In person, online, and up close: the cross‐contextual consistency of expressive accuracy