Where have they said this? My understanding is that they consider
test variability between re-taking WR/WIN tests as 3 words*. See bottom right of slide below.
View attachment 43362
In other words, if your score improves by 3 words, this still falls within the parameters of chance, not meaningful improvement.
*For the record, there is actually a lot of debate in the field as to what one should consider as the threshold for test variability. Thornton and Raffin, who Frequency Therapeutics have made previous reference to when describing their results, attempted to answer this question some years ago. For those who may be interested, check
this link out and scroll down to the subheading "
When is a difference really a difference?"