"Since the t-distribution is like the standard normal distribution but with a higher variance (smaller peak and fatter tails), this would seem to suggest that the sample standard deviation s systematically underestimates the population standard deviation, making it in fact a biased estimator."