BEST PRACTICES - MAPPING HEALTH DATACartographic FundamentalsMaps are strong visual tool summarizing complex information and patterns that my otherwise be missed when the data is in tabular form. When people look at a map, they are simultaneously interpreting a variety of information such as how the data were collected, how the data were analyzed, colour schemes, symbols, and other qualities such as geographical patterns, location or distances. While map interpretation depends in part on the map reader, understanding basic cartographic design processes will ensure that map readers take away your intended message after studying your map. Some key considerations when producing a map of health data include:
Mapping small and large areasMapping large sparsely populated areas in the same map as geographically smaller but more populated regions can have the result of the large sparsely population regions visually dominating the map. One way to address this issue is to employ the Statistics Canada Ecumene boundaries.
At a minimum, population ecumene’s only include populated areas with a minimum density of 0.4 persons per square kilometer, which is roughly about one person per square mile. Population Ecumene boundaries also offer an alternative approach for mapping rural areas. One of the limitations of mapping rural population areas is that these areas are usually large, but with very low population densities. Choropleth maps may misrepresent the ‘severity’ of a health condition in these areas as the uniform colour distribution implies that the rate is homogenous across the entire population. The Ecumene boundaries only contain the core population areas and provide a more accurate description of where the increased incidence risk is occurring. Another approach is to overlay a population density mask on top of the map. A mask refers to any item on a map that is excluded. Density masks are useful tools to use when mapping rural and urban areas simultaneously. For example, a choropleth map of two or more adjacent rural areas with the same rate may generate a serious visual impact on the severity of a health condition and keep the reader from seeing smaller areas that may have a similar incidence rate. Population density masks exclude can be created using a buffer tool. The buffer is created around each city or postal code. The area outside of the buffer is then used as a ‘blanket’ to mask areas outside of the radius. Rural areas may also be susceptible to ‘edge effects’. Edge effects occur as a result of placing a boundary over a number of sample points and using the sample points to quantify some type of value for the entire area. The concentration of sample points between neighbouring boundaries may skew the analysis if a large number of sample points are located near the edge of one boundary, but no sample points are located on the edge of the adjacent boundary. See ‘edge effects’ for additional examples. Data categorisationThere are numerous methods for classifying data and most GIS and mapping programs offer a selection that often includes quantiles (link to definition), equal intervals (link to definition), and Jenks (link to definition). Other choices include classifying by standard deviations and minimizing differences across boundaries. There is no one correct way to class a data set and different methods will produce different map patterns, especially if data are skewed or include extreme outliers. Examine the data histogram or graph of the data to assist in choosing classes. Generally, a sound approach is to start with a standard classification and adjust breaks to improve the map based on knowledge of the data and the audience. When maps are categorized into more than 10 classes it is difficult for the reader to interpret. A simple, clear-cut map with four or five classes may be better for an unsophisticated audience, inexperienced at reading graphics. Trained eyes may appreciate the extra information which seven or eight classes portray, therefore, it is important to keep your audience in mind. Whatever the choice, it is important to present the variation in the data and choose the most appropriate classification scheme to convey how a phenomenon varies geographically. Choosing appropriate colour schemesChoice of colours for displaying the data in your map will influence how your map is interpreted. When choosing colors for a map, the colour scheme should parallel the ordering in the data. For example, the lightest colours should be assigned to the lowest values and the darkest colours assigned to the highest values. Adding hue variation can help make it easier to see differences between color symbols. A number of useful and readymade colour ramps have been previously constructed and are available online through a program called ColorBrewer Examine the data histogram or graph of the data to assist in choosing classes. Generally, a sound approach is to start with a standard classification and adjust breaks to improve the map based on knowledge of the data and the audience. When maps are categorized into more than 10 classes it is difficult for the reader to interpret. A simple, clear-cut map with four or five classes may be better for an unsophisticated audience, inexperienced at reading graphics. Trained eyes may appreciate the extra information which seven or eight classes portray, therefore, it is important to keep your audience in mind. Whatever the choice, it is important to present the variation in the data and choose the most appropriate classification scheme to convey how a phenomenon varies geographically. ReferencesThe North American Association for central Cancer Registries has produced a handbook of best practices on the use of GIS with cancer data which contains best practices related to GeoCoding, Confidentiality, spatial analysis and cartography < back to Mapping Health Data main page |
