Generally speaking preXL2007 you are looking at SUMPRODUCT here, eg:
However this will prove expensive in performance terms if you have lots of data and you're generally speaking better off reverting to a Pivot Table based approach...
That said given your present layout of yes/no/na options and 59 question columns to surmise this is not going to work very well either... you would perhaps be better off revising your underlying table such that it works along the lines of a standard data table (that is to say Q becomes a field and you have one row of data per response).
In the attached I have gone through the process of reverting your initial data table into a more traditional layout via a multi consolidation PT process, the revised layout will make subsequent analysis far easier and more efficient and can be used as basis for a more traditional Pivot Table with ease.
This one-off task of converting the original data into the revised layout via transposition etc may seem long winded but I would say that if you're doing a decent amount of analysis it's likely to prove beneficial long-term.
Bookmarks