We had been using yesno as an equivalent to boolean, which has thrown off our reporting.
For example, we have a boolean dimension passed_test?, where there are many rows that do not have the input to this test. If we look at the percent of test failing using type yesno, it will give us
(# test failed + # did not take test)/
(#test passed + # test failed + # did not take test)
By treating the null values as no, it throws off the ratio, and we can’t see easily the number of people who have taken the test. We can simply not specify a type to get the required functionality, but then we lose the benefits of a yesno dimension (having the right predefined options on filters for example)
How have people who want to keep null values in their boolean data deal with this? Any help or direction would be appreciated.