GEOG 376 Project

DISCUSSION

Data Quality.
2001 census data was our primary data source. Problems associated with this data can arise from mis-interpretation of the data categories provided by Statistics Canada. When dealing with socio-economic data, these classification interpretations can generate very different results.

Census data is only based on a 20% sample. In other words, Statistics Canada only interviews one fifth of the population, and makes inferences with respect to the characteristics of the whole population based on that sample. Consequently, the data provided may not be representative of the entire population. Statistics Canada completes a census once every five years (01 and 06). As it is 2005 now, we are using the data collected four years ago. This means that the changes and trends which have developed in the last four years are not represented by this data source.

Other data issues include the fact that not all recent immigrants partake in the census process. Many have not settled in one place or are transient between locations. It is probably also reasonable to assume that Vancouver has a large number of illegal immigrants that are not represented with census data.

Geo-coding.
From the geo-coding process, it is evident the existing service centres are not located optimally to serve the immigrant population as a whole. While the location of existing service centres do appear to be located in areas of high immigrant settlement, there new areas of settlement that are not serviced (North and West Vancouver).

ArcGIS uses several parameters in identifying a location this includes ; ‘from left’, ‘from right’, ‘to left’ and ‘to right.’ It is problematic when we find that a address only has a left or right side and misses two of the parameters (scores only 25-65/100). As a result, we have to refer to paper maps in order to confirm the location, and at other times accept the low scores. More troubling are ‘unmatched’ results. In these circumstances, we must refer to paper maps, look at the attribute table of the reference street address, and search for the address we need. For example, if we find that the reference street address only goes up to 6900 and the existing service centre is on 7000, we assume that this discrepancy is an error in the reference address. We then manually adjust the error and match again.

Locating the new service centre.
After we overlay the existing service centres and the location of recent immigrants, it becomes clear that North Vancouver has a lot of new immigrants from Iran but no service centre for them. We then consider the bus route and accessibility in choosing its location. Through this process, we determine the location based on our own judgement and not on any statistically significant basis.

Ecological fallacy and modifiable areal unit problem (MAUP)
MAUP is always associated with socio-economic data. It occurs when researchers apply the characteristics of a group to a sub-group or individual, and infer results based on the arbitrary decided zones (areal units). These errors always accompany aggregated data, and we cannot avoid them unless we work at the individual level.

Because we tend to work with primary and absolute data (number of new immigrants in each DA) we reduce the impact of the MAUP. By distinguishing between recent immigrants and 'other' immigrants, we have ensured that the characteristics of the all immigrants are not applied to recent arrivals.

MAP COLLECTION