Categories
Accountability Education Governance Transparency

StatsCan’t be bothered with data requests

Promise today’s Globe write-up wasn’t co-ordinated with Stephen Gordon, who wrote a few days ago (in reference to the SEPH, but also more broadly)

Statistics Canada’s Attention Deficit Disorder
Stephen Gordon, Worthwhile Canadian Initiative January 23, 2014

Statistics Canada must be the only statistical agency in the world where the average length of a data series gets shorter with the passage of time.

That line is killer for those who’ve tried to get data from StatsCan in recent years. After ‘Data Liberation’ day in February 2012, hope was high. Then people started noticing CANSIM data tables discontinued, frozen or simply vanished.

A couple of points to add to Prof Gordon’s thoughtfully frustrated missive:

Having experienced it first-hand, StatsCan’s either sloppy or playing favourites, misrepresenting the availability of data to some (indicating it’s either not available or only available on a substantial ‘cost-recovery’ basis), while readily providing the same data to others gratis.

Dealing with a US statistical agency, like the Bureau of Labour Statistics or Census Bureau (which we have, on a number of occasions), really puts StatsCan’s poor service into perspective. Their US counterparts seem to bend over backwards to answer every question and provide detailed cross-tabs, many never published, just for the asking, sometimes sent to your inbox while you’re on the phone – even after advising them you’re not from the US.

And that’s without getting into the fact US, UK and other statistical agencies make available timely, complete public use microdata files (PUMFs) for all of their surveys, making it unnecessary to even call and ask about data in most cases. Contrast that with StatsCan: The last Survey of Household Spending PUMF was updated in 2009, and the file excludes key variables. While the Labour Force Survey is one of the few series the agency updates in a timely fashion, its PUMF has never included immigrant variables, despite the questions being part of the survey since 2006.

Despite assurances of a reply within a couple of business days, data requests sent to StatsCan sometimes go unanswered for weeks – often concluded with some variation of ‘it’s not available’ or ‘cost recovery’, and little effort to mask the punt.

Leave a Reply

Your email address will not be published. Required fields are marked *