The standardized coefficients for spanking rarely differed by more than.03 from the parallel coefficient for any alternative disciplinary tactic after initial differences on antisocial behavior were controlled for statistically.
Each endogenous variable, criminal risk, substance use, and sexual risk behaviors, were controlled for participant's age, gender, and race (data not shown).
The addictive behavior was controlled by Fagerström test for nicotine dependence questionnaire [ 16].
Once the variance in suicidal behavior due to age was controlled for, optimism, social support from friends, and social support from family significantly negatively predicted variance in participants' suicidal behavior, Δ R = .33, change in F 3, 122) = 19.72, p < .001.001
The association may indicate a propensity for risk behavior that could not be controlled for in the analysis.
All other analyses in Table 5 used the 6-item antisocial subscale as the outcome variable, varying only the comprehensiveness with which initial differences in antisocial behavior or externalizing problems were controlled for statistically.
Gender and age are the factors most strongly correlated with suicide-related behavior, so these variables were controlled for in all of the regression models; however, age and years of working as a psychiatrist were collinear, so only the latter variable was included in the analyses.
Nonetheless, a meta-analysis of 45 studies on the association of sexual abuse and self-injury only found a relatively weak relationship between self-injury and sexual abuse indicating that sexual abuse ceased to explain the variance in self-injurious behavior if the studies were controlled for other psychiatric risk factors [ 54].
Additionally, previous studies have reported the existence of many other variables connected to self-care behaviors that were not were controlled for in the present study.
