Response Bias and Reliability in Verbal Agreement Rating Scales: Does Polarity and Verbalization of the Middle Category Matter?