I think the correct answer from the choices listed above is the third option. If a statistic used to estimate a parameter is such that the mean of its sampling distribution is equal to the actual value of the parameter, then it is a biased estimator. Hope this answers the question.
