b. If the two-sample t test is incorrectly used to analyze paired data, what will happen to the t value if $\rho > 0$ (underestimated or overestimated)? Justify your answer.