Perhaps you could pad the data source with a zero then assign the symbol and then remove the padded data. I'm not that familiar with that aspect of tableau but I would say its worth a shot. You could test it by adding the padded data a second time to ensure that when a zero does appear in your data the assigned symbol stays.
In a calculation, this is not usually possible. Your two options are:
1. Create, then remove dummy data.
2. Adjust your calculation to throw the other results. This can get tedious with multiple values, but in this instance, you could just rewrite your condition to change = to <>. Then adjust the shape of "shape2" and flip your comparator back.
You are right Ben, it does get tedious, hence the question . The example I have in the question is a trivial one, I actually have to deal with case statements with multiple conditions. I wish Tableau had a way of assigning shapes based on the conditions in the statements rather than on the actual data.
I realized I neglected to ask if your [measure1] was aggregated, and just assumed it was. If this calculation is not aggregated, then you can, in fact, assign the values default shapes. If you are dealing with aggregates, however, this is currently not possible.