11. How To Not Lie With Statistics

                                                        How To Not Lie With Statistics

Back in the 70’s there was a popular Statistics textbook called “How to Lie with Statistics.”  It took the angle of exploring how perfectly true numbers can be manipulated to tell a perfectly false story, so it was an engaging and light-hearted text.

We feel like it is time to re-explore this concept, because these days there are a lot of statistics in the news - and a lot of them convey useless or misleading information.  We notice that this very often stems from a lack of considering phenomena on a “per capita” or “rate” basis. 

For example, we hear a reporter state that in 2019 China used the energy equivalent of 24 million barrels of oil while the U.S. used “only” 17 billion barrels, implying that it’s on China to clean up their act first. So what’s wrong with this statement? 

China’s population is about 1.4 billion, while the US population is around 330 million. When you divide energy used by population, you realize that the average person in China used 17 barrels equivalent, while the average person in the U.S. used 51 barrels equivalent. In other words, we actually use energy at nearly 3 times the rate of the Chinese. (In fact only a small number of tiny countries use more energy per person than we do.) 

This idea of “per capita” or rate basis measurement is usually much more revealing than giving the total number of any given thing.  Another example is measuring the rate of Coronavirus infections. 

At this moment the U.S. has the highest number of Coronavirus cases in the world. Some argue that this is because we are such a big country. That’s a good point, but it’s insufficient. We need to measure it on a per capita basis and compare it to some other countries.  How do we stack up there? 

At this writing, the U.S. has had 5.4 million cases. With a population of 330 million, this is 1.6% of the population, or about 1 out 62 people.  This 1.6% can also be referred to as the “rate” of infections. Only 7 nations have higher rates than this (and the total population of those countries is less than one tenth that of the U.S.).

Some other Covid-19 related quantities are also better stated on a “percent” or “per capita” basis.  The total number of active cases in the U.S. is currently almost 2.3 million, which is about 0.7% (seven tenths of a percent) of the population.  If this sounds small to you, think of it this way: One out of 140 people in this country are currently infected.  In this regard we have the highest active case rate in the world. 

Incidentally, in comparison to the rest of the country Maine is doing extremely well with  less than 0.03%, or one out of 3400 people with active cases, which is over 20 times better than the U.S. as a whole.

There have been claims that the U.S. has more Covid-19 testing than anywhere in the world. Again, we need to judge this statement on a rate basis.  It is true that the U.S. is second only to China in total tests, but when measured on a per population (per capita) basis, we come in nineteenth.  (By the way, the increased number of cases is not due to increased testing. It is because the disease has actually spread.)

Then there is the Covid death rate.  This can be measured in two ways. One, the total deaths compared to the total population; and two, the total deaths compared to resolved cases, sometimes called the “mortality rate.”

Here in the U.S. the total deaths have reached 170,000, or 0.05% of the population. In other words 1 out of every 2000 people in the U.S. has died of Covid-19. That’s compared to 1 out of 10,000 for the world as a whole.  We as a country are number 10 in that regard, but only marginally better than the worst-hit countries. 

As for the mortality rate: To get this, you take the total number of deaths and divide it by the total number of cases that have been resolved—either got better or died. For the U.S. that number stands at around 5.6%.  The Johns Hopkins Corona Virus Resource Center reports a lower number of 3.2%, as they divide by the total number of cases, not just the resolved cases. Both of these numbers are quite close to the world averages of 5.1% and 3.6%.

It should be noted that there is always uncertainty in the data from which an analysis like this comes.  The errors from those uncertainties tend to be systematically low. Here it is more likely that cases will be missed and undercounted, than that cases will be incorrectly diagnosed and overcounted, meaning that overall we could be doing worse than the data suggest.

Analyzing numerical phenomena on a rate basis isn’t always bad news, of course - it is simply far more informative and more difficult to deceive with!  And in an election season, it is more important than usual, so demand it from your news sources.

Sources:

bp-Statistical Review of World Energy https://www.bp.com/en/global/corporate/energy-economics/statistical-review-of-world-energy.html

Worldometer https://www.worldometers.info/coronavirus/

Johns Hopkins Corona Virus Resource Center https://coronavirus.jhu.edu/

Paul Stancioff, PhD., is a professor of Physics at the University of Maine Farmington who studies energy economics on the side.  He can be reached at pauls@maine.edu.  Cynthia Stancioff is an amateur naturalist/writer.