Great suggestions Emile. The use of scientifically established samples greatly makes the data credible. The common practice of talking to a few people is biased and not credible. however, a key challenge remains to reach these respondents. Phone surveys are now the common approach, however, the desired information should be collected should be very short because it is not possible to keep a respondent for a long period over the phone. The other means you have proposed are ideal but may be unsuitable for people in rural areas without access to the internet and electricity. SMS is also a great option but the information collected through this method is also very short. in addition, incentives must be provided to enhance participation. This also needs careful consideration so as not to bias response.