free statscan geo files?

[Tracey P. Lauriault]

Joe;

StatCan suppresses data if the sample size is too small.  This was the issues with the move toward dissemination areas versus enumeration areas for researchers in 2006.  Many of us would have preferred in many cases to have had the slightly larger geography and less suppression.  Yes it is possible to do some spatial auto-correlation but that does not lead to identifying individuals, especially since StatCan is incredibly rigourous with its de-identifying the data before aggregation.

During the census summer of 2010, lots of misinformation was being circulated about privacy, and as someone who is concerned about this, I have no concerns on that front with StatCan, this is where they excel to the point of overzealousness.  

If this is a hunch or a suspicion of yours that is one thing, people repeat this all the time but I have not yet seen anyone demonstrate this in real terms, some evidence and an algorithm to prove the statement would be useful. I just do not think the science is with you on this one Joe.  I and many people I know work with these data and we have yet to see this as a problem. This is something many people say, and to date I have not seen any examples to cause concern.

I am more concerned about some of the private sector data that I have seen that is way better at identifying individuals than StatCan will ever be, the only barrier to access those data are cost.  Google street view is much more nefarious on that front as are most social media sites in terms of what people release about themselves.  My Gmail account shares more with Google about me than what I put on the census form.  The sub-meter remote sensing data available in near real time is also an issue, and again, the barrier to access is cost.  Also, think of all the data your cell carrier has about you, near real time location being one of them, and that does not seem to cause people concern.  Also, anytime there is a credit check about you, someone has those data, think of credit card companies or freequent flyier points accumulated at the point of purchase.  Those are worth worrying about.

If someone wants to know something about you, StatCan's census will not be their dataset of choice, there are way better sources out there that are much more precise if you can afford the price.

Sincerely

Tracey

[Tracey P. Lauriault]

I also forgot!  The Geography files will probably not be free, and those are the golden eggs for analysis, and as we all know we all want that postal file and Digital Copyright Canada has been working on that for some time now!

http://www.digital-copyright.ca/blog/2

On Fri, Nov 25, 2011 at 10:06 AM, Tracey P. Lauriault <[hidden email]> wrote:

Joe;

StatCan suppresses data if the sample size is too small.  This was the issues with the move toward dissemination areas versus enumeration areas for researchers in 2006.  Many of us would have preferred in many cases to have had the slightly larger geography and less suppression.  Yes it is possible to do some spatial auto-correlation but that does not lead to identifying individuals, especially since StatCan is incredibly rigourous with its de-identifying the data before aggregation.

During the census summer of 2010, lots of misinformation was being circulated about privacy, and as someone who is concerned about this, I have no concerns on that front with StatCan, this is where they excel to the point of overzealousness.  

If this is a hunch or a suspicion of yours that is one thing, people repeat this all the time but I have not yet seen anyone demonstrate this in real terms, some evidence and an algorithm to prove the statement would be useful. I just do not think the science is with you on this one Joe.  I and many people I know work with these data and we have yet to see this as a problem. This is something many people say, and to date I have not seen any examples to cause concern.

I am more concerned about some of the private sector data that I have seen that is way better at identifying individuals than StatCan will ever be, the only barrier to access those data are cost.  Google street view is much more nefarious on that front as are most social media sites in terms of what people release about themselves.  My Gmail account shares more with Google about me than what I put on the census form.  The sub-meter remote sensing data available in near real time is also an issue, and again, the barrier to access is cost.  Also, think of all the data your cell carrier has about you, near real time location being one of them, and that does not seem to cause people concern.  Also, anytime there is a credit check about you, someone has those data, think of credit card companies or freequent flyier points accumulated at the point of purchase.  Those are worth worrying about.

If someone wants to know something about you, StatCan's census will not be their dataset of choice, there are way better sources out there that are much more precise if you can afford the price.

Sincerely

Tracey

[Danny van Gelder]

Apparently census socioeconomic and geographic data will be free!

http://embassymag.ca/dailyupdate/view/statistics_canada_to_make_all_online_data_free_11-24-2011

On 25 November 2011 10:15, Tracey P. Lauriault <[hidden email]> wrote:

I also forgot!  The Geography files will probably not be free, and those are the golden eggs for analysis, and as we all know we all want that postal file and Digital Copyright Canada has been working on that for some time now!

http://www.digital-copyright.ca/blog/2

On Fri, Nov 25, 2011 at 10:06 AM, Tracey P. Lauriault <[hidden email]> wrote:

Joe;

StatCan suppresses data if the sample size is too small.  This was the issues with the move toward dissemination areas versus enumeration areas for researchers in 2006.  Many of us would have preferred in many cases to have had the slightly larger geography and less suppression.  Yes it is possible to do some spatial auto-correlation but that does not lead to identifying individuals, especially since StatCan is incredibly rigourous with its de-identifying the data before aggregation.

During the census summer of 2010, lots of misinformation was being circulated about privacy, and as someone who is concerned about this, I have no concerns on that front with StatCan, this is where they excel to the point of overzealousness.  

If this is a hunch or a suspicion of yours that is one thing, people repeat this all the time but I have not yet seen anyone demonstrate this in real terms, some evidence and an algorithm to prove the statement would be useful. I just do not think the science is with you on this one Joe.  I and many people I know work with these data and we have yet to see this as a problem. This is something many people say, and to date I have not seen any examples to cause concern.

I am more concerned about some of the private sector data that I have seen that is way better at identifying individuals than StatCan will ever be, the only barrier to access those data are cost.  Google street view is much more nefarious on that front as are most social media sites in terms of what people release about themselves.  My Gmail account shares more with Google about me than what I put on the census form.  The sub-meter remote sensing data available in near real time is also an issue, and again, the barrier to access is cost.  Also, think of all the data your cell carrier has about you, near real time location being one of them, and that does not seem to cause people concern.  Also, anytime there is a credit check about you, someone has those data, think of credit card companies or freequent flyier points accumulated at the point of purchase.  Those are worth worrying about.

If someone wants to know something about you, StatCan's census will not be their dataset of choice, there are way better sources out there that are much more precise if you can afford the price.

Sincerely

Tracey

_______________________________________________
CivicAccess-discuss mailing list
[hidden email]
http://lists.pwd.ca/mailman/listinfo/civicaccess-discuss