I don't agree with the way Pew defined middle class. Here is what they did:
Pew defined middle class as households earning two-thirds to twice the overall median income, after adjusting for household size.
When looking at distributions, the convention is to consider the "average range" or "middle" to be elements of the population in the 16th percentile through 84th percentile range. In other words: The "middle" 68 percent.
By that convention the size of the middle class never changes. It's just a question of how that middle 68 percent is doing. And that middle 68 percent is doing better in terms of inflation adjusted household income than it was in 1971 even after the losses of the most recent few years.
You can tell that by looking at the "Supplemental Data" spreadsheet at
https://www.cbo.gov/publication/49440" onclick="window.open(this.href);return false;. It only goes through 2011 but hopefully we can agree that there's no way we've lost enough since 2011 to bring us back to 1979. And I think we can all agree that, in looking at the trend since the GAO started tracking things, 1979 was almost certainly higher than 1971.
In the supplemental data here is how average household incomes for each quintile of the population compared between 1979 and 2011 in terms of 2011 dollars:
There's no way around it. No matter where you go on the income distribution percentile scale someone at a given percentile in 2011 was WAY better off inflation adjusted income wise than they were in 1979. And look at the percentages by which they were better off. The averages for each quintile were higher by from 24% through 78%. The LOWEST quintile average was 40% higher in 2011 than it was in 1979.
Yes things have probably gone down some since 2011. But by better than 20%? Does anybody here believe that?
The linked article's cited conclusions are an artifact of the way Pew defined "middle class." While it wouldn't be possible to confirm it unless you had the income level of each individual percentile for each of the years you're comparing, the available evidence overwhelmingly suggests that people at any given percentile in 2011 were better off by a LOT than people at that same percentile were in 1979 and also by extension in 1971.
That's not deceptive or manipulated statistics at all. It is what is is. If you want to question the accuracy question the GAO and not me. And there really is no basis for defining the "middle class" as "households earning two-thirds to twice the overall median income, after adjusting for household size." That's totally arbitrary.
I guess at some point the convention of defining the "average range" as the middle 68 percent was arbitrary to a degree. It comes from the percent of elements that are within +/- one standard deviation of the mean in a normal distribution. But that is the convention and I don't think it's as arbitrary as what Pew did.
In any case I think it's a better way of looking at things. You define the "middle class" as some percentage of the population and look at how that "middle" percentage of the population is doing at any given time. The way Pew did it people could be doing better and you can say the "middle class" is shrinking. In fact that's exactly what happened in this case.