Scott Winship’s article about the Pew Research Center’s report, “The Lost Decade of the Middle Class,” opens by misstating one of our basic findings. He claims we say that “over the past 40 years, the middle class has grown poorer.” In fact, we say exactly the opposite. Our analysis of Census trend data shows that, from 1970 to 2010, the median annual household income of the middle tier of the population rose by 34 percent, after adjusting for inflation and changes in household size. Our report also shows that this finding is in sync with the public’s perceptions about its long-term economic mobility. Asked to compare their standard of living now with that of their parents at the same age, 60 percent of self-identifying middle-class respondents to our nationwide survey say theirs is better. Just 13 percent say it is worse (the remainder say it is the same).

While our report looked at these and other long-term trends, it focused mainly on changes over the past 10 years — the first decade of the post–World War II era in which the median household income of the middle-income tier declined (it fell by 5 percent); and also a time when the median wealth of the middle-income tier plunged (by 28 percent). Along with these economic findings from the Census, our survey found an increase in the economic anxieties of the public across a wide range of measures and a decline in the share that self-identifies as middle class. Thus, we began the report as follows: “Since 2000, the middle class has shrunk in size, fallen backward in income and wealth, and shed some — but by no means all — of its characteristic faith in the future.” Mr. Winship accuses us of “inventing” this decade of decline. We believe we’ve reported the facts and the public’s perceptions of the facts — which are in sync.

Mr. Winship also takes us to task for some of the methodological choices we made. For example, he suggests that we would have gotten different findings had we used different statistical boundaries to define the middle-income tier of the public. However, as we noted in the report, we ran simulations using many different boundaries; the basic trends were unaffected. He suggests that the use of the decade time frame distorted our analysis; we chose the decade time frame because median household income in the United States peaked in 1999 and has not returned to that peak since — the longest such stretch in our modern economic history. He also suggested that we would have gotten a different picture had we excluded from our analysis the Hispanics who have come to this country since 1970. Yes, we often add value to our findings by looking at variations among different sub-groups — by race, ethnicity, age, gender, immigrant status, marital status, etc. — and we did so in this report. But the overarching goal of this report was to draw a data-driven portrait of America’s middle class — all of it — and to look at how it has changed over time. We’re confident we achieved that goal.

— Paul Taylor is Executive Vice President of the Pew Research Center.