A Simulator of the Degree to Which Random Responding Leads to Biases in the Correlations Between Two Individual Differences
Personality and Individual Differences
Random responding can inflate Type I and Type II error rates (Huang, Liu, & Bowling, 2015b, Journal of Applied Psychology, 100). Type II error inflation often involves certain variables having Invalid Centered Responses And Valid Uncentered Responses (ICRAVUR; pronunciation: /aikreɪvər/). Although Huang et al. (2015b) offer a set of formulas for calculating the expected bias in a correlation when such variables are present, they do not offer a way to simulate the effects. We offer two sets of Monte Carlo simulations of ICRAVUR variables. Study 1 examines the correlation between narcissism and psychopathy—thought to be a large effect. The effect was inflated (by r = 0.16), comparable to what the Huang formulas forecast. Study 2 examines the correlation between secure attachment and self-esteem—thought to be a large effect. The effect was inflated (by r = 0.26), but this time the simulation result was larger than the forecast from the Huang formulas. Thus, our simulator offers a way to test tailored hypotheses about specific variables—sometimes yielding effects more extreme than the Huang formulas. We guide the readers through software, available at the first author's website, allowing for estimating the impact of ICRAVUR variables on any Pearson correlation.
Holtzman, Nicholas S., M. Brent Donnellan.
"A Simulator of the Degree to Which Random Responding Leads to Biases in the Correlations Between Two Individual Differences."
Personality and Individual Differences, 114: 187-192.
doi: 10.1016/j.paid.2017.04.013 source: https://www.sciencedirect.com/science/article/pii/S0191886917302635?via%3Dihub