*2.7. Statistical Analysis*

All analyses were conducted using SAS Version 9.4 (SAS Institute Inc., Cary, NC, USA). Univariate and bivariate methods were used to estimate counts and percentages. Bivariate associations in contingency tables were tested either using Fisher's exact tests or Fisher's exact tests with Freeman and Halton's adaptations for non-standard RowxColumn tables [36]. Monte Carlo estimation with 10,000,000 samples was used to calculate Fisher–Freeman–Halton statistics for tables within industry. Statistical significance was determined based on an alpha of 0.05 with Bonferroni corrections dependent on the number of questions in a given worker well-being domain or survey topic. Estimates were considered marginally significant if they met the alpha criteria of 0.05 but were no longer significant after Bonferroni correction. Analyses and categorization of questions about demographics, employment status, and during pregnancy included not reported as a category to account for nonresponse. For Total Worker Health well-being questions, directed towards people who were employed, on maternity leave, or recently employed, nonresponse was not included.
