‘Statistics Don’t Lie,’ But ... 3 Principles that Illuminate the Massive Downward Revision in U.S. Jobs

The U.S. Bureau of Labor Statistics (BLS) this week announced the largest negative revision to their employment estimates in at least 15 years. After initially estimating the U.S. economy added 2.9 million jobs between March 2023 and March 2024, they now estimate the number at closer to 2.1 million, a difference of 818,000 jobs. This means their initial estimate was off by 28% — a huge miscalculation, at least for anyone but a meteorologist. If a dentist were off by so much, he would stab his client in the eye.

While it’s not clear how the estimate could be so inaccurate, it’s more important to consider our response. The announcement triggered a predictable barrage of finger-pointing in all directions from public figures. Equally implausible accusations came from former President Donald Trump and Biden administration Commerce Secretary Gina Raimondo. Both are likely engaged in electoral posturing.

Trump declared that “the Harris-Biden Administration has been caught fraudulently manipulating Job Statistics.” It’s possible, of course, for political concerns to influence federal bureaucrats — even statisticians — so that they manipulate the data to serve a political agenda. But the implausibility of this claim lies in the actor behind the passive verb, “has been caught.” The BLS published the initial jobs estimate, and the BLS revised them. In effect, Trump is asking his audience to believe that the same people who were “fraudulently manipulating” statistics for political gain caught and exposed their own scam and did so weeks before the election. Without a compelling explanation for such a total about-face, this interpretation simply does not fit all the facts.

Equally implausibly, Raimondo suggested in an interview that the revised numbers had been fabricated by Trump. She responded to Trump’s accusations, “I don’t believe it, because I’ve never heard Trump say anything truthful.” Raimondo either was ignorant of, or feigned ignorance of, the contents of the scheduled BLS update, which had been released hours earlier. Despite the narrative crafted at the DNC, the Biden-Harris administration currently possesses the levers of power, while former President Trump has no power at all over the BLS number crunchers.

Politicians are not the only ones susceptible to bad takes on unwelcome economic news. The dramatic revision could also trigger ordinary citizens to feel gloomy or angry, or even erode what remaining trust they have in governmental institutions.

Lest we succumb to such temptations, it is wise to calibrate our expectations of job growth statistics (and other types of statistics) with a look at the big picture. You may have heard the saying, “Statistics don’t lie, but people do.” While this is true, I don’t think it helps to explain what happened here. Instead, I offer three other principles, in a related vein, which readers can apply to the U.S. jobs report estimate, as well as any other statistics they encounter.

1. Statistics Don’t Lie, but They’re Educated Guesses

Many figures that are tossed around in public discourse as concrete facts are not as solid as they seem. Economic snapshots, public opinion surveys, and exit polls all rely on statistics, a branch of mathematics that calculates probabilities and deals with very large numbers. Nearly everything in the social sciences relies on statistics, not hard numbers, to make inferences about the whole from a part (ideally a representative part).

Measuring a whole population is called a census; for large populations (like a whole nation), this is such a monumental and costly undertaking that our federal government only conducts it once a decade for the essential purpose of accurately determining the proper apportionment of congressional districts. Unless they have conducted a census, statisticians are always working with probabilities.

The more complex the issue, the more guesswork is involved. To derive any but the most basic information, social scientists must not only rely on probabilities but also make assumptions in order to interpret the data correctly. These assumptions often involve using something measurable but uninteresting as a “proxy” measurement for something interesting but unmeasurable (e.g. using DUI convictions as a proxy for the prevalence of drunk driving). If those assumptions are inaccurate, then the results based on them will be unreliable.

Tasks such as calculating the total number of jobs added in the U.S. over a 12-month period is an incredibly complex process. To do so, BLS economists must process thousands of data points and use that information to make inferences about millions of jobs. This is about as challenging as a blindfolded football announcer trying to provide live commentary on the game based on sound alone.

The BLS recognizes the limitations inherent to this task. Their initial calculation is called an “estimate.” Their standard practice is to issue a later “revision” to this estimate in February of the next year, and a “preliminary estimate” of that revision in August. This is because information that arrives later can help them to correct their earlier interpretation. The blindfolded announcer could do the same; “there must have been a turnover, because the plays are now progressing in the opposite direction.”

Most BLS revisions of job data are small, while the most recent was surprisingly large. This doesn’t mean that BLS economists suddenly forgot how to do their job. Rather, it means they finally got new information that contradicted previous information, allowing them to correct and refine the assumptions on which they make their calculations. The same is true with public opinion polls; just because some polls are wrong doesn’t mean the method is unreliable. A good pollster will do what the BLS has done and use later information to correct for earlier mistakes.

2. Statistics Don’t Lie, but They Don’t Tell the Whole Story

The Biden White House released a statement on June 28, 2023, claiming, “Bidenomics is already delivering for the American people. Our economy has added more than 13 million jobs ….” Their logic ran: job creation is a sign of a healthy economy, and jobs are being created, so the economy is healthy.

Yet the American people largely met the boasts about Bidenomics with skepticism and ambivalence. American families were still struggling with a prolonged period of high inflation and high interest rates. The number of jobs created did not tell the whole story — or even the primary story — about how the state of the economy affected most Americans.

Even if true, the simple statistic — 13 million jobs created — does not even tell the whole story about job creation. Many of those reappeared as the economy bounced back from the COVID pandemic lockdowns in 2020. Of the jobs that were created, many did not benefit American citizens; the BLS estimated that all job growth from February 2023 to February 2024 went to non-U.S.-born workers, while U.S.-born workers decreased by 741,000.

This is related to the saying, “statistics don’t lie, but people do.” One way people use true statistics to advance a false narrative is by telling only part of the story. However good the play-by-play announcer is — blindfold or no blindfold — he can’t tell about how this player powered through an injury during training, or that player overcame a difficult childhood, because those parts of the story simply aren’t visible on the field.

3. Statistics Don’t Lie, but They Aren’t Ultimate

A third caution about statistics is that they do not tell us about what is most important. Economic figures rise and fall. Opinion polls swing. Even elections are won and lost. Those who are elated by good news will be dejected by bad news. Win or lose, it will all dissipate like a frosty breath. “There is no remembrance of former things, nor will there be any remembrance of later things yet to be among those who come after” (Ecclesiastes 1:11).

The best things in life cannot be measured or statistically analyzed. Even if we try to find proxy values for family, happiness, or spiritual discipline, we can barely scratch the surface of their meaning.

The best knowledge is found in God’s Word, and the best wisdom is found in a lifelong pursuit of the fear of the Lord. What Paul says of physical knowledge applies to physical insight as well, “while bodily training is of some value, godliness is of value in every way, as it holds promise for the present life and also for the life to come” (1 Timothy 4:8).

Today may show a shocking downward revision of job numbers. Tomorrow they may be revised upward. Today your favored candidate may be leading or trailing, and tomorrow the electoral fortunes may be reversed. But “Jesus Christ is the same yesterday and today and forever” (Hebrews 13:8), and how we stand before him is far more important than what the statistics say.