Telephone Tag

Also, be on the lookout for words that sound specific but actually aren’t. Imagine reporting to be a game of “telephone tag”. The actual facts may involve something completely different than what the final person heard.

The classic example is the FBI statistic that:

You are more likely to be killed by a friend or a loved one than by a complete stranger.

Look at that statement. Where are the weasel words, and what parts are missing?

Weasel words: there is a subject, one statistical claim, one verb, and three nouns in that statement: “you”, “more likely to”, “killed by”, and “friend,” “loved one,” and “complete stranger”.

It is given as fact that this statement applies to “you”. You are more likely... But who is “you”? Studies are not generally performed on “you”, they are performed on other people. Those other people all have something in common that allowed them to be studied, or that made them attractive subjects to the researcher. If the study was performed on subjects who shared your relevant characteristics, you can apply a different level of relevance to your own daily life, than if it was performed on subjects who do not share your relevant characteristics. For example, “you are more likely to go to the beach on Sundays than read the newspaper” isn’t a useful statistic if you live in Minnesota and the study was performed on San Diegans.

The statistical claim “more likely to” most likely means that the person making the claim doesn’t have access to the original source material. They are either repeating what they heard a long time ago, or they are simply making up the statement on the spot.

The verb “killed by” sounds pretty airtight, and it probably is. There’s only one reasonable definition of “killed by”.

The word “friend” and the phrase “loved one” might be weasel words. How is the FBI going to define “friend” and “loved one”? The victim can’t tell them: the victim is dead. You should raise a mental red flag on those items.

“Complete stranger”, like “killed by” probably has a pretty tight definition. Someone the victim has never met before. But what is a “complete stranger”, compared to a “stranger”? There might be a clue here.

What’s missing from this factoid? Well, what’s between “friends and loved ones” and “complete stranger”? Isn’t there any possibility you might be killed by someone you know, but who isn’t a friend or loved one?

If you ask that question, you’ve come to the heart of the problem. Because the answer is yes, and that category is the biggest possibility. This statement comes from an ignorant or deliberate misinterpretation of FBI data.

Follow the logic. The FBI data says that a victim of murder is more likely to have been killed by an acquaintance than by a complete stranger. Newspapers (and people with a bias towards telling you you’re in danger from your loved ones) find the word “acquaintance” to be boring, and replace it with “friends and family”. This is a pretty big exaggeration but not totally unreasonable. The phrase then often gets converted to “friends and loved ones” because family ought to be loved ones and, well, it just sounds even better to have been killed by a loved one if you write headlines.

But what about those people between “acquaintances” and “complete strangers”? Turns out there aren’t any. The FBI definition of “acquaintance” is anybody who was not a complete stranger: anybody the victim was “acquainted with”. That is, if the victim and the killer had met or otherwise been acquainted with each other, even once before the killing, they are listed in the FBI crime statistics as “acquaintances”.

With this definition, it is almost a tautology that “acquaintances” are more likely to kill you than “complete strangers”. The following all fall into the definition of “acquaintances”: a drug dealer and a rival drug dealer; a gang member and a rival gang member. Most people who believe they have reason to kill you will have had to meet you at least once to come to that conclusion.

The weasel word “more likely to” should have alerted you to the possibility that the person reporting this data has not seen the original data, and thus has not been privy to the definitions and assumptions underlying that data. In conversational English it is probably only a white lie to equate “acquaintances” with “friends and loved ones”. In the context of this factoid, however, it completely misrepresents the finding.