CAHOST: An Excel Workbook for Facilitating the Johnson-Neyman Technique for Two-Way Interactions in Multiple Regression

Abstract

When using multiple regression, researchers frequently wish to explore how the relationship between two variables is moderated by another variable; this is termed an interaction. Historically, two approaches have been used to probe interactions: the pick-a-point approach and the Johnson-Neyman (JN) technique. The pick-a-point approach has limitations that can be avoided using the JN technique. Currently, the software available for implementing the JN technique and creating corresponding figures lacks several desirable features–most notably, ease of use and figure quality. To fill this gap in the literature, we offer a free Microsoft Excel 2013 workbook, CAHOST (a concatenation of the first two letters of the authors' last names), that allows the user to seamlessly create publication-ready figures of the results of the JN technique.

Uncovering the Moderating Role of Grit and Gender in the Association between Teacher Autonomy Support and Social Competence among Chinese Undergraduate Students

“…Although simple slope analysis is widely used to probe interactions, such an approach has been criticized, as identifying representative values of the moderator is subject to arbitrary guidelines for determining the values [ 66 ]. To compensate for this limitation, we also employed the Johnson–Neyman technique, which identifies regions of the significance of moderator values [ 67 ]. In these analyses, the significance level was interpreted at p < 0.05, and the 95% confidence intervals did not contain zero.…”

“Big brother is watching you”: surveillance via technology undermines employees’ learning and voice behavior during digital transformation

“…We chose the Johnson-Neyman technique (Johnson and Neyman 1936)-a socalled floodlight analysis-to display this significant interaction effect, as shown in Fig. 2 (using the CAHOST tool by Carden et al 2017). This technique identifies "regions in the range of the moderator variable where the effect of the focal predictor on the outcome is statistically significant and not significant" (Matthes et al 2010, p. 93).…”

The Luck of the Draw: How Attributions for Poverty Shape Support for Economic Inequality

“…Lastly, I predicted that there would be an interaction such that the observed effects on support for economic inequality and redistribution would be largest in participants who displayed the greatest increases in perceived knowledge of the causes of human behaviour (hypothesis 4.1) As noted in the pre-registration, I tested this hypothesis by implementing the Johnson-Neyman approach as outlined in Carden, Holtzman, and Strube [12]. Firstly, I ran a simple linear regression and found no interaction between condition and perceived knowledge of the causes of human behavior (b = .01, p = .96).…”

“…Firstly, I ran a simple linear regression and found no interaction between condition and perceived knowledge of the causes of human behavior (b = .01, p = .96). I then used the CAHOST excel software [12] to visually explore the interaction. The advantage of this excel software is that it allows me to explore the effect of condition on support for inequality across all levels of perceived knowledge of the causes of human behaviour, as opposed to splitting the data into high and low values of the moderator (Figure 5.1).…”