When you sample a population (hunters, in this case), you want the sample to be representative of the whole. One of the assumptions in such sampling is randomness in selection.
Hunter compliance in questionnaires is poor. One could send all questionnaires to all hunters. The return is unlikely to be better than 20%, and even with significant incentives like prizes, it is hard to get above 30%. I suspect that BC publishes hunter questionnaire stats. The amount of work generated by compulsory response is prohibitive.
With severely restricted budgets game managers have too few tools at their disposal.