| The real disproportionality story in Canada’s stimulus money |
|
|
|
| Written by Jim DeLaHunt | |
| Monday, 26 October 2009 | |
|
... a data analysis by two non-partisan software engineers from Vancouver. by Jim DeLaHunt <http://jdlh.com>,
This e-mail address is being protected from spam bots, you need JavaScript enabled to view it
, @jdlh; and Kaitlin Duck Sherwood <http://maps.webfoot.com>,
This e-mail address is being protected from spam bots, you need JavaScript enabled to view it
. The media are, as of October 2009, running stories like, “Stimulus program favours Tory ridings” (Globe&Mail), “Liberal, NDP ridings getting more than fair share…: analysis” (National Post). We’re engineers; we’re non-partisan; we wanted to see the data. The Government of Canada has a stimulus program website with some reports and an API. We wrote come code, called the API, captured the data including latigude and longitude for each of 6,424 projects. We used Ducky’s maps.webfoot.com’s geoinformatics tools to identify which riding contained the lat/long for each project. We made a big spreadsheet, and cross-tabluated by MP’s party and by province. There are two chapters to this tale. First, we think the news reports are missing the real disproportionality story in Canada’s stimulus money allocations, at least as revealed in this data. Second, this is a wonderful example of how open government data can empower citizens. More on both below.
The real disproportionality storySure enough, government ridings are getting more — slightly more — than their proportionate share of stimulus funding and projects. See the column breakdown, and the “partisan excess ratio” below them. But we think that’s not the big story. Also, Nunavut and the territories are getting a hugely disproportionate share (by population) of the funds and projects, and Atlantic Canada is doing very well also. But that looks like small (population) numbers effects. We think the real story is that Quebec is getting disproportionately less of the funding and the projects: 47% fewer dollars, and 65% fewer projects, than their population would justify. And Saskatchewan is making out disproportionately well: 85% more money, from 72% more projections, than their pro-rata share. This is a preliminary analysis. We probably have errors. Some of the data seems to be in error. We may have got the riding wrong for some projects, either because our riding boundary polygons were wrong, or our point-in-polygon algorithm got some edge cases wrong. Also, others may slice the data other ways, and find insights we missed. To help other interested researchers, we’re making this spreadsheet available under a Creative Commons license. The power of open government data: a small case studyStay tuned for a future report on our experiences using the “Canada’s Economic Action Plan” website and API, as a small case study of the power of open government data. About the spreadsheetYou can download our spreadsheet, in three format variants, from the following three links. We may update these documents from time to time.
The front tab, “Summary”, is a DataPilot (“pivot table”) drawn from project-by-project data in tab “Projects”. Feel free to look there if you want to drill down. The original data gave bucketed ranges for project value, so we made estimates in order to calculate sums of project value. See “Value estimates” tab for details. Source: Government of Canada, “Canada’s Economic Action Plan” website <http://actionplan.gc.ca/> and API. Unofficial copy. Non-commerical use permitted per <http://actionplan.gc.ca/eng/feature.asp?pageId=110>. Prepared using OpenOffice.org Calc. A version was also saved in Microsoft Excel format, but the Data Pilot ("pivot table") disappeared in the translation. The Summary tab is also saved in PDF format. |
|
| Last Updated ( Monday, 26 October 2009 ) |
canada-stimulus_jdlh_webfoot.pdf| Next > |
|---|