after reading DiCiccio and Efron's 1996 paper about the bias-corrected and accelerated bootstrap, I suspect line 51 to be erroneous.
U <- (sims - 1) * (mean(theta, na.rm=TRUE) - theta)
In the original paper (equation 6.7), it's not the i-th element in theta that has to be substracted from the mean, but rather the mean of theta when excluding the i-th element. If I am not mistaken, in this specific application, this error leads to the acceleration constant a having the wrong sign.
Hello there,
after reading DiCiccio and Efron's 1996 paper about the bias-corrected and accelerated bootstrap, I suspect line 51 to be erroneous.
U <- (sims - 1) * (mean(theta, na.rm=TRUE) - theta)In the original paper (equation 6.7), it's not the i-th element in theta that has to be substracted from the mean, but rather the mean of theta when excluding the i-th element. If I am not mistaken, in this specific application, this error leads to the acceleration constant
ahaving the wrong sign.With kind regards, stepls