A year or two ago, I became aware that Microsoft MapPoint 2002's Sized Circle data mapping feature often seriously misrepresents the relative values of the data it displays. I am wondering if this design flaw is still present in MapPoint 2004, or if it has been fixed.
In MapPoint 2002, a circle representing the value of 100 might be eight or ten times larger than a circle representing the value of 98. In fact, the size of each data circle bears no relationship at all to the underlying numbers - except that bigger circles represent bigger numbers.
This serious design flaw could mislead a business user of a sized circle data map to draw disasterously wrong conclusions about the relative importance of different sales regions, or mislead a public health official into thinking that the number of persons infected by an outbreak of a fatal epidemic in some areas is only an insignificant fraction of the real actual number - and thus divert desperately needed medications and vaccines from where they are truly needed.
To see a simple demonstration of MapPoint 2002's wild numeric distortions, try the following simple test.
Create a simple spreadsheet in Excel with the following values and then display them as sized circles in MapPoint
You will notice in the legend that the circle representing 50 widgets is roughly one-eleventh the size of the 100 widget circle, when it should be exactly one-half. (The largest circle has a radius of 33 pixels, giving it an area of 3,421 pixels. The 50-widget circle has a 10-pixel radius, yielding an area of 314 pixels.)
For a more dramatic illustration, replace the spreadsheet's widget count with 96, 97, 98, 99, and 100 ebola victims.
Look once again at the legend, and you will notice that the circle that now represents 98 is exactly the same size as the circle that formerly represted 50. According to the relative sizes of MapPoint 2002's data circles, Waldo County has recorded eleven times as many ebola virus victims than Oxford County has in the past 2-1/2 hours since the first discovery of this morning's bio-terrorism attack.
Again, going by the circle sizes displayed on the map, if we know that 100 people are infected in Waldo County, then it looks like only one person is dying in Lincoln (a 37-pixel circle compared to a 3,421 pixel circle), so we should let the local hospitals handle the tiny handful of cases in Lincoln and Cumberland on their own while we direct the entire federal and state reponse to Waldo, York, and a little bit of help to Oxford.
In the hypothetical emergency situation above, we could hope that someone would have had the presence of mind to refer back to the actual numbers that fed the emergency planning map, but then - why are we using MapPoint in the first place? Are we using MapPoint because we want to make bad judgements and serious mistakes?
Many months ago, I e-mailed two or three different people at Microsoft about this problem - with no response. It's hard to believe that a company as good as Microsoft would have been this lazy in writing the code to calculate the size of the area of each data circle - but they were.