The relationship between risk of bias criteria, research outcomes, and study sponsorship in a cohort of preclinical thiazolidinedione animal studies: a meta‐analysis