City Stats

Details of the Calculations

Table 1: Group Totals, Percentages, and Index of Dissimilarity (D)

This table includes the population totals and percentages for each ethnic group in the selected city, in accordance with the census categories available in the data for that year.*

In addition, Table 1 provides scores for the Index of Dissimilarity (D) for each group. D is the statistic most commonly used to measure segregation. It indicates the degree to which a group is evenly (or unevenly) distributed across the census tracts in a given city. D measures evenness by comparing a group's representation within individual census tracts to its representation in the city as a whole. The measure produces a number between 0 and 1 that can be interpreted as the proportion of the group's population that would need to change their area of residence to achieve an even distribution across all tracts. For example, people reporting German origins in Toronto in 1971 had a result of 0.1707, meaning that just over 17% of Germans would have had to relocate for the population to be spread evenly across the city. In contrast, people reporting Jewish origins in Toronto in 1971 had a score of 0.7241, meaning that over 72% would have had to relocate. Jews were significantly more segregated in Toronto in 1971 than people of German origins.

Table 2: Interaction Index (xP*y: The exposure of group x to group y)

The Interaction Index (xP*y) describes the social geographic surroundings of a randomly selected member of a given group (group x). The P* value (which also ranges from 0-1) indicates the probability that someone else selected from the same residential area (census tract) will be a member of group y. For example, the P* score for the exposure of Chinese (x) Vancouverites to Italian (y) Vancouverites in 1971 is 0.0545. This tells us that a randomly selected person living in the same census tract as a Chinese Vancouverite had a 6% chance of being Italian. You might think of this as a measure of strangers likely to be found at the bus stop, or waiting in line at the post office. For Chinese Vancouverites in 1971, Italians were a small but notable percentage of everyday social surroundings. The P* measure reflects population totals within the city. In 1971, for example, all Vancouverites had a high level of exposure to people of British origins, who were numerically dominant throughout the city. Accordingly, the P* values are not symmetrical. For example, the exposure of Chinese Vancouverites to those from the British Isles in 1971 was 0.4455, whereas the exposure of Vancouverites from the British Isles to Chinese Vancouverites was only 0.0256. Highlighted in black in each table is each group's exposure to itself (xP*x), which is also known as the isolation index.

Table 3: Adjusted P* (xP*y / Proportion Y: Interaction Index expressed as odds ratio)

The Adjusted P* featured on this site was created by our original website designer, Julius Davies. It is not a common statistic (although somewhat similar measures have been created by various scholars). This measure divides the interaction index by the total proportion of group y (Italian in the example above) in the population. This creates an 'odds ratio.' Adjusted P* indicates the extent to which being of a given ethnic origin (Chinese in our example above) increases or decreases your likelihood of living near members of group y. It generates a number that should be interpreted in relation to the number 1. If the adjusted P* results in a 1, then being in group x has no impact on living near group y. In the case of Chinese exposure to Italians in Vancouver in 1971, the adjusted P* is 1.9626, meaning that residents of Chinese origins were almost two times more likely to reside near Italian Vancouverites than we would have expected, given the proportion of Italians in the city. A number lower than 1 would indicate that being in group x reduces the likelihood of residing near group y.

* The data used for our calculations were provided by Computing in the Humanities and Social Sciences (CHASS) at the University of Toronto. With the exception of the 1981 and 1991 data, the tract data used are made available to users through the Canadian Census Analyser. In 1991 and 1981 we use more detailed ethnicity data provided to us by Laine Russ at CHASS. In 1991 and 2001 we include only individuals who indicated a single ethnic origin.