Abstract: How to use prior information effectively and reasonably in the Bayesian method is a key issue in small sample evaluation. For the evaluation of guidance precision of missiles, the article presents a new approach to construct the prior distribution in terms of prior data, and put forward a hybrid prior distribution after introducing the concept of prior data credibility. The hybrid distribution can overcome effectively the domination of prior information over current data through analyzing the posterior estimation of interesting parameters, while the credibility of the prior data can be obtained at the same time from posterior marginal distribution. Some basic principles for the application of prior data to engineering practice are introduced through calculating the mean square error (MSE) of posterior estimation.

Key words: small sample, Bayes method, hybrid prior distribution, credibility, estimation