The New York Times ran a surprising graph last week showing that the current economic recovery has been bad for white workers but a boon for workers of color. When we look at the numbers the way most economists would prefer, however, the data actually look disappointing for American workers across the board.

We’ve reproduced the Times’s graph in the left-hand panel of the chart below. The data show large job losses for white workers—down about 700,000—between November 2007 (the month before the recession began) and November of this year. Over the same period, employment of workers of color increased by millions, with African Americans and Asians both up by more than two million and Hispanics up by almost five million. The text accompanying the New York Times figure concluded that the data indicate “robust gains in employment” for these groups.

But, what is “robust” in this context? For this kind of analysis, economists generally prefer to compare changes in employment to changes in the corresponding population over the same period of time. To see why, imagine that an economy creates one million new jobs in a year, but the working-age population rises by two million people over the same period. Even with the all of the new jobs created, workers as a group would be less likely to have a job at the end of the year than at the beginning. We probably wouldn’t want to describe a level of job growth that left a smaller share of the population with a job as “robust.”

That is exactly what happened for workers of color in the United States since 2007. Employment grew, but the corresponding populations grew more. The right-hand panel of the chart below uses the same numbers in the original New York Times figure, but now expresses them as a share of the corresponding population. (See Figure 1.)

Figure 1

This picture is remarkably different. Once population growth is taken into account, we can see that employment rates fell for all four racial and ethnic groups in the analysis. The job growth for African American, Asian, and Hispanic workers emphasized in the newspaper’s account did not keep up with the growth in the corresponding population.

For three groups—Asian, Hispanic, and white workers—employment rates in the most recent data are still well below their pre-recession levels. For black workers, employment rates did not fall as much, but they are still below where they were in 2007. And before breaking out the champagne to celebrate the relative success for African American workers, keep in mind that the employment rate for black workers ages 16 and older in November 2016 was only 56.9 percent, trailing well behind whites (60.1 percent), Asians (61.0 percent), and Hispanics (61.9 percent).)

In a follow-up piece three days later, the newspaper acknowledged the importance of using employment rates to put employment numbers into proper context. But, the new piece argues that taking population into account does not change the original story much. The right-hand panel of the figure above suggests otherwise. The Great Recession and the current recovery have been tough for workers all around.

(For more discussion of these issues, see two analyses by the Center for Economic and Policy Research’s Dean Baker and one by The Washington Post.)