The COVID-19 pandemic has led to a globally unprecedented decline in transit ridership. This paper leveraged the 20-years daily transit ridership data in Chicago to infer the impact of COVID-19 on ridership using the Bayesian structural time series model, controlling confounding effects of trend, seasonality, holiday, and weather. A partial least square regression was then employed to examine the relationships between the impact of ridership and various explanatory factors. Results suggested: (1) COVID-19 pandemic exerted significant effects on 95% of transit stations, leading to an average 72.4% drop in ridership. (2) Ridership declined more in regions with more commercial lands and higher percentages of white, educated, and high-income individuals. (3) Regions with more jobs in trade, transportation, and utility sectors presented smaller declines. (4) Regions with more COVID-19 cases/deaths presented smaller declines in transit ridership. Findings provide a timely understanding of the significantly reduced ridership during the pandemic and help transit agencies adjust services across different socioeconomic groups and space to better constrain virus transmission.