Web Analytics

Comments From Peer Reviewers

I immensely liked this manuscript: it is a rare example of a scientific enterprise in which the authors, by means of a crystalline while rigorous style, drive the reader into the meaning of the proposed statistical techniques with a continuous reference to their clinical counterpart. The development of a sort of ‘supervised PCA’ able to deal with date sets plagued by missing data is a crucially important achievement that gives a direct solution to the ‘explainability’ issue that affects a great part of machine intelligence community. It is a solution that fosters a continuous exchange between experts in the field and data analysts and is a bright demonstration of how to develop a truly interdisciplinary science. It is not by chance that the same attitude was at the basis of the rise of modern statistics in the forst decades of the XX century.

This is an outstandingly innovative study. I think it will be of considerable interest to readers of the journal. I read it with great interest. Its originality is sky high, applying matrix factorisation methods to dyslexia diagnosis for the first time. It is also very strong methodologically – consider the time taken to undertake the 828 interviews on which the analysis is based!


I am very impressed by the very rare combination of the methodological rigour of the work undertaken together with the high technical level of understanding and analysis shown. Use of the entropy concept is of course a plus for readers of this journal! It is worth noting that this methodology could be used with benefit for a whole range of psychometric tests, with the Wechsler scales (IQ) being a natural one to explore since there is great interest in interpretation of such tests..


Information from Noise: Measuring
Dyslexia Risk Using Rasch-like
Matrix Factorization with a
Procedure for Equating Instruments