Most variables in a Quanvert dataset contain "Base" as a real category. This means that any question that is answered virtually always contains two categories, namely, {Base} and the answer that the respondent gave, for example, {Male}. As a result, the filter "gender={Male}" normally returns False for a Quanvert dataset because all responses will be {Base, Male}, {Base, Female}, or null. The expression that correctly selects only male respondents is:

A similar issue arises with questions that allow multiple responses, such as "Which soft drinks do you normally drink?". If you want to know how many respondents drink tea and nothing else then the expression you should use is "Usual={Base,Tea}" because Base will always be present in the case data if any answer is given.