How Political is Statscan?
The recent controversy over the long-form census has caused me to be a bit more suspicious of Statscan lately. Two recent events in particular have left me scratching my head.
First, as part of my doctoral dissertation research, I was trying to get ahold of (time series) social assistance statistics for all 10 Canadian provinces, namely social assistance rates and caseloads, going back to the mid-1980s (not exactly ancient history!). A social policy wonk I happen to know put me in touch with a former HRSDC employee who literally had to dig up hard copies of documents in a basement to help me out. After dusting them off, I asked: “Doesn’t Statscan have all of this data?” The answer was no. In fact, they wouldn’t even be able to provide it to me for a fee!
More recently, I’ve come across a 2010 Statscan publication that provides a very detailed look at salaries and salary scales of full-time teaching staff at Canadian universities. Thanks to this data, any member of the general public can very easily see, for example, that the median annual salary of a full professor at the University of Toronto is just over $156,000.
Call me a cynic, but this makes me suspicious. If more members of the general public knew how little social assistance recipients make, this would certainly cause some awkwardness for members of senior levels of government. (For what it’s worth, and not counting child benefits, a single adult without dependents on social assistance in Ontario now receives approximately half what they would have received in the mid-1990s, in real terms.) Such knowledge in the hands of the general public would likely put some added pressure on senior levels of government to spend more on anti-poverty efforts.
Yet, one can sensibly conclude that senior levels of government would be quite happy to have members of the general public know what tenured university professors make as salaries. (In fact, Mike Harris was so eager to have members of the general public know what senior public servants make that he brought in the Public Salary Disclosure Act in 1996.)  Such knowledge in the public domain certainly has the potential to take some pressure off of senior levels of government to increase funding to universities.
All of this raises a fundamental question for me: just how political is Statscan?
Well Nick. maybe you were not on this blog back during the last election, but that whole issue of arms length, neutrality, political bias came up, and one of the bloggers on here who worked at statcan was almost fired over sugesting tha potentially Statcan may not be as neutral as everybody thinks they are. Of course, there were a lot of remarks made on the blog that potentially this employee of statcan was just being cynical- however, it is precisely such evidence that you put forth that this employee was questioning, as to how neutral Statcan really was. Now with the government diving into this with both arms all over the surveys and methods, it sure does not bode well to think you will get national social assistance Stats anytime soon.
I agree with you 100% if we had those Statistics we surely could blow away the notions that somehow our social assistance is over generous or other such tea party like talk. We could also, like Andrew J. had suggested make better linkages between unemployment rates- EI benefits and social assistance benefits. You would think in any kind of nation that is trying to address such issues, data on all these aspects of unemployment would be quite useful. I once worker for a few years at a local Economic Development shop in a mid sized city in Northern Ontario, and social assistance Statistics were quite readily available, so it is not a matter of the stats not being there. It is the political will to actually have the numbers to take on the issue in a way that is much more encompassing that what we currently have going here in our fine nation.
Appalling, because when you think about it, this is undoubtedly the precise objective of what Mr.Harper has driving his quest to destroy the census. No numbers, no problem or at least a whole lot more difficult in raising the issue in a meaningful fashion.
As this former Statcan employee told me once a while back, the social assistance statistics, like the injury statistics, suffer from some deep seated political idea, that if you bury the stats, the backlash goes away. Since the workers comp stats have been handed over to the foxes in the AWCBs you get be sure that it is a whole lot more difficult to get those decent reliable stats than when statscan administered the file.
Statcan is truly one of the most politically charged departments within government- and you must realized that having the ability to define, construct and measure is the ultimate power over any policy. Harper is not a stupid man- in fact he and his cronies are some of the most devious we have had working in the PM’s office in some time. We need a tough mongoose to take this snake out.
pt
Nick, National Council on Welfare has an annual publication with the info you need. They’re who you need to get in touch with.
On salary disclosure – no one reads the Ontario salary disclosure documentation as closely as the faculty members themselves. It’s a great tool for negotiating a high salary (I am being paid $20,000 less than professor X, who I am identical with in every respect, see, look at our publication list).
Indeed executive salary disclosure legislation in the US has been blamed for spiraling executive salaries there, as every executive negotiates a salary “10% higher than the average of all executive salaries in my industry”.
Actually Statscan has published studies on trends in receipt of social assistance – in Perspectives a few years back and I am pretty sure you can in principle get data from the LAD, based on tax data.
Here is the link Nick:
http://www.statcan.gc.ca/daily-quotidien/041117/dq041117a-eng.htm
With no long form census, we won’t even have as much information as Globe readers on July 1, 1867.
See: http://www.themarknews.com/articles/2039-a-country-founded-on-statisticsch
A few years ago, when I was still a student, I tried to get information on Canada’s fiscal balance going back to Confederation, seeing as deficits were more the rule than the exception prior to McKenzie-King’s austerity régime of the late 1920s. Nothing like that is publicly available from the Government of Canada prior to 1930 (a quick google search suggests this is still the case, but I am not sure about that). The kind response from the StatsCan staffer who replied to my request was that they could provide me with the information for a substantial fee, but that otherwise I should see what I could figure out with the Data Liberation Initiative (DLI). Eventually, I was able to get the data, beautifully laid out in detailed tabular format, through personal contacts at the Department of Finance.
I don’t know what to think about why it is so hard to get information on such a simple and straightforward subject. Who makes the decision not to keep that kind of historical data available? Was a decision made that Canadians simply don’t have any interest in our fiscal finances before the Great Depression? Maybe there just is insufficient demand to keep such information available? If that is the case, as with the paucity of translation services in the Government of Canada, one has to wonder if the problem has really to do public-sector information management practices which, since the early 1990s, have opted for quasi-market provision of information. Ask for it or lose it. It seems to relate, I intuit, to a (mis)translation to document and data management of approaches used for bandwidth management in computer networks. That is, Government seems, particularly since Paul Martin’s time at Finance, to have moved to something like a Max-min fairness model of data management, but for public information in general rather than “data packets”. This was originally a strategy for reducing public investment in document management (as in “who needs librarians, the network administrators can manage access to all the information the public and civil servants require”). There is some value to this approach, but usage alone is not, from a public policy perspective, a fair way to value or “valorise” public information. Moreover, restrictions in access to information are self-reinforcing — the more frictions you must contend with in trying to access the information you want, the less you will request it, with the result that all but the most short-term and frequent information asks are gradually marginalized, or their supply is so limited as to create bottlenecks and underdevelopment. You get unbalanced growth in information development (in a kind of Hirschman-type situation). Civil servants and researchers “build from the small”, try to exploit hidden rationalities in access to data, but these are really just attempts to swim against the current. To extend the analogy further, what is really needed is an expansion of autonomous investment to allow for more balanced growth in policy and program relevent information, and to move to a higher equilibrium in information exchange. Interestingly, the only Department to keep its physical document registry after program review was the Department of Finance.
The tax data social assistance is very much filled with bias, as there are many who do not fill out taxes. the small area data based on tax data is very much an incomplete picture of the population, especially at the tails of the income distribution. the local municipal data are much more accurate as they follow the actual payouts. The trend monitoring that you mention Andrew, is based on the tax data which I do believe given the gravity of what we are monitoring is a poor proxy for income at the margin and is not an adequate substitute.
pt
oh and I just wanted to add, there is a political bias on many different issues when it comes to measurement. We know exactly how many loadings of railway cars we have by weekly measures, yet poverty measures are at best annual and three years behind.
I will say the health statistics have come a long way, but at the expense of many other statistical programs.
The current assault on the politics of Statistics is currently under a multifaceted attack. First it was the purge under Munir Sheik of the analytical capacity of statcan, (although given his resignation I would think much of that purge originated from outside the stat barn). Second, we now have the attack on the census, and finally once this frontal assault on the census is complete which is connected throughout the fabric of many other key surveys, we will have the third assault on statcan, its legitimacy through unreliable data will bring and end to many of its surveys. (at in terms of usefulness.)
I’ve enjoyed reading these comments.
Just to be clear, I do have *some* data, but it’s quite incomplete. With respect to (time series) benefit rates, the National Council of Welfare’s new interactive web site (launched last month) indeed provides that information for all 10 provinces and for all family types going back to 1986 inclusive, but not before. I cannot get said data for all Canadian provinces and family types going back to, say, the early 1970s.
As for caseloads, this is where it gets even tougher. These (time series) numbers for all Canadian provinces and family types are available, but I have to use three different sources. Plus, two of the years in the middle are missing, so I’ll have to impute those numbers. Moreover, the caseload data available for single unattached individuals does not provide a breakdown as to which ones are “employables” and which ones have “disabilities.”
As for the tax data discussed above, I think Paul T. does an excellent job in one of his comments of pointing out the limitations of relying on tax data for this population.
Yes Nick it sure is a shame that we do not have a decent set of statistics for such a life threatening experience. (our welfare system has become just that- life threatening)
I do know that if one is to seriously set about to make policy towards reducing poverty, having a solid statistical foundation for monitoring and feedback would actually be quite instrumental!?
We all know how sensitive Statcan can be about LICOS.
Yes indeed how does a nation have an effective series of emploment policies when it is missing such data.
Another huge area of political concern of statistics is the environment. With so much at stake, and the apparent neutrality of statcan, why are we collecting so little statistics on the environment. And what we do collect are basically quite loaded with a whole lot of politcial filters.
If Statcan is to continue on setting the example of a national statistical agency then we need to have a whole reworking of our environmental statistics and how we collect and report, analyze and disseminate such key data. Air quality, to water quality, robust co2 emissions, transportation statistics, production and land use, agriculture, forestry and a bunch more. We need these statistics to become a whole lot higher in the priority list. Not aure if many know this buts as late as 2002 Statcan budget for environment stats was less than 2.0 million a year and most of that was recycling environment Canada data. The situation is a bit better, but much more needs to be done in thius area. What is holding it up- politics!
Two key areas are beginning to take shape, povertty and environment. I do know I had a long debate once with a former professor Colin L., a friend of Leo, about how successfully the attempt at commodifying environment would be-I still believe they will strive further before massive change is more lucidly considered, till then denial and commidification will be the response, all based on income distribution and exposure to a less friendly climate, at least at the front and potentially into the beginning middle way through the changes. With out good quality statistics on either, we will see a minimization in the officialdom of this, but for those who already are feeling the adjustment pain- Statistics will not matter much.