Ok I will admit that I had mean and mode confused.

So let me restate this:

if you were to gather all of the raw data from your numerical tests and then compute the averages, having a few outliers at a number such as 195 versus the same group of outliers at a number such as 156 would affect the mean, but not affect the mode.

Since the curve is applied later, they have to find a way to make the mean, mode, and median 100, which means adjusting everybody's score so that it looks like a perfect normal distribution.

I am not privy to the way that they deal with statistical outliers for IQ testing, so I can only assume that their scores are summarily lowered so that they don't affect the placement of the mean with regard to the mode, which ensures that the number 100 remains meaningful over the generations.

However, I think with access to the raw data, one could see differences between mean and mode and median.

It also calls into question the accuracy of the number you obtain immediately after taking the test (especially if it is an extremely high or low number) since the curve has not been applied yet.