Reading the article, I fail to understand CASA's "Sampling Methodology" argument because they don't appear to have explained what their target market is nor how their sampling produces an unbiased assessment. CASA is however correct in its assessment that AOPA interference will bugger up the survey because angry old farts will self select to respond at a higher rate than satisfied customers. This is why most media polls are BS.
On Government consulting work it is possible to use the services of The Australian Bureau of Statistics (ABS) to help produce a representative sample. Give them the database of ARN holders along with your questionnaire and pay their fees. They can generate a random sample, post it out, de-identify the returns and give you your results. Simples. I've done it that way years ago.