Obviously some surveys are not useful, and the results of such surveys are scrutinized on the basis of the validity of the data collection methods, the number of respondents, the number of repeat questions, participant fatigue.
For example, a survey which interviews 10,000 brain cancer patients, has 100% respondents and finds that 53% of them felt happier if they listened to their favourite music during treatment, and this was compared to a finding that feeling happier during treatment can improve healing rates...then we have ourselves a valuable piece of information.
I wrote
an article that touches on the lengths survey makers go to to ensure accurate results. You might argue that the data is not 'scientific' but it can be statistically significant.