You might have noticed that there was none of the traditional beat-up over crime when the crime statistics came out last week even though reported crime was up 1.2%. The reason is pretty obvious. The right is on power now, so National, the Sensible Sentencing Trust, and the Herald have no interest in trying to generate public fear of crime and with it anti-government feeling.
Nonetheless, it did provide an opportunity for some journalists to display their problems with basic statistics. Take this mess from Stuff and NZPA:
“Violent crime in New Zealand continues to rise, largely driven by increasing family violence.
Across the region there were 1,200 less crimes in total recorded. That means every day last year three less people were affected by crime than in 2007.
Auckland City recorded the biggest decrease in total crime with four per cent fewer offences last year
One in every 10 Aucklanders is likely to experience a crime of some sort most of those unlucky ones being victims of burglary….
…just 760 people in every 10,000 being hit by crime”
Of course, it should read “1,200 fewer crimes” but the innumeracy is worse than the literacy.
Not all crimes that happen are recorded. So an increase in the number of recorded violent offences does not necessarily mean the number of actual violent offences increased. All the journalists had to do was add the word “recorded” twice to their first sentence to make it correct.
Likewise, 1,200 fewer crimes being recorded does not mean 1,200 fewer occurred. There may have been more, the same amount, even fewer or anything in between. So we cannot say “that means X” about the actual impact of crime. We certainly can’t say “that means every day three less (sic) people were affected by crime”. 1,200 goes into 365 3.3, not 3, times. Many offences ‘affect’ more than one person. Many offences occur as part of one criminal act and so ‘affect’ the same person(s) at the same time.All of which means we can have no idea how many more or fewer people were affected by crime in the last year.
To top it off, even if the reduction were three a day on average that’s completely different to saying three fewer every day as the article does. On some days there could be more than three fewer offences, on other days there could be less than three fewer or even more offences.
The crime rate in Auckland may be around 1000 recorded offences per 10,000 population but that does not mean one in every ten Aucklanders is likely to experience crime. Most crimes are experienced by a few people, who are victimised multiple times in a single criminal act and, often, repeatedly during a year. On the flip side, many crimes have more than one victim. So, again, just because the number of recorded crimes or rate of crime is known does not mean have any idea of the number of victims or rate of victimisation.
Even if the one in ten figure were correct, it wouldn’t mean “one in every 10” as the article states. Some randomly selected groups of ten would have more than one victim, others would have none. Same goes for “760 people in every 10,000”.
If journalists don’t understand statistics, they shouldn’t try to play fancy and pointless games with them. To the informed it merely reveals the journalists’ ignorance. For people who don’t know any better, the journalists just sows misinformation and misunderstanding.
– the mathemagician