Countering Conservative claims with a real calculation on real data
Update: Thomas Kwon asks on Facebook why I did not ask the question in simpler terms. The answer is that the Government usually interprets Order Paper questions to minimize the information they have to give you. So I used mathematics to be as precise as possible. And I also happen to know that they have precisely the data I am asking for, so the calculation will be possible.
I just tabled a written question for the House of Commons Order Paper, Q-161, which you can read here, and is reproduced at the bottom of this post. I think it might be the first Order Paper question to have mathematical formulae and to ask for a graph! I hope it sets a precedent. In any case, for somebody with a quantitative background, I also think it is a very natural line of inquiry.
Why ask this question?
People who work in corrections have long avoided double bunking of inmates whenever possible. The reason for this is that inmates may be more prone to disputes and violence and are more difficult to manage when they share cells. Limited rates of double bunking can be managed by the staff who know the inmates individually, but as a general rule it is to be avoided. In fact, double bunking doesn’t even meet the standard for inmates set by the United Nations.
Recently Conservative MPs have changed their messaging and have said two things:
1) They now embrace double bunking, which you can see in this exchange in question period:
The Conservative MP from Westlock-Saint Paul recently asked a question about double bunking to the Minister of Public Safety:
“Mr. Speaker, big union bosses at the Union of Canadian Correctional Officers are telling anyone who will listen that they want to ensure that convicted rapists and murderers have their own private accommodations while in prison. They say that it is troubling that these prisoners are forced to share cells. My constituents strongly disagree. Plenty of Canadians are required to share accommodations for a variety of reasons. Could the Minister of Public Safety please tell the House whether he agrees that these dangerous criminals should be treated better than college students?”
Rather than speak out against double bunking, the Minister responded in agreement with his Conservative colleague, saying:
“Let me also state that double-bunking is a completely normal practice used in many western countries, and studies, including a Canadian one, show that it has no impact upon the rate of violence in prison… We strongly believe that double-bunking is totally appropriate for convicted criminals.”
2) The government went on to claim in the House that “Corrections Canada, last year, completed a study that showed that the rates of double-bunking were not connected to the rates of violence.” This is strange because the sixth recommendation in the Annual Report of the Office of the Correctional Investigator 2012-2013 was the recommendation “that CSC’s inmate accommodation policy reinstate the principle that single occupancy is the most desirable and correctionally appropriate method of housing offenders.” Beyond that, the report explained that “The Prairie Region exemplifies these [double bunking] pressures. Over the past five years, the number of incidents of assault (including assaults on other inmates, visitors, and staff; inmate fights; and sexual assaults) increased by 60% (from 366 in FY 2008-09 to 586 in FY 2011-12). The number of use of force incidents increased by 48% (from 265 to 393) over the same period. In the last three years, there have been five inmate murders in the Prairie Region, accounting for more than half of all inmate homicides in federal penitentiaries.”
The problem is that the study quoted by the Minister has not been made public, and so it is impossible to verify.
What have you asked the Government, and why?
I have asked to government to do a precisely specified calculation of the correlation between excessive double bunking and the rate of violent incidents in maximum security institutions. As I mentioned above, corrections management have told me that they know how to deal with limited amounts of double bunking and so I focused on cases where the number of inmates is more than 10% over rated capacity, i.e. roughly speaking, more than 10% double bunking. I have assumed there is always a background rate of violent incidents, and so I’ve asked to correlate double bunking with the excess of violent incidents above this background rate.
I have “done my homework” and have a very good idea of what the result of this calculation will look like.
I hope to inject the debate over double bunking with some solid data. It is my contribution as a scientist-MP.
Here is the text of my Order Paper Question. The Government is required to answer it within 45 days so we should have an answer in February 2014.
Q-1612 With regard to violent incidents related to overcrowding in federal prisons:
(a) for each of the ten years from 2003-2004 to 2012-2013, and for each of the nine maximum security Correctional Services Canada (CSC) institutions, namely, Atlantic, Donnacona, Port Cartier, Quebec Regional Reception Centre and Special Handling Unit, Kingston Penitentiary, Millhaven, Edmonton, Saskatchewan Penitentiary maximum security unit, and Kent, what were the numbers of inmates;
(b) for each of the ten years from 2003-2004 to 2012-2013, and for each of the nine maximum security CSC institutions, namely, Atlantic, Donnacona, Port Cartier, Quebec Regional Reception Centre and Special Handling Unit, Kingston Penitentiary, Millhaven, Edmonton, Saskatchewan Penitentiary maximum security unit, and Kent, what were the rated capacities of each institution;
(c) if each of the 90 data points in part (a) is denoted by nij where i=1,10 runs over the ten years and j=1,9 runs over the nine institutions in the order given, and if each of the 90 data points in part (b) is denoted by cij, where i=1,10 runs over the ten years and j=1,9 runs over the nine institutions in the order given, then what are the values of the fractional excess of inmates over the rated capacity of each of the nine institutions, for each of the ten years, namely, fnij = (nij – cij)/cij;
(d) for each of the ten years from 2003-2004 to 2012-2013, and for each of the nine maximum security CSC institutions, namely, Atlantic, Donnacona, Port Cartier, Quebec Regional Reception Centre and Special Handling Unit, Kingston Penitentiary, Millhaven, Edmonton, Saskatchewan Penitentiary maximum security unit, and Kent, what were the numbers of violent incidents;
(e) if the 90 data points in part (d) are denoted vij, where i=1,10 runs over the ten years and j=1,9 runs over the nine institutions in the order given, what are the average numbers of violent incidents for each institution, averaged over the ten years, namely, Vavgj =(Σi=1,10 vij)/10;
(f) what are the values of the fractional excesses of violent incidents for each of the nine institutions, over and above each institution’s respective ten year average, for each of the ten years, namely, fvij = (vij – Vavgj)/Vavgj;
(g) what is the correlation between the fractional excesses of violent incidents and the fractional excesses of inmates over the rated capacity, for all combinations of years and institutions, for which the inmate population was more than 10% over the rated capacity, namely, the sample correlation coefficient between the set of all fnij such that fnij > 0.1, and the corresponding members of the set of all fvij such that fnij > 0.1; and
(h) what is the graph of all the pairs (fnij, fvij) which satisfy fnij > 0.1, plotted with the linear regression line?
To post comments, please go to my MP Facebook page: https://www.facebook.com/TedHsuMP/posts/10151903455734652