I've seen enough evidence to be sure that there's no way of dressing which drives the risk down to zero.

This seemed too obvious to mention to me. But to put in context of inferential distance, I've seen enough evidence to be sure that there's no way to eat or even act which drives the risk down lower than 1%, let alone getting near zero using only superficial changes in appearance like clothing.

This comes partially from groundbreaking-sounding study results like "overweight women don't actually have statistically-significant lower chances of being abused, even sexually!"¹, which isn't nearly as surprising when you approached the question from "How do rapists select their victims?" or more generally "Who rapes who and why?" instead of the default internal model that translates to "Which women would I (men) want to have sex with?".

¹. Read two studies to that effect years ago, do not remember sources. Strong possibility of cherry-picking / other biases.