1. Variables - Initially it is important to note that the factors chosen for this study were chosen subjectively by questioning those who are in a position to have some idea of the type of people who are dancers or who have children who dance. Some factors may not have been identified and therefore this could sway the final result. Other factors that were identified, such as dedication and desire, can not be calculated in this type of analysis.

2. Data - Because the data was collected from a variety of different sources, there were some initial problems with overlaying the layers and having them match up incorrectly. For example, at first the roads layer was almost an entire city out of place. This problem was due to the projection of each layer and was rectified using ArcToolbox to change projections and assign projections to the data. A lack of metadata for some of the layers proved to be a problem as well as there was no way of identifying the format that the layers were in, in the first place.

3. Conversions - A number of problems occured when converting the data. Initially, it was difficult to get the data to display based on the new fields calculated. However, this was solved using FME which converted data between Arc View shape files to IDRISI vector files. Data was also converted once in IDRISI from vector files to raster files. This constant changing of the data type causes problems. The amount of detail is reduced after numerous conversions leaving a rather generalized input once the images were finally used in the MCE analysis.

4. Constraints - Constraining the analysis using landuse and current studio locations is problematic because an area that may have been suitable would not be considered in analysis enough though it would have been a key location. For example, because of the buffering of the current highland studio locations, almost all of the city of Vancouver was not included in the analysis. This area is highly populated and there may have been an area in which more than one studio could have been successful.

5. Weightings - The weighting of each factor using the analytical hierarchical process is rather subjective as well. So, someone else could do a similar study and yield completely different results by having different weightings for each of the factors.

6. Compounding errors - By using mathematical operations to calculate new fields, errors are propagated throughout the analysis. This also occurs when joining fields, changing the projections and converting data.

7. Resolution - Assuring that the cell sizes were the same in all layers was one of the problems experienced throughout the analysis. Resolution also became a problem when designing the website as some of the images (either saved as jpegs or as bitmaps) did not show up well as they could have.