On 2017 Nov 23, Franck Ramus commented:

There is at least one more problem with this study that Mark Seidenberg did not point out in his critique (http://languagelog.ldc.upenn.edu/nll/?p=35144). It is that the two groups were tested sequentially: first the 30 controls, then the 30 dyslexics. This is generally avoided in experimental psychology (and biology as well), because this is so prone to experimenter bias (unwittingly influencing differentially the members of each group), as well as to apparatus bias (e.g., apparatus parameters changing at some point, making these parameters confounded with group membership).

On 2017 Nov 23, Franck Ramus commented:

There is at least one more problem with this study that Mark Seidenberg did not point out in his critique (http://languagelog.ldc.upenn.edu/nll/?p=35144). It is that the two groups were tested sequentially: first the 30 controls, then the 30 dyslexics. This is generally avoided in experimental psychology (and biology as well), because this is so prone to experimenter bias (unwittingly influencing differentially the members of each group), as well as to apparatus bias (e.g., apparatus parameters changing at some point, making these parameters confounded with group membership).