I’d suggest it is reasonable to suppose that someone with a grievance is more likely to answer the survey than someone whose only comment is that they are quite content. If that is correct one could suppose that the survey, if it has a bias, is more likely to be biased to underestimate the numbers of contented tenants.