Why do we have a significant reduction of journal full text retrievals when we compare Counter 4 JR1 Report vs. Counter 5 tr_j3 (Journal Requests by Access Type) of Title report?

Many of the journals included in Counter 4 JR1 report have been re-classified based on the data type they provide. e.g. Dissertations and newspapers. 
If you generate a Title Master report and filter out "Books" and "Multimedia" data types, you would find that the total number of requests is more comparable to that of JR1 in Counter 4.