RALEIGH — The income of the median American household fell by nearly 3% last year as the COVID-19 pandemic and subsequent regulations shuttered many businesses for months, closed others for good, and forced still other employers to cut back on hours and wages for the people they still employed.

Or so the official federal income statistic tells us. It includes wages and salaries, of course, as well as investment gains and unemployment-insurance benefits. However, it doesn’t include tax refunds, stimulus checks, or noncash assistance in the form of food or housing. If those forms of income were included, median household income in the United States went up in 2020, by 4%.

Similarly, the standard poverty measure from the U.S. Census Bureau hit 11.4% last year. That’s up a percentage point from 2019, representing about 3.3 million more poor Americans. But, again, the standard poverty measure leaves a lot of income out of its calculation. According to the bureau’s supplemental poverty measure, 9.1% of Americans were poor in 2020, a big drop from the 11.7% rate it reported in 2019.

Are these just examples of statistical fun and games? Hardly. How policymakers, opinion leaders, and the general public respond to economic issues cannot be attributed solely or even mostly to their own experiences. Decades of polls suggest that Americans tend to rate their own economic present and future much more positively than they rate the overall economy’s present and future. The latter is based more on what they see, read, or hear from news reports rather than their own personal experience, which is inherently limited.

It has never made sense to measure and track trends in personal incomes and poverty based on such narrowly circumscribed and unrepresentative “official” statistics. To say that a family would be poor if not for off-the-books income or government assistance, for example, is to convey useful information, to be sure. But it doesn’t tell us whether that family is actually living below the poverty line in terms of total cash, goods, and services received.

Indeed, even the Census Bureau’s supplemental poverty measure still leaves out too much. As economists Bruce Meyer of the University of Chicago and James Sullivan of Notre Dame have demonstrated convincingly in a series of published papers, the true poverty rate has averaged well below 5% in recent years.

Having written about this issue for a long time, I can say from personal experience that highly partisan analysts dislike hearing about it. Progressives say such alternative measures are nothing more than an attempt to wish away problems of poverty and income inequality. Conservatives say to include government benefits in income and poverty measures is to give too much credit to welfare policies that, they insist, have had little effect since the onset of the War on Poverty in the 1960s.

These faulty responses serve as an excellent illustration of why accurate statistics about economic conditions are so important. They aren’t just numbers. They are tools for depicting real facts on the ground. To measure the poverty rate without including all cash and noncash benefits the poor receive from the government is to misrepresent reality. Doubling or tripling welfare spending would, by this measure, have no effect on poverty. What a silly notion.

As for my fellow conservatives, we are better off distinguishing between measurements of living standards and measurements of self-sufficiency. The War on Poverty wasn’t simply a promise to redistribute income indefinitely, in order to alleviate immediate suffering. Its promise was that early interventions by government, ranging from preschool and child care to job training and housing assistance, would help more American families become self-sufficient over time.

Many programs did provide immediate relief, and continue to do so. You can measure their effects with accurate statistics, which show a dramatic decline in poverty since the 1960s. But for the most part, they didn’t produce self-sufficiency.

Poverty remains a big problem afflicting millions of people. Measuring it incorrectly doesn’t move us towards long-term solutions.