*4.3. Statistical Analyses*

For categorical data (building size, dichotomized threshold variables) *Legionella* positivity was compared using the Chi square test. Logistic models were used to evaluate *Legionella* positivity (presence/absence) and continuous variables (HPC, temperature, pH, free chlorine, calcium, magnesium, zinc, manganese, copper, and total organic carbon). A receiver operating characteristic (ROC) curve and the area under the curve (AUC) were generated to evaluate the utility of hot water return line temperature (dichotomized by recommended threshold) for predicting *Legionella* positivity. All statistics were performed using Stata version 13.0 (Stata Corp, College Station, TX, USA). A linear regression, and resulting R<sup>2</sup> value, was generated to evaluate the utility of HPC concentration for predicting *Legionella* positivity using Microsoft Excel.
