The effect of hypothesis framing, prior expectation and professional trait scepticism on experts’ and novices’ evidence selected in a complex task