<p>The traditional journal impact factor neglects the time difference between paper publication and citation, as well as the impact of prior information on journal quality when classifying and ranking journals. To address these issues, this paper proposes a Bayesian model for journal quality parameters. We develop a new journal impact factor and obtain Bayesian estimates of journal quality parameters. Furthermore, in cases where both the sample distribution and prior distribution of journal citation data are unknown, we employ the idea of credibility theory to confine the estimation of journal quality parameters to a linear function of the samples, and obtain the optimal linear Bayesian estimation of journal quality parameters under the criterion of minimum mean square error. The results indicate that the optimal linear Bayesian estimation of journal quality parameters can be expressed in the weighted average form of new impact factor and aggregated estimate, with the weight satisfying the property of credibility. Utilizing citation data from various journals, we obtain estimates for hyperparameters and empirical Bayesian estimates of journal quality parameters. Subsequently, we show the performance of the new estimators by comparing it with the traditional journal impact factor through numerical simulation. Finally, based on actual citation data from ten journals, we present the results, comparing traditional impact factors with the proposed new impact factors, Bayesian estimation, and credibility estimation.</p>