This dataset includes the Canadian Accessibility and Remoteness Index (CARI+) to measure rurality using travel time and distance to population centres across Canada.
Calculation of CARI+ Scores
The travel distance and time are measured from population-weighted DA centroids to the closest service centres, determined using population-weighted CSD (census subdivision) centroids. Both centroids are based on nested dissemination-block population counts. Population (service) centres were defined using community community-level CSD (census subdivision) boundary files grouped into six size categories based on the population cutoffs from Statistics Canada 2021 Statistical Area Classification.
The scores achieved within each category indicated relative accessibility and were capped at 3. The overall score for a DA was within a range of 0-18 and is the result of summing the scores for all six categories, followed by normalizing to be scaled to 0-1, with 0 as the easiest access and 1 as the hardest.
Category-specific CARI score for each DA (i):
If, ci > 3 → ci = 3
dk∧i represents the minimum distance between the representative point of a given DA (i) and the location (k) corresponding to a specific population size category. The score is calculated as the distance to the nearest service divided by the average of all minimum distances within that category. To reduce the impact of extreme values, scores are capped at 3.
Total CARI+ score, lower values indicate greater accessibility to service locations:
The total CARI+ score for each DA is calculated by summing the scores across all category-specific measures (M = 6 categories). This produces an overall index ranging from 0 to 18, where lower scores reflect better accessibility to service locations.
Calculation of CARI Index (normalized scores):
A normalized CARI+ score is then calculated using min–max feature scaling:
The normalized score, C′, ranges from 0 to 1, where 0 represents the highest level of accessibility. This normalization procedure was applied separately to both travel-time and travel-distance measures.
Classification of CARI Scores:
Multiple schemes were used to categorize the continuous CARI+ scores, including manual classification, equal intervals, quantiles, Jenks natural breaks and standard deviation.
Using manual classifications, the group boundaries would be determined using known population distribution using guidelines used by the Canadian government. Equal interval classification divides the scores into quintiles of the DAs, however it is not an accurate representation due to the data being skewed heavily to the left. Quantile categorization divides the data by placing the same number of DAs into each group, ensuring that each category is represented but resulting in variation across the score ranges. Standard deviation results in groups with fixed distances above and below the average CARI+ score. Though the uneven distribution of the data places the average in the “accessible” category. Finally, Jenks natural breaks groups together with similar values and separates groups where the differences are biggest. Resulting in minimal variation within a group, but more differences between them.
Analysis Steps
The DA and CSD Shapefiles were uploaded to ArcGIS Pro Version 3.0.3, and the DA centroids were merged with the 2021 population count. The Canadian Road Network File was processed in ArcGIS and generated through Network Analysis. The locations of different services were added to ArcGIS Pro. Using them and the centroids, the travel time and distance was generated. After being exported as CSV files, the information and results obtained were processed through Python version 3.10 to produce population weighted travel times and distances for each CSD.