Implicit in much of the research on political representation is the assumption that individuals with shared descriptive characteristics have both common interests and that they enact, or will try to enact, policies related to these interests once in office. Thus, the election of women into positions of power should result in policies that are substantively different than when men hold these same positions. Accordingly, descriptive representation is believed to give way to substantive representation. While politicians can represent their constituents in many different ways and be responsive to their needs through a variety of actions, from enacting policies to allocating resources addressing constituent need [1
], simply having more like-minded women in office does not automatically translate into the effective implementation of responsive policies. Although much research has documented divergent attitudinal positions and behavioral distinctions between women and men legislators (see for instance [3
]), the extent to which policy outcomes are affected by having women in positions of power remains ambiguous. Some scholars find that having women in office influences the types of policy outcomes enacted by governments (see for instance [4
]), while the research of others suggests a complicated or null relationship [12
]. Still others observe that differing institutional environments, like holding leadership positions, heading committees, or having seniority best explains which representatives may be better able to affect endorsed policies [13
We throw ourselves into the latter camp, arguing that policy responsiveness is nuanced and heavily dependent upon contextual factors. While women elected officials do tend to be more supportive of “women’s issues” than men, their influence is reflective of the contexts they inhabit. Indeed, scholars have recently begun to challenge the presumption of universal support by demonstrating distinctions in the ways in which women legislators behave (see [6
]). We continue in this vein by arguing that environmental factors play a substantial role in the successful implementation of policy. Given that women as a group fail to achieve majority status within any state legislature in the United States, their collective ability to address group interests is constrained in ways that majority group members do not experience. Within this setting, passing legislation encapsulating their interests should be particularly onerous. In light of their prospect for success, women should be most successful when they are able to convince others of the importance of an issue. An example of this occurs when substantial aggregate need is present among group members within the general population. As such, contextual state need coupled with the presence of legislators sensitive to the needs of specific populations should impact the allocation of scarce resources. Under these circumstances, it should be much easier to persuade non-group members of the issue’s pressing nature. Thus, policy responsiveness should be greatest in the face of extensive need; a relationship that should be enhanced with growth in women’s representation.
We explore these contentions in novel ways. First, we focus our attention on welfare policies involving health care influencing children, the elderly, and the disabled. Given that over half of all poor children live in families headed by women according the U.S. Census Bureau [19
], much of the literature examining the link between women’s representation and social welfare outcomes tends to concentrate on policies affecting children. But women’s interests are not confined to this particular domain. For instance, one in ten women relies on Medicaid for health services and 68 percent of seniors who are eligible for Medicare and Medicaid are women [20
]. Similarly, elder women are especially dependent upon government health care services as they are more likely to live longer and consequently more likely to live alone [21
]. The issue of health care also reflects women’s traditional role as caregiver—transcending partisanship in a uniquely gendered way [22
]. Consequently, we would expect that women legislators, regardless of political party, would be especially sensitive to these issues and would be more responsive to constituency needs than their male counterparts. By investigating elderly and disabled populations as relevant to women’s interests we expand upon past research. Moreover, understanding support for health care benefitting vulnerable populations is of great consequence given the depth of need and expense of meeting it. Medicaid is the largest single state expenditure and continues to grow. In fiscal year 2014, Medicaid accounted for 25.8 percent of all state expenditures or $460.5 billion nationwide, providing comprehensive and long-term medical care for more than 68 million individuals [23
We explore the extent to which increases in the representation of women affect spending on health care for the poor using information from states across time to create a dataset spanning a decade, thus giving us greater statistical power to substantiate our hypothesized relationships. As such, our analysis traverses time and unique circumstance. In doing so we find that women are indeed able to affect policy benefitting the interests of their group, but that they do so only when great need is apparent and when the number of women serving in the legislature is large. This effect holds independent of party. On the whole, our research demonstrates that representative bodies with more women spend more on health care for vulnerable populations benefiting the collective interests of women, but that this policy response occurs only under prescribed conditions. The findings are also relevant for scholars interested in examining ethnic minority group representation as analogous dynamics might occur and could be explored employing similar methods.
2. Descriptive and Substantive Representation
Over the course of the past few decades, women have slowly increased their presence within government. For instance, once comprising only 4.5 percent of total state legislators in 1971, women held 24.6 percent of the total seats in 2016, a five-fold increase [24
]. While far from reflecting the demographic realities of the American electorate, this diversification of institutions is believed to result in more varied life circumstances being reflected in public policy output, rather than the long-established male, white viewpoint. Moreover, the dynamics of the institution change as more and more women are elected. Academic inquiry has examined the impact of gender and patriarchy in politics, with increased attention being placed upon identity and the intersection of race/ethnicity, gender and other privileged categories, but has provided mixed empirical evidence.
Since Irene Diamond’s first empirical study, many have followed suit in examining the relationship between descriptive and substantive representation. Central to these inquiries is whether women representatives “stand for” women; whether African-Americans stand for African Americans, etc. [25
]. In her theory of presence, Phillips argues that distinct life experiences (such as child-rearing, educational and occupational experiences, exposure to sexual harassment and gendered violence) make women leaders uniquely qualified to represent the interests of women ([26
], pp. 67–68). Consequently, descriptive representation should lead to substantive representation. Specifically, sheer numbers (largely referenced as descriptive or demographic representation) are believed to augment policy (or substantive) representation ([27
], p. 622).1
Indeed in some early groundbreaking research, Thomas found that “women legislators [embraced] priorities dealing with issues of women, and children and the family, [while men did] not share this priority list” ([30
], p. 7). Additional research has observed that women are more likely to be supportive of legislation involving women’s issues [31
] as well as to serve on health and human services committees and to retain these positions as they rise in seniority rather than to migrate to committees on appropriations, finance or rules as men are inclined to do [7
On the whole, research does seem to suggest that women legislators are generally more interested in women and family-friendly issues than their male colleagues. This does not mean, however, that policy outcomes are affected by the presence of women in government. Action could be symbolic or women could lack power to implement policies that reflect their preferences. The evidence of the effect of descriptive representation on policy outcomes is decidedly mixed. As the number of women in elected office has increased over time, some research suggests that such increases may be related to changes in public policies. For example, through a systematic study of nineteen democracies from 1970 to 2000, Kittilson [36
] found that women’s parliamentary presence significantly influenced the adoption of maternity and childcare leave policies. Additional work established a connection between increases in the number of women representatives and women-friendly policies [35
]. Interestingly, a few researchers saw an effect on family policies when women comprised only a small minority [8
However, inconsistencies also abound. For example, many initial studies found little evidence to support the contention that increases in women’s representation led to differences in policy outcomes [25
]. And even though women legislators influenced regulation related to parental notifications of abortion, they did not affect the funding of abortions [39
], nor did they influence women’s health initiatives in the 1990s [13
] or shape policies related to domestic violence [14
]. Updating and refining the work of Hansen [41
], Cowell-Meyer and Langbein [12
] studied the extent to which growth in women’s representation affected 34 “women-friendly policies” in a highly comprehensive comparative state analysis. In doing so, they found that the percentage of women in legislatures was related to only eight of the identified policies, with five in the anticipated direction. For the clear majority of policies, the presence of women had no discernible effect. In another comparative state analysis, Reingold and Smith observed what also appear to be contradictory results. Depending upon which policies were examined the presence of women legislators had “a liberal effect, a conservative effect, or no effect” ([18
], p. 132); their intersectional approach (which examined both race and gender) produced more consistently liberal results. Overall, the findings from the literature exploring the main effect of women’s representation on relevant policy outcomes remain ambiguous.
3. The Importance of Context and the Aggregate Need of the Beneficiary Group
Failing to find consistent results, researchers have argued that presence is not sufficient. Rather, influence is thought to be conditioned by other factors such as institutional contexts involving party, leadership positions, deliberation rules, seniority, committee appointments, and external political environments [5
]. In a comparative study examining partisanship, market conditioning and politics at the state level, Zhu and Clark [45
] observed that Democratic-controlled legislatures mitigated the effect of racial diversity on health care inequality. Additional scholars have found that Democratic majorities influence Medicaid spending levels [46
] and the expansion of health care programs for the poor, notably Children’s Health Insurance Programs (CHIPs) [47
]. Another common theme in the literature examines the degree to which members of various groups hold chamber and committee offices as an indication of relative power and influence (incorporation models). For example, female legislators are more likely to be supportive of welfare policy than their male counterparts, however the results are contingent upon the majority party status of the women legislators as well as the incorporation of women on relevant legislative committees [6
]. Mendelberg et al. assert, “women’s descriptive representation does produce substantive representation but primarily under majority rule—when women are many, they are more likely to voice women’s distinctive concerns about children, family, the poor and the needy, and less likely to voice men’s distinctive concerns” ([48
], p. 291). In their multivariate study of TANF policies across the states in 1998, Reingold and Smith [18
] find that women’s influence, if present, is a result of a combination of absolute numbers as well as legislative incorporation into leadership (a finding consistent with Preuhs’ [49
] analysis of minority legislators).
Some scholars assert that in order for women to translate descriptive representation into policy representation, women need to reach a critical mass (meaning achieving a threshold in size to have a substantive impact), from which they can influence the dynamics of the institution as well as its output. In a quasi-experimental design, MacDonald and O’Brien [4
] find that gender impacts representation when a critical mass is achieved. Along the same lines, Berkman and O’Connor [39
] observe that if women reach a critical mass in the state legislature, they can impact parental notifications of abortions, but not funding. However, a good deal of uncertainty and questions have arisen involving the theory of critical mass.2
Notably, disagreement surrounds the needed threshold; research suggests the range is somewhere from 15 percent [50
] to 30 percent [51
] of the seats in a legislative body. Moreover, research has not resulted in consistent findings, leading some to question if the effect is theoretically sound (see for instance [28
]). Some argue, for instance, that when underrepresented minorities are “new” (Beckwith [52
] defines newness as a large increase over a short period of time resulting in greater visibility) they can be more influential than one would expect based upon their absolute numbers (see for instance [11
We follow along this vein, arguing that the status of women as a minority group plays an important role in conditioning when influence will transpire. According to gender role theory, women are less likely to be assertive and their concerns are given less attention when they are a numerical minority ([48
], p. 293, refers to this dynamic as a “minority status” hypothesis). Despite the numerous ways in which politicians can represent their constituents, policy responsiveness requires the passage of beneficial legislation, or the blockage of that which is harmful. This process is facilitated with growth in representation, and made more difficult by its decline. Being in the minority means that group members need others within the legislature to join them in passing beneficial legislation for their group. These non-members should be more willing to vote for legislation that they might not otherwise vote for when the effect of the policy is substantial. This occurs when legislation addresses considerable aggregate need among the population. With the presence of substantial need, even when comprising minority status, women should be better able to mobilize reluctant non-group members to align on the outcome. Indeed, MacDonald and O’Brien [4
] note that most models estimating the ability of women legislators to provide substantive representation on women’s issues omit constituency variables that impact decision-making, potentially biasing the analysis of the role of gender. Consequently, the theory that women “stand for” other women, providing a critical voice, is not as straightforward as one might think. These issues are constructed within larger political environments and impact representatives in fine ways which should lead to nuanced policy and allocation responsiveness. Specifically, when need is not substantial, women may not be able to create necessary coalitions to affect policy outcomes. Thus, women will gain traction with growth in need. This behavior will be conditional, however, on the number of women within a state legislature. Specifically, we expect that policies will be affected both when the beneficiary population is large and the number of women within a state legislature is substantial.
Policy outcomes will be affected by increases in the number of women across state legislatures, but only when there is also growth in the aggregate beneficiary population of a particular policy.
4. Defining Women’s Issues
There is disagreement not only about defining women’s interests but about whether or not uniform interests exist given the diversity of women in the United States.3
For the purposes of this study, we argue that there is value in examining gender as an influential factor, especially as it relates to support for elders, the disabled and vulnerable children. First, the history of social welfare in the United States is characterized as a decidedly female issue. According to the U.S. Census Bureau, in 2014 19.9% of children in the United States lived in poverty; over half of all poor children lived in families headed by women. In 2012, the poverty rate for female-headed households was 40.9% compared with 22.6% of male-headed households and 8.9% for families headed married couples [55
]. Figures for minority families were even higher than these averages. Similarly, elder women are especially dependent upon government health care services as they are more likely to live longer and consequently more likely to live alone [21
]. Living longer fosters higher rates of various disabilities and the need for medical care and assistance [57
]. But the income histories and net worth of women are characteristically lower than for men [58
], so paying for professional health and long-term care services is frequently difficult. Moreover, adult women often keep closer track and are more keenly aware of their parents’ situations than men, and they routinely bear the lion’s share of the care-giving provided by family members to elderly parents. Consequently, Medicaid’s coverage of personal care and home/community-based services is especially crucial to elder and disabled women, with one in ten women relying on Medicaid for health [18
]. According the Census Bureau, during the period under study, between seventeen and twenty-two percent of children in the United States (15.75 million in 2010) lived in poverty, making the Children’s Health Insurance Program (CHIP) especially salient [60
]. These services are particularly important for women of color who are especially vulnerable.
It is to be expected then that female legislators would be more sympathetic than their male counterparts to the long-term care needs and preferences of vulnerable populations like children, disabled adults, and elders because of the experiences of women voters as well their own personal histories with the stresses often accompanying family care giving [61
]. Hence if women legislators are likely to have an impact in an area of public policy, it is reasonable to assert that the impact should be discerned in the gendered realm of welfare policy that is focused on healthcare. Social welfare policies are also considered important women’s issues by the Institute for Women’s Policy Research (IWPR) [12
]. While much literature either looks at issues traditionally associated with women, including children, families, and health care, and feminist issues, such as abortion or violence, it is important to note that not all women are concerned about children and not all women support feminist policies.4
5. Methods and Analysis
In order to explore how spending for health care welfare programs aimed at children, the disabled, and the elderly differs by variation in the descriptive representation of women we pooled data from each of the 50 states comprising odd-numbered years between 1999 through 2009.5
All independent variables are lagged one time period, or legislative cycle, to allow for their substantive effect to manifest within the policy domain, so independent variables run from 1999 to 2007, and dependent variables range from 2001 to 2009. To examine these relationships, we estimate a random effects model using generalized least squares.6
This means that our analysis explores increases across states, not within states. We control for time-invariant factors using several variables as suggested by the literature and theory (see below). In addition, we include a lagged dependent variable to account for past expenditures explaining current spending as well as year dummies to control for national effects that may have influenced expenditures in similar ways across the states. Furthermore, even with our numerous controls, due to the possibility of correlation across time between states, we follow the lead of others and use robust standard errors clustered around states throughout the analyses. See the Appendix A
, Table A1
, Table A2
and Table A3
, for descriptive statistics on each of the variables.
With respect to children, we draw on average state Child Health Insurance Program (CHIP) expenditures per resident younger than 18 living in poverty. We compare the results for CHIP expenditures with three measures of state elder/disabled care and include average state Medicaid hospital care expenditures per enrollee, average state Medicaid personal care expenditures per enrollee, and average state Medicaid expenditures on home health services per enrollee.7
These home health expenditures represent the Medicaid funds used to maintain elders and disabled adults needing help with various activities of daily living in a variety of community, rather than institutional (i.e., nursing facility) settings. Personal care expenditures also help people with disabilities, the elderly and those suffering from chronic conditions. States had great flexibility in designing these programs. In 1998–1999, for example, per capita expenditures ranged from $0.02 to $91.21 [66
]. During the period under study, many states limited personal care services and in 2004 and 2005 several states froze reimbursement rates for hospitals [67
]. Overall, states spent less on CHIP than they did on any of the services provided to the elderly and disabled through Medicaid. Since CHIP is not an entitlement program, states have more flexibility and can require co-pays, limit benefits and determine eligibility.8
State expenditures on CHIP averaged $137 annually with a maximum of $872. In sharp contrast, spending on personal health care averaged $6520 per year with a maximum of $11,569 across the same period. The figures for hospital care averaged $2322 annually with a maximum of $4464, and averaged $232 per year with a maximum of $1268 for home health care. The dependent variables were mildly correlated. For example, spending on home health had a 0.30 correlation with spending on CHIP, a 0.36 correlation with spending on hospital care, and a 0.48 correlation with personal health care spending. The largest correlation, 0.74, between hospital care spending and personal health care was the only one to exceed 0.55. On the whole, spending on the various programs appears to be nuanced and not necessarily tied to spending on other similar programs (see Table A1 in the Appendix A
for additional statistics).
Our independent variable of primary interest is women’s descriptive representation. The literature on aggregate-level women’s representation employs three general approaches for translating descriptive representation into policy influence (e.g., [12
]). Various studies use the simple percentage of women in a state’s legislature on the assumption that, as this percentage grows, so too does the capacity for influencing substance. Other researchers think that women must be present in a “critical mass” [70
] in order to have much influence on legislation. Still others examine how long women have had a substantial presence in a legislature with the expectation that greater presence will become conventional or routine [30
]. Although percent representation may not translate precisely into policy responsiveness, we adopt the simple percentage option here, mindful of its potential limitations. During the time of our investigation, women held an average 22.9 percent of the seats within state legislatures, with a minimum of 7.9 percent and a maximum of 40.8 percent.
Because pressure from individuals directly benefiting from a governmental program is likely to effect spending along with increased need prompting a greater positive state response [72
], we include the size of a program’s target population,
or the level of state need for the program. We would expect that the size of the beneficiary population would be particularly relevant, in our case, since children and older persons are often held in positive regard by both public officials and ordinary citizens [74
], so spending on policies benefitting them should elicit greater support. This variable is measured using the percentage of a state’s population that is younger than 18 living in poverty (for CHIP) and the percentage of a state’s population that is 65 or older living in poverty (for Medicaid home health, personal health, and hospital care spending for the elderly and disabled).9
6. Control Variables
Because it is plausible that expenditures on health care are actually a reflection of institutional backdrops and voter predilections and not the manifestation of group preferences among legislators we add a variety of controls to capture state environment (see Appendix A
, Table A4
, for a correlation matrix of all continuous independent variables). Using theory as well as past precedent within the literature as a guide, we include controls reflecting variations in political, socio-economic, and demographic conditions that likely influence policy responsiveness. These variables serve a statistical purpose as well, potentially mitigating against biased estimates. Given these ends, we include variables that assess the natural proclivity of states to invest in social welfare programs. In order to account for differences in outcomes being explained by party affiliations [28
], like the tendency of Democrats to favor greater spending, we gauge state Democratic legislative control using an updated version of Ranney’s [77
] index which captures the degree of preponderance of seats held by Democrats [78
]. Higher values in the measure indicate a larger Democratic presence. Similarly, spending on health care policies benefiting the poor may not reflect the presence of large numbers of women, but the fact that the women present are Democratic. To control for this factor, the model includes a measure capturing the percentage of women who identify with the Democratic party (Democratic women). Larger values indicate a higher percentage of women within a state legislature being part of the Democratic party. Because competitive political environments have been found to be associated with more generous welfare benefits [46
], party competition is included and captured using Holbrook and Van Dunk’s measure (based upon a folded version of the Ranney index) [78
]. Greater interparty competition is reflected with higher values. While there is correlation between party competition and partisan control, Shufeldt and Flavin demonstrate both should be incorporated into models as they capture distinct elements [81
]. Variation in policy outcomes have also been tied to differences in ideological and cultural environments [18
], with more liberal states supporting more permissive spending. This is accounted for with measures of state liberalism, culture, and region in our model. State liberalism is assessed using Berry et al.’s revised state ideology measure (see Berry et al. [82
]) with higher values indicating greater liberalism. Culture is gauged using Elazar’s index (which ranges from 1 for traditional states to 20 for more moralistic states).10
Region is a simple dichotomous variable, classifying states based on a South/non-South distinction, with the former serving as the baseline category. The last institutional element included in our model captures the ability of the state to pay for social welfare policy initiatives. Material capacity is important for the execution of costly programs [72
] and is measured via state tax capacity (total taxable resources per capita) with states indexed to the national average and higher values indicating greater available resources.
In addition to capturing institutional elements of the state environment, we also attempt to assess constituent factors that likely affect spending on health care spending for vulnerable groups. In doing so, we control for citizen ideology using Berry et al.’s revised citizen ideology measure (see Berry et al. [82
]) with higher values signifying greater liberalism.11
Similarly, environments containing constituents who are generally more concerned with women’s issues might be more supportive of social welfare spending benefiting them [28
]. As such, we control for women’s participation levels and educational attainment (percent of state with a Bachelor’s degree or greater).
And lastly, we include a measure capturing the racial/ethnic composition of the state. Theoretical justification and empirical realities tend to be mixed leaving us with ambiguous expectations. Some scholars assert that growth in the size of a minority population leads to diminished support for policies that benefit them (racial resentment theory) [89
] with more racially diverse states providing less generous welfare benefits [49
] and lower Medicaid expenditures [83
]. Conversely, others argue that increases in the size of minority groups diminish racial hostility because of interracial contact with attendant growth in support for policies that disproportionately benefit the minority group [93
]. Regardless, including the state’s percentage of minorities controls for theoretically relevant factors.
Looking first at the main effects (Table 1
), we find little evidence to suggest that increases in women’s representation within state legislatures translates into greater expenditures on select welfare policies. While the percentage of women appears to affect CHIP spending, it is in an unanticipated direction.12
As the percent of the legislature comprised of women grows, a negative effect on spending manifests. Even though the effect is not large, it is somewhat unexpected. We believe that the negative results can be explained by the minority status of women legislators which makes it difficult to translate sheer presence into policy outcomes on such a large scale. No statistically-meaningful results transpire for any of the relationships involving the elderly and disabled populations. In fact, like CHIP funding, the direction of each is negative. This is merely suggestive, though, given that none of the coefficients are statistically significant. Taken together, the results imply that women as a group play a modest or insignificant role in affecting select welfare policies involving health care when considered alone. We would argue, however, that this is in part due to failing to account for the behavior of legislatures with large numbers of women responding to the aggregate need of state populations. While the model controls for the size of need as measured by the beneficiary population, it fails to examine its interactive effect with the presence of women in the legislature. This is important because, as a group, women should be better able to influence non-members to align on preferred policies when great situational need is present. This impact, however, will be greater with increased presence.
The size of the beneficiary population, in isolation, also fails to influence spending for both CHIP and home health care. It is significant, however, for personal health care and slightly so for hospital care. This suggests that need alone may not be enough to encourage greater spending across all policies. It’s unclear why it is significant for some but not others though. One of the most consistently significant effects involved the level of education within the state. In general, as the percentage of the population with a Bachelor’s degree or higher grows, so too does support for health care policies benefitting the poor across the different domains, except for home health spending. While many of the remaining controls are in the expected direction, few consistently register as significant. For example, Southern states are generally less likely to support greater spending. However, the relationship is only statistically significant and in the proposed direction for hospital care spending and CHIP. Similarly, citizen liberalism tends to have a positive relationship with spending, as does state liberalism, although they are only significant and in the correct direction for spending on home health and CHIP respectively. The remaining controls are a bit more irregular. For instance, increased party competition predicts decreased spending on CHIP, running counter to the literature, but increased spending on home health. The effect on the other dependent variables is insignificant, however. Interestingly, state racial composition does not have a consistent effect. While a statistically significant negative relationship between racial context and spending on personal health care manifests, it does not appear for the other variables. Future work may be necessary to explore the conditions of this relationship further as other research has found this variable to be significant.
The inability of our analyses to substantiate a relationship between the size of the representative population and diverse welfare policies involving health care for the poor lends evidence to the notion of a limited connection. However, we argue that failing to account for behaviors responsive to context has led to some of the null results here and within the literature. As such, we next explore the conditional relationship of beneficiary size and legislative representation on health care spending targeted toward vulnerable populations involving children, the disabled, and the elderly.
Looking at the coefficient on the interaction term, it is quickly evident that the presence of women in the state legislature, in conjunction with aggregate need, influences spending levels (Table 2
). In general, legislatures with more women tend to be responsive to context in ways that reflect the need of the state. For example, as the percent of women in the legislature and the percent of poor children in the state both go towards their maximum values, spending on CHIP increases by roughly $40 per recipient—about one-third the average amount of $137. Similarly, when the percent of women in the state legislature and the percent of vulnerable adults are held at their maximum value, state expenditures on hospital care spending are likely to grow by $236, or about 10 percent of the average amount of $2322. The effect is comparable for personal health care spending and home health spending, although the latter just misses standard levels of statistical significance. These effects hold when controlling for the partisanship of the women legislators.13
Thus, party politics are not driving the relationship. However, when the population of women is large but the recipient group is small, less average funding is secured across the board. That is to say, states with large percentages of women legislators do not appear to act similarly across all contexts and environments but are rather responsive to aggregate state conditions. If legislatures were to be supportive regardless of target population size, then average funding on health care for vulnerable populations would be uniform across states notwithstanding this particular context. Rather, legislatures with large numbers of women appear to respond to the size of the need across the state. Because women never comprise a majority under either condition, it suggests that men align with women’s interests when great need is present to build support for heightened allocations.14
Moreover, because more spending occurs when the beneficiary group is large and
the number of women representatives is substantial, influence is predicated on growth in the presence of women. We also find, however, that in the absence of women, as the recipient group grows larger, there will be less funding for health care for these vulnerable groups. Thus, without the presence of women, individuals benefiting from these categories of health care spending will see less money, implying that women legislators are crucial for increases in spending to occur.15
On the whole, the results suggest that our initial inconclusive findings, as well as those within the literature, can likely be attributed to failing to consider the influence of aggregate need upon the responsiveness of institutions in allocating resources to select populations. When we include this dynamic in our statistical model through interacting the size of the beneficiary population with the degree of women’s descriptive representation, we find significant and meaningful relationships. Modeling the nuanced manner in which legislatures respond to these public policies appears necessary for capturing its natural complexity. Interestingly, the findings also suggest that holding positions of power is not necessary for women to influence policy outcomes. They merely need to be numerous and in an environment with substantial aggregate need.
As before, the level of education at the state level is consistently predictive of state expenditure levels. In general, when the percentage of the population with a Bachelor’s degree or higher grows, so too does support for health care spending for these vulnerable groups across the various areas. Southern states tend to support decreased spending across the various policies, a relationship that is statistically significant for hospital care. The degree of state liberalism continues to be a significant predictor of CHIP spending, however liberal tendencies among the general population only mildly predict home health spending. Again, somewhat unexpectedly, state party competition affects CHIP spending in a negative direction. Greater party competition within a state environment indicates decreased spending on welfare benefiting children, contrary to the literature. While many of the remaining controls are in the expected direction, few consistently register as significant.
8. Conclusions and Discussion
The findings of our research carry significance in a variety of ways. First, we examine past research on the effect of women’s representation on policy outcomes, speculate why results have been inconclusive, and develop a novel theoretical framework in resolving inconsistencies. Acknowledging that women political leaders tend to be more supportive of “women’s issues” than men, we argue that their influence is not automatic but rather dependent upon context. Having women present in the legislature may alter the dynamic, making institutions potentially more responsive to state aggregate need, but given that women are consistently in the minority, policy responsiveness manifests only under prescribed conditions. One such condition occurs when aggregate need is substantial. This is likely because just enough members of the majority align with women on these policy outcomes to elicit greater spending. Arguing that the inconclusive findings, both our own initially and within the literature, have arisen as a result of failing to acknowledge this situational factor, we include the dynamic in our statistical model by interacting aggregate need, as determined by the size of the beneficiary population, with the degree of women’s descriptive representation. This enables us to capture the nuanced manner in which legislatures with more women approach these public policies. Modeling this behavior with interaction terms, we find that average spending on these components of health care substantially when both the recipient population and the number of women within a state legislature also grow. Importantly, this effect holds even when controlling for the partisanship of the women legislators suggesting that gender is of greater importance than party labels. In addition, when the population of women in the state legislature is large but the recipient group is small, less average funding is secured in the interactive models. In the same vein, spending on welfare policies involving health care for vulnerable populations does not increase significantly with the growth of women in the legislature as a main effect. Taken together, these findings suggest that presence alone does not guarantee policy influence. If this were the case, then average funding on health care spending for these vulnerable populations would be uniform across states regardless of the size of the beneficiary population. It would also be the case that growth in the representation of women would translate into increases in spending on these policies. The findings suggest, rather, that great need, in conjunction with growth in presence, is necessary for spending to increase. Because women do not have the votes to go it alone in these circumstances, some members of the majority must align on these priorities.
Furthermore, in the absence of women in the legislature, vulnerable populations would be more at risk. When women are elected in significant numbers and the need is great, higher amounts of money are spent on vulnerable children, disabled adults and elders. Our findings indicate that in the absence of women, as the recipient group grows larger, there will be less funding for health care funding for these vulnerable populations. Thus, without the presence of women, individuals benefiting from these spending programs will see less money. Similarly, large populations alone do not guarantee increases in spending. This suggests that the presence of women matters, even if it is conditional upon substantial need.