Methodological considerations re: Toxicological analysis of George’s marvellous medicine: literature review
Dear Editor,
I read with interest 'Toxicological analysis of George’s marvellous medicine: literature review', by Johnson and Davies. (see: BMJ 2020;371:m4467). Their investigation is not only comprehensive, but also timely, and of global importance.
With regret, I must question one aspect of their statistical analysis.
Where Johnson and Davies introduce the concept of 'rainbow juice', they assert that there are a total of '9 ingredients that have a colour in their names'. When I conducted my own analysis of the open-source data provided, I coded 10 such ingredients. I compared my codes to Johnson and Davies', and determined that they have omitted 'Superwhite washing powder' as an ingredient containing a colour. Consequently, were one to add 'superwhite washing powder' and eight other ingredients to the pan, one would not be awarded a gold 'rainbow juice' achievement badge for adding 'only the 9 ingredients that have a colour in their names'.
This brings me to two relevant methodological considerations. The first is that the authors discounted 'Superwhite washing powder' because they consider white to be a shade, not a colour. However, this assertion is refuted by their inclusion of 'Black peppercorn', which, were shades a consideration, would likely fall into the same category. The second methodological consideration is predicated upon compound words; 'Superwhite washing powder' is distinct from other elements in this category, such as 'Purple pills for hoarse horses' and 'Golden gloss shampoo' due to the presence of a morpheme (more specifically, a prefix), 'super'. However, this assertion is inconsistent with the authors' position on 'Thick yellowish liquid for cows', which has been classified as an ingredient that has a colour in its name in spite of the inclusion of a morpheme (in this case, a suffix).
I thank the authors for their otherwise most excellent analysis, and for their contributions to this important field of research.
Rapid Response:
Methodological considerations re: Toxicological analysis of George’s marvellous medicine: literature review
Dear Editor,
I read with interest 'Toxicological analysis of George’s marvellous medicine: literature review', by Johnson and Davies. (see: BMJ 2020;371:m4467). Their investigation is not only comprehensive, but also timely, and of global importance.
With regret, I must question one aspect of their statistical analysis.
Where Johnson and Davies introduce the concept of 'rainbow juice', they assert that there are a total of '9 ingredients that have a colour in their names'. When I conducted my own analysis of the open-source data provided, I coded 10 such ingredients. I compared my codes to Johnson and Davies', and determined that they have omitted 'Superwhite washing powder' as an ingredient containing a colour. Consequently, were one to add 'superwhite washing powder' and eight other ingredients to the pan, one would not be awarded a gold 'rainbow juice' achievement badge for adding 'only the 9 ingredients that have a colour in their names'.
This brings me to two relevant methodological considerations. The first is that the authors discounted 'Superwhite washing powder' because they consider white to be a shade, not a colour. However, this assertion is refuted by their inclusion of 'Black peppercorn', which, were shades a consideration, would likely fall into the same category. The second methodological consideration is predicated upon compound words; 'Superwhite washing powder' is distinct from other elements in this category, such as 'Purple pills for hoarse horses' and 'Golden gloss shampoo' due to the presence of a morpheme (more specifically, a prefix), 'super'. However, this assertion is inconsistent with the authors' position on 'Thick yellowish liquid for cows', which has been classified as an ingredient that has a colour in its name in spite of the inclusion of a morpheme (in this case, a suffix).
I thank the authors for their otherwise most excellent analysis, and for their contributions to this important field of research.
Rebecca E Glover
Competing interests: No competing interests