some variable are represented as sparse, whereas they were full in earlier matlab versions running the same program. Why is this happening, and is there a way to fix it short of invoking the full command in every case?

I've recently done a partial update (from Matlab 2008 to Matlab 2010, but the issue likely persists with newer versions), and am experiencing the following errors in code that previously ran cleanly. The output from several subroutines, including scalar, vector and matrix outputs, are returned to the calling routine in sparse representation, whereas they used to be returned as full. This happens with some, but not all, output variables from the same routine. This causes subsequent commands to crash. For example, some of these variables are passed to fprintf, which does not accept sparse-representation input. I could, of course, go back through my code and coerce the full representation with the full command, but this seems inefficient as there are hundreds of instantiations. So, does anybody know either a) what is causing the program to choose sparse output for variables that were formerly full, or b) a global method to turn this off that doesn't require using the full command repeatedly?