Corrections to Nelson (2023): DP_norm and D_KLnorm are Not Wrong on Pi at All

ABSTRACT

This paper mainly discusses two computational errors in Nelson (2023), which demonstrate that part of his conclusions regarding two dispersion measures are flawed.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1. In this very specific case, where all digit bins are equally large – ¹/₁₀₀₀ – the normalization formulae from Gries (2008) and Lijffijt and Gries (2012) amount to the same results; the R code I give here is the one that would work in the more general case where not all corpus files are equally large.

2. Also, the paper of mine he cites for D_KLnorm as a dispersion measure used the binary log, not the natural log Nelson’s equation uses.