Honest question on polling

[Tsing Tao]

i genuinely curious about the methodology in polling political races. if anyone has any insight, i'd appreciate it if they could share.

i look at a particular election race in my home state. i'm going to use fictitious numbers as an example. it'll show candidate A winning by 9 points in the latest poll, and the talking heads will yammer on about how this is a comfortable lead. the problem is, i ask myself, can the poll actually properly determine the reality of who has what lead and how much it is? i want candidate B (in the above example). no one has ever asked me who i am voting for, no one has ever polled me. no one has ever polled anyone i know where i live, work or spend my spare time. all of the people i know are in a similar state of mind, and want candidate B. but none of us have ever been asked. are we the "stealth" voters who swing elections at the last minute, causing the talking heads to say "Candidate B came out of no where in a surprising victory" or are we just another group that isn't counted - and there's likely a large group of candidate A people out there thinking the same thing?

 

[bigdavediode]

If you're into trading, you should have a basic statistical background and understand sampling a normal distribution.

 

[Tsing Tao]

i am, and do. but that's not the question, although perhaps the same question can be applied to trading in terms of how many times traders improperly sample. but leaving trading out of it, it would seem that more and more polls predict final results incorrectly.

regardless, it would appear this forum lacks the knowledge to discuss it, just as i lack the knowledge to answer my own question.