Problems:

• Due to the poor possibilities of IDRISI to work with datatables, I had a hard time to prepare my data. I already described this problem in the section Data Preparation and Manipulation.

• As there was no help available (IDRISI Help is not very helpful let alone the IDRISI Manual and our Tutor does not know the program) it took a lot of time and nervs to find out about sometimes only tiny things. (Getting unexpectedly stuck at those 'considered-to-be-easy' steps made a time planning for the project extremely difficult not to say almost impossible).

Error Sources:
     +  I used Data of the Census 1996 and my literature is from 1997 - thus my data and background 
         sources are relatively new. 
     +  I worked with Enumeration Areas - as I think an appropriate size for my purposes.
     +  I was able to bring my data almost exactely in the form that my literature refers to. 

     +  The literature data I used is data which was collected in the United States. It was collected for all 
         216 Mio members of the society, which provides a broad and reliable base. Although, the factors I 
         calculated refer to data of the U.S., the correlation between the socio-economic factors and drugs 
         should be generally valid - at least for a culture that is as close to the U.S. American cultrue as the 
         Canadian is (even though the Canadians keep negotiating that).
 

     -  I won the impression that there are errors in the Census data itself, as I detected a few values that 
        seem to be highly unlikely. For example is the unemploymentrate extremly high in some areas. There 
        are quite a few EAs with more than 90% unemployment. 
 

Methodology:

     +  I multiplied with factors, added and built averages. I did not provide arbitrary 'Reclass-Categories' or 
         Weighing Factors or boundaries. Therefore, I think the Analysis itself does not contain significant 
         error sources. 

Conclusion:

I found that the error sources for Part I of my project are to be neglected. 
I always worked very close to my background literature when I was building data categories and calculated weighing factors, the results should be theoretically reliable. 
It is another question if theory reflects reality - and if the variables are suitable to draw a picture of drug usage. My results show: the soico-economic criteria are expressive and theory seems to correspond with the real situation in Vancouver. 
 
 
 
 

                            Part II:   Market Analysis for Dealers 

      -  The basic maps I used show police and skytrain stations. I digitized these maps - thus  they contain 
          the inaccuracy that occurs with digitizing. 

      -  As there is no literature about markets for dealers, all factors I worked with are a personal choice. 

 WLC:

      -  A buffer of 150m around police stations is my constraint factor. 150 m is an arbitrary value. It is 
         simply based on my feeling what distance you should keep from an police station in order to reduce 
         the probability to bumb into a police man. 

      -  The weighing factors I provided for Number of Customers (0.55), distance from police stations 
          (0.35) and proximity to skytrain stations (0.15) are arbitrary. I made the choice based on personal 
          thoughts about 'what is how important'. There is no literature base (of course not).
 

Conclusion:

Another person could have chosen different factors, another buffer value and other weighing factors which would have created a different result. 
Notwithstanding, I made my choices for particular reasons, that I explaint (see Methodology). Keeping these explanations and my purposes in mind the map still possesses expressiveness.