Re-Examining the Statistical Methodology and Onomastic Claims of Gregor and Blais' Argument from Name Popularity

TL;DR

This paper critically examines Gregor and Blais's statistical methods and onomastic claims, identifying flaws and proposing improvements, including new data, to better understand name distributions in ancient texts.

Contribution

It exposes flaws in Gregor and Blais's statistical approach, advocates for the chi-squared test, and introduces 87 new onomastic findings to refine historical name distribution analysis.

Findings

Gregor and Blais's first method is statistically invalid.

The chi-squared goodness-of-fit test is more appropriate for their analysis.

87 new onomastic data points are identified for future research.

Abstract

In 2024 Gregor and Blais published a JSNT article using two different statistical methods to conclude, contra Bauckham (2017), that selected Apocryphal texts and the Babylonian Talmud "do not correspond to the distribution among first-century Palestinian Jews statistically significantly worse than the distribution in Gospels-Acts" and "the two corpora paradoxically align better in some respects". In this paper, we show that the first method is statistically invalid, and the second is the wrong tool for the job. This is in alignment with the critique of Van de Weghe and Wilson (2024) and in support of their use of the chi-squared goodness-of-fit test which established name occurrences in the Gospels and Acts, as opposed to Gregor and Blais` uniform, apocryphal, or Talmudic corpora, "fit into their historical context at least as well as those in the works of Josephus". Regarding this historical context, helpful insights are provided by Gregor and Blais regarding potential distortions within the onomastic reference distribution, and this article suggests a way forward, addressing orthographic issues, sample biases, several problems with the implementation of Gregor and Blais` inclusion criteria, and 87 new onomastic finds from ossuaries, ostraca, and documentary papyri that need to be incorporated into the lexicon. While legitimate concerns are raised by Gregor and Blais, several problems with their own onomastic datasets are also discussed.