Since I never use inline CSS, there's little motivation.

Anecdotal evidence is not evidence, and correlation is not cause -- nothing personal, just two epistemic principles I believe in (and with good reason). And another one: experiments with small sample sizes are useless (except if followed by ones with big sample sizes).

But for this example, it hardly matters. Weaning oneself off inline styles is a Good Idea anyway.