To motivate people to use bikes for transportation, cities are shifting their focus from constructing isolated bike lanes to building interconnected bike networks. The effectiveness of these networks is measured by their level of connectivity, specifically how easily individuals of all ages and abilities can reach their destinations by bike. However, the quantification of connectivity varies, including methods like graph analysis and destination analysis. Despite significant investments at the network level, few studies have explored the impact of these networks on safety. Moreover, there is a lack of research providing guidance on the most effective method for quantifying connectivity in safety analysis. Our study aims to understand the relationship between safety and various connectivity measurements at the neighborhood level by comparing different connectivity metrics. We calculated three sets of connectivity indices based on: (1) graph analysis of bike infrastructure networks, (2) graph analysis of low-stress street networks, and (3) destination analysis of low-stress street networks. Using a negative binomial regression model, we examined the correlation between bike crashes and connectivity indices across 125 block groups in Santa Barbara and Goleta, California. The results from the three connectivity indices show conflicting associations with bike safety. Our analysis suggests that using graph analysis of low-stress street network is the most effective approach. We conclude that (1) enhancing bike network coverage improves bike safety, but increased network complexity, which disrupt the network, may negate these benefits; (2) better ridership data are needed to account for the induced ridership effect of connectivity and fully understand the benefits of a connected network.