Lies, damned lies, and statistics

With the Duffy trial starting tomorrow, I took a #cndpoli stroll through Twitter this morning to see what folks had to say.

Predictably, there were lots and lots Duffy tweets. Everyone loves a public hanging, especially when the noose is around the neck of a disgraced and appointed for life politician with a 30 year history of nasty to vicious journalism. Though, that's not what drew the ire that caused this post. Regardless of verdict, the Duff's duff is cooked. What is also cooked, is this tweet:

Incarceration rate under Chretien (116-118 per 100K) was *higher* than Harper's highest rate (114 in 2012) http://t.co/8br0r77ccT #cdnpoli
— Neil Edmondson (@NeilJEdmondson) April 6, 2015

  The source of these figures is sound*, but the conclusion is either drawn in ignorance or with intent. 

	1999	2000	2001	2002	2003	2004	20061*	20082*	20113*	20124*
United States	682	699	700	701	714	723	738	756	743b	730a
New Zealand	149	149	145	155	168	168	186	185	199	194
England & Wales	125	124	125	141	142	141	148	153	155	154
Scotland	118	115	120	129	132	136	139	152	155	151
Australia	108	108	110	115	117	120	126	129	133a	129z
Canada	118	116	116	116	108	107	107	116	117c	114b

Yes, in 1999, the incarceration rate was 118 per 100,000 people. Yes, in 2012, the incarceration rate was 114 per 100,000 people. Yes, the first figure is definitely higher than the second figure. But declaring a conclusion based on a single data set is like declaring a plague over because the infection rate is decreasing. On one hand, the plague could be abating. On the other, most of the population could be already dead.

So let's take a look at a complementary data set, crime rates:

1999	2000	2001	2002	2003	2004	2005
7,694.33	7,606.79	7,586.73	7,512.11	7,770.25	7,599.62	7,325.04

								2006	2007	2008	2009	2010	2011	2012
								7,243.98	6,899.26	6,616.55	6,442.36	6,136.98	5,755.69	5,588.11

In 1999, there were 7,694 Criminal Code incidents per 100,000 people. In 2012, there were 5,588 Criminal Code incidents per 100,000 people. That's a drop of 2,106 per 100,000, or 2.1 per 100.

According to StatsCan and the federal government, "with only 5,588 incidents per 100,000 people in 2012, Canada's national crime rate reached its lowest level since 1972. The rate of violent crime in Canada was 1,190 incidents per 100,000 people, its lowest level since 1998. The rate of property crime, for its part, was 3,414 incidents per 100,000 people, the lowest since 1998."

Crime is down and down significantly. Got it. Let's go back to those incarceration rates.

In 1999, the incarceration rate was 118/100,000 and the crime rate was 7,694/100,000 or 0.118/100 to 7.7/100. In 2006, the crime rate dropped to 7,244/100,000 or 7.244. If we were to use 1999 incarceration rates and 2006 crime rates to predict the 2006 incarceration rate, we'd get a value of 0.1056342066 per hundred or 106 per 100,000. It's a difference of 1/100,000 from the actual incarceration rate of 107/100,000 or 0.001/100. It's damn close.

If incarceration rates remained reflective of crime rates, a crime rate of 5.6/100 in 2012 predicts a 2012 incarceration rate of 0.088/100. The actual incarceration rate for 2012 is 0.114/100. Based on 2012 crime rates, we incarcerated an extra 26 people per 100,000.

So while it is true that incarceration rates are technically lower under Harper than they were under Chretien, factoring for crime rates means incarceration is up an additional 9,189 people in 2012. Each male inmate costs an average of $111,042 per year, each female inmate $214,614. Just using male inmate costs, this cost Canadians an extra $1,020,364,938 in 2012. In contrast, it costs $31,148 per year to maintain an offender in the community.  

Mark Twain popularized the saying "there are three kinds of lies: lies, damned lies, and statistics." The nice thing about statistical lies is that they are easily refuted with more data.

The tough-on-crime agenda that began in 2006 and continues to be the Conservative fundraising battle cry is a different kind of lie. Unfortunately statistics have no place in this lie, because statistics require analysis and reason and there is no reason, no analysis to be found in ideology.  There is certainly no reason or analysis to be found in locking up an extra 9,000 people at when crime rates are at historic lows.

*Looking at the readily available (and actual not predicted) statistics for 2013, the crime rate has fallen to 5,191 per 100,000 or 5.2/100. The 2006 incarceration/crime rate stats predicts an incarceration rate of 0.07667545555 or 0.077 per 100 people. The actual incarceration rate for 2013 is 0.118. Exactly what it was in 1999 with a crime rate that's 2,503 lower per 100,000 or 2.5 per 100.