Hmmm so much for being gone - the new article prompted a least a quick lunchtime note.
Changing the percentile definition just moves you up one row higher in the percentile table than previously. Not like the doubling from 99.0 to 99.5; it just bumps up by a single row - this doesn’t change the high part of the tables much at all, but could change the center of the tables a fair amount.
Example:
Old Table
73 99.3%
72 99.2%
71 99.1% Scored better than 99.1% of people
70 99.0%
69 98.9%
New Table
73 99.4%
72 99.3%
71 99.2% Scored as well as or better than 99.2% of the people
70 99.1%
69 99.0%
Your percentile value shown in the table has risen by 0.1%
At the high end, each row changes by less than 0.1%. So there’s percentile inflation, but it is very small.
However, in the center of the table, moving up one row could increase the apparent percentage by several points. It is much more of an concern there.
This isn’t much of an issue for us in trying to determine cutoffs.
The user vs national is more questionable.