Thematic Maps 101


  1. Introduction
  2. Indicators
  3. Standards
  4. Data distribution types
  5. Geographies
  6. Custom geographies


Thematic maps provide a visual impression of comparative data for different geographic areas.  This is usually done for selected indicators such as average salary, percentage of homeowners living in an area, or a composite indicator that is a mathematical combination of several indicators. Different colours represent ranges of values. This can be a single colour with different shades of the same colour or A range of colours such as red-yellow-green.

  • G5 barSingle colour is best with ranges for 3 to 5 colours. A single colour is appropriate for linear data such as number of people in a geographic area.
  • RYG barMulti-colour theme works well with indicators that are compared to a standard. In the work done on this site three shades of red, a neutral yellow and thee shades of green are used to compare a specific indicator across various geographies.

Some of the variables considered when making a thematic map include such things as:

  • An indicator of interest to the policy or program being considered
  • Is there a “better” or “worse” for indicator values. For example, higher income levels are associated with improved quality of life and low incomes present many challenges in daily living. Green for better, red for worse and yellow for an average or neutral value is used here.
  • What is the “standard” to be used for determining an average value. How should it be applied across geographic levels or over time? What judgements or assumptions are embodied in the various choices?
  • What unit of geography should be used?
  • What distribution method should be used?

Details on these choices are shown below as well as some examples of applying them.


Judicious use of indicators that are directly relevant to a policy, program or project is better than a lot of statistics chosen to make a point. Understanding the policy intent or program objectives is an important first step.  Choosing statistics and indicators that inform the issue being explored will be more productive.  Often the lack of information or direct statistical evidence is an important observation. Composite indicators, such as the shelter cost index described under Measures of Poverty can give good additional insights. Measuring something like community engagement is difficult and therefore a proxy indicator may have to be used.  Voter turnout is considered to be a reasonable proxy measure for community engagement.

Return to Contents


When preparing the data for a thematic map, as with any graphical presentation of statistics, a number of judgement calls are required. Generally statistics can be relied on, but it is too easy to present them in a way that distorts a quick visual impression of actual differences. Unemployment rate and the daily stock market changes are examples of these. Monthly changes in the unemployment rate and daily changes in the stock market are generally quite small, but because of the way the data charts are presented they appear to be quite large.  Similarly, presentation of thematic data can appear to be very different depending on how the data is presented. In the section below showing thematic mapping a few of the differences are demonstrated.

Average value: For the comparative maps on this site, HRM average value for a given indicator is used as the base or neutral value (yellow). Generally, the HRM and Canadian averages are comparable and the local reference point was considered to be a better choice. The Nova Scotia average is also considered a good choice for a base.

Data Range: For a given indicator, each geography has its own range of values and these can change over time. Other factors such as gender or family type can also contribute to the range of values for an indicator. For most of the maps presented on this site, the DA values for HRM were used to establish a custom set of ranges for each indicator. These were then used for each of the other geographic levels. The DA values had the widest range and by using the same range set it is possible to get a visual comparison of the same data indicator for different geographies. It also highlights the masking effect of larger geographic areas on neighbourhood differences.

A similar approach can be taken for values over multiple time periods. Thematic maps for historical changes of persons living alone and voter turnout show dramatic differences over an extended time period by using a single range set over the  time period covered. A highly regarded professor from England was invited to make a presentation on his research to Dalhousie University staff and students several years ago. His research found significant gender differences on a particular indicator. His thematic maps in the presentation used different range sets for each of the female and male results. This gave an interesting visual presentation. If he had used a combined range set the visual impression of the differences would have been far more striking and obvious.

There is no right way to do this, just different ways.  Interpreting and Using the Thematic Maps provides more details on DWPilkey Consulting’s current approach to thematic mapping.

Return to Contents

Data distribution types

When using a geographic information system (G.I.S.) to create thematic maps, one of the first options is to choose one of several methods for displaying the range. MapInfo Pro offers the following options.  The samples shown with each of these options is based on the Low Income Measure – After Tax (LIM-AT) for Nova Scotia CDs. For more about this measure see Measures of Poverty. The same data set is used for all of the following maps.

LIM-AT 2016 Equal Count

Equal Count

Equal Count has close to the same number of records in each range.

LIM-AT 2016 Equal Ranges

Equal Ranges

Equal Ranges divides records across ranges of equal size. Cumberland and Pictou are among those that changed colour, i.e. in a different relative category.

LIM-AT 2016 Natural Break

Natural Break

Natural Break creates ranges according to an algorithm that uses the average of each range to distribute the data more evenly across the ranges. It distributes the values so that the average of each range is as close as possible to each of the range values in that range. This ensures that the ranges are well-represented by their averages, and that data values within each of the ranges are fairly close together.

LIM-AT 2016 Standard Deviation

Standard Deviation

When you create ranges using Standard Deviation, the middle range breaks at the mean of your values, and the ranges above and below the middle range are one standard deviation above or below the mean.

LIM-AT 2016 Quantile


Quantiling (divided into equal-sized, adjacent, subgroups) enables you to build ranges that determine the distribution of a thematic variable across a segment of your data.

LIM-at TM cust rng 2016CDsm

Custom Range – DA derived

You can also define your own ranges using Custom. This is the method preferred by DWPilkey Consulting.  Its application for our work is described more fully in Interpreting and Using the Thematic Maps

In this case, HRM is in the average or yellow range with four other counties in the same range. This representation is closest to Quantiling and quite different from the other methods.

LIM-at cust rng 2016CD-CT

Custom Range with CT overlay on HRM

By itself, it seems strange, but when CT and DA level analysis is added for HRM it demonstrates the averaging effect of larger communities. The map on the the right overlays CT data for HRM. This demonstrates the disparity of income among areas within HRM and the sharper contrast with other parts of the province.  Pockets of more economically advantaged areas also exist within other counties.

LIM-AT 2016 custom HRM core ct

Custom Range for HRM by CT

Zooming in on HRM core shows the sharper contrast that was masked by using CD level data.  Further unmasking is achieved by looking at the DA based map below.

LIM-AT 2016 custom HRM core da

Custom Range for HRM by DA

See the thematic maps for  a better quality representation of these last two maps.

Return to Contents


Statistics Canada Geographies in NS

In addition to data being released at the national, provincial and territorial level, Statistics Canada also releases data at various levels within each province and territory.  Electronic geographic files that correspond with the data are used in the production of thematic maps. The Hierarchy of standard geographic areas for dissemination, 2016 Census shows the overall structure and number of each for Canada and each province and territory. Geographic areas by province and territory, 2016 Census is a table that shows number of each for Canada and each province and territory.

A brief description of those used on this site for Nova Scotia is shown in the following.  Five of the 17 or more geographies available are used in the work demonstrated on this site.

  • Census division (CD) – The 18 CDs in Nova Scotia are the same as the county boundaries. They range in population size from 7,100 to 403,400.
  • Census subdivision (CSD) – The 96 CSDs are parts of counties. Each Indian Reserve is a CSD. They range in population size from 80 to 24,900 with Cape Breton at 94,300 and HRM at 403,100 as outliers. Eleven CSDs have data supressed.
  • Census metropolitan area (CMA) – HRM is the only CMA in Nova Scotia.  For HRM, the CD, CSD, and CMA are all virtually the same data. HRM CMA has a population of 403,400.
  • Census tract (CT) – The 98 CTs are for HRM. Five of the Census Tracts (Indian Reserves) had the data suppressed to protect confidentiality. Except for six small ones the CTs range in size from 1300 to 9600 people.
  • Dissemination Area (DA) – This is the smallest level of geography for which data is released by Statistics Canada. Nova Scotia has 1658 DAs with most ranging in size from about 200 to 3300 (one outlier in HRM has 6,800 people in the DA). Five DAs had less than 100 people and another 15 had between 100 and 200 people. The larger DAs are likely the result of rapid growth after establishment of the DA boundaries which is several years before each Census. Twenty-eight DAs have the data suppressed, and some mid-size DAs have selected data elements suppressed.  Various income categories and details of particular small groups are among the most likely to be suppressed.

Return to Contents

Custom geographies

When Nova Scotia Community Counts was active, it modelled Statistics Canada CSD, CT and DA data into almost 300 communities as well as other geographies of interest such as Community Health Boards. The geographic work initiated under Community Counts has been refined by Dr. Mikiko Terashima’s collaborative work in developing the Community Cluster Report.

In HRM, United Way Halifax has provided strong program support in three “Action” neighbourhoods – Spryfield, Dartmouth North and Fairview.  As part of United Way’s Poverty Solutions research, thematic mapping showed two additional rural areas as having higher than average number of people living with low income – HRM Rural East and Preston Area. Data was modelled for all five of these custom geographies and the UW Census Snapshot 2016 March 2018 was prepared.

Return to Contents