To be or not to be associated: power study of four statistical modeling approaches to identify parasite associations in cross-sectional studies