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

The regressions described above incorporate dummies to account for gender-based differences for women recipients, donors, and both. But what happens when one instead estimates the regressions broken down by partisan-gender type? This appendix provides the results for four separate sets of regressions for each of four donor types: Republican men, Democratic men, Republican women, and Democratic women. These results are substantively similar to the results reported in the paper, where reported instead are the results of two partisan models (one for Democrats, the other for Republicans) using dummies to account for the gender of the donor.

First, consider Table 2.A.1, which reports the regression results for both Democratic and Republican men. One can directly compare this table to Table 2.1 described previously. Accounting for woman donors using dummies allows a direct comparison of coefficient values for the Woman Recipient variable from Table 2.A.1 with that of Table 2.1. This is because both report the coefficient for a man potential donor and a woman potential recipient. Indeed, the coefficients are virtually identical. In fact, the difference between the two coefficients for Republican men is a tiny 0.0003.

Comparing the values for Table 2.A.2, which reports the results for women donors, with those in Table 2.1 is not quite as simple, but is very straightforward. To take the dummy variables into account properly, one must add the two dummies, plus their interaction, to arrive at a value that is the equivalent of the Woman Recipient coefficient reported in Table 2.A.2. In all four cases, the coefficients for both the decision and amount regressions are the same sign. The results, however, are not as strikingly similar for women as they are for men. Interestingly, for both Democratic and Republican women, the added coefficients from Table 2.1 report a weaker effect than in Table 2.A.2 for the Decision equation (0.05937 vs. 0.1763 for Republicans, -0.0622 vs. -0.2024 for Democrats). But at the same time, the added coefficients for the Amount equation from Table 2.1 report a stronger effect than in Table 2.A.2 (0.3709 vs. 0.1344 for Republicans, -0.1652 vs. -0.1019 for Democrats). In other words, pooling men and women in the same equation estimates a weaker effect for the gender of the recipient in the Decision equation, but a stronger effect for the gender of the recipient in the Amount equation.

But what may be more important than the coefficients themselves are the predicted values these statistical models generate. To consider this, Figures 2.3-2.5 are replicated using the coefficients from the partisan-gender models reported here. Comparing Figure 2.A.1 with Figure 2.3 from the manuscript, it is clear, unsurprisingly, that the predictions for men, both Democratic and Republican, are indistinguishable – the figures are virtually identical. Differences do exist for the predicted probabilities for the women. But note that the effect of estimating separate partisan gender models is that the probability of receiving a donation decreases for both Republican and Democratic women. The main result – that Republican women are more likely to give to women than Democratic women are – remains firmly intact.

Moving to Figure 2.A.2 (compare with Figure 2.4) and Figure 2.A.3 (compare with Figure 2.5), one can see that not only are the predicted values for men virtually identical, but so are the predicted values for women. This is because the figure takes into account both the probability of getting a donation, and the amount contributed, if a contribution is made. The striking similarities of these two figures provide strong evidence that the results reported in the paper are robust to these different regression model specifications.

“The Republicans have used their women very differently than the Democrats. Poor Pat Schroeder was never in charge of a committee in Congress, but a great many of the Republican women have been.” – Congresswoman Sue Kelly (R-NY) (quoted in Dodson 2006: 54)

When the Republican Party took over the majority of the House of Representatives in 1994, they did so in part at the expense of some of the Democratic women who came to Congress in 1992’s much touted-Year of the Woman. Six freshman women members of the Democrat’s 1992 cohort lost in the 1994 “Republican Revolution”: Leslie Byrne (D-VA), Maria Cantwell (D-WA), Karan English (D-AZ), Marjorie Margolies-Mezvinsky (D-PA), Lynn Schenk (D-CA), and Karen Shepherd (D-UT). At the same time, though, seven newly-elected Republican women – Helen Chenoweth-Hage (R-ID), Barbara Cubin (R-WY), Sue Kelly (R-NY), Sue Myrick (R-NC), Andrea Seastrand (R-CA), Linda Smith (R-WA), and Enid Greene Waldholtz (R-UT) – helped to form the new Republican majority. Indeed, Republican women gained a net of five seats, whereas Democratic women lost a net of five seats. This represents a 29 percent increase in the number of women in the Republican Party, a 15 percent decrease in the number of women in the Democratic Party. These changes would surely have implications for the treatment of women in Congress. Furthermore, as noted in Chapter 2, conventional wisdom would suggest that the treatment of women would suffer because the Republican Party is traditionally considered to be antagonistic toward the agenda of feminists. For example, a speaker at a National Organization for Women rally held soon after the 1994 election labeled the Gingrich-led Republican majority “a conservative crusade against women’s rights” (Wilgoren 1995).

Yet these findings presented in the previous chapter, and the sentiments Representative Kelly expressed in the quotation that begins this chapter, call this conventional wisdom into question. Those findings imply that treatment of women might actually improve, because women in the Republican majority constitute a token minority group rather than a non-token minority group. According to the tokenism logic, Republican men, feeling less threatened by their Republican women colleagues than Democratic men were, would actually treat women better than the Democratic men of the previous majority had. Indeed, this conjecture is borne out when one considers the treatment of women in the 104^th Congress. Two women – Susan Molinari (R-NY) and Barbara Vucanovich (R-NV) – entered Republican leadership at the top ranks, as Republican Conference Vice Chair and Republican Conference Secretary, respectively. In contrast, the Democrats had only one woman at that level of leadership in the 103^rd Congress, the result of 1992’s Year of the Woman. Two Republican women were also named to the powerful Rules Committee. The committee is generally reserved for loyal party members with high seniority, yet one of the women, Enid Greene Waldholtz, received the assignment in her first term in office, a virtually unprecedented occurrence.¹⁶ Speaker Newt Gingrich also held frequent meetings with Republican women, showcasing his interest in integrating women into the party (Dodson 2006: 54). Both Democratic and Republican women noticed the change. Barbara Rose Collins (D-MI) perceived the same phenomenon that Congresswoman Kelly noted in the quotation that begins this chapter: Republican women were fast-tracked into powerful assignments, whereas long-serving Democratic women never received such assignments. Congresswoman Collins conceded: “This is the only good thing I can say about the Republicans: They really showcased their women…Some people who had one or two terms under their belts were given very plum assignments. Democrats never do that.” (quoted in Dodson 2006: 54).

These advancements for Republican women came at a price, however. Tokenism theory argues that majority group (men) members act to mentor minority group (women) members, providing them with special benefits, but only if the minority group members accept their roles as tokens. By its very definition, the tokenism relationship reinforces the dominance of the majority group. This is, of course, possible only when minority group members are sufficiently compliant to the wishes of the majority group members whose favor they are attempting to curry, much as that seen in the case of Susan Molinari (R-NY), discussed in the previous chapter. Because of this, the newly-elected Republican women had to ensure that they did not represent a threat to their majority men colleagues. Indeed, several early choices of the Republican women are consistent with the behavior of token minority group members seeking to diffuse the threat they represent to the majority. The first term Republican women House members, for example, voted unanimously to deny funds to the Congressional Caucus for Women’s Issues, along with the 28 other caucuses, including the Congressional Black Caucus and the Congressional Hispanic Caucus (Gertzog 2004). Furthermore, they made clear that they were willing to pay fealty to the more conservative wing of the Republican Party. To that end, six of the women in the newly elected freshman class helped to honor Rush Limbaugh at a dinner meant to honor him for his efforts to secure the majority for the Republicans. The women presented him with a plaque inscribed “Rush Was Right.” When she presented the plaque to Limbaugh, Barbara Cubin (R-WY) invoked one of Limbaugh’s favorite epithet for feminists when she promised: “There’s not a femiNazi among us.” (Merida 1994). Similarly, Linda Smith (R-WA) referred to the League of Women Voters as the League of Women Vipers (Micklethwait and Wooldridge 2004: 284). Furthermore, these differences were not lost on the Democratic women: “They are anti-choice, they are not environmental, I don’t know what they are,” (Dodson 2007: 50), said one Democratic woman about her Republican women colleagues. The goal of the current chapter is to provide a general theory that explains why majority group members value minority group members more highly when the size of the minority is small and how ideological preferences can serve to strain that relationship.

The previous chapter revealed that Democratic men in the U.S. House of Representatives are substantially less supportive of their fellow partisan women colleagues than are Republican men House members. Overall, Republican women receive five times more leadership PAC funds than Democratic women do. In tandem, these findings provide strong evidence of tokenism behavior that suggests a fundamental chasm between the public actions of legislators and their private treatment of their colleagues. Put another way, neither electing more women nor championing women’s issues implies that women themselves will receive better treatment from their legislator colleagues. This is reflected in women’s prospects both through advancement to service in leadership positions and through the more subtle colleague valuations leadership PAC donations decisions capture. The finding has important implications for the translation of descriptive representation of minorities into substantive representation. Indeed, increased descriptive representation may actually lead to decreased substantive representation, since both the organizational treatment and the status of minority group members within the institution suffers as their ranks increase.

Although the tokenism logic is central to how members of minority groups in political organizations are treated or valued, it cannot address the conditions under which these tokenism effects are most pervasive. This is a critical issue in need of careful, rigorous analysis since tokenism effects should critically vary depending upon the extent to which minority group members’ preferences align with those of their colleagues, particularly in settings such as legislatures, where ideological agreement is what makes collaboration possible. Individual legislators value those colleagues who are ideologically closest to them more highly. (Kanthak 2007; Currinder 2008) and those legislators who are ideologically closest to their party are also likely to be favored both with leadership PAC contributions (Kanthak 2007; Cann 2008) and valued committee positions (Cox and McCubbins 1993; Leighton and Lopez 2002; Kanthak 2004). In addition, as learned from the previous chapter, Republican women are more supportive of one another than are their Democratic counterparts, evidence that runs counter to the predictions of classical tokenism theory. The classical theory implies that Democratic women, because of their greater numbers, ought to treat each other better than do their Republican women counterparts, who comprise a traditional token minority.

The remainder of this chapter advances a novel theory that directly addresses both of these issues. The claim is that legislators weigh two aspects of colleague valuation: (1) the benefits or threats associated with the opposing group’s size and (2) the existence or deficiency of ideological affinity with individual colleagues. As a result, this chapter offers a unified theory of colleague valuation whereby individual legislators use diversity as a means to differentially discount colleagues whose policy preferences diverge from their own. That is, the tokenism logic advanced in the previous chapter is extended to account for the extent to which any pairwise set of colleagues that involve a ‘valuator’ (i.e., donor) and the ‘valuatee’ (i.e., the recipient) share policy perspectives. Next, the analysis begins with a substantive discussion of the issues motivating our theoretical model.