METHODOLOGY

DEMOGRAPHIC CRITERIA

The new library should be in an area with a high number of people who fit the demographic profile of a 'typical' library user. The six different demographic criteria I used for this project were selected based on the Vancouver Public Library 2000 Survey of Users and Non-users, the survey conducted by CGT Research International that I mentioned earlier. Since it is the absolute number of typical users that is important, as opposed to the relative, I used the actual number of people representing the ideal characteristic of a typical user, as opposed to the number in relation to the total number of people per enumeration area. The six demographic criteria used were:

1. Age:
Generally, library users are younger than non-users. According to the survey, 48% of library users were between the ages of 18 and 44 ( 4 ). As mentioned earlier in the 'Data' section, the values used for this factor were derived by adding the more specific age category values from the census data.




2. Gender:
A disproportionate number of women identified as library users. 53% of users were women, while only 37% of non-users were women ( 5 ).

3. Education:
Library users tend to have attained higher levels of education than non-users. 48% of users have a university degree or higher, and an additional 14% have some university education        ( 6 ).

4. Income:
On average, library users have higher household incomes than non-users ( 7 ). For this factor, I used the average income value per enumeration area.

5. English:
Library users are somewhat more likely than non-users to speak English at home, although the difference is not substantial ( 8 ).

6. Chinese:
Aside from English, the language spoken the most by library users at home is Chinese (both Cantonese and Mandarin). This is also an important factor in terms of attracting more people to use the library: according to the survey, when non-users were asked what would motivate them to use the library, the third-most frequent response was 'more Chinese books' ( 9 ).

After assigning 'avl' files to the Enumeration Areas raster file for each of the six demographic criteria, I needed to reclassify each layer so that that they would be standardized to byte-level range (0-255), making it possible to use them for Multi-criteria Evaluation. As a guide, I displayed each factor in Arcview using the natural breaks classification and six categories, then reclassified the data in Idrisi based on the values in Arcview. Below are the reclass tables for each of the six factors.

Tables used to reclassify demographic criteria layers.

Once I had created reclassified layers for each factor, I used the FUZZY module to convert the values into byte format. I used 0 as the first control point and 5 as the second (the minimum and maximum values, respectively). The demographic criteria were now ready to be used for MCE.

Below are the 3 different maps showing the steps I took to create the layer 'womenfuzzy' as an example.

OTHER CRITERIA

Aside from demographic criteria, there were three other factors used for the locational analysis. Two of these involved proximity to public transportation routes: 1) bus routes, and 2) Skytrain stations. Aside from being close to people who are regular users, the library should also be accessible by public transit. The final factor was the presence of already existing libraries. Even if there is a high concentration of typical library users in a given area, it clearly makes no sense to build a new library in this area if one already exists.

In my original project design, transportation routes was going to be a factor, based on the need for a new library to be accessible to vehicles circulating material between branches. I was, unfortunately, not able to use this factor due to limitations of data and software quality (see Methodological and Operational Problems ).

For the transit routes factor, I needed to reclassify the data to create a boolean image, with a value of 1 representing transit routes.

I then used the DISTANCE module to create a layer showing the distance values from transit routes.

Finally, I used FUZZY to create an image showing suitability based on proximity to transit routes, using the j-curve function. Ideally, the new library should be within 50 meters of a transit route, and anything beyond 300 meters would be unsuitable, so I used these as the two control points.

For the proximity to skytrains factor, I used the exact same steps, although I extended the areas of suitability by using 1000m and 3000m as control points a and b respectively. This is based on two assumptions: 1) skytrains have a much greater passenger capacity, and 2) future population growth will generally be concentrated to some extent at skytrain stations, especially the new ones. This is apparent from the high-density developments that have occurred at already existing stations, including the Main Street and Metrotown stations.

Finally, I created the distance from Libraries layer, again reclassifying the data to give all libraries the same value, then creating a distance layer using the DISTANCE module. I used the FUZZY module with the j-curve function again, only this time I used the 'monotonically increasing' function, since suitability increases as distance from already existing libraries increases. For the control points, I used 1000m and 1500m, based on minimum distances between already existing libraries (Arcview came in handy to quickly query these distance values).

The final step before performing Multi-Criteria Analysis was the weighting of the nine different factors, then calculating the weights for the 9 factors and creating a decision support file 'libdecision'. Below I have provided 1) a screen capture of the pairwise comparison matrix 'libdecision' that I created, and 2) the table derived from calculating the factor weights from this matrix.

The comparison matrix created to perform MCE.



Table showing weight values for MCE.

CARTOGRAPHIC MODEL

The final piece for the methodological section of this project is, of course, the cartographic model. Originally, I managed to get it all into one Illustrator File, but because I needed to add a step later (see next section), a second model for that step was added below.