Using item‐level analyses to better understand the consequences of partialing procedures: An example using the Dark Triad. Issue 4 (13th November 2019)

• Record Type: Journal Article
• Authors: Vize, Colin E.; Collison, Katherine L.; Miller, Joshua D.; Lynam, Donald R.
• Is Part Of: Journal of personality. Volume 88:Issue 4(2020)
• Page Start: 719
• Page End: 734
• Publication Date: 2019-11-13
• Subjects: Dark Triad -- Machiavellianism -- narcissism -- partialing -- psychopathy; Psychology -- Periodicals; Character -- Periodicals; Personality -- Periodicals; Personality; 155.2
• DOI: 10.1111/jopy.12521
• Languages: English
• ISSNs: 0022-3506

Abstract

Abstract: Objective: Partialing procedures are frequently used in psychological research. The present study sought to further explore the consequences of partialing, focusing on the replicability of partialing‐based results. Method: We used popular measures of the Dark Triad (DT; Machiavellianism, narcissism, and psychopathy) to explore the replicability of partialing procedures. We examined whether the residual content of popular DT scales is similar to the residual content of DT scales derived from separate samples based on relations with individual items from the IPIP‐NEO‐120, allowing for a finer‐grained analysis of residual variable content. Results: Profiles were compared using three sample sizes (Small N = 156–157, Moderate N = 313–314, Large N = 627–628) randomly drawn from a large MTurk sample ( N = 1, 255). There was low convergence between original and residual DT scales within samples. Additionally, results showed that the content of residual Dirty Dozen scales was not similar across samples. Comparable results were found for short Dark Triad‐Machiavellianism, but only in the moderate and small samples. Conclusion: The results indicate that there are important issues that arise when using partialing procedures, including replicability issues surrounding residual variables. Reasons for the observed results are discussed and further research examining the replicability of residual‐based results is recommended.