This study compared the differences in the impact of the COVID-19 pandemic on transit ridership depending on the length of the analysis period. First, we tested the relationship between the forecast period and the robustness of the Bayesian structural time series model using data only for the period prior to the COVID-19 pandemic, and confirmed that extending the analysis period beyond that of previous studies provides a larger estimate of the short-term impact of the COVID-19 pandemic. We also analyzed differences in short-term effects by attribute. Next, the long-term impact of the COVID-19 spread was analyzed by attribute using local linear trend components extracted by component decomposition of Bayesian structural time series analysis. Especially, we newly introduced a method to quantitatively analyze the time series differences in the local trend components of the two attributes using Bayesian structural modeling and confirmed that it is possible to show time-series differences that are difficult to discern by simple comparison.