A May post from the left-leaning Economic Policy Institute gained some attention this week. Written by Valerie Wilson, it offers an interesting explanation for why unemployment among African-Americans has fallen slower than that for white workers since the start of the recession. Wilson's answer: Black workers have been more likely to keeping trying to find jobs, meaning that they don't drop out of the workforce. And that means a higher unemployment rate.

Wilson offers this graph.

There is a link between unemployment and workforce participation, as we explained earlier this month. If you have 100 people in the workforce and 80 are employed, your unemployment rate is 20 percent. If 10 people retire or stop looking for work, 80 of the 90 people are now employed, and your unemployment rate drops to 11 percent without anyone getting a new job.

The problem is that Wilson's explanation doesn't appear to hold up over time.

Looking at employment data since 1972, black unemployment has been higher than white unemployment in every month. In every month save one (June 1972), the participation rate for blacks has been lower. Here is the value for each January over the past 42 years.

That suggests that the problem may be more deeply rooted than the participation rate. But let's look at her numbers.

Wilson's graph doesn't compare one year to the next. It compares April 2014 to the average unemployment and participation numbers for 2007 to get her comparison above. We did the same comparison, going back to 1980. We compared April 1980 to the average values of 1973, and so on -- giving us this.

At the bottom, those last four lines, you see Wilson's graph, turned to the side. A bigger change in unemployment for blacks, as at right, and a smaller drop in participation, as at left. But look two years above that. Blacks saw a slightly bigger increase in unemployment between 2012 and 2005, but a bigger drop in participation.

In fact, over these 35 years, Wilson's correlation — higher unemployment linked to a smaller drop in participation, or vice versa — only holds up 16 times, less than half.

The gap in employment between whites and blacks is endemic, and not strongly correlated to labor force participation. If Wilson's theory were right, it's hard to see how this would be the case.

As for Wilson's point that in April black unemployment was 2.2 times higher than white unemployment: Over the past 509 months (since 1972), that's been the case 231 times.