Ontario Marginalization Index (ON-Marg)       Ontario Marginalization Index (ON-Marg)
 ON-Marg FAQs


  Q: What is the Ontario Marginalization Index?

The Ontario Marginalization Index (ON-Marg) uses data from 2001, 2006, 2011 and 2016 to illustrate levels of marginalization across the province. ON-Marg focuses on four dimensions that contribute to the process of marginalization: residential instability, material deprivation, dependency and ethnic concentration. ON-Marg makes this information available for various geographic units (like census tracts and dissemination areas), allowing you to examine levels of marginalization in specific areas. For example, you could use ON-Marg to look at the level of residential instability for a particular census tract.

  Q: Who created ON-Marg?

The Canadian Marginalization Index (CAN-Marg) was developed in 2006 by researchers at the MAP Centre for Urban Health Solutions at St. Michael’s at St. Michael's Hospital in Toronto and McMaster University in Hamilton. ON-Marg is an Ontario-specific cut of the CAN-Marg data. The 2011 and 2016 versions of ON-Marg were created through a collaboration between MAP Centre for Urban Health Solutions and Public Health Ontario.

  Q: What is marginalization?

Marginalization is the process by which individuals and groups are prevented from fully participating in society. Marginalized populations can experience barriers to accessing meaningful employment, adequate housing, education, recreation, clean water, health services and other social determinants of health. Both community and individual health are deeply impacted by marginalization.

  Q: What is an index?

The goal of an index is to combine discrete yet related variables into a single, broader measure that can be used for applications including research, advocacy, policy design and program implementation. For example, in order to provide a measure of 'residential instability,' ON-Marg uses seven separate census variables including the proportion of people living alone, and the proportion of multi-unit housing.

  Q: What is a deprivation-based index?

A deprivation-based index focuses on factors that undermine individual and area health. Prominent examples of deprivation indexes include the Townsend Index of Material Deprivation (Scotland) and the New Zealand Deprivation Index. In Canada, we have seen the adoption of the Socio-Economic Risk Index, the Socio-Economic Factor Index, the INSPQ Deprivation Index for Health and the Vancouver Area Neighbourhood Deprivation Index (VANDIX). While all of these indexes contain valuable information, ON-Marg is unique in that it includes the four distinct dimensions of marginalization, and allows these to be explored at various local levels of geography.

  Q: How did you choose the four dimensions of marginalization used in ON-Marg?

We pulled together 42 different census variables based on historical and contemporary theoretical perspectives on inequality and marginalization in Canada. Some of these variables are also used in other current deprivation-based indexes. We then used a statistical method called 'factor analysis' to reduce the number of variables from 42 to 18. The analysis pointed to four themes of marginalization, each associated with some of these variables. Finally, we named these themes, and they became the four 'dimensions' that make up ON-Marg: residential instability, material deprivation, dependency and ethnic concentration.

  Q: Why is 'ethnic concentration' a domain of ON-Marg?

When census variables related to the process of marginalization were analyzed through factor analysis, two came out as particularly relevant to area-level marginalization: proportion of recent immigrants and proportion of people identifying as visible minorities. We called these two variables together 'ethnic concentration.' Some studies have found, however, that ethnic concentration can be correlated with positive health outcomes. The domains for ON-Marg were chosen due to their impacts on the overall process of marginalization in society. At the same time, it is clear that the ethnic concentration domain has varying - and often positive - impacts on health outcomes.

  Q: What are some applications for the Ontario Marginalization Index?

ON-Marg allows users to explore the relationship between specific outcomes and marginalization rates at the area level. Potential uses include:

  -  Predicting the kinds of health or social services that may be needed in a specific area;

  -  Monitoring inequities in an area over time, and evaluating interventions;

  -  Researching the relationship between marginalization, health and/or other outcomes for local

  Q: Where can I get the data and more information?

To access the data and ON-Marg User Guide please refer to the "ON-Marg" page at www.OntarioHealthProfiles.ca
Ontario Community Health Profiles Partnership (OCHPP)


  Q: How can I choose the right dimensions for my study?

Your choice of dimensions will be informed by your working hypothesis and knowledge of the population and issues in your study area. It is also important to consider potential applications. For example, if you are interested in using ON-Marg to design a program or service, this might inform your choice of dimensions. In general, we recommend using as many relevant dimensions as possible to explore the relationship between health outcomes and the different aspects of marginalization.

  Q: How do I explore the relationship between outcomes and area-level marginalization?

ON-Marg can be used to measure associations between area-level marginalization (e.g. geographic units) and various types of outcomes. Some commonly studied outcomes include rates of particular diseases, mortality, health-related behaviours and levels of health care provision and uptake. The process often involves transferring outcome measures from points data (for example postal codes) to dissemination areas (DA) or census tracts (CT) with the use of such tools as Postal Code Conversion File (Statistics Canada) or other methods of geographic conversion. Once the dimensions of the index and outcome measures are defined for the geographic unit, they can then be linked and analyzed.

  Q: Does ON-Marg work equally well in urban and rural settings?

While individual census variables that fall under the four domains such as 'proportion of multi-unit housing' or 'proportion of five-year recent immigrants' may be considered more relevant for urban settings, as a whole, the four dimensions are designed to reflect various aspects of marginalization in both urban and rural settings. Please note: census data is suppressed for some sparsely populated areas (see below).

  Q: Why isn't there a census indicator of Aboriginal status included in ON-Marg?

Aboriginal status refers to a set of Canada Census variables including Aboriginal Ancestry and Aboriginal Identity. It refers to persons from North American Indian, Metis or Inuit groups. The Canada Census incompletely enumerates people living on First Nations reserves and in Aboriginal settlements. This means that Aboriginal people are undercounted in the Canada Census. Undercounting is most problematic in Northern areas, but can potentially be problematic in Southern Ontario as well. For example, in the context of the 2006 census, 18 reserves and Aboriginal settlements declined enumeration entirely. In addition, when numbers are very small for a particular area, Census Canada will suppress the data, meaning that many census variables for Northern populations, and especially undercounted Aboriginal populations, will not be populated. As a result of the above, while 'Aboriginal status' was included as one of the original census variables fed into factor analysis, it did not emerge as a separate factor related to marginalization.

The potential impact of area marginalization will be underestimated by any of the deprivation/marginalization indexes that do not account for Aboriginal populations. We recommend that those who wish to use the index also include one of the census indicators of Aboriginal status as an additional variable when questions arise that are specific to marginalization and Aboriginal populations.

  Q: What are the factor scores and quintiles shown in the data files for each dimension of

Please note: Factor scores and quintiles are dimension-specific and cannot be used to obtain a summary score for all four dimensions.

Each domain of marginalization represented in the index is broken into quintiles. Quintiles represent different degrees of marginalization within a specific domain starting with 1 (least marginalized) and up to 5 (most marginalized). Each group contains a fifth of the geographic units in question. Quintiles allow for comparisons of marginalization levels in a given dimension across Ontario (ON-Marg).

Those wishing to identify more detailed differences in marginalization across the study area should use the factor scores. Factor scores define level of marginalization for each dimension in each single geographic unit. They do not represent absolute measures of marginalization, but are relative to each other. A higher factor score represents a higher degree of marginalization.

  Q: Can I calculate quintiles for my own study area?

Yes. To do this, gather the original ON-Marg factor scores by census tract or dissemination area. Include all the census tracts or dissemination areas that make up the custom study area. Order the census tracts or dissemination areas by factor scores; then divide them into five equal groups. The group with the lowest factor score would be assigned 'Q1,' representing the lowest level of marginalization, while the group with the highest factor scores would be assigned 'Q5,' representing the highest level of marginalization.

  Q: Can I aggregate the index to larger geographic units?

Yes. To calculate the index value for a larger unit such as an urban area, factor score values from DAs or CTs within that unit need to be multiplied by the corresponding populations in those DAs or CTs and then added together. This sum needs to be then divided by the total population from all included DAs/CTs. This procedure is described in detail in the User Guide.

Caution: Weighted averages can disguise heterogeneity within large geographic areas.

  Q: What are the common geographic units suitable for ON-Marg?

ON-Marg is most accurate in illustrating marginalization using small geographic units (e.g., dissemination areas and census tracts). Users of the index should apply their own judgment when aggregating index dimensions to larger areas as described in the previous section. Some larger areal units commonly used in Ontario in the health context include:

  •  Ontario LHINs (n=14)

  •  Ontario Sub-Regions (n=76)

  •  Public Health Units (n=36)

  •  Census Subdivisions (n=575)

  •  Census Divisions (n=49)

  •  Aggregate dissemination area (n=1685)

  •  Forward Sortation Areas (n=513)

  Q: Why is there no data for some dissemination areas and census tracts?

To ensure data quality and privacy, Census Canada does not release census information for areas with low response rates or with low population or household counts. For more information on (2006) census data quality and confidentiality standards please visit:


  Q: What does it mean when some individual census variables within a given dimension are
       'reverse coded'?

Sometimes, the data that's most relevant to ON-Marg is framed oppositely to the census data. For example, 'residential instability' tends to be higher in areas with a higher percentage of dwellings that are rented. The original census variable records the percentage of dwellings that are owned. As a result, in order to have this variable correlate positively with the domain of 'residential instability,' it needed to be reverse-coded prior to factor analysis.

  Q: Does the index show area or individual characteristics?

ON-Marg values (factor scores and quintiles) are assumed to be capturing the general characteristics of a given area. Some research uses area-based measures as a proxy for individual-level data when none is otherwise available. This, however, can produce an, 'ecological fallacy,'-a situation where general information about a group or area is used to incorrectly characterize the characteristics of an individual. For example, some people living in areas with high overall rates of residential instability might own their homes. Using the smallest areal unit possible (dissemination area) diminishes this measurement error.