Your firm is attempting to determine the effect of sensitivity training on employee behavior. To do so, it has collected data for each employee on the number of times he/she has been reprimanded for insensitive behavior in the past year (Reprimands_i) and whether that employee receives sensitivity training the prior year (Training_i). When regressing Reprimands on Training, the estimated effective training is an average reduction in Reprimands of 0.21.

1. Argue why Training is likely endogenous in this regression.
2. What is the likely sign of the bias in your estimated effect of Training on Reprimands?