Psychophysical Threshold Estimates in Logistic Regression Using the Bootstrap Resampling

We propose the non-parametric bootstrap resampling algorithm for the problem of psychophysical threshold estimates. We use the logistic regression with guessing rate and the log-likelihood ratio test statistics of two samples for testing the hypothesis by using the bootstrap resampling. We apply our algorithm to the visual acuity test, and show that the bootstrap resampling is useful for the problem of the two-sample test when the numbers of observations are not identical between the two samples. We also propose the bootstrap algorithm for one-sample testing to certify the values of parameters and threshold obtained by logistic regression.