The media has often repeated statistics that show that pay for the average worker has remained basically flat since 1979, even as worker productivity has soared. Left-leaning politicians like Massachusetts Senator Elizabeth Warren have argued that if the minimum wage had increased along with worker productivity, we’d have a $22 per-hour minimum wage. And if the household income of the middle class—those making between the 20th and 80th percentiles of income—had merely kept pace with worker productivity, the average middle class household would be bringing in $17,890 more than it was in 2010, according to the Economic Policy Institute.

That’s a lot of money for the middle class to be leaving on the table, and it makes for good politics to appeal to people who feel the system is stacked against them. The problem is the statistics used to make the case often oversimplify what exactly is happening in the economy. Take, for example, this chart, also from the Economic Policy Institute, which purports to show the relationship between worker productivity and worker pay:

Looking at this chart, it appears as if a fundamental change in the economy occurred during the 1970s. For the political left, this is convenient, because it fits into a larger narrative about the decline in the power of labor unions, the deregulation movement which began around the same time, and the decline in manufacturing employment. But putting together such a chart requires a lot of assumptions, including which employees you choose to include in the wage measure and the measure of price growth used to gauge inflation. Use a different set of assumptions and you get a very different picture. Take a look at this chart from Scott Winship at the Manhattan Institute:

Unlike the EPI numbers, Winship uses a slightly different measure of inflation to adjust both productivity and compensation for inflation. He also includes the pay of all workers, not just non-supervisory employees. Finally, he adjusts the productivity to better measure what we’re trying see, which is, “whether workers are being fairly compensated.” Normally, we measure productivity by dividing GDP by the total number of hours worked. But if we’re trying to see what is happening in the workforce, it doesn’t make sense to include things like depreciation of plant and equipment in our measure of productivity growth, as both workers and owners benefit from those things. It also doesn’t make sense to include income from rent, which requires little labor to make productive.

Real wage growth figures can also be tricky. By many measures, like inflation adjusted hourly earnings, the median worker is worse off today than he was in the 1970s. But as Aaron McNay, an economist for the state of Montana, pointed out to the The Wall Street Journal, that changes significantly depending on what measure of inflation you use. When McNay measures real average weekly pay when looking at inflation as seen through personal consumption expenditure (how much households are spending), it turns out that average worker pay is at an all all-time high.

Finally, there is a debate in the economic community over whether the share of national income going to workers versus owners of capital is on the decline. For decades, economists thought that the share of GDP going to both groups was stable over time, but in recent years, increasing evidence suggests that owners are taking a bigger share of the pie. But again, this all depends on what sorts of metrics you use. Recently a 26-year old graduate student from MIT showed that the decline in the labor share is, contrary to what Thomas Piketty has argued, a product of appreciating real estate rather than the increased power of owners over labor.

Few economists would argue that income and wealth inequality across the rich world aren’t growing. And as the well off capture more of the gains of economic growth, that is necessarily going to hurt the median worker. But the magnitude of the average workers’ plight is very difficult to measure in any precise way. And the solutions that politicians have proposed to help the average worker—from raising the minimum wage, to across-the-board tax cuts, to changes in education policy—are, to say the least, not silver bullets.

In fact, real change to our system is more likely to come from the bottom up, with local governments and community colleges working with corporations to craft a supply chain for labor, and with individual workers learning to make better decisions about how they gain skills. Sure, something like smart tax reform can only do the economy good, but don’t expect big policy changes to solve a problem that we have difficulty even measuring.