Abstract: In this paper, we propose several different bootstrap algorithms, to get approximate confidence intervals for the parameters of an ergodic Markov chain, each of which can have k possible outcomes. Small-sample comparisons are used to select the best intervals for different parameters. An illustrative example which analyzes a farmers' tenure behavior pattern data set of Taiwan is given.
Key words and phrases: Bootstrap, hitting time, Markov chain, stationary distribution, transition probability.