For this anniversary, we are republishing this piece originally published in 2016. — The editors

Today — Aug. 26 — is Women’s Equality Day. On this day in 1920, the 19th Amendment to the U.S. Constitution became law. Suffragists had worked for decades to secure American women the right to vote — and won.

Yet almost immediately, observers began to declare that women’s suffrage was a failure. By the mid-1920s, popular magazine headlines like these were common:

“Is Woman-Suffrage a Failure?” (The Century, 1924)

“Is Women’s Suffrage a Failure?” (Good Housekeeping, 1924)

“Are Women a Failure in Politics?” (Harper’s, 1925)

Critics attacked women’s suffrage on many grounds. Perhaps most damning was the claim that the vast majority of women simply failed to vote. Scholars repeated this conclusion in American politics textbooks and histories for generations.

But the evidence for this claim was extremely limited. Women and men did not fill out pink and blue ballots, so official voting records generally cannot tell us the number of women who voted. While states now maintain records of who votes in which elections, in the 1920s this type of information was either not recorded, not preserved, or did not include the sex of the voter. Today we also track who votes through exit polls and public opinion surveys, but those tools either didn’t exist or weren’t reliable in the 1920s.

We figured out what proportion of women and men voted. Here’s how.

In our book, “Counting Women’s Ballots: Female Voters from Suffrage through the New Deal,” we get around this problem by using a statistical tool known as ecological inference, which lets us combine information from the U.S. Census about the population (in this case, the number of voting-age women and men) with information from the voting record (such as the number of votes cast for each party) to estimate the percentage of women and men who voted and for which parties in a sample of 10 states. (If you’re interested, our book goes into detail about the evidence that shows you can rely on our estimates.)

Did women vote in the first years after suffrage?

Overall, critics were right: Few women turned out to vote in the first elections after they got the right to do so. In 1920, just 36 percent of women, on average, turned out to vote across our 10 states, compared with 68 percent of men, a 32-point gap. Apparently inexperience and persistent beliefs — whether women’s or their families’ — that voting wasn’t appropriate for women kept turnout low.

But these averages mask a more complicated story about variation from state to state. The figure below summarizes the estimated turnout of women and men in our 10 states, arranged from lowest to highest turnout.

You can see that fewer than 10 percent of women turned out to vote in Virginia, and barely 20 percent in Massachusetts and Connecticut. But in Missouri and Kentucky, more than 50 percent of women turned out to vote in the first election in which they were eligible. In those states the turnout gap between women and men is only around 25 points. In Connecticut, by contrast, the gap is more than 40 points.

Why did so many women vote in Missouri and Kentucky?

In the early 20th century, most states were either very Republican (North and West) or very Democratic (South), a legacy from the Civil War. Who was going to win in those states was rarely in question: Democrats in Virginia held a 23-point advantage in the 1920 presidential election, and Republicans took Massachusetts by a margin of more than 40 points.

In contrast, Missouri and Kentucky were both more competitive. In fact, Democrats won Kentucky by less than 1 percent in 1920. Close competition meant that every vote counted. Parties and candidates had good reasons to mobilize every potential voter, including women. And when elections were close, women turned out to vote — not to the same extent as men, but closer, and at rates that are comparable to contemporary turnout. Consider that 57 percent of voting-age Americans actually went to the polls in 2012; roughly that percentage of women voted in Kentucky in 1920.

Why did so few women vote in Massachusetts, Connecticut and Virginia?

There’s another reason that some states had lower female turnout: They had significant barriers to voting. Massachusetts and Connecticut required voters to take literacy tests. Connecticut and Virginia had long residency requirements. Virginia levied a poll tax, among other discriminatory practices directed at African Americans and poor whites. These practices reduced turnout among men as well — but suppressed the votes of newly enfranchised women even more.

These laws probably had an outsized impact on women of color. In 1920, nearly 30 percent of Virginia’s eligible voters were African American. We have every reason to believe that, despite determined efforts to register and vote, the vast majority of African American women were unable to, contributing to dismal turnout rates statewide. For these women, effective enfranchisement was delayed for decades.

When races were close, and when there were fewer restrictions on voting, women turned out in large numbers, even in the first elections in which they could. That made for a dramatic difference in female votes cast between the high-turnout and low-turnout states, for a more than 50-point difference between women in Virginia and women in Kentucky in 1920. That difference between women in different states is larger than the difference between men and women in any one state.

In other words, context affected turnout as much or more than gender. Where women first had the opportunity to vote mattered at least as much as the fact that they were women.

Turnout lessons from elections nearly 100 years ago

Here’s what we can learn from the supposed failure of the first women voters. Voting may be an individual responsibility but the actions of parties and other organizations make a big difference in getting us to go to the polls. And that mobilization is most likely to happen in places where the race is close — precisely the kinds of contests that are again increasingly scarce in U.S. politics.

Further, laws that make it harder to vote weigh most heavily on those people who are already marginalized. Rather than asking if potential voters — women in 1920 or other groups today — are failing to fulfill their civic responsibilities, we might do better to ask if the political system is failing its citizens.

Kevin Corder is professor of political science at Western Michigan University.

Christina Wolbrecht is professor of political science at the University of Notre Dame.