Potential Errors:

Data Sets:

• The base maps used for this project were ArcView Shapefiles converted into Idrisi raster images.  They came in a data set downloaded from the GeoCommunity website.  The data set came with metadata, which indicated that were originally from Esri's ArcView county data coverage.  For the purpose of this project the data is of sufficient accurancy.

• The accurancy level of the results and census data, however cannot be as easily verified.  The results were compiled from the preliminary results published in the USA Today.  The county coverage from the entire country has some missing data.  In the project the county data was only used for Florida, which being a smaller data set appears to be complete and correct.  The census data being originally from the U.S census bureau should be correct, however because it was downloaded using a query request on a webpage there is no way of knowing if the data downloaded is correct, because i've never seen the original data set.

• The Federal Election Commission data set that was used for country-wide (by State) analysis was the official election results and I would not expect it to contain errors.

Analysis:

• I would not expect the actual images produced using Idrisi to contain errors.  In this project the errors in the resulting analysis would likely be found in the conceptual design.  Bias would be one potential cause of error.  The amount of new coverage on this event means that even in the design stage, considerable knowledge about the election would lead to focusing on certain variables or locations (rejected ballots and florida).  This could cause the analysis to miss other important trends or patterns elsewhere in the data or country.

Project After-Thoughts:

This is where I get to discuss the shortcomings of my project and possible future extensions.  The longest part of this project was getting started.  Not physically, I started looking for data early in the semester.  The problem I found was I did not have enough knowledge or sophisication in dealing with natural phenomena, which seemed to be best suited for Idrisi's tools of analysis.  Eventually I decided to do my project on the U.S election partly because I could find enough data and partly because I also understood the data.  I found for my purposes, simple analysis (visualizing trends using choropleth maps) was sufficient.  I find this disappointing because in lab we started to see that Idrisi is capable of much more complex analysis.  I tried to conceive of analysis related to my data that might be able to exploit some of these features, however my project was not suited to them.

Idrisi itself presented quite a challege.  None of the data came in Idrisi's format and it was difficult to get it in. In addition once data was in Idrisi (databases) it was tedious to edit or manipulate it.  As a result, most of the project was spent converting and customizing the data I had collected, to work in Idrisi.  Most of the data could have been used without much work in ArcView, which would have allowed much more time for the analysis and webpage design stages.