move forward with the negative binomial model for consistent
and e cient results.

11The likelihood ratio test for each baseline model and it’s respective full model suggests that the

full model is the more appropriate model speciﬁcation.

59

Figure 4.3 Training Program Participation

Women

Men

500

400

300

i

s
t
n
a
p
c
i
t
r
a
P

 
f
o
 
t
n
u
o
C

500

400

300

No

Yes

Participated in Training Program

urban, the count of unavailable seats in an election year, the region, the state level party organization
structure and the percent of democrats in the state. The variables speak to previous literature that
suggests the opportunity structures in which women emerge for political candidacy (see Ondercin
and Welch, 2009; Palmer and Simon, 2010; Sanbonmatsu, 2010). In the case of the ﬁrst baseline
model, the signiﬁcant indicators for women running for o ce are the presence of a campaign focused
training program in the state and the as well as region12, party structure, and the unavailable seats
in the legislature - in the negative direction. Similarly, the second baseline model indicates that
women focused programs are a signiﬁcant indicator for seeing women candidates in the race as
well. The aforementioned control variables are signiﬁcant in the negative direction as well.

Given that there is some evidence to support the idea that the presence of the programs matter, I
12In particular, the south is signiﬁcantly di erent from the Midwest and Northeast

60

turn to the full models - models 3 and 4 respectively. Models 3 and 4 di er from the baseline models
in that I include measures to account for party competition, term limits, divided government, public
opinion, policy output, and time ﬁxed e ects. These variables are important to parse out the e ect
that training programs have on state level candidacy as an indicator in comparison to other features
of a state that might contribute to a more welcoming environment for women as candidates. I ﬁrst
look to the results in Model 3. I ﬁnd that campaign focused training programs are not a signiﬁcant
indicator of women’s candidate emergence. What is signiﬁcant in this model is public opinion
about women as candidates as well as state level policy outcomes. Again, the count of unavailable
seats in th legislature is signiﬁcant and in the negative direction as is party structure. The region
variable is positive and signiﬁcant in this model - which suggests that the west is signiﬁcantly
di erent from the South. That is, women are more likely to be candidates in the Western region.
Turning to Model 4, I ﬁnd that training programs focused on women is a signiﬁcant indicator of
women running for o ce at the state level. Similar to Model 3, unavailable seats in the legislature,
region - more speciﬁcally the Western region, party structure, public opinion and policy outcomes
are signiﬁcant predictors of women’s candidate emergence. Model 4 di ers from Model 3 in that
term limits in a state is a signiﬁcant and negative predictor of women’s emergence.

To further examine the impact of women focused training programs on women running for
o ce, I look to the predicted values from the model. I show the average marginal e ect of a woman
focused training program in Figure 4.4. The ﬁgure shows that as the number of programs increase
the predicted count of woman running for o ce increases as well. In particular, an increase in
the standard deviation (centered around the median) in the number of women focused programs
increases the count of women candidates by 7.9.

While the number of women focused programs in a state provides some evidence of the
importance of training programs, mere presence of these programs may simply be an indicator
for systems of advocates that seek to see women in political o ce. That is, the presence of these
programs does not necessarily mean that they have an impact on the people who complete these
programs. For this reason, I look to data from the Center for American Women and Politics

61

Table 4.1 The Impact of Training Programs on Candidacy at the State Level

Dependent variable: Candidates

Campaign Focused Program

Women Focused Program

Legislative Professionalism

Urban

Holdover

West

Midwest

Northeast

Traditional Party Organization

Percent Democrat

Ranney Index

Term lImit

Opinion about Women as Candidates

Divided Government

Policy Liberalism _median

Year Fixed E ects

Constant

Observations
Log Likelihood
✓
Akaike Inf. Crit.
Note:

Model 1
0.165⇤⇤
(0.073)

 0.149
(0.349)

0.004
(0.003)
 0.093⇤⇤⇤
(0.006)

0.022
(0.110)
 0.835⇤⇤⇤
(0.107)
 0.500⇤⇤⇤
(0.126)
 0.101⇤⇤⇤
(0.028)
 0.001
(0.006)

Model 2

0.121⇤⇤⇤
(0.024)
 0.454
(0.346)

0.003
(0.003)
 0.096⇤⇤⇤
(0.006)
 0.125
(0.111)
 0.845⇤⇤⇤
(0.104)
 0.553⇤⇤⇤
(0.123)
 0.110⇤⇤⇤
(0.027)
 0.00001
(0.006)

4.403⇤⇤⇤
(0.255)

4.523⇤⇤⇤
(0.251)

Model 3
 0.084
(0.068)

 0.409
(0.354)

0.001
(0.003)
 0.102⇤⇤⇤
(0.005)

0.472⇤⇤⇤
(0.108)

0.282
(0.158)
 0.002
(0.126)
 0.085⇤⇤⇤
(0.025)
 0.013⇤⇤
(0.006)
 0.088
(0.281)
 0.074
(0.070)

0.920⇤⇤⇤
(0.184)
 0.073
(0.061)

0.238⇤⇤⇤
(0.057)

X

(0.117)

1.293⇤
(0.746)

Model 4

0.092⇤⇤⇤
(0.022)
 0.640⇤
(0.346)
 0.001
(0.003)
 0.103⇤⇤⇤
(0.005)

0.277⇤⇤
(0.111)

0.141
(0.153)
 0.085
(0.124)
 0.101⇤⇤⇤
(0.025)
 0.009
(0.006)
 0.190
(0.276)
 0.142⇤⇤
(0.069)

0.906⇤⇤⇤
(0.177)
 0.067
(0.059)

0.189⇤⇤⇤
(0.056)

X

(0.114)

1.519⇤⇤
(0.720)

400

 1,816.540
2.351⇤⇤⇤ (0.196)
3,653.079

400

 1,806.468
2.497⇤⇤⇤ (0.210)
3,632.935

392

 1,727.261
3.523⇤⇤⇤ (0.311)
3,498.523

392

 1,719.348
3.717⇤⇤⇤ (0.332)
3,482.695

⇤⇤p<0.05; ⇤⇤⇤p<0.01

62

Figure 4.4 Predicted Count of Candidates Given the Number of Women Focused Programs

l

e
u
a
V
 
d
e
t
c
d
e
r
P

i

120

100

80

60

40

0

2

4

Number of Programs

6

8

Recruitment Study, a data project conducted by Sanbonmatsu, Carroll and Walsh (2015). This data
is insightful as it focuses on individual level factors and choices of current legislators and city level
executives. Again, the main independent variable is a dichotomous measure of whether or not the
survey respondent participated in a candidate training program and the dependent variable is the
measure I constructed to capture strategic behavior.

In seeking to examine if participation in a candidate training program is associated with its
participants exhibiting strategic behavior, I use ordinary least squares models.13 I present three
models in Table 4.2. A baseline model that considers personal factors that might cause a candidate to
13I also estimate ordinal logit models in which the direction and signiﬁcance of the following
variables discussed remains the same. I choose to discuss the ordinary least squares models for
ease of interpretation. I present the ordinal logit models in the appendix.

63

behave strategically (Model 1), a full model in which I consider personal and external factors (Model
2), and an alternating least squares optimal scaling (ALSOS) model to determine if the additive
scale (the dependent variable) is appropriately capturing the categorical spacing in the dependent
variable. The use of ALSOS answers the fundamental measurement question of whether or not the
variable optimally scaled. It does this by selecting data values that both maximize R squared and
retain the measurement characteristics of the included variables (Jacoby, 1991).

I start by addressing the results in Model 1. In Model 1, I control for gender, party identiﬁcation,
education, race (white or non-white), lack of previous experience as an o ceholder, and whether or
not the decision to run was inﬂuenced by desire and someone else’s suggestion to do so. I ﬁnd that
there is a signiﬁcant and positive e ect for gender and participation in a candidate training program.
More speciﬁcally, participation in a candidate training program is associated with strategic behavior
as is being a woman. Lack of political experience and the decision to run variable are also both
signiﬁcant. Running for the ﬁrst time is associated with less strategic behavior, while politically
ambitious individuals who receive the suggestion to seek o ce is associated with more strategic
behavior.

Because there is some evidence of an association between strategic behavior and candidate
training program participation, I move forward with the full model (Model 2). The di erence
between Model 1 and Model 2 is that I add additional controls that may account for strategic
behavior - either intentionally or unintentionally. For instance, an individual may decide to run
because she is dissatisﬁed with politics at large or her own incumbent. This dissatisfaction may
stem from real events that might give her the cause to raise a strong challenge against the incumbent
For this reason, I control for dissatisfaction as a motivation for running for o ce. I also consider
the fact that learning how to be strategic may not be a product of candidate training program
participation. It could be the case that other political experiences provide this important lesson. I
include a measure for whether the respondent is involved with a political party at the local or state
level as well as if the respondent has engaged in campaign work. Finally, strategic entry may not
be a product of individual choice, but could be a product of active party recruitment. I include a

64

Table 4.2 The Impact of Training Programs on Strategic Entry

Dependent variable: Strategic Entry
Strategy

Strategy_os

Democrat

Woman

College graduate

Graduate education

Candidate Training Program Participation

First O ce

Decision to Run

Dissatisfaction

Campaign Work

Involved in Party

Active Party Recruitment

Constant

Observations
R2
Adjusted R2
Residual Std. Error
F Statistic
Note:

Model 1
 0.243⇤⇤⇤
(0.092)

0.163⇤
(0.089)

0.092
(0.123)

0.004
(0.117)

0.202⇤
(0.109)
 0.312⇤⇤⇤
(0.106)

0.271⇤⇤⇤
(0.099)
 0.407⇤⇤⇤
(0.109)

4.017⇤⇤⇤
(0.158)

Model 2
 0.242⇤⇤⇤
(0.093)

0.094
(0.091)

0.005
(0.125)
 0.111
(0.119)

0.190⇤
(0.110)
 0.302⇤⇤⇤
(0.108)

0.261⇤⇤⇤
(0.100)
 0.368⇤⇤⇤
(0.110)

0.474⇤⇤⇤
(0.100)

0.152⇤
(0.092)
 0.020
(0.089)

3.730⇤⇤⇤
(0.180)

Mdel 3
 0.181⇤
(0.093)

0.099
(0.091)

0.011
(0.125)
 0.127
(0.119)

0.212⇤
(0.110)
 0.344⇤⇤⇤
(0.108)

 0.313⇤⇤⇤
(0.110)

0.514⇤⇤⇤
(0.100)

0.175⇤
(0.092)

0.052
(0.089)

3.675⇤⇤⇤
(0.179)

928
0.093
0.082

1.320 (df = 916)

8.524⇤⇤⇤ (df = 11; 916)

928
0.094
0.084

1.319 (df = 917)

9.515⇤⇤⇤ (df = 10; 917)
⇤p<0.1; ⇤⇤p<0.05; ⇤⇤⇤p<0.01

971
0.059
0.051

1.340 (df = 962)

7.524⇤⇤⇤ (df = 8; 962)

65

measure in consideration of this.

As it pertain to the results for Model 2, I ﬁnd that the indicator for candidate training program
attendance is not signiﬁcant; however there are other results here that are important to consider.
I will note that the variable for candidate training program attendance is signiﬁcant in a 90%
conﬁdence interval, as is being involved in a political party. In this model, gender is no longer a
signiﬁcant predictor of strategic behavior, but being a Republican is; the variable for partisanship is
signiﬁcant and in the negative direction. Again, education is not a signiﬁcant predictor and active
party recruitment is not signiﬁcant either. Having a lack of political experience is also signiﬁcant
in this model and in the negative direction. Furthermore, the indicator for the decision to run
being based on one’s political ambition and encouragement from others is signiﬁcant here as well.
Finally, dissatisfaction as political motivation is negatively related to strategic behavior and having
worked on a campaign is positive and signiﬁcant here. The results of the model seems to suggest
that the rules of the electoral game are learned by working on campaigns.

Finally, I look to the ALSOS model (Model 3) to examine how the categorization of the additive
scale might be inﬂuencing the model results. Model 3 looks very similar to Model 2 in that there
are few di erences in the sign and signiﬁcance of the variable coe cients. The only di erence
is that partisanship is no longer signiﬁcant in this model. However, there is an important note to
be made about the ALSOS model as can be seen in Figure 4.5. Figure 4.5 represents the optimal
transformation of the dependent variable. The relatively straight lines at the values of 0 and 1 as
well as 3 and 4 indicates that individuals with these scores were essentially no di erent from each
other in behavior and these data values are essentially the same. Thus, it cannot be assumed that
categories in the dependent variable are equally spaced. It should be noted that the ALSOS model
was estimated with this optimized index. This suggests that the ALSOS additive scale is capturing
an optimal categorization of strategic behavior given the data.

66

Figure 4.5 Optimal Scale of the Dependent Variable (Strategic Behavior)

Optimal Transformation for Variable: model.response(model.frame(form, data))

l

s
e
u
a
v
 
d
e
a
c
s
 
y

l

l
l

a
m

i
t

p
O

6

4

2

0

0

2

4

Original data values

6

Discussion

While the literature on political ambition has progressed, there is still much to be learned about
the path to elected o ce for current o ceholders and potential o ceholders. The most research
has moved toward examining individuals in the pre-emergence stage, but the study of individuals
that participate in candidate and leadership training programs has not been strongly considered.
although see (Hennings, 2011; Stock, 2012; Schneider and Sweet-Cushman, 2018; Sanbonmatsu,
2015). The study of the programs could provide further insight into candidate’s strategic decision
making to enter the electoral arena. Further examination of the training program participants might
also prove useful in better understanding women’s path to political candidacy. Additionally, since
many training programs target women in particular for public o ce, the success rate of these

67

programs might speak to where women in particular emerge and for what o ces they choose to
run.

While I was able to show that training programs, particularly training programs geared toward
women, have an impact on women running in state legislative elections, I do not ﬁnd evidence
that these programs are signiﬁcantly associated with strategic entry. The fact that these training
programs do have a positive impact on women’s candidacy in Table 4.1 may speak to the important
role that training programs play in recruiting women to run.

There is certainly more work to be done in this area though. This can be seen in the lack of
literature about these programs as well as data availability. there are also limits to the conclusions
that can be made from this data. While a training program may be present in the state, this is not
an indication of whether potential women candidates use the program as a resource. Furthermore,
there may be di erences in the resources and types of instruction these programs provide across
states that better prepare some candidates and not others.

This study does not address the partisan a liation of some of these programs across states.
Some states have training programs that are speciﬁcally geared toward women of one party as
opposed to another. For this reason, the use of individual level data was meant to ﬁll in some of the
gaps that are missed with aggregate data at the state level. The individual level data gives an idea of
how these programs inﬂuence individual candidacy decisions amongst current o ceholders. The
ﬁndings here do suggest that there may be a relationship between training program participation
and strategic behavior, but data that better addresses the question of the relationship between the
two is necessary for better model speciﬁcation as well as to address other factors that could play a
role in the strategic behavior of individuals seeking o ce.

Additionally, while the survey responses from the current o ceholders are instructive, these
legislators represent the hits and not the misses - i.e. the candidates that were able to convert their
candidacy to an electoral win. This is not the case for all candidates. Participation in a training
program is not a guarantee of electoral success. Thus, it is likely not the case that every potential
candidate that completes a training program is able to win o ce. With that being said, future

68

work should focus on how all participants in training programs fare in the electoral arena after the
completion of these programs.

69

CHAPTER 5

WAITING TO RUN: CONSIDERING THE EXPERIENCES OF WOMEN OF COLOR AS

LATENT CANDIDATES

Introduction

Figure 5.1 Percent of Women in State Legislatures

Percent of Women Who are WOC

l

i

t

e
r
u
a
s
g
e
L
e
h

 

25
20
15
10
5

Total Percent Women

t
 

n

i
 
t

n
e
c
r
e
P

25
20
15
10
5

1999 2001 2003 2005 2007 2009 2011 2013 2015 2017

Year

Although the number of total women serving in political o ce has become relatively static
(Sanbonmatsu, 2006), the growing number of women of color amongst the ranks of women in
o ce is the exception (Smooth, 2006; Pinderhughes et al., 2006). This idea is reﬂected in Figure
5.1, a depiction of the percent of women in state legislatures. From 1999 to 2017, the percentage
of total women legislators has not changed very much. However, the percent of women state
legislatures that identify as a person of color has increased.
In fact, in 2017, women of color
represented 24% of women state legislators and given the number of women of color candidates
that have emerged in the 2018 election cycle this percentage is likely to increase.

70

Despite the growing number of women of color in state legislatures, we are still learning
about them as candidates and o ceholders. There is certainly recognition of the necessity to
consider both race and gender in American politics, but the race and ethnicity politics and gender
politics literatures have largely developed independently as it pertains to candidate emergence.
This issue is most glaring as it pertains to women of colors’ political ambition, although see
(Sanbonmatsu, 2015). Though the literature on women’s candidate emergence attempts to speak
for all women, there is reason to believe that women of color have unique experiences and challenges
as candidates (Smooth, 2014; Sanbonmatsu, Carroll and Walsh, 2009; Brown, 2014a) as well as
distinct considerations as latent candidates (Sanbonmatsu, 2015).

In this study, I explore women of colors’ understanding of the process of running for o ce and
how this shapes their political ambition using 3 semi-structured interviews. In chapter 1, I provided
evidence that participants of the program are more likely to run for o ce than non-participants,
even when gender and race are considered. For this reason, I focus on three latent candidates who
are former participants of the Michigan Political Leadership Program. Although not all of the
women I interviewed have run for o ce, they desire to do so. By conducting interviews, I highlight
and provide context to the considerations that women of color, who are latent candidates, bring to
bare on their decision to run as well as when is the right time to run.

In what follows, I provide a discussion of what we know about political ambition from the
respective gender and race and ethnicity politics literatures as well as where the gaps in knowledge
are about women of color. Next, I describe the selection criteria for participants of the study and
provide some detail about the participants.
I, then move to an analysis of major themes in the
interviews and ﬁnally lay out conclusions from the interview ﬁndings.

Theoretical Framework

The gender politics literature as well as the race and ethnicity politics literature both have taken steps
to understand candidate emergence in American politics. Prior work on the electoral success of

71

women indicates that they are just as likely to win as men (Fox and Carroll, 2006), but the patterns
of women’s electoral success varies across states (Sanbonmatsu, 2010; Wilcox and Norrander,
2005) and the relative growth in women o ceholders overtime has plateaued (Hennings, 2011;
Sanbonmatsu, 2010). Similarly, there is evidence that minority candidates can and do win in
non-majority minority districts (Highton, 2004; Juenke, 2014; Shah, 2014), but electoral success
is conditioned by the presence of minority candidates willing to run for o ce (Juenke and Shah,
2016, 2015).

In addition, both sets of literature have identiﬁed challenges for women and minorities in
seeking political o ce. Personal factor such as self-e cacy (Preece, 2016; Kanthak and Woon,
2015) and feeling qualiﬁed to run (Fox and Lawless, 2004) matter for women’s electoral entry as
does support from political parties (Sanbonmatsu, 2010; Niven, 1998). The literature on minority
candidates has largely focused on external barriers to o ce seeking. The larger narrative around
minority candidates points to the electoral racial cost that they pay with prejudiced White voters
(Sigelman et al., 1995; Terkildsen, 1993; Moskowitz and Stroh, 1994; Krupnikov, Piston and
Bauer, 2015). While minority candidates are more likely to be elected in places with a majority-
minority population (Marschall, Ruhil and Shah, 2010; Branton, 2009), they are less likely to
pursue o ces that would require white voter support, such as at the Congressional level (Johnson,
Oppenheimer and Selin, 2012; Highton, 2004). These ﬁndings comport with work that suggests
that the racial composition of the district determines the likelihood of candidate emergence (Shah,
2014). Furthermore, there is evidence that barriers to political o ce, such as ﬁnances and political
inexperience, inﬂuence political ambition expression amongst minority groups as well (Reny and
Shah, N.d.).

While we have learned a great deal about women and minority candidates distinctly, there is
much to be learned about how the combination of race and gender inﬂuence political candidacy. In
fact, it can be argued that having multiple identities uniquely situates women of color and that by
emphasizing gender and race, respectively, we essentially erase their narratives (Hancock, 2007b;
Simien, 2007). For this reason, I take an intersectional approach in this work to understand how both

72

race and gender inﬂuence latent candidacy and to center the voices of women of color. However,
intersectionality does more than just consider multiple identities.
It takes into account the role
of power and institutional structures that often marginalize individuals who do not neatly ﬁt into
one category or another (Crenshaw, 1991; Collins, 2015). Furthermore, intersectionality theory
is sensitive to the idea that the inﬂuence of an individual’s multiple identities may vary given the
context (Hancock, 2007a).

Though we are still learning how the intersectional experiences of women of color inﬂuence their
political behavior, there have been substantial contributions to the literature that focus speciﬁcally
on this group. Women of color face racial, cultural, and gendered challenges to their political
participation (Brown, 2014a). Furthermore, Black women’s political participation, in particular,
has been noted to be higher than expected given the group’s relative rank on the factors that are often
associated with political activity (Farris and Holman, 2014). As elected o cials, Black women
largely account for the gains in African American o ceholders over time (Smooth, 2006); however,
women of color have signiﬁcantly increased among the ranks of elected o cials (Pinderhughes
et al., 2006). Attaining political o ce does not come without challenges for women of color though
(Smooth, 2014), women of color are less likely to be recruited and supported by political parties
(Brown, 2014a; Sanbonmatsu, 2006). In addition, there is an open question as to whether women
of colors’ intersecting identities disadvantage or advantage their electoral ambition (Philpot and
Walton, 2007; Bejarano, 2013; Githens and Prestage, 1977; Sanbonmatsu, 2015). However, in
examining the narratives of Black women candidates, Brown (2014b) ﬁnds that the answer to the
question of whether these women are advantaged or disadvantaged by their multiple identities is
dependent on the context.

I seek to move the literature on women of color as candidates forward by examining women of
color who are latent candidates. These women are in a unique position to discuss what motivates
their desire to run for o ce as well as what has inhibited their decision to move forward as
candidates. To date, there is only other work that pays particular attention to the latent candidacy
of women of color (Sanbonmatsu, 2015). This work di ers from that of Sanbonmatsu 2015, in

73

that I focus on the individual experiences of women who have participated in a leadership training
program rather than the leaders of such a program. The critique from Dittmar (2015a) is well taken
here in that candidate and leadership training programs may prepare women to run for political
o ce, but do not necessarily address the institutional contexts in which these women are expected
to run (Dittmar, 2015a,b). By capturing the narratives of women of color with nascent ambition, I
hope to speak to what these women see as the barriers to their electoral entry.

Data

The data for this study derives from semi-structured interviews with three former participants of the
Michigan Political Leadership Program (MPLP). The participants of MPLP present a good starting
point for identifying women of color who have shown interest in running for o ce as each of the
women have expressed a desire to run for o ce. I adapted some questions from the CAWP 2008
Recruitment Study (Sanbonmatsu, Carroll and Walsh, 2015) to qualitatively examine the political
ambition of these women of color. The interview questions are included in the Appendix as . The
overarching questions that I used to guide the interviews included ones that got at interest in running
for o ce, who encourages these women’s interest in political o ce, and their capacity to support
a run for political o ce.

The women included in the study were selected based on three characteristics - their party
a liation, racial identity, and whether or not they ran for o ce. Also key here is that there
is variation in the time since program completion for each of the participants. One participant
completed the program in 2015, another in 2006, and the last in 2005. Two of the participants
identify as Black and the other identiﬁes as Latina. Furthermore, at the time of program entry, two
identiﬁed as Democrats and the last as Independent. Hereafter, the participants will be referred to
by their pseudonyms Annette, Molly, and Whitney.

Annette was part of the 2015 MPLP cohort. At the time of program entry, Annette was 32 years
old had not previously sought o ce, but desired to and she had obtained a law degree. In addition,

74

Annette identiﬁed as Independent and hailed from one of the major cities in Michigan. As part
of her application, Annette completed an essay in which she described the outcome she hoped for
from participating in the program. In it, she described a desire to broaden her experiences, increase
her knowledge of policy, and learn the necessary skills to take next steps in Michigan politics.
Furthermore, when asked about a policy issue that was important to her, Annette named crime as
an important issue that needed to be addressed.

Molly was part of the 2006 MPLP cohort. She was a 26 year old college graduate when she
participated in the program and lived in one of Michigan’s major cities. Unlike Annette, Molly
was very much involved in her local community. She listed herself as a member of a local board
and had participated in a Hispanic leadership program. As part of her expected outcome from
participating in the program, Molly hoped to learn about creating laws and how to a ect changed as
a politician. The policy issue that was most important to Molly was education. Like Annette, Molly
desired to run for o ce and after program completion did so. She ran for a University governing
board position and a school board position after that. She was unsuccessful in both o ce seeking
attempts.

Whitney was part of the 2005 MPLP cohort. She was 29 years old when she completed the
program and was also a college graduate. Like Molly, Whitney was involved in her community. She
served in a local appointed position in her city and listed herself as a member of Alpha Kappa Alpha
Sorority, Inc. Whitney’s expected outcome from MPLP was to hold elected o ce and learn about
leadership skills. In addition, Whitney described her experience working as a campaign consultant.
The issue Whitney described as a major concern was health care. Like Annette, Whitney did not
seek o ce after completing the program.

Analysis

In moving forward with the analysis of the interviews, I describe the major themes that emerge, as
well as points of departure, and the current context in which each woman now lives. I previously

75

provided a description of each women’s experiences as a program participant and in the interview
I ask each woman to reﬂect on her participation in MPLP and whether or not that participation was
meaningful. In addition, I ask about their political ambition and what, if anything, has changed to
inﬂuence their decision to run.

Interest in Politics

Each of the women interviewed discussed di erent paths to politics, but the common thread was
that someone in their lives made politics personal. For Whitney, her mother played a pivotal role.
When her mother volunteered for political campaigns, she brought Whitney along and Whitney
developed an interest from there. Not only that, she went on to work on other campaigns and
to serve as a legislative aid for two state legislators. Although, Annette describes a high school
government class as the point in which she became interested in politics, at di erent points of the
interview she highlights the importance of her parents in shaping her interest in politics as well.
Because Annette’s parents were independent, she was encouraged to explore both parties and their
stances to develop her own perspective. While Whitney and Annette described positive experiences
that encouraged them to be interested in politics, Molly’s interest derived from a more negative
situation. Molly had an uncle, whom she was very close with, become incarcerated at a young age.
This situation and her ﬁrst job’s involvement in ex-o ender re-entry spurred her interest in criminal
justice reform. In particular, she was concerned about how mass incarceration disproportionality
impacts men of color.

Struggles with Political Identity

Although all three women are Democratic leaners, they all discussed a struggle with their political
identity. This struggle was most apparent with Annette. At the time of her application she described
herself as independent, but in the interview she described a time in her life in which she identiﬁed
as a Republican. As previously mentioned, Annette’s parents were independent, but her father was

76

also a pastor. For Annette, her faith played a strong role in her politics and in college she became
conservative because of it. The point of challenge came when she had a personal realization that
Republican party ideals were not what she saw in the real world. She would argue that her faith
still plays a role in her politics, but in a di erent way. In fact, she now identiﬁes as a Democrat and
would describe herself as an active member.

Similarly, Whitney and Molly describe a shift in their thinking about political identity as well.
Whitney now identiﬁes as an Independent, as opposed to a Democrat. Her decision stems from
her concern about how the parties handle issues and the stances they take. She suggests being a
registered Independent better suits her perspective. On the other hand, Molly still identiﬁes as a
Democrat, but she is dissatisﬁed with the party. In the interview, she describes a frustration that the
party leaves people of color, particularly women of color out of the conversation. She highlights
the party’s choice to support a progressive white woman, in a winnable race for a state house seat,
instead of another candidate.

Barriers for Women in Politics

In discussing challenges for women in running for o ce, Whitney, Annette, and Molly all point
to di erent factors that inﬂuence women’s experiences in running for o ce. Whitney suggests the
barrier involves attitudes about women in leadership. For her, women leaders are in precarious
positions in that they may not be fully accepted or they are more open to scrutiny. For Annette and
Molly, money, resources, and networks can be a deterrent to political candidacy. Molly points to the
fact that women tend to make less than men, which may inhibit women’s ability to run a campaign.
In addition, women may not have access to the “bro network", which Molly points to as excluding
women from being in the places where decisions are made. Annette makes a similar observation
about networks in that she not only makes the case that networks matter, but also that the "legacy of
politics does not beneﬁt women." Here, Annette mentions political families and name recognition.
Her example is John Conyers Jr. and how his political legitimacy comes from his shared name with
his father, John Conyers. Women can lose that name recognition when they change their name in

77

marriage.

Barriers for Women of Color in Politics

When asked about barriers for women of color in politics, all three women indicate that women of
color face greater challenges than their white counterparts. For Whitney, these challenges are not
tangible, but related to attitudes about women of color in leadership positions. She suggests that
women of color have to be four times as good as their male counterparts to convince people of their
ability to lead. In addition, she points to a hierarchy of sorts when she expresses the idea that people
see color before they see gender among minority groups. Molly points to similar attitude based
challenges. She emphasizes how women are portrayed in the media. She provides an example of
a time in which she worked on a campaign for a black woman candidate who was portrayed as
angry and aggressive in the media, while her white woman opponent was portrayed as demure.
Molly similarly points to needing to conform to White norms of beauty and professionalism as
problematic.

Beyond how women of color might be perceived, Molly highlights her concern about health.
She is aware of the statistics on chronic health conditions for women of color and makes the case
that women of color running for o ce might have to consider health risk factors. In a similar vein,
Annette focuses in on disparate economic and educational opportunities for women of color as a
deterrent to o ce seeking. She makes the statement that “Doors or dollars win races". By this
she means that women of color have to be ﬁnancially stable to both mount a funded campaign and
devote the time necessary to campaign e ectively.

Value of Leadership Training Program Experience

As it pertains to the value of the leadership training program experience for each woman, all
three valued the non-partisan aspect of the program. Whitney considers MPLP to have been a
good opportunity to have conversations with people who think di erently than she does. Molly

78

also appreciated the di erent view points that was exposed to during the program experience.
In addition, she valued the ability to build relationships with her cohort, given the length of the
program, 1 and the policy discussions. Annette expressed being overwhelmed by everything she
learned, but considered the emphasis on policy, not partisanship, her favorite aspect of the program.
Molly does have a critical perspective on the program though. She is the only one of the three
who would not consider participating in MPLP again if given the chance. Molly participated in the
program when MPLP fellows were fully funded and did not have to make a ﬁnancial investment to
attend the program. She expressed that the cost would have been a barrier for her. Furthermore,
she notes the lack of discussion about the systems and structures that are barriers - particularly for
women of color in seeking political o ce.

Race appeared to be a deﬁning factor in each woman’s experience in the program, as opposed
to gender, even when asked about their experiences as women of color. Because MPLP is gender
balanced (12 men and 12 women chosen as fellows), none of the three women felt that their
gender mattered during their program experience. For Molly, her race was most salient during
the conversation about criminal justice reform. The questions raised about the disproportionate
number of minorities in the prison system made her conscious of the disparity. Whitney felt that
race was salient because people had expectations of what her beliefs would be. Annette took an
active role in raising discussions about race because she did not feel race was discussed enough.
She indicated that the program missed opportunities to insert race into the conversation, especially
when it came to the program weekend in Detroit.

Waiting to Run

Although all three women still expressed an interest in running for o ce, there was a sentiment
of waiting for the right time to run. This means something a bit di erent for each of the women
though. For Annette and Whitney, the idea of the "timing being right" was centered around them
1MPLP is 10 months long and the MPLP fellows meet once a month for a weekend learning

experience.

79

both having children. However, Whitney’s reason for waiting was about her children’s ages and
their well being. She expressed that being a "dedicated public servant" required late nights and
weekends and she was not willing to take that time away from her family. Annette also indicated
that the position had to be right for her family, but expressed a concern about public exposure of her
children as well. She cited an incident in which she posted a picture of her children showing support
for a local o ceholder and this picture was used by the opposition in a negative and racialized way.
This was very upsetting for her and made her more cautious of how politics can negatively target
children. Though Molly has had unsuccessful o ce seeking attempts, she does plan to run again.
This time, she is waiting to run until it is strategically time to do so. The current o ceholder will
be term limited out and she will be faced with an open seat election.

While Molly has her mind set on the position she will be running for, the same cannot be said
for Annette and Whitney. Molly is in the beginning stages of mounting a campaign for a county
commission seat. In preparation for the race, she is building her network and letting people know
that she is running for o ce. Annette had some ideas about political positions she might seek
though. She has thought about running for a mayoral position or a judgeship. When asked about
why she would run for a judicial position, she cited her passion for criminal justice and equity as
well as having a law degree.

Each of the participants were asked about MPLP’s role in their decision to run for o ce and
all three expressed that the program reinforced their commitment to seeking a political position.
Whitney went so far as to say that the program gave her a toolbox of skills to run and she felt guilt
that she has not used them yet. Along with Molly expressing that MPLP made her think critically
about running for o ce, she also had some critiques. She indicated that the network she built
through MPLP was good for camaraderie, but did not necessarily translate into local support in her
election attempts. In addition, she thought that the program did not do enough to teach tactics for
running in local elections. Finally, Molly felt that the best learning experiences she had in learning
how to run a campaign were working on other people’s campaigns.

80

Discussion and Implications

The 2018 primaries have drawn more attention to women of color candidates than ever before.
Because of the positions for which these women are running and the sheer number of them that
are emerging, it is more important than ever to be able to contextualize what it means for them
to run for o ce and the barriers they face in doing so. There is already some work that suggests
that the cost-beneﬁt analysis involved in the decision to run can be a deterrent for women of color
(Shames, 2015), but this work is unique in that I present the narratives of women of color that
bring di erent experiences and perspectives to the table in desiring to run for o ce. That each of
these women has completed the MPLP program not only shows that they are interested in being
a political o ceholder, but also shows that they have taken concrete steps to do so. Throughout
each of their interviews, these women spoke to how their gender and race have factored into their
experiences as well as the experiences of other women in politics.

Of note here is that each of these women have familial experiences that have sparked their
interest in politics. Though we might call this political socialization, the discussion in the literature
often focuses on positive experiences of parents and institutions sharing political knowledge and the
virtues of citizenship. We speak very little about the possibility that political socialization perhaps
does not look like this for all groups. Molly’s point of entry into her political interest was especially
poignant because she learned through the absence of a family member and had to contend with the
reason for this absence as well as the systematic pattern it represented in communities of color. I
think this raises the question of how many women of color might identity with Molly’s story as
opposed to that of Annette and Whitney. No no one experience is better than the other here, but it
is important to understand how women of color come to the political table and why they decide to
stay.

The fact that each of the women grappled with their satisfaction and attachments to political
parties speaks to a larger story about where women of color, and more generally, people of color
stand within the party structure. This is especially an interesting point to be made when women

81

of color are largely running as Democrats, but may not feel that the party is completely on their
side. There is already evidence that women of color are less likely to be recruited to run for o ce
(Sanbonmatsu, Carroll and Walsh, 2009), but there is less discussion about how women of color
might conceptualize their position within the party as a factor in their decision to run. Scola Scola
(2013, 2007) shows that women of color tend to emerge as candidates in di erent spaces than their
white women counterparts and future work should examine the factors that go into decision. It
should also not be missed that Molly indicated that the Democratic party in her area was more
likely to choose a white, liberal woman than making another choice.

Although many of the barriers that Molly, Whitney and Annette highlighted for women and
women of color are not new, they do have some insights that are worth thinking about. The
importance of name recognition and being aligned with traditional leaders is a concept that Gillespie
(2010) mentions in her book in seeking to classify Black political leadership. This might serve
to disadvantage women of color, and women more generally, who do not have those connections
and make network building even more important for these women to be electorally viable. This
ties into Whitney’s statement that women of color have to work harder than their counterparts to
get ahead. What is the cost of working harder for women of color though? Molly’s concern about
her health is important to consider here. Women of color are at higher risk for a number of health
conditions. Future would should consider how health matters for women candidates, particularly
women of color candidates. There was some indication during the 2016 election that concerns
about Hillary Clinton’s health disadvantaged her. How might this look at lower levels o ce and
are the perceptions the same?

Finally, moving forward, we must think critically about the role that training programs play for
women of color. Though these programs may be serving to increase women’s self e cacy (see
Hennings, 2011; Stock, 2012) and make them committed to run for o ce, there must be some
consideration of the institutional and structural barriers that they may face in the campaign context.
This is not to say that there are not programs that speciﬁcally speak to the needs of women of color
(see Sanbonmatsu, 2015), but that perceptions of women of color as candidates, their networks,

82

and their ability to build campaign support may ﬁgure prominently in their ability to succeed.
Future work should address how program curriculum matters for women of colors’ outcomes after
program completion.

83

CHAPTER 6

CONCLUSION

At the outset of the dissertation, I laid out two main points: running for o ce is a form of political
participation, albeit a rare one, and that as a form of participation we can apply Verba, Schlozman
and Brady’s model of civic voluntarism 1995 to understand who runs for o ce.
In addition, I
argued that candidate and leadership training programs are a mechanism through which individuals
gain the resources, motivation, and encouragement to participate. In each chapter, I have sought to
present a cohesive narrative about the beneﬁts of participation in candidate and leadership training
programs, what these programs mean for who is deﬁned as a potential candidate, as well as the
impact of the training program experience on women and their political ambition.

The original data collection e ort for the ﬁrst chapter was made possible by Michigan Political
Leadership Program and the access to the total applicant pool of the program from 2003 to 2015.
The nature of data provided leverage to examine the electoral trajectory of individuals who were
accepted into the program as well as those who were not. Furthermore, the matching design and the
positive association between program participation and the likelihood of running for o ce speak
to the causal implications of the results.

Although both ﬁeld experiments discussed in Chapter 2 were unsuccessful, this presents an
innovation and an important learning experience in the application of Verba, Schlozman, and
Brady’s theory. Political participation is a form of participation, but it is not equatable to the act
of being a consistent voter in state and local elections. By including the universe of active voters
as potential participants in the ﬁrst experiment, I was able to speak to the expanse of the candidate
eligibility pool. Not everyone will be latent candidates, but there are some. Participation in this way
seems to require a di erent type of motivation and perhaps even a di erent form of encouragement.
The examination of the impact of training programs across states also involved an original data
collection e ort to both identify training program presence at the state level as well as to classify
what curriculum these programs o er its participants via website analysis. While available work

84

has focused on a select number of programs to understand the inﬂuence of training programs, the
results here show that training program presence in a state is positively associated with a higher
count of women running for o ce at the state level. Furthermore, the results of the individual level
analysis provides some support for the idea that participation in training programs leads to strategic
political entry. Although, there is stronger support for working on a campaign and being involved
in the political party as indicators of strategic behavior.

The semi-structured interviews with women of color were particularly insightful as there are
few studies that examine women of color as latent candidates. By focusing on women who have
both completed a training program and have expressed interest in running for o ce at two time
points, I was able to make some substantive conclusions about what they perceive to barriers for
themselves as well as for other women of color. In addition, the analysis brings forth points that
are rarely considered, especially as it pertains to women of color as candidates.

There is certainly more work to be done in this area. This is noted in the lack of literature about
these programs as well as the available data. My dissertation presents the ﬁrst major undertaking
to collect data on these programs across states as well as to identify the impact of these programs
across states. I do not address the partisan a liation of the programs in the analysis though. Some
states have training programs that are speciﬁcally focused toward women of one party as opposed
to another. While other programs have a bi-partisan focus and may or may not accept both women
and men. In addition, training program participation might also be associated with espousing a
speciﬁc set of beliefs. The Michigan Political Leadership Program is one such example of a bi-
partisan program that includes both men and women and considers racial and geographic diversity
in participant selection.

The results of the analysis here have important implications for the candidate emergence litera-
ture. The common thread throughout is that candidate and leadership training programs play a role
in who runs for o ce and because this program experience matters the universe of latent candidates
is potentially larger than the usual types of individuals we would think of to run. Furthermore, the
analysis suggests that while these programs do good things for women by providing them will skills

85

and the conﬁdence to run, there are curriculum based changes that could be made to better speak
to the needs of women of color.

Theoretical Implications

I think the contribution of this work will largely be based on providing a greater understanding of
candidate training programs and how they might be inﬂuencing the behavior of its participants.
Thus far, I have only seen two dissertations that focus on candidate training programs. Both focus
on how these programs operate and use self report measures alone as evidence of their e ectiveness.
By situating this work in the candidate emergence literature as well as the literature on political
participation, I am using the existing knowledge from these two areas to show how candidate
training programs matter for potential candidates.

This work is one of the few that asserts that candidates and leadership training programs play
a role in politics - particularly in who run for o ce. I would argue that these programs can have
implications for representational outcomes as well. This point is especially salient as a reﬂection on
the current political climate and the fact that people across identity groups are looking to candidate
and leadership training programs to learn how to get involved in politics. In just looking at the
applications of the people who apply to the Michigan Political Leadership Program, as one example,
they do not necessarily ﬁt the traditional candidacy pool. This suggests that there is a point of
departure between people who are interested in running for o ce and the ones who are represented
in the legislature. Certainly, I ﬁnd evidence that individuals who participate in a training program
are more likely to run for o ce than those who do not. However, the ones who are not participating
in these programs still have a desire and unrealized political ambition.

Past literature on political candidacy has largely focused on o ceholders. This presents an
interesting conundrum for the study of candidate emergence as the emphasize is inadvertently
placed on the successful cases. What may or may not be a barrier for potential candidates could have
a ected o ceholders in a di erent way. Furthermore, the discussion about factors that inﬂuence

86

women’s candidacy do not always consider intersectional identities. The desire to generalize what
factors deter all women may serve to erase the voices of the women at the margins. For this reason,
it was important that I include interviews, particularly interviews that centered the perspectives of
women of color. I not only wanted to speak to their participation in training program, but also their
perceived barriers in seeking o ce. I found those interviews to be both thought provoking and
meaningful in terms of the ways in which the literature can move forward in understanding women
of color who are latent candidates and those who are actively pursuing political o ce.

Practical Implications

In a practical sense, this work represents an example of a partnership between practitioners and
academics. This is certainly not the ﬁrst example of this, but one instance in which the operations
of training programs inform the theory about candidate emergence. Future partnerships with
organizations like the Michigan Political Leadership Program might yield new insights about latent
candidates as well as serve to provide some best practices in latent candidate recruitment.

In thinking about the current wave of political activism and interest in political o ce, candidate
and leadership training programs are perfect vehicles to understand individuals’ motivation to seek
political positions. At a time when a record number of women are running for o ce, we can
only speculate on why they made the decision to run right now. Many of these programs have
an application process in which applicants must describe what drives them to run for o ce and
what issues they are passionate about. These application responses might be used to build and
test theories on this wave of political ambition expression and how we might operationalize such
an explanation to induce political ambition at di erent time points. Furthermore, this presents an
opportunity to both understand what potential candidates need to learn to be successful as well as
calibrate program curriculum decisions with the structural and institutional factors that serve to
deter women, in particular, from running for o ce.

87

APPENDICES

88

APPENDIX A

FIRST APPENDIX

Table A.1 Balance Statistics for Matched Sample

distance
Men
Women
POC
College
Higher Degree
Interview Score

Means Treated Means Control SD Control Mean Di  eQQ Med
0.1827
0.0000
0.0000
0.0000
0.0000
0.0000
9.0000

0.3089
0.4952
0.5048
0.2556
0.3770
0.1118
75.5176

0.3253
0.4952
0.5048
0.2556
0.3770
0.1118
76.7160

0.1625
0.5008
0.5008
0.4369
0.4854
0.3157
20.5634

0.0163
0.0000
0.0000
0.0000
0.0000
0.0000
1.1984

eQQ Mean

0.1609
0.0053
0.0053
0.0160
0.0904
0.0798
11.0638

eQQ Max
0.2061
1.0000
1.0000
1.0000
1.0000
1.0000
70.0000

Table A.2 Percent Balance Improvement for Matched Sample

distance
Male
Female
POC
College
Higher Degree
Interview Score

Mean Di .
91.9106
100.0000
100.0000
100.0000
100.0000
100.0000
92.7772

eQQ Med
22.2377
0.0000
0.0000
0.0000
0.0000
0.0000
18.1818

eQQ Mean
20.6708

-Inf

-3.7234
37.7660
-3.7234
2.7593
33.9018

eQQ Max
22.3503

-Inf
0.0000
0.0000
0.0000
0.0000
5.4054

Table A.3 Sample Size of the Control and Treated Groups

Control Treated

All
Matched
Unmatched
Discarded

391
313
78
0

195
188
7
0

89

Table A.4 The Impact of Political Training Programs on Running for O ce

Dependent variable:

Ran_O ce1

Unmatched (1) Matched (2)
1.047⇤⇤⇤
(0.226)

1.000⇤⇤⇤
(0.218)

Unmatched (3) Matched (4)
1.451⇤⇤⇤
(0.346)

1.214⇤⇤⇤
(0.318)

Accepted

Prior O ce

College

Higher Degree

POC

Involved with Community Group

Involved with Party

Women

Interview Score

Desire

Age

Year2004

Year2005

Year2006

Year2007

Year2008

Year2009

Year2010

Year2011

Year2012

Year2013

Year2014

Year2015

Constant

Observations
Log Likelihood
Akaike Inf. Crit.
Note:

1.695⇤⇤⇤
(0.294)

0.371
(0.234)
 0.236
(0.328)
 0.127
(0.235)
 0.499
(0.347)

0.900⇤⇤⇤
(0.274)
 0.059
(0.218)

0.006
(0.005)

0.888⇤⇤
(0.403)

0.032⇤⇤⇤
(0.010)

0.438
(0.408)
 0.033
(0.506)

0.817⇤
(0.469)

0.175
(0.498)

0.805⇤
(0.475)

0.191
(0.500)

0.286
(0.616)

0.914
(0.589)

1.790⇤⇤⇤
(0.321)

0.364
(0.250)
 0.369
(0.370)
 0.134
(0.265)
 0.639⇤
(0.386)

0.970⇤⇤⇤
(0.294)

0.032
(0.238)

0.0001
(0.007)

1.057⇤⇤
(0.444)

0.034⇤⇤⇤
(0.011)

0.760⇤
(0.439)

0.286
(0.537)

1.054⇤⇤
(0.509)

0.514
(0.530)

1.373⇤⇤⇤
(0.521)

0.845
(0.551)

0.663
(0.698)

1.399⇤⇤
(0.649)

0.933⇤
(0.539)
 0.762
(0.575)
 1.245⇤
(0.707)
 1.160⇤
(0.690)
 4.314⇤⇤⇤
(0.795)

1.557⇤⇤
(0.644)
 0.784
(0.622)
 1.204
(0.808)
 0.860
(0.722)
 4.411⇤⇤⇤
(0.955)

586

501

 285.696
619.392

 247.287
542.573
⇤p<0.1; ⇤⇤p<0.05; ⇤⇤⇤p<0.01

0.378⇤
(0.212)
 0.362
(0.301)
 0.296
(0.214)

 0.181
(0.196)

0.007
(0.005)

0.582⇤
(0.323)

0.044⇤⇤⇤
(0.009)

0.420⇤
(0.224)
 0.431
(0.335)
 0.272
(0.239)

 0.100
(0.212)

0.004
(0.006)

0.691⇤⇤
(0.347)

0.046⇤⇤⇤
(0.009)

 4.041⇤⇤⇤
(0.646)

 3.981⇤⇤⇤
(0.767)

586

 324.701
667.403

501

 285.402
588.804

90

This table displays the full models depicted in the main body of the paper as Table 2. Here the ﬁxed e ects for year are
displayed with accompanying coe cients and indicators for statistical signiﬁcance.

Table A.5 The Impact of Political Training Programs on Running for O ce

Dependent variable: Ran For O ce

Accepted

Prior O ce

College

Higher Degree

(1)

0.682⇤⇤⇤
(0.229)

(2)

0.761⇤⇤⇤
(0.246)

0.075
(0.229)
 0.443
(0.315)

0.065
(0.251)
 0.577
(0.351)

(3)

1.072⇤⇤⇤
(0.355)

2.001⇤⇤⇤
(0.338)

0.091
(0.258)
 0.314
(0.354)

0.758
(0.512)

0.355
(0.551)

0.982⇤⇤
(0.498)

0.448
(0.528)

(4)

1.318⇤⇤⇤
(0.400)

2.252⇤⇤⇤
(0.397)

0.051
(0.288)
 0.423
(0.404)

0.956⇤
(0.542)

0.647
(0.608)

0.920⇤
(0.538)

0.537
(0.561)

1.397⇤⇤⇤
(0.510)

1.530⇤⇤⇤
(0.547)

0.709
(0.545)

0.704
(0.665)

1.348⇤⇤
(0.629)

1.314⇤⇤
(0.609)
 1.624
(1.154)
 1.281
(0.818)
 1.097
(0.725)
 16.402
(1,083.199)

0.058
(0.751)
 0.075
(0.285)

0.223
(0.729)
 15.161
(830.678)

0.513
(0.860)
 0.508
(0.397)

0.902⇤⇤⇤
(0.303)
 0.103
(0.237)

0.023⇤⇤⇤
(0.008)

0.886⇤
(0.486)
 4.573⇤⇤⇤
(0.923)

0.933
(0.605)

0.655
(0.800)

1.619⇤⇤
(0.682)

1.397⇤
(0.824)
 15.965
(703.274)
 2.646⇤⇤
(1.287)
 0.896
(0.772)

0.015
(1.211)
 0.193
(0.330)

0.544
(1.648)
 15.745
(1,389.364)

 0.741
(0.454)

0.821⇤⇤
(0.336)
 0.037
(0.265)

0.021⇤
(0.012)

1.538⇤⇤⇤
(0.573)
 5.070⇤⇤⇤
(1.257)

501

398

 237.605
531.210

 191.074
434.147
⇤p<0.1; ⇤⇤p<0.05; ⇤⇤⇤p<0.01

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

Arab American

Asian

Black

Hispanic

Native American

Other

Involved in Community Group

Involved with Party

Female

Interview Score

Desire to Run

Constant

Observations
Log Likelihood
Akaike Inf. Crit.
Note:

 15.699
(1,131.286)
 0.391
(0.700)
 0.156
(0.255)

0.131
(0.687)
 15.615
(940.649)
 0.181
(0.711)

 0.763
(1.158)
 0.214
(0.290)

0.604
(1.455)
 13.671
(584.758)

 0.172
(0.210)

0.024⇤⇤⇤
(0.008)

0.438
(0.380)
 3.246⇤⇤⇤
(0.735)

501

 278.984
583.969

 0.147
(0.232)

0.028⇤⇤
(0.012)

0.788⇤
(0.434)
 3.964⇤⇤⇤
(1.087)

398

 231.457
484.914

91

This table displays an alternative speciﬁcation for race from the one speciﬁed in the main body of the paper. The
variable for race in the body of the paper is dichotomous and the one displayed here contains 7 categories for race.
The category for white is left out of the model as a point of comparison. As indicated in a footnote in the paper, this
speciﬁcation for race is not signiﬁcant.

Table A.6 The Impact of Political Trainings Programs on Running for O ce: Matched Interaction Models

Dependent variable: Ran for O ce

Gender Interaction Model

Race Interaction Model

Accepted

Prior O ce

Women

POC

Involved in Community Group

Involved in Party

College

Higher Degree

Interview Score

Desire

Age

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

Accepted x Women

Accepted x POC

Constant

Observations
Log Likelihood
Akaike Inf. Crit.
Note:

This table displays an alternative speciﬁcation for the interaction models speciﬁed in the main body of the paper. Here
the ﬁxed e ects for year are displayed with accompanying coe cients and indicators for statistical signiﬁcance.

⇤p<0.1; ⇤⇤p<0.05; ⇤⇤⇤p<0.01

1.311⇤⇤⇤
(0.407)

1.788⇤⇤⇤
(0.321)
 0.096
(0.310)
 0.134
(0.266)
 0.646⇤
(0.386)

0.972⇤⇤⇤
(0.294)

0.362
(0.251)
 0.347
(0.371)
 0.0002
(0.007)

1.051⇤⇤
(0.444)

0.034⇤⇤⇤
(0.011)

0.759⇤
(0.439)

0.286
(0.537)

1.036⇤⇤
(0.510)

0.498
(0.531)

1.360⇤⇤⇤
(0.521)

0.846
(0.551)

0.604
(0.705)

1.393⇤⇤
(0.649)

1.548⇤⇤
(0.645)
 0.798
(0.624)
 1.245
(0.810)
 0.868
(0.720)

0.302
(0.466)

 4.327⇤⇤⇤
(0.966)

501

 247.076
544.152

1.394⇤⇤⇤
(0.368)

1.789⇤⇤⇤
(0.320)

0.030
(0.239)
 0.234
(0.348)
 0.653⇤
(0.387)

0.969⇤⇤⇤
(0.294)

0.362
(0.251)
 0.366
(0.369)

0.0002
(0.007)

1.040⇤⇤
(0.446)

0.034⇤⇤⇤
(0.011)

0.760⇤
(0.439)

0.269
(0.538)

1.037⇤⇤
(0.510)

0.510
(0.530)

1.361⇤⇤⇤
(0.522)

0.841
(0.551)

0.659
(0.698)

1.415⇤⇤
(0.650)

1.564⇤⇤
(0.645)
 0.772
(0.623)
 1.206
(0.807)
 0.881
(0.723)

0.237
(0.528)
 4.358⇤⇤⇤
(0.961)

501

 247.186
544.371

92

APPENDIX B

SECOND APPENDIX

Figure B.1 Text of Postcard 1- Control

DO	YOU	WANT	TO	HAVE	AN	IMPACT	ON	MICHIGAN	POLITICS?	

LEARN	THE	SKILLS	TO	RUN	FOR	OFFICE	

	

Join	our	program	and	become	a	fellow!	The	Michigan	
Political	Leadership	Program	at	Michigan	State	
University	is	a	10-month	bipartisan	fellowship.	We	bring	
together	elected	officials	and	professionals	to	teach	
Michigan’s	future	leaders	the	knowledge	and	skills	they	
need	to	serve	their	communities.	Fellows	travel	around	
the	State	learning	practical	politics,	public	policy,	
leadership,	and	governance.	

	

	

	
Our	graduates	are	twice	as	likely	to	run	for	elected	office	and	three	times	as	
likely	to	win.	Eleven	alumni	currently	serve	in	the	Michigan	Legislature.	
Applications	are	due	by	September	9th.	Learn	more	or	apply	today:	

Visit:	ippsr.msu.edu/mplp1	
Email:	mplp1@msu.edu		

Call:	517-353-0891	(Mention	Card1)	

Figure B.2 Text of Postcard 2 - Treatment 1

WE	HAVE	IDENTIFIED	YOU	OUT	OF	7	MILLION	MICHIGAN	VOTERS	
YOU’RE	IN	THE	TOP	5%	MOST	POLITICALLY	ACTIVE	AND	ENGAGED	

	

You’ve	shown	you	care.	Now	it’s	time	to	be	a	leader!	
The	Michigan	Political	Leadership	Program	at	Michigan	
State	University	is	a	10-month	bipartisan	fellowship.	We	
bring	together	elected	officials	and	professionals	to	
teach	Michigan’s	future	leaders	the	knowledge	and	
skills	they	need	to	serve	their	communities.	Fellows	
travel	around	the	State	learning	practical	politics,	public	
policy,	leadership,	and	governance.		

	
Our	graduates	are	twice	as	likely	to	run	for	elected	office	and	three	times	as	
likely	to	win.	Eleven	alumni	currently	serve	in	the	Michigan	Legislature.	
Applications	are	due	by	September	9th.	Learn	more	or	apply	today:	

Visit:	ippsr.msu.edu/mplp2	
Email:	mplp2@msu.edu		

Call:	517-353-0891	(Mention	Card2)	

	

	

93

Figure B.3 Text of Postcard 3 -Treatment 2

WOMEN	ARE	HALF	THE	STATE	BUT	ONLY	20%	OF	THE	LEGISLATURE	

ISN’T	IT	TIME	FOR	WOMEN	TO	TAKE	THE	LEAD?	WE	NEED	YOU!	

	

	

We	need	a	woman	like	you	to	be	one	of	our	leaders.	
The	Michigan	Political	Leadership	Program	at	Michigan	
State	University	is	a	10-month	bipartisan	fellowship.	We	
bring	together	elected	officials	and	professionals	to	
teach	Michigan’s	future	leaders	the	knowledge	and	
skills	they	need	to	serve	their	communities.	Fellows	
travel	around	the	State	learning	practical	politics,	public	
policy,	leadership,	and	governance.	

	
Our	graduates	are	twice	as	likely	to	run	for	elected	office	and	three	times	as	
likely	to	win.	Eleven	alumni	currently	serve	in	the	Michigan	Legislature.	
Applications	are	due	by	September	9th.	Learn	more	or	apply	today:	

Visit:	ippsr.msu.edu/mplp3	
Email:	mplp3@msu.edu		

Call:	517-353-0891	(Mention	Card3)	

Figure B.4 Text of Email - Control

	
	
Subject	Line:	Your	Friend	thinks	you	have	what	it	takes	to	be	a	Michigan	political	leader.	
	
	
	
Dear	____________,		
	
Your	friend,	____________,	thinks	you	should	run	for	office.	
	
If	your	friend,	________,	thinks	you	have	what	it	takes,	then	you	should	too.		
	
Do	you	want	to	learn	more	about	running	for	office?	Michigan	State	University’s	Michigan	
Political	Leadership	Program	(MPLP)	can	help	you	learn	the	skills	to	run.		
	
The	Michigan	Political	Leadership	Program	at	Michigan	State	University	is	a	10-month	
bipartisan	fellowship.	We	bring	together	elected	officials	and	professionals	to	teach	
Michigan’s	future	leaders	the	knowledge	and	skills	they	need	to	serve	their	communities.	
Fellows	travel	around	the	State	learning	practical	politics,	public	policy,	leadership,	and	
governance.	
	
Our	graduates	are	twice	as	likely	to	run	for	elected	office	and	three	times	as	likely	to	win.	
Eleven	alumni	currently	serve	in	the	Michigan	Legislature.	Applications	are	due	by	
September	8th.	Learn	more	or	apply	today:	
	
	

Visit:	ippsr.msu.edu/mplp	
Email:	mplp1@ssc.msu.edu	

Call:	517-353-0891	

	

94

Figure B.5 Text of Email - Treatment

Subject	Line:	You	have	what	it	takes	to	be	a	Michigan	political	leader.	
	
Dear	____________,		
	
I	think	you	should	run	for	office.________________________________________________	
__________________________________________________________________________	
__________________________________________________________________________.		
	
If	I	think	you	have	what	it	takes,	then	you	should	too.		
	
Do	you	want	to	learn	more	about	running	for	office?	Michigan	State	University’s	Michigan	
Political	Leadership	Program	(MPLP)	can	help	you	learn	the	skills	to	run.		
		
The	Michigan	Political	Leadership	Program	at	Michigan	State	University	is	a	10-month	
bipartisan	fellowship.	It	brings	together	elected	officials	and	professionals	to	teach	
Michigan’s	future	leaders	the	knowledge	and	skills	they	need	to	serve	their	communities.	
Fellows	travel	around	the	State	learning	practical	politics,	public	policy,	leadership,	and	
governance.	

	

Graduates	are	twice	as	likely	to	run	for	elected	office	and	three	times	as	likely	to	win.	
Eleven	alumni	currently	serve	in	the	Michigan	Legislature.	Applications	are	due	by	
September	8th.	Learn	more	or	apply	today:	
	

Visit:	ippsr.msu.edu/mplp	
Email:	mplp2@ssc.msu.edu	

Call:	517-353-0891	

	
	
	
Your	Friend,		
	
________________	
	

95

APPENDIX C

THIRD APPENDIX

Table C.1 Cronbach’s Alpha of Items Used to Create Strategy Measure

raw alpha
0.37
Conﬁdence Boundaries

Reliability if an item is dropped:

Member of Party
Open Seat
District Composition
Prior Experience
Supported by Party
Financial Resources

Item Statistics

Member of Party
Open Seat
District Composition
Prior Experience
Supported by Party
Financial Resources

Non missing response frequency for each item

Member of Party
Open Seat
District Composition
Prior Experience
Supported by Party
Financial Resources

std.alpha G6(smc)
0.41

0.36

average r
0.084

S/N
0.55

ase
0.029

mean
0.66

sd
0.23

lower
0.31

alpha
0.37

upper
0.42

rawalpha
0.17
0.24
0.24
0.30
0.52
0.39

std.alpha G6(smc)
0.18
0.31
0.26
0.37
0.51
0.45

0.16
0.23
0.23
0.28
0.51
0.39

average r
0.037
0.056
0.056
0.073
0.170
0.114

S/N alpha se
0.039
0.19
0.30
0.036
0.036
0.29
0.033
0.40
0.023
1.02
0.64
0.029

r.cor
0.681
0.459
0.545
0.307
-0.191
0.075

r.drop
0.389
0.291
0.290
0.206
-0.150
0.048

mean
0.68
0.67
0.58
0.53
0.67
0.81

sd
0.47
0.47
0.49
0.50
0.47
0.39

n
1086
1086
1086
1086
1086
1086

0

0.32
0.33
0.42
0.47
0.33
0.19

raw.r
0.66
0.59
0.60
0.54
0.19
0.33

1

0.68
0.67
0.58
0.53
0.67
0.81

std.r
0.65
0.58
0.58
0.52
0.19
0.39

miss

0
0
0
0
0
0

As referenced within the text of Chapter 3, I estimated the reliability of the strategy measure.

96

Table C.2 Factor Analysis of Items Used to Create Strategy Measure

Variable
Member of Party
Open Seat
District Composition
Prior Experience
Supported by Party
Financial Resources

MR1 MR2
-0.01
0.89
0.10
0.44
0.58
-0.10
0.08
0.21
0.99
-0.02
0.04
-0.02

h2
0.80
0.19
0.37
0.04
1.00
0.00

u2
0.20
0.81
0.63
0.96
0.00
1.00

com
1.00
1.11
1.05
1.29
1.00
1.50

SS loadings
Proportion Var
Cumulative Var
Proportion Explained
Cumulative Proportion

1.38
0.23
0.23
0.58
0.58

1.02
0.17
0.40
0.42
1.00

MR1
MR2

1.00
-0.20

-0.20
1.00

As referenced within the text of Chapter 3, I performed factor analysis on the items used to create a strategy measure.

97

Table C.3 The Impact of Training Programs on Candidacy at the State Level: Model with Time E ects

Displayed

Dependent variable: Candidates

(1)

0.165⇤⇤
(0.073)

 0.149
(0.349)

0.004
(0.003)
 0.093⇤⇤⇤
(0.006)

0.022
(0.110)
 0.835⇤⇤⇤
(0.107)
 0.500⇤⇤⇤
(0.126)
 0.101⇤⇤⇤
(0.028)
 0.001
(0.006)

(2)

0.121⇤⇤⇤
(0.024)
 0.454
(0.346)

0.003
(0.003)
 0.096⇤⇤⇤
(0.006)
 0.125
(0.111)
 0.845⇤⇤⇤
(0.104)
 0.553⇤⇤⇤
(0.123)
 0.110⇤⇤⇤
(0.027)
 0.00001
(0.006)

Campaign Focused Program

Women Focused Program

Legislative Professionalism

Urban

Holdover

West

Midwest

Northeast

Traditional Party Organization

Percent Democrat

Ranney Index

Term lImit

Opinion about Women as Candidates

Divided Government

Policy Liberalism _median

1998

2000

2002

2004

2006

2008

2010

Constant

4.403⇤⇤⇤
(0.255)

4.523⇤⇤⇤
(0.251)

(3)
 0.084
(0.068)

 0.409
(0.354)

0.001
(0.003)
 0.102⇤⇤⇤
(0.005)

0.472⇤⇤⇤
(0.108)

0.282
(0.158)
 0.002
(0.126)
 0.085⇤⇤⇤
(0.025)
 0.013⇤⇤
(0.006)
 0.088
(0.281)
 0.074
(0.070)

0.920⇤⇤⇤
(0.184)
 0.073
(0.061)

0.238⇤⇤⇤
(0.057)

0.034
(0.117)
 0.040
(0.118)
 0.061
(0.119)
 0.030
(0.124)
 0.038
(0.127)
 0.051
(0.132)
 0.091
(0.135)

1.293⇤
(0.746)

(4)

0.092⇤⇤⇤
(0.022)
 0.640⇤
(0.346)
 0.001
(0.003)
 0.103⇤⇤⇤
(0.005)

0.277⇤⇤
(0.111)

0.141
(0.153)
 0.085
(0.124)
 0.101⇤⇤⇤
(0.025)
 0.009
(0.006)
 0.190
(0.276)
 0.142⇤⇤
(0.069)

0.906⇤⇤⇤
(0.177)
 0.067
(0.059)

0.189⇤⇤⇤
(0.056)

0.049
(0.114)
 0.069
(0.115)
 0.098
(0.116)
 0.121
(0.122)
 0.121
(0.125)
 0.180
(0.131)
 0.216
(0.134)

1.519⇤⇤
(0.720)

400

 1,816.540
2.351⇤⇤⇤ (0.196)
3,653.079

400

 1,806.468
2.497⇤⇤⇤ (0.210)
3,632.935

392

 1,727.261
3.523⇤⇤⇤ (0.311)
3,498.523

392

 1,719.348
3.717⇤⇤⇤ (0.332)
3,482.695

⇤⇤p<0.05; ⇤⇤⇤p<0.01

Table 5 displays the full models depicted in the main body of the paper as Table 1. Here the ﬁxed e ects for year are
displayed with accompanying coe cients and indicators for statistical signiﬁcance.

Observations
Log Likelihood
✓
Akaike Inf. Crit.
Note:

98

Table C.4 The Impact of Training Programs on Strategic Entry: Ordered Logit Models

Dependent variable: Strategic Entry

Democrat

Woman

College graduate

Graduate education

Candidate Training Program Participation

First O ce

Decision to Run

Dissatisfaction

Campaign Work

Involved in Party

Active Party Recruitment

0|1

1|2

2|3

3|4

4|5

5|6

Strategy

Model 1
 0.129⇤
(0.070)

0.141⇤⇤
(0.068)

0.063
(0.094)
 0.002
(0.090)

0.163⇤⇤
(0.109)
 0.246⇤⇤
(0.106)

Model 2
 0.302⇤⇤⇤
(0.125)

0.101
(0.122)
 0.008
(0.169)
 0.166
(0.160)

0.282⇤
(0.149)
 0.385⇤⇤⇤
(0.147)

0.2071⇤⇤⇤
(0.077)

0.339⇤⇤⇤
(0.136)

 0.368⇤⇤⇤
(0.150)

0.645⇤⇤⇤
(0.136)

0.200⇤
(0.125)

0.003
(0.120)
 5.576
(0.625)
 3.038
(0.291)
 1.312
(0.251)
 0.321
(0.245)
0.834
(0.247)
2.249
(0.258)

 2.754
(0.223)
 1.781
(0.136)
 0.910
(0.122 )
 0.328
(0.120)
0.355
(0.120)
1.156
(0.124)

980

3255.731
3281.731

928

3043.84
3077.84

Table 6 displays an alternative speciﬁcation of the models from Table 2 in the body of the paper. The models in Table
2 are ordinary least squares models. The models displayed here are ordered logit models.

Observations
Residual Deviance
AIC
Note:

⇤p<0.1; ⇤⇤p<0.05; ⇤⇤⇤p<0.01

99

APPENDIX D

FOURTH APPENDIX

1. Can you pinpoint an experience or event that sparked your interest in politics?

2. Please tell us about who informs your perspective on politics.

a) As a child, who talked to you about politics? What do you talk about?

b) Who do you talk about politics with now? Would these same people encourage your

interest in running for o ce?

3. Do you perceive any barriers for women in politics? What would those barriers be, if any?

4. Do you perceive any barriers that speciﬁcally apply to women of color in politics? What

would those barriers be, if any?

5. Please tell us about your experience in MPLP.

a) Do you still keep in contact with fellow from your class? If so, in what context?

b) Would you participate in MPLP again?

c) Did your race play a role in your experience as a MPLP fellow?

d) Did your gender play a role in your experience as a MPLP fellow?

e) Did you have an intersectional experiences as a MPLP fellow?

6. Before you applied to the MPLP program, had you worked on a political campaign or served
on the sta  of an elected o cial? If so, what was the gender identity of the candidate for the
political campaign and/or legislator you worked for?

7. Have you run for o ce since applying/completing the MPLP program?

a) If not.....

100

i. Do you see yourself running for o ce sometime in the future?

ii. What factors are inﬂuencing your decision to not run for o ce?

iii. What role did the MPLP program play in your decision to not run for o ce?

iv. If you had the necessary political support and opportunities, is there one position

in particular that you would like to hold?

v. Were you discouraged from running for o ce at one point? If so, who discouraged

you?

vi. Were you encouraged to run for o ce? If so, who encouraged you?

b) If so......

i. What factors inﬂuenced your decision to run for o ce the ﬁrst time?

ii. What role did the MPLP program play in your decision to run for o ce?

iii. Where you discouraged from o ce at any point? If so, who discouraged you?

iv. Did you win o ce?

v. Were you encouraged to run for o ce? If so, who encouraged you?

101

BIBLIOGRAPHY

102

BIBLIOGRAPHY

Ai, Chunrong and Edward C Norton. 2003. “Interaction terms in logit and probit models.” Eco-

nomics letters 80(1):123–129.

Aldrich, John H. 1995. Why parties?: The origin and transformation of political parties in America.

University of Chicago Press.

Alter, Charlotte. 2018. “A Year Ago, They Marched. Now a Record Number of Women Are Running

for O ce.”.
URL: http://time.com/5107499/record-number-of-women-are-running-for-o ce/

Atkeson, Lonna Rae. 2003. “Not all cues are created equal: The conditional impact of female

candidates on political engagement.” Journal of Politics 65(4):1040–1061.

Bedolla, Lisa Garcia and Melissa R Michelson. 2012. Mobilizing inclusion: Transforming the

electorate through get-out-the-vote campaigns. Yale University Press.

Bejarano, Christina E. 2013. The Latina Advantage: Gender, Race, and Political Success. Univer-

sity of Texas Press.

Black, Gordon S. 1972. “A theory of political ambition: Career choices and the role of structural

incentives.” American Political Science Review 66(1):144–159.

Bond, Jon R, Cary Covington and Richard Fleisher. 1985. “Explaining challenger quality in

congressional elections.” The Journal of Politics 47(2):510–529.

Boyd, Richard W. 1989. “The e ects of primaries and statewide races on voter turnout.” The

Journal of Politics 51(3):730–739.

Branton, Regina P. 2009. “The importance of race and ethnicity in congressional primary elections.”

Political Research Quarterly 62(3):459–473.

Bratton, Kathleen A and Kerry L Haynie. 1999. “Agenda setting and legislative success in state

legislatures: The e ects of gender and race.” The Journal of Politics 61(3):658–679.

Broockman, David E. 2014a. “Do female politicians empower women to vote or run for o ce? A

regression discontinuity approach.” Electoral Studies 34:190–204.

Broockman, David E. 2014b. “Mobilizing candidates: Political actors strategically shape the
candidate pool with personal appeals.” Journal of Experimental Political Science 1(2):104–119.
Broockman, David, Nicholas Carnes, Melody Crowder-Meyer and Christopher Skovron. 2014.
Who’sa Good Candidate? How Party Gatekeepers Evaluate Potential Nominees. In The Annual
Meeting of the American Political Science Association, Washington DC.

Brown, Nadia E. 2014a. “Political participation of women of color: An intersectional analysis.”

Journal of Women, Politics & Policy 35(4):315–348.

103

Brown, Nadia E. 2014b. Sisters in the Statehouse: Black Women and Legislative Decision Making.

Oxford University Press.

Burrell, Barbara C. 2010. Political parties and women’s organizations: Bringing women into the

electoral arena. Second ed. Cambridge University Press.

Butler, Daniel M and Jessica Robinson Preece. 2016. “Recruitment and perceptions of gender bias

in party leader support.” Political Research Quarterly 69(4):842–851.

Caldeira, Gregory A and Samuel C Patterson. 1982. “Contextual inﬂuences on participation in US

state legislative elections.” Legislative Studies Quarterly pp. 359–381.

Calhoun-Brown, Allison. 1996. “African American churches and political mobilization: The

psychological impact of organizational resources.” The Journal of Politics 58(4):935–953.

Campbell, David E and Christina Wolbrecht. 2006. “See Jane run: Women politicians as role

models for adolescents.” Journal of Politics 68(2):233–247.

Canon, David T. 1990. Actors, athletes, and astronauts: Political amateurs in the United States

Congress. University of Chicago Press.

Canon, David T. 1994. “„The Social Bases of Legislative Recruitment “.” Encyclopedia of the

American Legislative System 1:321–342.

Canon, David T. 1999. Race, redistricting, and representation: The unintended consequences of

black majority districts. University of Chicago Press.

Carroll, Susan J and Kira Sanbonmatsu. 2013. More women can run: Gender and pathways to the

state legislatures. Oxford University Press.

Caughey, Devin and Christopher Warshaw. 2018. “Policy preferences and policy change: Dy-
namic responsiveness in the American states, 1936–2014.” American Political Science Review
112(2):249–266.

Chong, Dennis and Reuel Rogers. 2005. “Racial solidarity and political participation.” Political

Behavior 27(4):347–374.

Citrin, Jack, Donald Philip Green and David O Sears. 1990. “White reactions to black candidates:

When does race matter?” Public opinion quarterly 54(1):74–96.

Collins, Patricia Hill. 2015. “Intersectionality’s deﬁnitional dilemmas.” Annual Review of Sociology

41:1–20.

Cox, Gary W and Michael C Munger. 1989. “Closeness, expenditures, and turnout in the 1982 US

House elections.” American Political Science Review 83(1):217–231.

Crenshaw, Kimberlé. 1991. “Mapping the margins: Identity politics, intersectionality, and violence

against women.” Stanford Law Review 43(6):1241–1299.

104

Crowd Counting Consortium. 2017.

URL: https://sites.google.com/view/crowdcountingconsortium/home

Crowder-Meyer, Melody. 2013. “Gendered recruitment without trying: how local party recruiters

a ect women’s representation.” Politics & Gender 9(4):390–413.

DiCiccio, Thomas J and Bradley Efron. 1996. “Bootstrap conﬁdence intervals.” Statistical science

pp. 189–212.

Dittmar, Kelly. 2015a. “Encouragement is not enough: Addressing social and structural barriers to

female recruitment.” Politics & Gender 11(4):759–765.

Dittmar, Kelly. 2015b. Navigating gendered terrain: Stereotypes and strategy in political cam-

paigns. Temple University Press.

Dittmar, Kelly. 2018a. “"Pink Wave": A Note of Caution.”.

URL: http://www.cawp.rutgers.edu/footnotes/pink-wave-note-caution

Dittmar, Kelly. 2018b. “Putting the Record Numbers of Women’s Candidacies into Context Putting
the Record Numbers of Women’s Candidacies into Context Putting the Record Numbers of
Women’s Candidacies into Context.”.
URL:
context

http://www.cawp.rutgers.edu/footnotes/putting-record-numbers-womens-candidacies-

Dolan, Kathleen. 2006. Women candidates in American politics: What we know, what we want to

know. In meeting of the Midwest Political Science Association.

Dolan, Kathleen and Kira Sanbonmatsu. 2009. “Gender stereotypes and attitudes toward gender

balance in government.” American Politics Research 37(3):409–428.

Dolan, Kathleen and Lynne E Ford. 1997. “Change and continuity among women state legislators:

Evidence from three decades.” Political Research Quarterly 50(1):137–151.

Dolan, Kathleen and Timothy Lynch. 2015. “Making the connection? Attitudes about women in

politics and voting for women candidates.” Politics, Groups, and Identities 3(1):111–132.

Farris, Emily M and Mirya R Holman. 2014. “Social capital and solving the puzzle of Black

women’s political participation.” Politics, Groups, and Identities 2(3):331–349.

Fowler, James H. 2006a. “Connecting the Congress: A study of cosponsorship networks.” Political

Analysis 14(4):456–487.

Fowler, James H. 2006b. “Legislative cosponsorship networks in the US House and Senate.” Social

Networks 28(4):454–465.

Fowler, Linda. 2010. Analytical Perspectives On Politics: Candidates, Congress, and the American

Democracy. University of Michigan Press.

Fowler, Linda L. 1993. Candidates, Congress, and the American democracy. University of

Michigan Press.

105

Fowler, Linda L and Robert D McClure. 1990. Political ambition: Who decides to run for Congress.

Yale University Press.

Fox, Richard L and Jennifer L Lawless. 2004. “Entering the arena? Gender and the decision to run

for o ce.” American Journal of Political Science 48(2):264–280.

Fox, Richard L and Jennifer L Lawless. 2005. “To run or not to run for o ce: Explaining nascent

political ambition.” American Journal of Political Science 49(3):642–659.

Fox, Richard L and Jennifer L Lawless. 2010. “If only they’d ask: Gender, recruitment, and political

ambition.” The Journal of Politics 72(2):310–326.

Fox, Richard L and Jennifer L Lawless. 2011a. “Gaining and losing interest in running for public

o ce: the concept of dynamic political ambition.” The Journal of Politics 73(02):443–462.

Fox, Richard L and Jennifer L Lawless. 2011b. “Gendered perceptions and political candidacies: A
central barrier to women’s equality in electoral politics.” American Journal of Political Science
55(1):59–73.

Fox, RL and SJ Carroll. 2006. “Gender and Elections: Shaping the future of American Politics.”

New York .

Fraga, Bernard L. 2016. “Candidates or districts? Reevaluating the role of race in voter turnout.”

American Journal of Political Science 60(1):97–122.

Gay, Claudine. 2001. The e ect of black congressional representation on political participation. In

American Political Science Association. Vol. 95 Cambridge Univ Press pp. 589–602.

Gerber, Alan S and Donald P Green. 2000. “The e ects of canvassing, telephone calls, and direct

mail on voter turnout: A ﬁeld experiment.” American political science review 94(3):653–663.

Gerber, Alan S, Donald P Green and Christopher W Larimer. 2008. “Social pressure and voter
turnout: Evidence from a large-scale ﬁeld experiment.” American Political Science Review
102(1):33–48.

Gerber, Alan S, Donald P Green and Christopher W Larimer. 2010. “An experiment testing the
relative e ectiveness of encouraging voter participation by inducing feelings of pride or shame.”
Political Behavior 32(3):409–422.

Gerber, Alan S, Donald P Green and Ron Shachar. 2003. “Voting may be habit-forming: evidence

from a randomized ﬁeld experiment.” American Journal of Political Science 47(3):540–550.

Gillespie, Andra. 2010. Whose Black politics?: cases in post-racial Black leadership. Routledge.
Gimpel, James G, Frances E Lee and Joshua Kaminski. 2006. “The political geography of campaign

contributions in American politics.” Journal of Politics 68(3):626–639.

Githens, Marianne and Jewel Limar Prestage. 1977. A portrait of marginality: The political

behavior of the American woman. Addison-Wesley Longman Ltd.

106

Goodli e, Jay. 2001. “The e ect of war chests on challenger entry in US House elections.” American

Journal of Political Science pp. 830–844.

Green, Donald P. 2004. “Mobilizing African-American voters using direct mail and commercial

phone banks: A ﬁeld experiment.” Political Research Quarterly 57(2):245–255.

Green, Donald P and Michael Schwam-Baird. 2016. “Mobilization, participation, and American

democracy: A retrospective and postscript.” Party Politics 22(2):158–164.

Hamm, Keith E and Robert E Hogan. 2008. “Campaign ﬁnance laws and candidacy decisions in

state legislative elections.” Political Research Quarterly .

Hancock, Ange-Marie. 2007a. “Intersectionality as a normative and empirical paradigm.” Politics

& Gender 3(2):248–254.

Hancock, Ange-Marie. 2007b. “When multiplication doesn’t equal quick addition: Examining

intersectionality as a research paradigm.” Perspectives on politics 5(1):63–79.

Harris, Fredrick C. 1994. “Something within: Religion as a mobilizer of African-American political

activism.” The Journal of Politics 56(1):42–68.

Harris, Fredrick C, Valeria Sinclair-Chapman and Brian D McKenzie. 2005. “Macrodynamics of
Black political participation in the post-civil rights era.” The Journal of Politics 67(4):1143–1163.
Hennings, Valerie M. 2011. Civic selves: Gender, candidate training programs, and envisioning

political participation PhD thesis University of Wisconsin–Madison.

Hero, Rodney E and Robert R Preuhs. 2010. “Black-Latino political relationships: policy voting

in the US House of Representatives.” American Politics Research 38(3):531–562.

Highton, Benjamin. 2004. “White voters and African American candidates for congress.” Political

Behavior 26(1):1–25.

Ho, Daniel E, Kosuke Imai, Gary King and Elizabeth A Stuart. 2007. “Matching as nonparametric
preprocessing for reducing model dependence in parametric causal inference.” Political analysis
15(3):199–236.

Hogan, Robert E. 2008. “Policy responsiveness and incumbent reelection in state legislatures.”

American Journal of Political Science 52(4):858–873.

Holder, Sarah. 2017. “The New Wave of Local Candidates Running for O ce.”.

URL: https://www.citylab.com/life/2017/11/new-wave-of-local-candidates/544853/

Huckfeldt, R Robert and John Sprague. 1995. Citizens, politics and social communication: Infor-

mation and inﬂuence in an election campaign. Cambridge University Press.

Huckshorn, Robert Jack and Robert C Spencer. 1971. The politics of defeat: Campaigning for

congress. Univ of Massachusetts Press.

107

Hutchings, Vincent L and Nicholas A Valentino. 2004. “The centrality of race in American politics.”

Annu. Rev. Polit. Sci. 7:383–408.

Iacus, Stefano M, Gary King and Giuseppe Porro. 2012. “Causal inference without balance

checking: Coarsened exact matching.” Political analysis 20(1):1–24.

Jacobson, Gary C. 1981. “Incumbents’ advantages in the 1978 US congressional elections.”

Legislative Studies Quarterly pp. 183–200.

Jacobson, Gary C. 2001. “A House and Senate divided: The Clinton legacy and the congressional

elections of 2000.” Political Science Quarterly 116(1):5–27.

Jacobson, Gary C and Samuel Kernell. 1983. “Strategy and choice in congressional elections.”.
Jacoby, William G. 1991. Data theory and dimensional analysis. Number 77-78 Sage.
Jensen, Jennifer M and Wendy L Martinek. 2009. “The e ects of race and gender on the judicial

ambitions of state trial court judges.” Political Research Quarterly 62(2):379–392.

Johnson, Gbemende, Bruce I Oppenheimer and Jennifer L Selin. 2012. “The House as a Stepping
Stone to the Senate: Why Do So Few African American House Members Run?” American
Journal of Political Science 56(2):387–399.

Juenke, Eric Gonzalez. 2014. “Ignorance Is Bias: The E ect of Latino Losers on Models of Latino

Representation.” American Journal of Political Science 58(3):593–603.

Juenke, Eric Gonzalez and Paru Shah. 2015. “Not the usual story: the e ect of candidate supply
on models of Latino descriptive representation.” Politics, Groups, and Identities 3(3):438–453.
Juenke, Eric Gonzalez and Paru Shah. 2016. “Demand and Supply: Racial and Ethnic Minority

Candidates in White Districts.” Journal of Race, Ethnicity and Politics 1(1):60–90.

Juenke, Eric Gonzalez and Robert R Preuhs. 2012. “Irreplaceable legislators? Rethinking minority

representatives in the new century.” American Journal of Political Science 56(3):705–715.

Kam, Cindy D. 2007. “Implicit attitudes, explicit choices: When subliminal priming predicts

candidate preference.” Political Behavior 29(3):343–367.

Kanthak, Kristin and Jonathan Woon. 2015. “Women don’t run? Election aversion and candidate

entry.” American Journal of Political Science 59(3):595–612.

Karpowitz, Christopher F, J Quin Monson and Jessica Robinson Preece. 2017. “How to elect more
women: Gender and candidate success in a ﬁeld experiment.” American Journal of Political
Science 61(4):927–943.

Kazee, Thomas A. 1994. Who runs for Congress?: ambition, context, and candidate emergence.

Cq Pr.

Kenny, Christopher B. 1992. “Political participation and e ects from the social environment.”

American Journal of Political Science pp. 259–267.

108

Krupnikov, Yanna, Spencer Piston and Nichole M Bauer. 2015. “Saving Face: Identifying Voter

Responses to Black Candidates and Female Candidates.” Political Psychology .

Landsbaum, Claire. 2017. “First They Marched, Now More Than 13,000 Women Are Planning to

Run for O ce.”.
URL:
o ce.html

https://www.thecut.com/2017/02/an-unprecedented-number-of-women-plan-to-run-for-

Lawless, Jennifer L. 2012. Becoming a candidate: Political ambition and the decision to run for

o ce. Cambridge University Press.

Lawless, Jennifer L and Richard L Fox. 2005. It takes a candidate: Why women don’t run for o ce.

Cambridge University Press.

Leighley, Jan E. 1990. “Social interaction and contextual inﬂuences on political participation.”

American Politics Quarterly 18(4):459–475.

Leighley, Jan E. 1995. “Attitudes, opportunities and incentives: A ﬁeld essay on political partici-

pation.” Political research quarterly 48(1):181–209.

Leighley, Jan E and Arnold Vedlitz. 1999. “Race, ethnicity, and political participation: Competing

models and contrasting explanations.” The Journal of Politics 61(4):1092–1114.

Maestas, Cherie D, Sarah Fulton, L Sandy Maisel and Walter J Stone. 2006. “When to risk it?
Institutions, ambitions, and the decision to run for the US House.” American Political Science
Review 100(2):195–208.

Maisel, L Sandy, LL Fowler, RS Jones and WJ Stone. 1990. “The Naming of Candidates: Recruit-
ment or Emergence.” The Parties Respond: Changes in the American Party System pp. 137–159.
Maisel, L Sandy and Walter J Stone. 1997. “Determinants of candidate emergence in US house

elections: An exploratory study.” Legislative Studies Quarterly pp. 79–96.

Mansbridge, Jane. 1999. “Should blacks represent blacks and women represent women? A

contingent" yes".” The Journal of politics 61(3):628–657.

Marschall, Melissa J, Anirudh VS Ruhil and Paru R Shah. 2010. “The new racial calculus: Electoral
institutions and black representation in local legislatures.” American Journal of Political Science
54(1):107–124.

Matsubayashi, Tetsuya. 2010. “Racial environment and political participation.” American Politics

Research 38(3):471–501.

Matthews, Donald R. 1984. “Legislative recruitment and legislative careers.” Legislative Studies

Quarterly pp. 547–585.

McClurg, Scott D. 2006. “The electoral relevance of political talk: Examining disagreement and
expertise e ects in social networks on political participation.” American Journal of Political
Science 50(3):737–754.

109

Miller, Joanne M and Jon A Krosnick. 2004. “Threat as a motivator of political activism: A ﬁeld

experiment.” Political Psychology 25(4):507–523.

Minta, Michael D and Nadia E Brown. 2014. “Intersecting Interests: Gender, Race, and Con-
gressional Attention to Women’s Issues.” Du Bois Review: Social Science Research on Race
11(2):253–272.

Moncrief, Gary F. 1999. “Recruitment and retention in US legislatures.” Legislative Studies

Quarterly pp. 173–208.

Moncrief, Gary F, Peverill Squire and Malcolm E Jewell. 2001. Who runs for the legislature?

Prentice Hall.

Mondak, Je ery J. 1995. “Competence, integrity, and the electoral success of congressional

incumbents.” The Journal of Politics 57(4):1043–1069.

Morehouse, Sarah M and Malcolm E Jewell. 2004. “States as laboratories: A reprise.” Annu. Rev.

Polit. Sci. 7:177–203.

Moskowitz, David and Patrick Stroh. 1994. “Psychological sources of electoral racism.” Political

Psychology pp. 307–329.

Nickerson, David W. 2008. “Is voting contagious? Evidence from two ﬁeld experiments.” American

Political Science Review 102(1):49–57.

Nickerson, David W, Ryan D Friedrichs and David C King. 2006. “Partisan mobilization campaigns
in the ﬁeld: Results from a statewide turnout experiment in Michigan.” Political Research
Quarterly 59(1):85–97.

Nilsen, Ella. 2017. “The women, people of color, and LGBTQ candidates who made history in the

2017 election.”.
URL: https://www.vox.com/policy-and-politics/2017/11/8/16622884/women-minorities-lgbtq-
candidates-made-history

Niven, David. 1998. “Party elites and women candidates: The shape of bias.” Women & Politics

19(2):57–80.

Norton, Edward C, Hua Wang, Chunrong Ai et al. 2004. “Computing interaction e ects and

standard errors in logit and probit models.” Stata Journal 4:154–167.

Ondercin, Heather L and Susan Welch. 2009. “Comparing Predictors of Women’s Congressional
Election Success: Candidates, Primaries, and the General Election.” American Politics Research
37(4):593–613.

Ondercin, Heather, Susan Welch, Sue Thomas and Clyde Wilcox. 2005. “Women and elective

o ce: past, present, and future.” Women and elective o ce: past, present, and future .

Palmer, Barbara and Dennis Simon. 2001. “The political glass ceiling: Gender, strategy, and

incumbency in US house elections, 1978-1998.” Women & Politics 23(1-2):59–78.

110

Palmer, Barbara and Dennis Simon. 2003. “Political ambition and women in the US House of

Representatives, 1916-2000.” Political Research Quarterly 56(2):127–138.

Palmer, Barbara and Dennis Simon. 2010. Breaking the political glass ceiling: Women and

congressional elections. Routledge.

Philpot, Tasha S and Hanes Walton. 2007. “One of our own: Black female candidates and the

voters who support them.” American Journal of Political Science 51(1):49–62.

Pinderhughes, D, P Lien, C Hardy-Fanta and C Sierra. 2006. “Gender, Race, and Descriptive
Representation in the United States: Findings from the Gender and Multicultural Leadership
Project.” Journ. of Women and Public Policy 28(3/4):14.

Pitkin, Hanna Fenichel. 1967. The concept of representation. Univ of California Press.
Platt, Matthew B. 2008. “Participation for what? A policy-motivated approach to political activism.”

Political Behavior 30(3):391–413.

Preece, Jessica Robinson. 2016. “Mind the Gender Gap: An Experiment on the Inﬂuence of

Self-E cacy on Political Interest.” Politics &amp; Gender 12(1):198–217.

Preece, Jessica Robinson and Olga Bogach Stoddard. 2015. “Does the message matter? A ﬁeld
experiment on political party recruitment.” Journal of Experimental Political Science 2(01):26–
35.

Preece, Jessica Robinson, Olga Bogach Stoddard and Rachel Fisher. 2016. “Run, Jane, run!

Gendered responses to political party recruitment.” Political Behavior 38(3):561–577.

Preuhs, Robert R and Eric Gonzalez Juenke. 2011. “Latino US State Legislators in the 1990s:
Majority-Minority Districts, Minority Incorporation, and Institutional Position.” State Politics &
Policy Quarterly 11(1):48–75.

Prewitt, Kenneth. 1970. “Political ambitions, volunteerism, and electoral accountability.” American

Political Science Review 64(01):5–17.

Prewitt, Kenneth and Heinz Eulau. 1971. “Social bias in leadership selection, political recruitment,

and electoral context.” The Journal of Politics 33(02):293–315.

Rauch, Jonathan and Raymond La Raja. 2017. “Re-engingeering politicians: How activist groups

choose our candidates - long before we vote.”.
URL:
choose-our-politicians-long-before-we-vote/

https://www.brookings.edu/research/re-engineering-politicians-how-activist-groups-

Reny, Tyler and Paru Shah. N.d. “New Americans and the Quest for Political O ce.” Social Science

Quarterly. Forthcoming.

Rohde, David W. 1979. “Risk-bearing and progressive ambition: The case of members of the

United States House of Representatives.” American Journal of Political Science pp. 1–26.

111

Rosenstone, Steven J and John Hansen. 1993. Mobilization, participation, and democracy in

America. Macmillan Publishing Company,.

Rozell, Mark J. 2000. “Helping women run and win: Feminist groups, candidate recruitment and

training.” Women & Politics 21(3):101–116.

Sanbonmatsu, Kira. 2002. “Gender stereotypes and vote choice.” American Journal of Political

Science pp. 20–34.

Sanbonmatsu, Kira. 2006. “State Elections: Where Do Women Run? Where Do Women Win?”

Gender and Elections: Shaping the Future of American Politics pp. 189–214.

Sanbonmatsu, Kira. 2010. Where women run: Gender and party in the American states. University

of Michigan Press.

Sanbonmatsu, Kira. 2015. “Electing women of color: The role of campaign trainings.” Journal of

Women, Politics &amp; Policy 36(2):137–160.

Sanbonmatsu, Kira, Susan J Carroll and Debbie Walsh. 2009. “Poised to run: Women’s pathways
to the state legislatures.” Center for American Women and Politics, Eagleton Institute of Politics,
Rutgers University .

Sanbonmatsu, Kira, Susan J. Carroll and Debbie Walsh. 2015. “Center for American Women and

Politics (CAWP) Recruitment Studies, 2008.”.

Schlesinger, Joseph A. 1966. “Ambition and politics: Political careers in the United States.”.
Schneider, Monica C and Jennie Sweet-Cushman. 2018. “Campaign Trainings for Women: Col-

lecting Puzzle Pieces for a Run for O ce.”.

Scola, Becki. 2007. “Women of color in state legislatures: Gender, race, ethnicity and legislative

o ce holding.” Journal of Women, Politics & Policy 28(3-4):43–70.

Scola, Becki. 2013. “Predicting presence at the intersections: Assessing the variation in women’s

o ce holding across the states.” State Politics &amp; Policy Quarterly 13(3):333–348.

Scott, Jamil and Corwin Smidt. 2017. “State Support for Women in Politics and its Impact on

Female Candidate Emergence.”.

Seligman, Lester G, Michael R King, CL Kim and RE Smith. 1974. Patterns of recruitment: a

state chooses its lawmakers. Cambridge Univ Press.

Shah, Paru. 2014. “It takes a Black candidate: A supply-side theory of minority representation.”

Political Research Quarterly 67(2):266–279.

Shames, Shauna L. 2015. “American women of color and rational non-candidacy: when silent
citizenship makes politics look like old white men shouting.” Citizenship Studies 19(5):553–569.
Shingles, Richard D. 1981. “Black consciousness and political participation: The missing link.”

American Political Science Review 75(1):76–91.

112

Sigelman, Carol K, Lee Sigelman, Barbara J Walkosz and Michael Nitz. 1995. “Black candidates,
white voters: Understanding racial bias in political perceptions.” American Journal of Political
Science pp. 243–265.

Simien, Evelyn M. 2007. “Doing intersectionality research: From conceptual issues to practical

examples.” Politics & Gender 3(2):264–271.

Sinclair, Betsy. 2012. The social citizen: Peer networks and political behavior. University of

Chicago Press.

Smooth, Wendy. 2006. “Intersectionality in electoral politics: a mess worth making.” Politics &

Gender 2(03):400–414.

Smooth, Wendy. 2014. “African American Women and Electoral Politics: Translating Voting
Power into O ce-holding.” Gender and Elections: Shaping the Future of American Politics
pp. 167–189.

Stock, Dayna M. 2012. Making Politics Personal: Leadership Programs as a Tool for Developing

Political Interest and E cacy in Young Women PhD thesis University of Missouri–St. Louis.

Stokes-Brown, Atiya Kai and Kathleen Dolan. 2010. “Race, Gender, and Symbolic Representa-
tion: African American Female Candidates as Mobilizing Agents.” Journal of Elections, Public
Opinion and Parties 20(4):473–494.

Summers, Juana. 2018. “Candidates of color get o  the sidelines in the age of Trump: ’The soul

of America is at stake right now’.”.
URL: https://www.cnn.com/2018/01/27/politics/candidates-of-color-age-of-trump/index.html

Swers, Michele L. 2005. “Connecting descriptive and substantive representation: An analysis of

sex di erences in cosponsorship activity.” Legislative Studies Quarterly 30(3):407–433.

Tate, Katherine. 1991. “Black political participation in the 1984 and 1988 presidential elections.”

American Political Science Review 85(4):1159–1176.

Terkildsen, Nayda. 1993. “When white voters evaluate black candidates: The processing impli-
cations of candidate skin color, prejudice, and self-monitoring.” American Journal of Political
Science pp. 1032–1053.

Thomas, Sue and Susan Welch. 1991. “The impact of gender on activities and priorities of state

legislators.” Western Political Quarterly 44(2):445–456.

Verba, Sidney, Kay Lehman Schlozman, Henry Brady and Norman H Nie. 1993. “Race, ethnicity
and political resources: Participation in the United States.” British Journal of Political Science
23(4):453–497.

Verba, Sidney, Kay Lehman Schlozman, Henry E Brady and Henry E Brady. 1995. Voice and

equality: Civic voluntarism in American politics. Vol. 4 Cambridge Univ Press.
Verba, Sidney and Norman H Nie. 1972. Participation in America. Harper & Row.

113

Verba, Sidney, Norman H Nie and Jae-on Kim. 1987. Participation and political equality: A

seven-nation comparison. University of Chicago Press.

Voss, D Stephen and David Lublin. 2001. “Black incumbents, white districts: An appraisal of the

1996 congressional elections.” American Politics Research 29(2):141–182.

Wahlke, John C, Heinz Eulau, William Buchanan and LeRoy C Ferguson. 1962. “TheLegislative

System.”.

Whitby, Kenny J. 2000. The color of representation: Congressional behavior and black interests.

University of Michigan Press.

Whitby, Kenny J. 2007. “The e ect of black descriptive representation on black electoral turnout

in the 2004 elections.” Social Science Quarterly 88(4):1010–1023.

Wilcox, Clyde and Barbara Norrander. 2005. “Change in Continuity in the Geography of Women

Legislators.” Women in Elective O ce pp. 176–196.

Zaller, John. 1998. “Politicians as prize ﬁghters: Electoral selection and incumbency advantage.”

Politicians and party politics pp. 125–85.

114