Quite simply you are wrong.
Using your logic then every survey ever is flawed unless it included the whole population. (And even then it may not be accurate as some respondents may not always tell the truth).
There is significant science and research that stands behind methods of surveys. They can make allowances for bias. They can make allowances for income level, age, political persuasion, etc. And of course they can identify levels of confidence in the results based on the sample size.
If you are disputing these figures, doing so on the basis that you think the survey sample is small is about as poor a reason as you could use. This is a long running, well respected survey, that has produced results backed up by other information, and which nobody with any credibility has disputed.
I put more stock in the professionals and statisticians behind the range of surveys that are taken every day than some bloke on the internet.
But by all means, please produce your own evidence on why crime is not actually falling.