In a post here on Wednesday, I criticized the New York Times and the Clinton campaign for relying on graphs put out by the Economic Policy Institute that purport to show that over the last few decades, productivity has grown much faster than wages. I cited Scott Winship of the Manhattan Institute, who corrected for several flaws of the EPI graph—I noted, in particular, that when considering wages it did not use the same inflation adjustments or count the same workers as when considering productivity—and found no such long-term trend.

The post prompted an energetic pushback on twitter from the president of EPI, Larry Mishel. He made one reasonable point: Winship’s version of the graph tells you how average wages have stacked up against average productivity, but it’s worth knowing how median wages have performed against median productivity. It could be that very high earners have pulled up the average figures, but the guy in the middle of the wage distribution has not been getting raises commensurate with his productivity gains. This point is so reasonable that it has, in fact, been made by Winship in the course of criticizing EPI’s graph.

That graph doesn’t tell us that this possibility is a reality: that the median worker’s wages have been falling behind his productivity for decades. It doesn’t tell us anything useful. The Times and the Clinton campaign should quit using it (and, in the case of the Clinton campaign, misrepresenting it).