Political Parties, Legislatures, and the Organizational Foundations of Representation in America

The central regression results of interest provide inferential support for the partisan tokenism hypothesis. Specifically, Republican men are relatively more supportive of Republican women members than are Democratic men of their own partisan women colleagues. This is true for both the likelihood of providing a donation to women colleagues (α₁^R > α₁^D), and the amount given when a donation is made (β₁^R > β₁^D). In a distinct yet related vein, Republican women legislators are more likely to receive a donation from Republican men legislators than are their co-partisan men colleagues (α₁^R > 0), and when they do, it is for a significantly larger amount (β₁^R > 0). Democratic women MCs, however, are less likely to obtain a donation from Democratic men MCs relative to co-partisan men colleagues (α₁^D < 0), and when they do, they receive a significantly smaller amount (β₁^D < 0). Interestingly enough, the statistically insignificant regression coefficient corresponding to the Gender Recipient × Gender Donor interaction variables reveal that in all cases, women donors do not act differently from men donors in their valuation of women partisan colleagues. This evidence is consistent with evidence that questions the attainment of a critical mass in legislatures (e.g. McAllister and Studlar 2002; St. Germain 1989) because it depicts Democratic women as failing to value each other more highly that Republican women do, a pattern that is contrary to the patterns critical mass theory predicts.

Because the substance underlying these partisan-gender coefficients are not directly interpretable from the entries reported in Table 2.1, included are simulated impacts of various partisan-gender donor and recipient groups on leadership PAC contribution decisions and amounts, respectively. This is done by manipulating the values for the covariates of theoretical interest (i.e., Woman Recipient, Women Donor, and Woman Recipient × Women Donor), holding all other variables at their mean values. The simulations account for the control variables in Table 2.1, including Size of Party and Ln(Total PAC Gifts) that capture “scale-based” partisan differences in leadership PAC support independent of the partisan tokenism hypothesis.

The first set of simulations calculates the expected probability of a woman MC receiving a leadership PAC donation by each partisan-gender donor type, culled from the Republican and Democratic donors’ Probit equations, holding all other variables at their mean values. The results of these simulations appear in Figure 2.3. These regression models predict that a typical Republican man MC is about three times more likely to provide a woman colleague with a leadership PAC donation (0.0518, or an expected 204 donations) than a typical Democratic man MC is for his woman colleague (0.0166, or an expected 65 donations). This evidence provides strong support for the partisan tokenism hypothesis. External to the theory’s prediction, these simulated probabilities also reveal that Republican women are relatively more supportive of women colleagues than are their Democratic counterparts, by nearly a factor of 1.67 to 1 (0.0449 vs. 0.0271). This, in turn, suggests that the House Republican party caucus treats women members better than the House Democratic party caucus – irrespective of the gender of those members serving as ‘treaters.’ Again, this is a question upon which the extant literature is silent, since their research designs preclude the measurement of individual colleague valuations in political parties.

The second set of simulations, depicted in Figure 2.4, analyze per-donor expected average dollar contributions made to women House members, adjusted for differences in the number of leadership PACs various partisan-gender groups hold. As a result, these graphical figures exploit information from both the donation decision and conditional donation amount equations to provide statistical estimates regarding both the absolute and relative per donor valuation of women House members by each partisan-gender donor group. Furthermore, these sets of figures allow us to better assess the substantive effects of gender on colleague valuation decisions. Consistent with the partisan tokenism hypothesis, these dollar amount estimates reveal that Republican men typically contribute nearly 6 times more leadership PAC funds to women colleagues than do their Democratic counterparts ($2856.20 vs. $484.47). Moreover, Republican men MCs provide greater relative support to women colleagues vis-à-vis Republican women MCs at a rate of about 9 to 1 ($2856.20 vs. $326.84). Democratic men’s and women’s relative per-donor expected dollar support ratio for women members is almost 3.5 to 1 ($484.47 vs. $144.16). Taken in tandem, these findings demonstrate that Republican men value women colleagues more highly than Democratic men in both absolute and relative terms.

The final set of simulations calculates the expected total dollars a typical member of each partisan-recipient group receives, holding all other variables at their mean values.¹⁵ The results of these simulations appear in Figure 2.5. These regression models once again provide strong support for the partisan tokenism hypothesis. Specifically, these models predict that a typical Republican woman legislator receives $3183.04 per cycle in leadership PAC funds in our sample period, whereas a typical Democratic woman receives only $628.63 in such funds. This partisan-gender gap in expected total resources favors an average Republican woman Member of Congress vis-à-vis an average Democratic woman Member of Congress by a factor of more than 5 to 1. Interestingly enough, the typical Republican woman MC fares 1.5 times better than the typical Republican man MC ($3183.04 vs. $ 2066.38) whereas the average Democratic man Member of Congress fares slightly more than twice as well as the average Democratic woman ($1206.85 vs. $628.63). Besides the fact that the results are squarely consistent with the partisan tokenism hypothesis, our findings indicate that tokenism-based colleague valuation differences are also manifested in the relative treatment of men and women in each party caucus.

Given the results presented here, it is unsurprising that the first woman to be elected to the upper echelons of party power was a Republican: Lynn Martin of Illinois, who was elected to the position of Caucus Vice-Chair in 1985 (Office of History and Preservation 2007). Prior to that, Democratic women served in the leadership of their party, but only as the appointed Democratic Caucus secretary, a position that had been reserved for specifically for women members. When the position was eliminated in 1987, it had, in fact, never been held by a man (Gertzog 1995: 108). By means of comparison, Nancy Pelosi entered the Democratic leadership at the start of the 107^th Congress in 2001, when she became Democratic Whip (Office of History and Preservation 2007). By the time the Democrats elected their first woman to their leadership – Barbara Kennelly as Caucus Vice Chair at the start of the 104^th Congress– the Republicans were electing their second: Susan Molinari, whose quotation began this chapter.

Two years after becoming the fifth highest-ranking Republican in the House, Republican presidential candidate Bob Dole tapped Representative Susan Molinari (R-NY) to give the keynote address at the Republican National Convention. In contrast, the Democrats chose Evan Bayh, the white male governor of Indiana, to give their convention’s keynote address the same year. Tellingly, Molinari spoke at the Republican convention on the same night as African-American Colin Powell: The only speaker at the entire Democratic convention that year who was not a white male was First Lady Hillary Rodham Clinton.

Molinari, whose voting record on issues such as labor, the environment, and feminist issues had taken a decidedly rightward turn since she joined the Republican Party leadership in the U.S. House (Sleeper 1996), was not notified that she would keynote before Dole made the announcement on the “Larry King Live” television program. A CNN crew tracked Molinari down in a restaurant in Buffalo, where she was attending a campaign event for her husband, Representative Bill Paxon (R-NY). Larry King asked Molinari if her more liberal views on abortion would heal divisions in the party that were otherwise highlighted by Dole’s pro-life stance. Molinari chose not to discuss her own pro-choice views, but rather to turn the question around to praise Dole personally: “I think Senator Dole binds the wounds in this party. There’s no doubt that this is a man who has the capacity, the vision and really has the heart to bind these people together.” (Nagourney 1996). Molinari plays the role of the token here – she receives benefits from the majority, but must be careful not to raise their ire. It is telling, as well, that Molinari’s statement was not the product of coaching by the Dole team. Off-the-cuff, she was able to support her candidate while at the same time masking her own ideological differences with him.

The differences between the two parties in how they treat their women members call into question the prevailing wisdom about the effects of including diverse groups in legislative bodies. Increasing descriptive representation of minority groups in elected assemblies is intended to enhance the substantive (policy) representation of their interests (e.g., Mansbridge 1999; Phillips 1995). It is accepted wisdom that the Democratic Party is the preferred political party for advancing women citizens’ policy interests (e.g., Box-Steffensmeier, DeBoef, and Lin 2004; Frankovic 1982; Kaufman 2002). A natural corollary to this line of reasoning presupposes that the Democratic Party is a more hospitable organizational environment for women than is the Republican Party. The fact that Democrats publicly support laws mandating the equal treatment of women would seem to imply that they would at the very least support equal treatment of women in their own workplace, if only for fear of falling afoul of the laws they hope to pass.

But the protection of one’s own group status may trump these preferences favoring the equal treatment of women. Under partisan tokenism theory, even if the Democratic Party is more supportive of women’s policy issues when compared with the Republican Party, the dominant majority group in the Democratic Party – men – will be less supportive of women colleagues than will their Republican counterparts. Furthermore, Figure 2.2 depicts the implications of this effect in matters other than campaign contributions by demonstrating that in recent years, the proportion of Republican women serving in powerful legislative positions has been greater than the proportion of Democratic women. This behavior is driven neither by the hypocrisy of Democratic men nor by the virtue of Republican men, but rather by the relative scarcity of women legislators within each party caucus. Because Democratic women are better proportionally represented in the U.S. House than are Republican women, Democratic women pose a relatively more serious threat to their majority group – Democratic men – than do Republican women to their majority group – Republican men. Likewise, Republican men MCs treat their partisan women colleagues better than their Democratic men counterparts since there are too few Republican women MCs to pose a threat to the majority group’s (Republican men’s) status in their legislative party caucus. Because Republican support of women colleagues is based on their relatively small numbers, one would expect to see a decline in their generosity toward same-party women legislators as their size increases, akin to the decline in the Democratic Party. Similarly, should the ranks of Democratic women increase to the point that men are no longer the dominant majority, one would expect to see the backlash toward women wane, since that backlash is linked to fear of losing that dominance.

Our statistical evidence provides unambiguous support for the tokenism logic in explaining how men value their women colleagues. These results suggest that a significant chasm exists between the public action and private treatment of women. These findings, therefore, provide an explanation for the patterns observed in Figure 2.2: The failure of Democratic women to achieve greater leadership numbers than their Republican women counterparts comes because of their greater numbers, not in spite of it. For example, our statistical analysis of leadership PAC contributions reveals that Republican men contribute almost six times more money per donor to women colleagues than do their Democratic counterparts. Our analysis of member to member intra-party campaign donations also indicates that a typical Republican woman Member of Congress receives more than five times the amount of leadership PAC funds in our sample period than does a typical Democratic woman Member of Congress. These findings not only run counter to conventional wisdom regarding the relationship between political parties and women, but also highlight the disconnect between substantive representation and private treatment of organizational members that political commentators presume are linked tightly together (Raines 1983; Sommers 1997), a link that is central to reaping the substantive rewards of increasing descriptive representation (Dodson and Carroll 1991; Dovi 2002; Mansbridge 1999; Phillips 1995).

This chasm between public actions and private treatment by members of elected assemblies has tangible adverse consequences for understanding the link between descriptive and substantive representation of historically under-represented and minority groups. Preuhs (2006) has convincingly demonstrated that the role minority group members play in a political organization is of crucial importance in determining their impact on policy decisions. If so, then our findings suggest that Democratic women MCs are hindered in their attempt to fully realize their policy influence because Democratic men implicitly sanction them for the simple fact that their sheer numbers represent a formidable threat – even though they are clearly a minority group not close to attaining majority status. In turn, the tokenism behavior men legislators display is indicative of a latent “glass ceiling” – or “marble ceiling,” as Speaker Pelosi herself has dubbed it (Chaddock and Sappenfield 2002) – that limits the capacity of women to become a stronger presence in elected assemblies. Therefore, this inverse relationship between proportion of women legislators within a party caucus and their private treatment from men colleagues may help to explain why the translation between descriptive representation and substantive representation may often entail considerable slippage (Hero and Tolbert 1995; Weldon 2002). Moreover, such tokenism behavior may also explain the lack of evidence for critical mass (e.g. Hedge, Button, and Spear 1996; McAllister and Studlar 2002; Reingold 2000; Rosenthal 1998), and recent calls to halt the search for the elusive condition (Grey 2006; Childs and Krook 2006, 2008, 2009). On the one hand, the evidence presented in this chapter may offer insight into the location of this critical mass. Because there are different gender-based contribution patterns among Democrats and Republicans, this suggests that critical mass is located between the proportions of women in the Democratic and Republican Parties (about 0.06 and 0.20, respectively). On the other hand, the fact that Democratic women do not treat each other better, despite being beyond the point at which good treatment from men is halted, is indicative of a much more complicated critical mass problem exists than the literature has previously suggested.

The example of Representative Susan Molinari’s rise within the ranks of the Republican party organization noted earlier in this section highlights not only the different approaches the two parties take to gender inclusiveness, but also the importance of ideology in the tokenism relationship. Molinari receives benefits from the relationship – a leadership position and a national stage at the convention – that are not as readily available to her more numerous Democratic women colleagues. Yet at the same time, those benefits rely heavily on her willingness to avoid straying too far from the Republicans’ ideological fold. Molinari provides benefits to the majority men of her party because she deflects criticism that the Republicans’ descriptive characteristics are too homogeneous. But her men colleagues receive those benefits only because Molinari opted to avoid discussing abortion and other issues that would have highlighted the differences between her and her men colleagues. To construct a unified theory of tokenism, then, one must account for these effects of ideology on colleague valuation.

The current chapter provides strong evidence that the tokenism logic is valid for explaining the treatment of women in the U.S. Congress. But gender relations alone cannot provide an entirely accurate portrayal of colleague valuation in legislatures. Furthermore, well-tested means of measuring and understanding ideology in legislative settings allows us to test the robustness of Kanter’s (1977) conception against a competing notion that perhaps agreement among colleagues on important workplace issues will trump any tokenism effect. The next chapter proposes a more nuanced view of tokenism-based colleague valuation in which these valuation decisions are best understood through the lens of ideological proximity. Specifically, that chapter examines how gender diversity affects how colleagues value one another based upon the extent to which they share policy preferences. Attention now turns to addressing this puzzle.