At Level 1, this is obviously not true. For murders in particular, I would expect it to be especially not true. Murders are easy to count, so I would expect the data to be very accurate. While anecdotes about murders tend to be very newsworthy and so we read about them and might feel like they're more frequent when they are not.
But at Level 2, I do think there's a kernel of truth to this (Jeff's statement), because sometimes it is hard to measure the same thing that people actually mean when they are talking about something like crime.
For example, in this article: There are "two really visible crises" in the downtown area: homelessness and open-air drug use. "And honestly, people conflate that with crime, with street safety," she said."
So basically the author is saying people feel unsafe about homelessness and drug use, but those aren't really crimes (people "conflate" them with crimes), and so those don't count as higher crime. So people are mistaken about the crime rate. Fair enough, but that's a technicality of what we are actually measuring, VS what people are feeling about it when you ask them about "crime" or safety or something.
Just wanted to make sure I understand this right - Since Jeff B made that one statement, anecdotes are a better measure? And the second part of it talks about measurement, not data.
People are irrational and we give in to irrational fears. If I heard about one incidence of someone who is homeless attacking a bystander, I would be afraid of all homeless people. I do not have enough data to analyze (nor do I care to analyze) how many attacks have been by homeless vs others. My gut reaction drives the anecdotes and that is not always the reality.
I will give one point to this statement though - you cannot fight an emotional reaction with logic.
In most places, your car will almost never get broken into. In a few cities, this will happen basically all the time.
The averages are removing information about how crime is distributed and focusing on violent crime when people are more worried currently about property crime. This uneven distribution and the rise in property crime does a lot explain the difference between these national averages and people's personal experiences of crime.
While the plural of anecdote isn't data, it's not some mystery why people look at local temperatures to decide how to dress themselves, instead of the national average, nor are they statistically illiterate for doing so. Nobody with any sense is going outside in a blizzard without a coat just because the national average temperature is warm.
The same is true with crime. It is not evenly distributed and that variance defines people's lived experience with it.
Yes, but you are still looking at data. I never said there are no local variances. However, extreme dependence on your lived experience should not result in policy decisions.
I am taking a more extreme example just to illustrate - If I could not find a job, does not mean there are no jobs for anyone. If I face hardship, does not mean everyone faces hardship.
I am not saying ignore local variances.. however, it is still not anecdote driven. The original article was about crimes dropping in US measured across different cites and not that there is no crime.
My point is that people hearing about crime will still have an emotional reaction and might not reflect the reality. Anecdotes get tainted and exaggerated by emotions. JeffB's statement mean little in this case.
> The original article was about crimes dropping in US measured across different cites and not that there is no crime.
No, it was about violent crimes dropping and comparing that to a survey about people worried about all crime. Given that property crimes are infamously up in some places with mass looting events, those people are not being irrational or ignoring data to believe what they do.
The real data point we should worry about from a policy perspective is that there are a relatively small number of people who do a lot more crime than everyone else.
> Unless the survey was biased towards people living in areas of high crime, I am not sure how to draw the conclusion you drew.
You don't need any such bias to understand there's a difference between all crime & violent crime and that comparing the two measures of entirely different things is not valid.
> The last sentence seems to be a dog whistle for something else.
No, Jeff Bezos has it right. There is some nuance to his data analysis approach and I recommend listening to his interview on the Lex Fridman podcast to understand what he was getting at.
