A Half-Truth Is the Best Lie

There is an old saying that “A half-truth is the best lie”. Today, the internet makes it much easier to find useful information and statistics than ever before. Unfortunately, it is also easier to find half-truths and outright misinformation.

Long before the introduction of the internet, in 1954 journalist Darrell Huff published a book entitled “How to Lie With Statistics”. Though statistics play a significant role in shaping our perceptions, often these numbers can be misleading, even if presented by reputable sources. In his book, Darrell Huff outlined some of the top techniques that are used to mislead the public and make statistics support a particular conclusion. And the numbers don’t have to be fabricated, as factual data can be used to twist the truth either intentionally or mistakenly.

For example, every day we use the word “average”, but average can refer to a number of different values, primarily the mean or the median. Using the mean as an average can be misleading if the data set contains major outliers, which occurs often when reporting average salary figures. Let’s say you have 10 employees, two of which earn $20,000, six earn $70,000 and two earn $300,000.  Calculating the mean would deliver an average salary of $106,000 per year.  But using the median provides a more realistic average of $70,000. 

Here’s a real-life example: Joe Biden declared that one did not need a college degree to earn the average salary of $110,000 in the semiconductor industry. The “half-truth” here is that the figure includes all jobs within that industry, including the highly paid positions that do require a college degree. As pointed out by PolitiFact in 2024, people without college degrees in the semiconductor industry actually earn closer to $40,000, those with an associate degree earn up to $70,000, and those with graduate degrees make up to $160,000. The highest salaries skew the mean average and don’t tell the whole story.

Conflating percentage point changes is another inconspicuous way to use data to sway public opinion on any given topic. Here are some real-life examples of how that can easily happen:

A 2023 headline read that a study found that eating too much red meat was linked to a 50% increase in the risk of Type 2 diabetes. That sounds quite alarming to most of us, but if the risk of Type 2 diabetes in the general population is 3%, then a 50% increase would bump up the risk by 1.5% to 4.5%, a much less daunting statistic. 

Another example was a report in 2018 from the Florida GOP that the average murder rate in Tallahassee increased 52% under Mayor Andrew Gillum. That sounds disastrous, but according to PolitiFact in 2002-2009, before Gillum became mayor, the murder rate in Tallahassee was 4.6 murders per 100,000 citizens. During his tenure from 2010 to 2017,  the murder rate increased to 7 murders per 100,000 citizens. This translates to a percentage change of 52%, but the actual murder rate was still under 0.1 per 1,000 citizens.

Another method is that of using an unrelated piece of information to support a claim. As Darrel Huff wrote, “If you can’t prove what you want to prove, demonstrate something else and pretend that they are the same things.”

In 2009, an ad for Kellogg’s cereal claimed that eating their Frosted Mini-Wheats could boost children’s attentiveness by nearly 20%. But the study did not compare Kellogg’s cereal to any other cereal or breakfast food, and the control group in the study actually received no breakfast at all. After being sued, Kellogg agreed to drop the claim, as the only thing their study had proved was that children who eat breakfast are nearly 20% more attentive than those who do not.

Another way to lie with statistics is to confuse correlation with causation, implying a cause-and-effect relationship between two elements only because they appear to be associated. For example, people who own ashtrays are more likely to get lung cancer, so ashtray ownership and lung cancer rates are correlated, but one does not cause the other. Smoking is the factor that influences both ashtray ownership and the increased rate of lung cancer.

Leaving out crucial context is another way to lie with statistics. A perfect example is a 2020 study by George Mason University, in which researchers using data on 1.8 million hospital births in Florida between 1992 and 2015, claimed that Black infants were three times more likely to die during and after birth when cared for by White doctors rather than Black doctors.

In August of that year CNN reported: “The mortality rate of Black newborns in the hospital shrunk by between 39% and 58% when Black physicians took charge of the birth . . . which laid bare how shocking racial disparities in human health can affect even the first hours of a person’s life.” 

That claim was even cited by Supreme Court Justice Ketanji Brown Jackson in her dissent in Students for Fair Admissions v. Harvard, in which the majority of the Court ruled that race-based affirmative action practices in universities were unconstitutional.  An amicus brief filed by the Association of American Medical Colleges in that case read that “For high-risk Black newborns, having a Black physician is tantamount to a miracle drug: it more than doubles the likelihood that the baby will live, yet due to the enduring and significant underrepresentation of minorities in the health professions, many minority patients will not receive care from a racially diverse team or from providers who were trained in a diverse environment.”

Only that wasn’t true. Harvard researchers at the Institute of Family Studies found that the study “did not control for the impact of very low birth weights (i.e., under 1,500g) on newborn mortality. . . . Although these types of births are rare, they occur more frequently in the Black population, and they account for a very high fraction of mortality.” 

“It turns out that a disproportionately large number of Black newborns with very low birth weights are attended by White physicians. . . once we control for the impact of very low birth weights on mortality, the estimate of the racial concordance effect is substantially weakened and becomes statistically insignificant in models that account for other factors that determine newborn mortality. In other words, the newborns attended by White and Black physicians are not random samples. Black newborns with a very low birth weight are disproportionately more likely to be attended by White doctors than by Black doctors. Those newborns are also more likely to have a low chance of survival.”

It was also reported that a note written in the margin of the George Mason study by lead author Brad N. Greenwood showed he knowingly omitted his study's findings that White children were less likely to die under the care of White doctors than Black ones. “I’d rather not focus on this. If we’re telling the story from the perspective of saving black infants this undermines the narrative,” Greenwood wrote.

Understanding how statistics can be manipulated helps us discern truth from fiction.  Critical thinking and a skeptical approach to statistics will help us recognize deceptive tactics, better evaluate information, and make more informed decisions.