I suppose it's a matter of preference, although making the "lower half" all zero is sensible.
P(A and B are first two) = P(B and A are first two)
The way the sample solution excludes the double counting and only shows the unique entries. You can use the fact that they sum to 1 as a check. Alternatively you'll have everything twice so it sums to 2, not quite as neat and more cluttered.
Re hardcoding.
Almost certainly you'll have to expand your spreadsheet to handle the extra data, and it's far better (for reasons of checking, and using the model for something else later) that you don't hard code anything.
That being said, the spreadsheet doesn't carry too many marks and the audit trail and summary (where the marks are) depend on a working answer (of some sort, whether right or wrong)
So you might lose some audit marks, but if you can't do it otherwise, you can hard code it, finish your documents (and get marks to pass) and come back to fix it later.
Beware though - hardcoding sometimes takes longer and its easier to make errors, and harder for reviewer to follow.
Last edited by a moderator: Jan 7, 2010