(1) Mask Deaths
Claim. Masks cause death.
Fact. 70% of people who died wore a mask.
Analysis. “People who wear masks” is not a random sample: it is not “people in general.” On the contrary, inside the “people who wear a mask” sample is a subset of people who wear masks because they are (a) close to death, and/or (b) stupid. Are those near-death people the only ones who died (was the mask not a factor)? The stat doesn’t tell us that. It cannot say, because there are too many variables that are not controlled. The statistic would be helpful only if it showed that, all things being equal, masks will kill you. For that, the statistic would have to tell us that in the 70% were regular people—kids, healthy, etc.—not just people who were dying anyways.
Conclusion. The 70% could have died from being already near death.
(2) Private School
Claim. Private school causes student success.
Fact. A higher percentage of kids who go to private school are successful (compared to kids at public school).
Analysis. A private, selective school is a program that accepts only smart, highly prepared students. A high percentage of the students who complete the program go on to be successful. So the program advertises the students’ success as proof that the program works. But does the program work? Or is the high success rate simply a result of the program being loaded with highly prepared students who are going to succeed no matter where they go to school?
Conclusion. The kids who get into private schools in the first place are already well-prepared and would succeed regardless—private school or not.
(3) Blackness and Crime
Claim. Being black makes you prone to crime.
Fact. Blacks are about 13% of USA’s population, but commit more than half the nation’s violent crimes. (More specifically, black men are about 6% of the population and account for the majority of reported violent crimes.)
Analysis. A tiny fraction of people—black or otherwise—commit violent crimes. True enough most reported violent crime happens by members of that 6%—but a tiny fraction of the 6%. For example, there are something like a million violent crimes in the USA every year. Imagine if:
(A) only black men commit those crimes;
(B) every crime is commited by a different black man (to implicate the highest possible number of black men, as opposed to the reality which is that one criminal is usually responsible for several crimes).
Even in that crazy scenario, that means there are about one million black criminals.
The USA has something like 330 million people. Blacks are about 13%—about 43 million people. Black men are about 6%—about 20 million people. In the crazy and unrealistic scenario that blacks commit 100% of the nation’s violent crime, and the most possible blacks are involved—recall: that is about one million black criminals. Still, that is one million blacks out of 43 millions, or 2.3%. That means, even in the crazy scenario—97.7% of black people commit no violent crimes. Or, focusing on black men: one million out of 20 million would be 5%. Meaning 95% of blacks commit no violent crimes.
Conclusion. Saying “blacks are 13% (or %6) of the population but commit half the violent crime”: this leaves out that the vast majority of people—black or not—commit no crime. And the statement—while factually true—wrongly implicates the vast majority of blacks who commit no crimes.
(But remember: the statement “blacks are only X% of population but commit X% of crime” is said not to claim that all or most blacks are criminals; instead, the statement is almost exclusively used in response to hysterical claims that black overrepresentation in arrests and imprisonment proves “systemic racism.”)
—Moral of the story—
The purpose of statistics is to explain how something happened: How did the people die? How did the students succeed? Why did the crime happen?
The goal of statistics is to predict how to cause, avoid, or manage future outcomes: how to prevent death; how to increase student success; how to stop crime.
Accurate statistics is all about (a) the “sample” of the population, and (b) a full analysis. A random sample isolates a variable (e.g. cause of death, or cause of success, or cause of crime). But it can be highly misleading to use an “unrepresentative” sample. And without a full analysis, statistics are useless at best—but usually counterproductive.