PROFILING POLITICAL LEADERS AT-A-DISTANCE

The central assumption in profiling political leaders is that the language that leaders use approximates who they are: what people say reflects how they think about the world, and how they think about the world reflects what they do (Suedfeld, Guttieri, and Tetlock Reference Suedfeld, Guttieri, Tetlock and Jerrold2003). Government agencies (e.g., the Central Intelligence Agency) and academics have long used leaders’ words to understand their priorities and policy positions. Given a lack of direct access to political elites and inspired by personality psychology, “at-a-distance” approaches have become the primary way for providing reliable, generalizable data for leadership studies (Schafer Reference Schafer2000). Although early leadership research was psycho-biographical and idiosyncratic, the state of the art has evolved into systematic quantitative content analyses using words as data.

The operational code (George Reference George1969, Reference George1979; Leites Reference Leites1951; Schafer and Walker Reference Schafer and Walker2006; Walker Reference Walker1990) has been used most widely among at-a-distance approaches. Most OCA research uses prepared speeches delivered in a public setting to profile an individual; these speeches, according to Schafer and Walker (Reference Schafer and Walker2006), provide greater insight into a leader’s beliefs than spontaneous remarks.Footnote ¹ Contemporary OCA runs the “Verbs in Context System” (Walker, Schafer, and Young Reference Walker, Schafer and Young1998), which is a suite of coding schemes in the ProfilerPlus automated coding software (Levine and Young Reference Levine and Young2014). OCA focuses on the verbs in a text corpus and their attributions regarding the exercise of power to the self and others to construct quantitative indices that correspond to multiple beliefs (Schafer and Walker Reference Schafer and Walker2006). The operational code, representing a leader’s belief system, has two components: (1) the five philosophical beliefs about the political universe in which leaders find themselves and the nature of the “other” they face in this environment; and (2) the five instrumental beliefs that reflect the image of “self” in this political universe and the best strategies and tactics that leaders could use to achieve their objectives (George Reference George1979; Walker Reference Walker1990). The corresponding beliefs “explain diagnostic and choice propensities of the agents who make foreign-policy decisions” (Schafer and Walker Reference Schafer and Walker2006) and offer causal mechanisms in explaining policy choices.Footnote ²

Norming groups serve as a baseline to compare leaders’ beliefs to an average profile of their peers (Schafer and Walker Reference Schafer and Walker2006).Footnote ³ This allows researchers “to put the scores into perspective by determining how they compare with those of other leaders” and “to judge whether the particular leader’s traits are unusually high or low or about average” (Hermann Reference Hermann and Post2003, 204). In other words, the norming group is “the basis for comparison” (Schafer and Walker Reference Schafer and Walker2006, 43) and “functions as a proxy for ‘the average leader’” (Schafer and Walker Reference Schafer and Walker2006, 45). Hermann’s (Reference Hermann1980, Reference Hermann and Post2003) and other leadership trait analysis (LTA) research illustrate this practice, in which the comparison groups have been regional, cultural, or country specific. “[W]hat group of leaders to use as the comparison—or norming—group” is, therefore, an important decision in this vein of research. The reference group constitutes the benchmark to determine “increased likelihood of potential impact [of a trait] on a leader’s political behavior” (Post and Walker Reference Post, Walker and Jerrold2003, 404). These norming groups help to identify which aspects of a leader’s political beliefs are abnormal or differ significantly from the average. The norming group, accordingly, directly provides the context to a leader’s personality. Indeed, as Hermann (Reference Hermann, Klotz and Prakash2008, 161) stated, “[T]his norming group is viewed as defining what is more usual—the norm—among a particular group of leaders (as opposed to more extreme or different).”

We take issue with the fact that it is rarely known who is in the norming groups. Existing average profiles are almost exclusively of male leaders (table 1). Until the introduction of the Psychological Characteristics of Leaders (PsyCL) dataset (Schafer and Lambert Reference Schafer and Lambert2022), there had been little transparency in leadership profiling approaches regarding precisely who made the norming groups—only the N was reported. For example, whereas Hermann’s (Reference Hermann and Post2003) LTA reference groups are representative of world leaders, they do not indicate the gender composition of their samples. The OCA research similarly is bound by this: two of the most-used comparison groups (Malici and Buckner Reference Malici and Buckner2008; Schafer and Walker Reference Schafer and Walker2006) do not suggest gender composition of their respective samples. The recently released norming group, the PsyCL dataset (Schafer and Lambert Reference Schafer and Lambert2022), offers a transparent, significantly large collection—albeit a heavily male- (and Western-) dominant group.Footnote ⁴ Because the reference group conditions how a leader deviates from the “norm,” in existing personality approaches this norm evidently becomes the characteristics of male leaders. Ideally, the norming group should include “other relevant individuals” to the individual profiled (Dyson and Billordo Reference Dyson and Billordo2004, 115). In fact, if textual data are available, there is an inherent opportunity in leadership profiling “to create one’s own reference group of logically grouped leaders” (Dyson and Billordo Reference Dyson and Billordo2004, 116; emphasis added). Accordingly, the development of a new norming group is crucial for providing contextualized political-personality profiles of a new generation of leaders—and, specifically, women leaders. The average American female leaders’ beliefs outlined herein provide preliminary insight into their baseline beliefs in a largely masculine political landscape.

Table 1 Existing Operational Code Reference Groups

Notes: N equals the number of individuals in each norming group. Standard deviations are in parentheses. The post-1945 American presidents in PsyCL (alphabetically by last name) are GHW Bush, GW Bush, Carter, Clinton, Eisenhower, Ford, Johnson, Kennedy, Nixon, Obama, Reagan, Truman, and Trump (2016–2020). The women leaders in PsyCL are Aquino, Arroyo, Ciller, Gandhi, Kumratunga, and Thatcher.

BRINGING GENDER IN PROFILING POLITICAL LEADERS

There are distinct attributions to female and male leaders (Enloe Reference Enloe2013; Hooper Reference Hooper2001; Shepherd Reference Shepherd2010; Tickner Reference Tickner1992). Research suggests that female leaders tend to be more collaborative than their male counterparts (Eagly and Johnson Reference Eagly and Johnson1990), more empathetic (Kamas and Preston Reference Kamas and Preston2021), and less likely to be overconfident—a trait that historians and political scientists point to as a major cause of war (Johnson et al. Reference Johnson, McDermott, Barrett, Cowden, Wrangham, McIntyre and Rosen2006). Many of these gender-based studies occur in corporate and academic environments rather than in politics, given the underrepresentation of women politically (Bark et al. Reference Bark, Escartín, Schuh and van Dick2016). It is plausible that the political processes remove more traditional “feminine women” and that female political leaders may adhere to the practices of their male peers in politics (Elsesser Reference Elsesser2023). This self-selection bias means that female politicians may be fundamentally different than other women and/or women leaders. A more accurate reading of male and female leaders suggests that “both male and female decision makers could exhibit either masculine or feminine attributes” (Brummer Reference Brummer, Mello and Ostermann2023, 247). However, there is limited research regarding differences between male and female political leaders due to the systematic underrepresentation of women in governments globally.

Women leaders endure unique challenges in the male-dominated political world: they face more obstacles when running for office (Jalalzai Reference Jalalzai2008), including sexist voter attitudes (Ratliff et al. Reference Ratliff, Redford, Conway and Smith2017), increased party pressure to exit primaries in tight races (Lawrence and Rose Reference Lawrence and Rose2010), and sexist attacks in the media (Groch-Begley Reference Groch-Begley2016). To succeed, female leaders must consciously defy gender stereotypes or overcome the “double-bind”—that is, female leaders who are more feminine are viewed as weak and those who are more masculine are perceived as unlikeable. Schramm and Stark (Reference Schramm and Stark2020) found that female politicians frequently overcorrect against the stereotype that women are peacekeepers by being more likely to initiate political conflicts than their male counterparts. Female politicians are plagued by gender stereotypes whether they present themselves as more traditionally masculine or feminine.

In a male-dominated world of politics, female leaders may have to compromise certain aspects of their principles to succeed. To push against sexist stereotypes, female politicians may exhibit masculine-perceived behavior. They may engage in “strategic use of gender” to balance these double standards and fight against the stereotypes of women being too emotional or less competent in leadership roles (McThomas and Tesler Reference McThomas and Tesler2016). They do this by emphasizing more masculine traits and downplaying more feminine traits. Historically, female candidates have had to “act like men” to get elected; in more recent years, female candidates have begun to openly discuss their gender as a value rather than a hurdle to overcome (Mekouar Reference Mekouar2018). However, there remains pressure within political parties to adhere to masculine norms regarding toughness when it comes to crime and security and to maintain the status quo for gender roles and expression. Indeed, Jalalzai (Reference Jalalzai2008, Reference Jalalzai2014) remarked that the political executive is associated with masculinity “particularly as one moves up the hierarchy” and at its apex is the chief executive (i.e., the head of government), where the perception of masculine traits therefore should be observed most often (Jalalzai Reference Jalalzai2008, 209).

Even an abject summative input from gender in politics scholarship is absent in at-a-distance studies of political leaders.Footnote ⁵ When our research was underway, Brummer’s (Reference Brummer, Mello and Ostermann2023) brief discussion was the first to explore gender in leadership studies research. Comparing leadership traits of a small sample of male and female leaders, he found “little support for the biologically rooted, systematic differences” between the two sexes. The only exception in LTA was the belief in the ability to control events, for which male leaders had a significantly higher score. Brummer (Reference Brummer, Mello and Ostermann2023) focused on gender but did not provide a new norming group—he implied that this is mostly due to the time commitment required to collect textual data.

With this as background, our study develops a preliminary benchmark profile of American female politicians. We are pleased to present a new average profile of American women leaders and make new data available for existing and future profiling of leaders. This is a timely undertaking because a recent expansion of interest in the elites in IR arguably has not found its way to scholarship in American politics.Footnote ⁶ The American two-party system and absence of term limits give the overwhelmingly male incumbents an advantage for reelection, and there remains a gender disparity in American politics (Bjarnegård Reference Bjarnegård2010; Sanbonmatsu Reference Sanbonmatsu2020). However, the growing number of female leaders at the national level makes this effort timely and meaningful. Moreover, for leadership studies, this constitutes an important step forward. With the expansion of textual corpus availability and coding capabilities, as well as the release of recent large datasets, exploring variations between various groups likely will be a future trajectory in this field (Schafer and Lambert Reference Schafer and Lambert2022, 9).

DATA: THE TEXT CORPUS

Our primary data are from the Iowa State University’s (2022) Archives of Women’s Political Communication. We focused exclusively on American women political leaders who held the offices of Vice President, Senator, Representative,Footnote ⁷ and Cabinet MemberFootnote ⁸ from January 2000 to June 2022. We included American female politicians’ speeches while they were in a national office.

All speeches are exclusively political in their content, come from diverse venues, and represent an array of topics. The text corpus excludes anecdotal stories, commencement addresses, congratulatory remarks, jokes, and opening and closing remarks. To reduce duplicity, the archive does not include campaign stump speeches, which are repetitive in nature. Following the operational code of the customary focus on using prepared speeches, we did not use spontaneous statements (e.g., interviews, debates, and town halls) in our analyses.

The final text corpus in our largest collection included 870 speeches from 172 leaders, totaling more than one million words. This is our “All Women” profile. A subgroup of 52 individuals’ average profile (i.e., 775,649 words in 563 speeches) covers Democratic and Republican women with at least four speeches in our corpus. We named this profile “Women Leaders” and shared the entirety of the 52 leaders’ operational code indices in our dataset (Kesgin and McCormick Reference Kesgin and McCormick2025). For additional women-peers comparisons, we present multiple profiles of American women political leaders by their roles—that is, in the Cabinet, Congress, House, or Senate. To calculate operational code indexes, we aggregated the corresponding text corpus for either the entire group or an individual.Footnote ⁹

We used ProfilerPlus Version 7.3.18 (Levine and Young Reference Levine and Young2014) to generate the operational code scores. This analysis focused on the three master beliefs: P-1, nature of the political universe; P-4, control over historical development; and I-1, direction of strategy.Footnote ¹⁰ These beliefs are the basis for which all other philosophical and instrumental beliefs are calculated.

PROFILES OF AMERICAN WOMEN POLITICAL LEADERS

Table 2 displays our findings. The average contemporary American female politician holds a relatively friendly view of the world (P-1), has a high belief in her control over historical developments (P-4), and is cooperative in her direction of strategy (I-1). All three indicate higher scores—except in one instance—than the existing norming groups’ averages reported previously in table 1 (which, to reiterate, are heavily if not exclusively male).

Table 2 The American Women Leaders Dataset

American women politicians are different from the average post-1945 American presidents, specifically in their P-1 and P-4 scores. Compared to the 13 men’s average profile, they have a less friendly view of the world (P-1). The two profiles differ most noticeably in their P-4 values: American women leaders’ belief about their control over historical development are consistently higher than the other norming groups. These results contradict the well-documented gender-confidence gap and Brummer’s (Reference Brummer, Mello and Ostermann2023) finding that male leaders had a higher belief in their ability to control events. One explanation for this high level of confidence and perceived control over historical development is that it may be caused by gender-based barriers in participation that result in only the most ambitious and qualified women reaching leadership positions in their field. Other than the P-4 values, there is only a minor difference in the American women leaders control group in comparison to the existing male-centric control groups. This suggests that there is not as much difference in the master beliefs of male and female leaders as theoretically would be expected. This finding is consistent with Brummer’s (Reference Brummer, Mello and Ostermann2023) conclusions. Finally, the American women leaders’ I-1 score is higher than the 13 male presidents’ score, indicating their cooperative direction of strategy.

The small PsyCL women’s group and the American women differ from one another in their average profile. The existing datasets and our American women leaders dataset, unfortunately, do not allow for further statistical analyses at present, which is due to lack of data transparency in older datasets or differences in units of analyses.

The relative stability of all three indices between the All Women and Women Leaders profiles potentially suggests that these are important benchmarks. Concurrently, our findings indicate that, by their average scores, there are nuances among American women leaders. Women leaders in Cabinet roles differ from other sub-profiles: the former are more optimistic about the world and more cooperative, but they believe that they have less control over historical development. Our preliminary analyses suggest that, on average, Republican women (in the House or the Senate) compared to Democratic women also exhibit a similar trend. T-tests across these different groups did not reveal statistical significance to report. We project, however, that there may be statistically significant nuances as other researchers advance this inquiry.

Figures 1a and 1b display the 52 American women leaders’ individual operational code scores. We observed in each of two clusters by party affiliation that few individuals emerge as outliers from their respective party’s cluster. Our replication data publish and share all philosophical and instrumental belief scores for the 52 individual leaders (Kesgin and McCormick Reference Kesgin and McCormick2025).

Figure 1 (a) American Women Leaders’ P-1 and I-1 Indices by Party; (b) American Women Leaders’ P-1 and P-4 Indices by Party