The Voter Participation percentage has been useful in identifying potential problems with participation or node failure within the system. However, it has also yields some false negatives. The primary reason for this is due to the number of blocks we are tracking compared to the number of nodes we have. We only the past 50 blocks of voting history in general. As a result, we are missing some of the picture, as we don't see the full leader rotation and how that might affect the participation.
@Ancient123 has suggested and requested that we track two full rotations of history instead for a more holistic view and understanding of a node's overall health.