Statistical Approaches for Highly Skewed Data: Evaluating Relations between Maltreatment and Young Adults’ Non-Suicidal Self-injury

Abstract

Objective: Clinical phenomena often feature skewed distributions with an overabundance of zeros. Unfortunately, empirical methods for dealing with this violation of distributional assumptions underlying regression are typically discussed in statistical journals with limited translation to applied researchers. Therefore, this investigation compared statistical approaches for addressing highly skewed data as applied to the evaluation of relations between child maltreatment and non-suicidal self-injury (NSSI). Method: College students (N = 2,651; 64.2% female; 85.2% nonwhite) completed the Child Abuse and Trauma Scale and the Functional Assessment of Self-Mutilation. Statistical models were applied to cross-sectional data to provide illustrative comparisons across predictions to a) raw, highly skewed NSSI outcomes, b) natural log, square-root, and inverse NSSI transformations to reduce skew, c) zero-inflated Poisson (ZIP) and negativebinomial zero-inflated (NBZI) regression models to account for both disproportionate zeros and skewness in the NSSI data, and d) the skew-t distribution to model NSSI skewness. Results: Child maltreatment was significantly and positively related to NSSI frequency in the raw, transformation, and zero-inflated models, but this relation was negative in the skew-t model. Conclusions: These findings highlight the importance of using zero-inflated models rather than transformation approaches to address data skew. Moreover, whereas the skew-t distribution has been used to model skewed non-clinical data, this study suggests that the skew-t approach may not be well-suited to address skewed clinical data.

Emotion Dysregulation and Eating Disorder Symptoms: Examining Distinct Associations and Interactions in Adolescents

“…Using this regression, the occurrence of the behavior (zero-inflated part) and the frequency of the behavior for those estimated to not represent excess zeros (Poisson part) are examined separately in the same model. Using zeroinflated models is recommended for use with highly skewed clinical data (Gonzalez-Blanks et al, 2020), including eating disorder symptoms (Schaumberg et al, 2018).…”

Bullying Victimization: Investigating the Unique Contribution of Homophobic Bias on Adolescent Non-suicidal Self-injury and the Buffering Role of School Support

“…To examine the associations of traditional and homophobic victimization with the probability of engaging in NSSI, as well as the frequency of NSSI once initiated, we conducted zero-inflated Poisson (ZIP) regression analyses in Mplus 8 (Muthén & Muthén, 2017), controlling for the non-independence of observations. This approach has been demonstrated to be particularly suitable for analyzing highly-skewed variables with an overabundance of zeros (Lambert, 1992), such as the frequency of NSSI in community samples (Gonzalez-blanks et al, 2020). Descriptive statistics to check for the presence of overdispersion and zero-inflation and comparisons with competing models (i.e., negative binomial and zero-inflated negative binomial models) are reported in supplementary materials.…”

Childhood maltreatment, psychiatric symptoms, and suicidal thoughts among adolescents receiving substance use treatment services

“…Competing indirect associations were evaluated using the distribution of the product approach, using bias‐corrected bootstrap (2,000 replications) confidence intervals (Preacher & Hayes, 2008). This method was selected to maximize statistical power for detecting significant mediation effects and to address data non‐normality which is common in studies of self‐report health risk behaviors (Gonzalez‐Blanks, Bridgewater, & Yates, 2020).…”