Some variations on the standard theoretical models for the h-index: A comparative analysis

TL;DR

This paper compares various statistical models for estimating the h-index, highlighting the limitations of traditional assumptions and proposing more realistic parameter specifications for improved prediction accuracy.

Contribution

It introduces a statistical modeling approach to evaluate and improve the estimation of the h-index, challenging previous fixed-parameter assumptions.

Findings

Traditional models often rely on unrealistic parameter assumptions.

Alternative distributions provide better fit for bibliometric data.

The proposed models improve h-index prediction accuracy.

Abstract

There are various mathematical models proposed in the recent literature for estimating the h-index through bibliometric measures, such as number of articles (P) and citations received (C). These models have been previously empirically tested assuming a mathematical model and predetermining the models parameter values at some fixed constant. Here, by adopting a statistical modelling view I investigate alternative distributions commonly used for this type of point data. I also show that the typical assumptions for the parameters of the h-index mathematical models in such representations are not always realistic, with more suitable specifications being favorable. Prediction of the hindex is also demonstrated.