May 27, 2015
Earlier this month we published our analysis of the Hardest Places to Live in Canada.
However, because of gaps in the quality and availability of data across our nine health and wealth indicators, we ended up with a rather incomplete picture. Nevertheless, the map sparked some good discussion and follow-up questions about our data.
Today, we are releasing the data (download here) we used to build our map of the Hardest Places to Live in Canada, in the hopes of providing a platform for further research and analysis.
As you browse through the data file, below are a few things to keep in mind.
First, as our original post explains, we combined four years’ worth of Canadian Community Health Survey (CCHS) data, 2010 to 2013, to achieve a sufficient sample size. The annual CCHS data files contain the following number of cases:
- CCHS 2010: 63,191;
- CCHS 2011: 63,542;
- CCHS 2012: 62,103; and
- CCHS 2013: 64,346.
The combined 2010-2013 CCHS data file contains a total of 253,182 cases. The combined 2011-2012 data file used for the food security indicator contains a total of 125,645 cases.
Second, we calculated our rankings based on a three-step process:
- Standardized the scores for each indicator using the number of median absolute deviations (MAD) from the median. This measure is less sensitive to outliers than the typical approach to standardization, which is to take the number of standard deviations from the mean.
- Averaged the standardized scores across each indicator.
- Ranked the health regions based on their average standardized scores. Regions with lowest average score are considered to be doing worse and regions with highest average score are considered to be doing better.
The first sheet in the data file provides the raw data and the coefficient of variation (CV). According to Statistics Canada, CVs are the best overall measure of data quality. In social statistics, a CV between 0 and 16.5 percent is considered good quality. A CV between 16.51 percent and 33.3 percent is considered marginal quality and should be used with caution. A CV greater than 33.3 percent is considered poor quality and should not be used. Statistics Canada will not publish estimates with a CV greater than 33.3%. We included data in our analysis with a CV between 0 and 33.3%.
The second sheet contains the standardized rankings for all the communities with the MAD score. The third sheet provides a technical description of each variable.
More related to this topic
- We’re speaking more languages, but is our landscape more diverse?
- The curious case of Boaty McBoatface: Why trust between publics and governments needs to flow both ways
- How can delivery units work in a federation like Canada?
- Trudeau signals his commitment to open government by releasing his ministers’ mandate letters
- Hardest Places to Live Data Release
Mark D. Jarvis
May 25, 2015