Data were presented as mean and/or median ± standard deviation of raw and/or natural log-transformed oxytocin levels by diagnosis (PWS or unrelated siblings). Natural log-transformed oxytocin levels were utilized to meet normality criteria required for analysis using simple analysis of variance (ANOVA) and general linear model adjusting for diagnosis, gender, age and body mass index (BMI). Data falling below detection limits of the Milliplex assay were replaced with one half of the minimum detectable level of oxytocin (4.675 pg/ml) as reported in previous studies using this methodology [Ashwood et al., 2010 ; Manzardo et al., 2012 (link)]. Final transformed data met statistical criteria for the assumption of normality with equal variance and near linear residual plots. The frequency of oxytocin levels below detection limits were statistically analyzed by PWS diagnosis and gender using Fisher Exact Test. Statistical analyses and descriptive statistics were generated with SAS statistical analysis software version 0.4 (SAS Inc., Cary, NC) with P-values <0.05 were considered significant.