When you've normally had 50% and higher response rates and new technology drops you to a 9% response rate at best, you really no longer have something that is under statistical control and you can't say it's reliable.
The fact you would even begin to argue with me on the point is a demonstration you don't know anything about statistical sampling, whether for opinion, profiles or quality control
If you select the right kind of sample it doesn’t matter if the response rate is 9%. It matters what comprises the 9%. The fact is that the legitimate polls predicted the outcome of the election fairly accurately. Somehow, they managed with the 9% response rate.