Detect Aberrant Behavior in Test Data

Detect several types of aberrant behavior, including answer copying, answer similarity, change point, nonparametric misfit, parametric misfit, preknowledge, rapid guessing, and test tampering.

aberrance

The aberrance package contains a collection of functions for detecting several types of aberrant behavior, including:

• Answer copying, using statistics such as the $\omega$ statistic (Wollack, 1997).
• Answer similarity, using statistics such as the $GBT$ statistic (van der Linden & Sotaridona, 2006) and the $M4$ statistic (Maynes, 2014).
• Change point, using statistics such as the likelihood ratio test-based statistic, the score test-based statistic, and the Wald test-based statistic (Shao et al., 2016; Sinharay, 2016; Tu et al., 2023).
• Nonparametric misfit, using statistics such as the $ZU3$ statistic (van der Flier, 1982) and the $H^T$ statistic (Sijtsma, 1986).
• Parametric misfit, using statistics such as the standardized log-likelihood statistic (Drasgow et al., 1985) and its various corrections (Bedrick, 1997; Gorney et al., 2024; Molenaar & Hoijtink, 1990; Snijders, 2001).
• Preknowledge, using statistics such as the signed likelihood ratio test statistic (Sinharay, 2017).
• Rapid guessing, using methods such as the custom threshold method (Wise et al., 2004; Wise & Kong, 2005), the normative threshold method (Wise & Ma, 2012), and the cumulative proportion correct method (Guo et al., 2016).
• Test tampering, using statistics such as the erasure detection index (Wollack et al., 2015; Wollack & Eckerly, 2017) and its corrected versions (Sinharay, 2018).

Installation

Install the released version from CRAN:

install.packages("aberrance")

Alternatively, install the development version from GitHub:

# install.packages("devtools")
devtools::install_github("kyliegorney/aberrance")