Analysis with locked segments vs. locked segments excluded

These are two analysis done on the same project in the same version of Trados. The question is why is there a variance between the word counts for unlocked segments when performing both analyses (with and without excluding the locked segments.) Is it because, if you don't clarify, Trados still analyzes locked segments when performing an analysis? I’m trying to explain this to a client. Thank you very much for your time!