Well, yes and no. In reality, that pay gap narrows when you start to account for variables like age, education, and experience. A study released today by career review site Glassdoor finds that when you control for those factors—as well as company, industry, geography, firm size, and job title—two thirds of the pay gap disappears.

Here’s a closer look at how each set of variables impacts the gap. The chart below shows how, as you factor adding in more variables, the average pay gap keeps shrinking, going from 24% down to just 5.4%. Translation: When a man and a woman work in the same type of job, in the same city, and have the same levels of experience, the difference in pay is only about 5 cents.

However, before we start jumping for joy, let’s look at the data in another way. Those five cents represent how much more money men make for no reason at all—or at least no reason economists can find. In fact, about a third of the overall U.S. pay gap is “unexplained.” This difference has typically been attributed to gender bias, says Glassdoor chief economist Andrew Chamberlain.

“If you asked an economist 25 years ago, he or she would say [those five cents represent] discrimination. Today, we have a more nuanced view. It could be workplace bias, but it could also be workplace data that we don’t see: race and ethnicity, kids, an elderly parent.”

To make matters even more complicated, while the “unexplained” part of the pay gap isn’t necessarily caused by bias, the “explained” portion—the two-thirds of the pay gap that economists can account for by looking at things like industry and education—very well could be.

“Just because we can explain why there’s a gap it doesn’t mean that this is due to [women’s] choice,” says Chamberlain. In other words, there may be larger reasons why a woman has a lower level of education or works in lower paying field than her male cohorts. Perhaps she was steered away from certain majors when she was in school, or can’t work in an industry like law or finance because of care-taking responsibilities.

“If you think about it for five minutes, it’s clear that most of these things aren’t choice but systemic,” says Chamberlain, a former economics instructor at U.C. San Diego.

Chamberlain is the first to admit that there are limits to what we can learn from sweeping studies like this one (Glassdoor looked at the salaries of 505,000 U.S. employees). Researchers are all “walking around an elephant in the room and measuring it differently, and it’s never zero,” he says. In other words, there will always be a part of the gap that cannot be accounted for statistically.

So if stats don’t cut it, how do we figure out what’s causing that gap? One answer is experimentation, says Emilio Castilla, a professor at the MIT Sloan School of Management. “The beauty of experiments is that you can establish a causal link” between people’s work and how they are rewarded.

In one study, Castilla ran an experiment in which he randomly assigned managers fictional profiles of male and female employees— with the same exact backgrounds and levels of performance—and asked them to allocate bonuses. What he found was that women were systematically awarded lower bonuses, even though managers were told to only look at an employee’s merit.

Because these are fictional scenarios, we can’t blame the lower pay the woman received on her desire for flexibility, her poor negotiation skills, or her need to take care of the kids. The answer here can only be bias.

Now let’s try and explain that.