Women hold a minority of seats in every state legislature in the United States. That is true today and has been true every day since the first legislature was organized in 1619 in what would become the United States. While many ethnic, religious, and social groups are underrepresented in political office, few can match the disproportionate level or the persistence of women's underrepresentation (Menefield 2001). The mystery of women's underrepresentation is even more distinct because unlike other political minorities, women, of course, are actually a population majority.
This study explores the role of political elites in the candidacy creation process with a unique approach. Here, the decisions and experiences of women and men who declared their candidacy for state legislative office are studied. Importantly, this group includes those who won office, those who lost, and the overlooked but crucial group who declared their candidacy but dropped out before election day. Studying all declared candidates provides a unique window not just on the recruiting process but on the negative recruiting process, that is, the effort of political elites to stop people from running, or to channel them into untenable political circumstances.
With both election data and a survey of declared candidates, this study addresses central issues regarding the interaction of gender and recruitment. To wit, do women receive less encouragement to run, do women receive encouragement to run in less attractive districts, and are women apt to follow such advice?
THE SUSPECTS
Scholars have pursued answers for this puzzling history of underrepresentation for decades. Most conclude that there are three main suspects: women, voters, and political elites.1
Other entities, such as the media (Niven 2005) and campaign contributors (Theilmann and Wilhite 1989) have come under scrutiny, but they are generally not seen as the most significant impediments to women's representation.
Voters
If voters are hesitant to support women candidates, there will clearly be few women officeholders, and given the meager incentive to run under such circumstances, there will presumably be fewer women candidates. Evidence, however, suggests that voters play little role in stymieing the political advancement of women, especially for down ballot offices such as state legislative seats. That is, controlling for political party and incumbency status, there is little discernible effect of candidate sex on voter preferences (Bernstein 1986; Burrell 1994; Darcy and Schramm 1977; Darcy, Welch, and Clark 1994; Ekstrand and Eckert 1981; Fox and Oxley 2003; Gaddie and Bullock 1995; Huddy and Terkildsen 1993a; Sigelman and Welch 1984).2
Similarly, the Fox and Lawless (2004) survey of professional men and women found that among those who had run for office, there was no difference between men's and women's likelihood of winning.
Women
Regardless of the voters' inclinations, if women lack interest in politics, or lack the background and resources necessary for candidacy, then there will be few women candidates and few women officeholders. This was a compelling explanation, particularly in the 1970s, for the paucity of women officeholders (Diamond 1977; Dubeck 1976). Two strains of research undercut the theory's practical applicability, however. First, rising numbers of women in the professions (such as law), which have produced the most political officeholders, did not alleviate the gender imbalance in candidacies (Welch 1978). Second, studies measuring political interest found women to have as much or more than men (Clark, Hadley, and Darcy 1989).
Compelling recent research makes it ever more clear that a modest candidacy rate is not the product of women's collective lack of political interest (Elder 2004; Fox and Lawless 2004). Richard Fox and Jennifer Lawless conducted the Citizen Political Ambition Study, a national survey of people in professions that tend to produce political candidates. Examining men and women of comparable background, they find (2004, 267–68) that men were much more likely to have considered running for office (men 59%, women 43%) even though women were not less but more interested in politics than men (high interest in local politics: men 41%, women 49%; high interest in national politics: men 31%, women 41%). They conclude that the obstacle to women's candidacy and officeholding clearly lies in the recruitment process.
Political Elites
Voters may be willing to support women, and women may be equally or more interested in politics, but that is not enough. If the political infrastructure is hostile to women as candidates, they may never receive encouragement to run, and in the process, countless potential political futures could be cut short.
Several recent studies of candidacy patterns offer indirect evidence suggesting the heavy hand of political elites weighing down women's political inclinations. David Lublin and Sarah Brewer (2003) studied officeholders in county politics in several southern states. They found that women rarely held executive positions, law enforcement positions, or positions of considerable independent authority. Instead, elected women were typically found in “process-oriented jobs that have relatively little discretion” (Lublin and Brewer 2003, 391), such as county clerks. They conclude that women are channeled away from seeking desirable political positions, and that the infrastructure of political elites tolerates women “in areas where men do not want the jobs” (ibid., 391). Kira Sanbonmatsu (2002) studied women's candidacies for state legislature, and concomitantly found that where political parties were strongest and the positions more desirable, fewer women ran for office. Richard Fox and Zoe Oxley (2003) considered the candidacy patterns of women in statewide office. They find that women are particularly unlikely to run for stereotypically masculine positions such as attorney general, and more likely to seek positions such as education commissioner, with feminine or gender-neutral responsibilities. Given that they find no election pattern in which women are handicapped in winning votes for masculine offices, Fox and Oxley conclude (2003) that the recruitment process must account for this ongoing imbalance.
A few studies hone in more directly on the experiences of potential political candidates. Fox and Lawless's survey, for example, also speaks clearly to this point. Among their respondents, men were more likely than women to have been encouraged by political elites to run for office. The importance of this distinction is great, given that Fox and Lawless's results show that absent encouragement, neither men nor women are likely to consider running for office, but with encouragement, the vast majority of both men and women consider a political candidacy.
This author's (Niven's 1998) surveys of women holding local office and county political party leaders documented significant bias against potential women candidates. When asked about their experiences with party leaders, and the likelihood that they might seek state legislative office in the future, the majority of women (64%) replied that party leaders discouraged women's candidacies, primarily by channeling women into low-profile roles, though some reported that politically ambitious women were openly belittled. Indeed, when asked about their notion of a strong legislative candidate, the typical party chair described someone with masculine traits. And when asked directly about who they considered the best future prospects in their area for a legislative campaign, party chairs named twice as many men as women.
Admittedly, there are studies suggesting that women are treated equitably in the recruiting process (for example, Moncrief, Squire, and Jewell 2001). Those studies are largely based on the experiences of women in office, a highly unrepresentative sample that effectively eliminates from scrutiny precisely those individuals who have been successfully discouraged from running.
Why would political elites be biased against women when they should have a self-interested desire to encourage quality candidates regardless of sex? Fox and Oxley (2003) argue that it is not that women are thought ill-suited for politics, but that stereotypes continue to guide assumptions of “personality traits and policy competencies” that encourage political elites to view women as suited for some offices and not others (ibid., 847). Niven (1998) finds evidence of in-group preferences, in which party leaders hold slight but distinct preferences for candidates who resemble themselves on a whole range of features, such as occupation and personality traits. Given that the vast majority of party leaders are men, this results in a host of sometimes subtle and sometimes overt efforts to encourage the candidacies of men over women.
Psychological research has found that in-group favoritism is perpetuated even when such behavior is disadvantageous for the in-group (Brown, Collins, and Schmidt 1988). Beyond self-interest, many would imagine that education and expertise would deter people from falling prey to perceptual bias. Mary Jackman, however, has argued that education and position do not liberate people from intergroup negativism (Jackman and Muha 1984). In fact, when you believe yourself to be knowledgeable about something, you can more easily convince yourself that your decisions are based on expertise, not prejudice. Lee Jussim (1990) argues that those with high belief certainty have a low accuracy motivation because when they believe they are right, they do not invest much effort in seeking objective confirmation. When political elites are presented a potential female candidate, if their instincts tell them she would not make a good candidate, they could immediately assume that their decision is based on political knowledge and give the situation little further thought.
Both theory and nascent evidence suggest a need for continued development of evidence of how the process works. It is, as Fox and Oxley (2003) note, a difficult process to view because it takes place largely or entirely outside the public eye, but advancing an understanding of recruitment and its effects on women's candidacy rates nevertheless requires that we look closely.
METHODS
Fox and Lawless (2004) and Niven (1998) offer insight into the precandidacy phase, including in their samples a range of people who will someday run for office, as well as people who may never run for office for reasons having nothing to do with political elites. Lublin and Brewer (2003), Sanbonmatsu (2002), and Fox and Oxley (2003) offer a perspective on the end stage, the outcome of the interaction of ambition and recruitment that results in particular candidacies. This study advances our understanding by offering a focus on what takes place between those two points.
The study of all declared candidacies, by definition, provides a sample of people who want to run for office. These are people who are serious enough to take the procedural step of filing candidacy papers. Yet many take that step without ultimately staying in the race.
The relevance of these declared candidates who drop out, and the comparison between their experiences and those of candidates who persist, is that these may be the single most valuable people to understand if women are to increase their candidacy numbers. That is, practically speaking, finding women willing to run for office is an enormous undertaking, the metaphorical equivalent of reading the entire phone book one line at a time to find a particular listing. Applying the laws of physics to politics, noncandidates must be started from a state of inertia, a daunting task, while declared candidates are already running and need only not be stopped from continuing.
This study examines the experiences of all declared candidates for the state house and state senate in Florida in 2000 and 2002. After an examination of the basic candidacy data from the Florida Division of Elections, survey results from candidates who dropped out and who remained in the race are provided.
In January 2003, a survey was sent out to all major party candidates for the Florida state legislature from 2000 and 2002 who dropped out (men and women), all women who became official candidates, and a random sample of half of the male candidates who became official candidates.3
The survey is restricted to major party candidates (Democrats and Republicans) for two reasons. Of the 160 seats in the Florida house and senate, 160 are held by the two parties. Second, the requirements for running vary for nonparty candidates, including a provision that allows “write-in” candidates to file and run for office without meeting the requirements of major party candidates.
Candidates who did not respond were sent up to three follow-up letters. Those whose addresses had changed were sought by a search through publicly available databases.4
Candidates were required to list an address with the state when they declared their candidacy. Most provided home addresses; others provided campaign addresses. Candidates were also required to list an address for a campaign treasurer. If searches for the candidate's address were unsuccessful, up to two letters were sent to the treasurer's address.
Inclusion of candidates for whom no valid address could be established reduces the overall response rate to 48%, including 47% for men and 50% for women. There were no statistically significant differences between the respondents and the nonrespondents in terms of sex or the decision to remain in the race (although, by design, there was an oversample of women overall and of dropouts).
Florida is a significant state in which to study this dynamic as it is close to average in its gender representation. (According to the National Conference of State Legislatures, 24% of Florida legislators are women, a rate that ranks 22d among the 50 states.) In addition to being a geographically large and populous state, Florida also includes an array of political traditions, ranging from traditional southern politics evident in the state's Panhandle and northern reaches, to Sunbelt suburban politics in the state's center, to the politics of transplanted northern liberals and retirees in the southern region of the state. County political party organizations range from the highly organized and professionally staffed operations in Broward County to the more somnambulant practices in Hardee County (Hill, MacManus, and Moreno 2004). Organizationally, both parties remain vigorous statewide, fueled by the competitiveness of presidential and U.S. Senate contests and by the recent evolution of the state from Democratic dominance to Republican dominance in elections for state office (ibid.).
The state house has 120 seats, elected in single-member districts of about 133,000 constituents. The state senate has 40 seats, elected in single-member districts of about 400,000 constituents. Professionalized legislative campaigns have become the norm, with individual candidates spending at least $100,000 in competitive house races and sometimes as much as several million in competitive senate races (Hill, MacManus, and Moreno 2004). Overall, Republicans have dominated the legislature in recent years, capturing about two-thirds of the seats in both the house and senate.
Based on the implications of previous research, three major areas of concern are explored. First, are women subject to more discouragement and less encouragement to run than men? Second, are women channeled away from the most attractive races? Third, are women more susceptible to recruitment messages than men?
RESULTS
Candidacy Rates
The most basic observation about these state house and state senate candidacies is that there is a significant population of dropouts. In Florida, there were 944 declared candidacies for the state legislature in 2000 and 2002. Of those, 113 (12%) dropped out in various stages before appearing on the final ballot.
In Florida, to declare a candidacy requires that forms be submitted to the Department of State establishing the intent to run and naming a campaign treasurer. To become an official candidate, persons must either pay a fee of approximately $1,600 or submit petitions signed by approximately 1,000 (for house seats) to 3,000 (for senate seats) registered voters from within the district asking that the candidate be placed on the ballot. With the primary election in September, and the general election in November, the vast majority of dropouts left their race before the mid-July deadline to submit petitions or pay the fee. However, several dropped out in August and one waited until late September.6
Betsy Benedict dropped out September 27, 2000, as a candidate for state house District 56, despite having secured the Democratic nomination and having spent $130,000 on the campaign, most of it out of her own pocket. She was in the midst of a divorce and was facing foreclosure proceedings, and she cited the need to attend to her personal life, as well as the likelihood that her personal life would become a political issue (St. Petersburg Times, September 28, 2000).
To what extent are women more likely to drop out than men? The numbers are nearly indistinguishable. Of the 737 male candidates, 88 (11.9%) dropped out. Of the 207 female candidates, 25 (12.1%) dropped out. On the surface, the gender dynamic in dropouts appears nonexistent.
Looked at another way, according to these figures, women comprise 21.9% of declared legislative candidates. Surely that is an uninspiring total in a state where 25% of the seats are already held by women. There is little hope for increases in women's representation, let alone significant increases in women's representation, if their candidacy rates sink below their current representation level. Nevertheless, women also comprise 21.9% of official candidates, that is, after removing dropouts, women remain proportionally equal to their position as declared candidates.
Removing third party candidates from the analysis, which is a meaningful step in a state in which zero seats are held by third party members or independents, offers a similar picture. Among men, 607 declared their candidacy, with 75 (12.4%) dropping out. Among women, 176 declared their candidacy, with 18 (10.2%) dropping out. Here, we see that women major party candidates are slightly less likely to drop out than are men. Thus, women comprise 22.5% of the declared major party candidates, and 22.9% of official major party candidates.
What appears to be a process seemingly unaffected by gender dynamics changes significantly when one examines the nature of the races from which men and women are dropping out. Two-thirds of the women dropped out from races their party would go on to win. For men, 57% dropped out of races their party would win.7
A difference that is statistically significant, p < .05 using chi-square.
A difference that is statistically significant, p > .01 using T-test.
Assuming a certain baseline strategic motivation in all political candidates (see, for example, Jacobson and Kernell 1981), why would women be disproportionately more likely to drop out of attractive electoral circumstances? The implication, it would seem, is that the recruiting process differentially responds to women and men.
Survey Results
Declared candidates, both women and men, were asked if any political or community figures had encouraged them or recruited them to run, and if any such figures had discouraged them from running or encouraged them to drop out. The results, depicted in Figure 1, show that the majority of both men and women received encouragement, with women reporting slightly higher rates of positive response. While not a huge disparity, the results also show that women received more negative response to their prospective candidacies, a difference that reaches statistical significance.
Recruiting messages received by men and women state legislative candidates (N = 249).
Given the pattern displayed in the candidacy results, though, the more interesting question than how much encouraging and discouraging recruitment took place is what candidate attributes attracted that recruitment. That is, can we see evidence of women being dissuaded from running in attractive races while men were channeled away from hopeless causes?
Four logistic regression models were constructed to separately assess the determinants of encouraging and discouraging recruitment for men and women. Following previous work on the subject of recruitment and candidacy decisions (Miller 1986; Niven 1998; Rule 1981), independent variables included in the models are elected office experience (0 = none, 1 = other office, 2 = incumbent), previous election loss (0 = none, 1 = loss in any previous election), party strength (% of vote for candidate's party in last election for that office), political party (0 = Democrats, 1 = Republicans), age (candidate age in years), children (0 = candidate does not have children, 1 = has children), political occupation (0 = candidate does not work in typical political field, 1 = candidate does work in political field). The political experience variables presumably speak to electability, and presumably should be positive for elected office experience (that is, attracting encouraging recruitment and repelling discouraging recruitment) and negative for previous election loss. The political party variable will suggest whether this process varies between Democrats and Republicans. The party strength variable will demonstrate whether candidate recruitment is affected by the desirability of the race. The personal variables, such as age and children, speak to a concern in research (Blair and Henry 1981; Sapiro 1982) that women are disadvantaged by family life while men are either unaffected or advantaged by it. Finally, the political occupation variable suggests whether there is an advantage to having a more typical career background, such as law, politics, or other publicly oriented work.
Three out of the four models perform strongly, correctly assessing at least 85% of the cases and improving on the null hypothesis by at least 20% (Table 1). The model assessing encouraging recruitment for women offers only a modest improvement over chance alone, however. This suggests either a lack of a systematic basis for positive recruiting messages to women, or perhaps, that women have received encouragement from gender-oriented groups who welcome women's candidacies without regard to the factors tested here.
Attributes that draw recruitment among state legislative candidates (logistic regression)
The only variable that achieves statistical significance in all of the four models is party strength. Notably, comparing the party strength coefficient for encouraging recruitment of men to encouraging recruitment of women shows the slope in the opposite direction. The same is true for the variable when comparing discouraging recruitment of men to that of women. In other words, men are more likely to report encouraging recruitment efforts in districts where their party has a strong electoral base, while women are more likely to report encouraging recruitment efforts where their party is weak. Men are more likely to receive discouraging recruitment messages where their party is weak, while women are more apt to be discouraged from running where their party is strong.
To illustrate the scope of the difference, the odds ratio (not shown) for the party strength model is applied while holding all other variables constant. For the purposes of the demonstration, a strong district is assumed to offer a 20 point victory by the candidate's party, and a weak district a 20 point loss. As Table 2 illustrates, the disparities are sizable. The typical man receiving an encouraging recruitment message is in a strong district, while the typical woman is in a weak one. The typical man receiving a discouraging message is in a weak district, while the typical woman is in a strong one.
Estimated likelihood of drawing recruitment based on sex and party strength
Most other variables in the model appear to have little measurable effect. The exceptions are previous election loss and previous election victory. A previous loss seems to sully a candidate's prospects, as it reduces the likelihood of encouraging recruitment for men and increases the likelihood of discouraging recruitment for women. As one would expect, a previous election victory increases the likelihood of encouraging recruitment for women and decreases their likelihood of discouraging recruitment. Oddly, previous election victory is associated with a reduced likelihood of encouraging recruitment for men. This result suggests the possibility that previously successful men are self-starters and therefore willing to advance their candidacies irrespective of encouragement. Indeed, anecdotally Linda Fowler and Robert McClure (1989) found that women were more dependent on external validation of their candidacies when weighing whether to enter a political race, while men tended to assume the strengths of their candidacies regardless of others' reactions.
The survey results are supportive of that conclusion. Respondents were asked how strongly they were affected by the recruiting messages they had received, and provided a 5 point response scale, with endpoints labeled “1 = Not at All” and “5 = Very Strongly.” The mean score for men was 3.3. The mean score for women was 4.1 (p < .05 using T-test). This pattern can serve only to amplify the effect of recruitment imbalance. Not only are women prodded away from strong districts and toward weaker ones, but they are more likely to consider those messages significant. As both Niven (1998) and Fox and Lawless (2004) found, women are simply unlikely to run in the face of elite discouragement.
SUMMARY AND DISCUSSION
Consideration of the experiences of all declared candidates for the Florida state legislature in 2000 and 2002 provides several important pieces of evidence on why women remain in the minority of that state's legislature, and an even smaller minority of that state's legislative candidates.
According to state election figures, about 12% of declared legislative candidates drop out before election day. The dropout rate is quite similar for men and women. However, men are disproportionately more likely to drop out of races their party was unlikely to win, while women are disproportionately more likely to drop out of races their party was unlikely to lose.
A survey of declared candidates offers a clear explanation for this anomaly. Men receive encouragement from political elites to run in favorable districts and discouragement from political elites to run in unfavorable districts. Women receive the opposite messages. Moreover, women report being more apt to value the input they receive from political elites.
Obviously, this study does not provide evidence of how women in general are treated by the political process. Instead, it is a depiction of how women on the front lines of legislative candidacy are treated in Florida. Importantly, though, the findings here are consistent with the thrust of recent research that highlights an absence of women from particular political offices, and posits a significant role for political party leaders and other political elites in reducing women's candidacy rate.
In addition to offering a unique look at gender dynamics in the recruiting process, these results suggest the need for greater attention to the concept of negative recruitment. For the most part, recruitment has been studied as a question of determining who is encouraged to run. These results suggest that just as important are the two prongs of negative recruitment, that is, who is discouraged from running and who is encouraged to run in hopeless circumstances.
The candidates who were lost as drop outs were on the cusp of an official candidacy. They represent an efficient source of additional women candidates. They need not be convinced of their interest in politics or in running for office, but only of their need to continue. In Florida, keeping all the declared women candidates in the race would mean increasing not only the number of women running but also the number of women running in winnable districts.
APPENDIX: CANDIDATE SURVEY
Did anyone from a political party, political group, or community group encourage or recruit you to run for the state legislature, or encourage you to stay in the race for the state legislature?
Yes
No
Did anyone from a political party, political group, or community group discourage you from running for the state legislature or discourage you from staying in the race for the state legislature?
Yes
No
Was your ultimate decision to formally declare or withdraw your candidacy for the state legislature influenced by any positive or negative messages or feedback you received from a political party, political group, or community group?
1 = Not at All
2
3
4
5 = Very Strongly
Before your most recent candidacy for the state legislature had you held an elective office?
No
Yes
If Yes, which office?
school board
city council/commission
county commission
countywide office
State House
State Senate
Other_________________
Before your most recent candidacy for the state legislature had you ever lost an election for office?
No
Yes
What year were you born?
Do you have children?
Yes
No
Other than serving in elective office, what is your primary occupation?
