The 2018 elections promise to be the “Year of the Woman,” with more women planning to step into local, state and federal elections than ever before.

This represents a significant change. The United States has some of the lowest female political representation in the world. Only 24.8 percent of state legislature seats are occupied by women. As more women consider entering politics in response to Donald Trump’s sexist remarks during the election and the historic Hillary Clinton loss, findings from our study on voter attitudes offer a warning: Candidates shouldn’t assume women will vote for other women.

This was evident in the 2016 presidential election. Hillary Clinton worked to appeal to female voters but performed poorly among white women. Some have argued Clinton’s personality caused her inability to emotionally connect to voters.

However, our research indicates that Clinton’s failure to capture the white female vote is, in part, based in something more fundamental – marriage.

The fate of women

We used data from the American National Election Study, which has collected data on American voters’ attitudes since 1948. In 2012, more than 2,000 women were asked: “Do you think that what happens generally to women in this country will have something to do with what happens in your life?”

Those who answered “yes” were then asked to report the extent to which what happens to others affects them. We used this measure to identify “gender-linked fate,” or the extent to which women see their futures as tied to those of other women. We found that married white and Latina women were less likely to view their fate as tied to other women.

We then tested whether this impacted their political attitudes. We found that when married white women felt disconnected from other women, answering “no” to the question above, they were less likely to identify as a Democrat, and more likely to hold conservative political views.

In contrast to these married white and Latina women, single and divorced whites and Latinas were more likely to see their futures tied to other women. As a result, they were also more likely to identify as Democrats and liberals.

Black women, regardless of marital status, were most likely to see their futures as tied to those of other women and consistently voted democratically and held progressive attitudes.

In part, this captures a difference in message. While the Republican Party has focused their efforts to redress gender inequality in a platform that emphasizes family values, Democrats have focused more explicitly on equalizing opportunity by reducing institutional gender discrimination.

Single and divorced women resonate more with the Democrats’ message, with four times as many respondents in our data reporting that the Democratic Party did a better job looking out for women’s interests than the Republican Party.

So why does marriage alter white and Latina married women’s political alliances? And, why do black women not follow the same trend?

You and me: Marriage and changing behaviors

Research suggests that marriage generally shifts individuals’ attitudes and behaviors. For example, evidence shows married women become more conservative on gender-related issues over the course of their marriage and perceive themselves as having less in common with other women. In part, this captures the fact that many married couples become more similar to each other in their attitudes and behaviors.

Someone could reasonably ask, why does marriage make married women more conservative, rather than making men more feminist? It’s a matter of power and resources.

Women consistently earn less money and hold less power, which fosters women’s economic dependency on men. This dependency increases if women reduce employment and rely on husbands’ earnings following the birth of a child. Thus, it is within married women’s interests to support policies and politicians who protect their husbands and improve their status.

Some married women perceive advances for women, such as lawsuits to mitigate pay discrimination, as coming at the expense of their male partners. In part, this captures the shift in married women’s alliances from the individual to the marital union. Women who depend on their own income are more supportive of feminist issues such as abortion, sexual behavior, gender roles and family responsibilities, which widens the political gap between single and married women.

It follows that politicians cannot expect married women to vote as a block on women’s issues.

The only exception to this rule may be black women. An existing body of research shows that blacks are better able to identify systematic forms of discrimination because of their experiences with it. Thus, they are more likely to see their futures as tied to other blacks. Our study shows this extends to gender as well. Black women are better able to identify gender discrimination regardless of marital status and, as a consequence, they vote more progressively.

Lessons for female politicians

Given that married women make up about 30 percent of the electorate, what lessons might 2018 hopefuls draw from our research?

First, targeting messages to the demographics of the audience can make a difference – and this includes race, class and marital status.

Second, don’t assume that married women will connect to other women based on a notion of shared womanhood. Rather, feminist messages of discrimination and sexism may be more compelling to women who shoulder disproportionate levels of inequality, poverty and job insecurity – single, divorced and black women.

Finally, messages about economic struggles should be expanded to the family level, to better capture the challenges of married couples.

As we deepen our understanding of women’s voting patterns, the 2018 elections may prove a momentous expansion of women’s political representation.

Leah Ruppanner receives funding from the Australian Research Council.

Christopher Stout and Kelsy Kretschmer do not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond the academic appointment above.