You are studying whether longer prison sentences lead to fewer violent crimes (assault, murder, etc.) committed after being released from prison. You regress the number of violent post-release crimes (crimes;) on the number of years in prison (prison_years;), years since release (post_release_years;), and a constant: crimes; = Bo+ B1 prison_years; + B2 post_release_years; + 6; Consider another variable indicating the person's number of prior criminal offences (prior arrests, convictions, etc). Assume that those who have more prior offenses are more likely to commit crimes in the future AND are given longer sentences. How does excluding this variable from the model affect your estimate of the coefficient on prison sentence length? O A. This would cause upward bias (the sign of the bias will be positive). O B. This would cause downward bias (the sign of the bias will be negative). O C, This would cause no bias (the sign of the bias will be zero). O D. This could cause bias, but we do not know if it would be upward or downward bias (the sign of the bias could be positive or negative)