164
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
this post was submitted on 17 Oct 2023
164 points (100.0% liked)
chapotraphouse
13546 readers
832 users here now
Banned? DM Wmill to appeal.
No anti-nautilism posts. See: Eco-fascism Primer
Gossip posts go in c/gossip. Don't post low-hanging fruit here after it gets removed from c/gossip
founded 3 years ago
MODERATORS
This wasn't a random sample though
Maybe not but I don't see anything in their brief methodology section quoted above to indicate it wasn't a random sample of mobile phone users. What makes you think it wasn't?
A random sample of mobile phone users isn't a random sample, because you're only going to get the people who answer texts from strangers.
It's called "non-response bias" and it's a huge part of the reason political polling doesn't work. It's strong enough to render almost any sample from a phone survey non-representative.
Yeah I understand that there's a difference between the sampled population and the actual population of interest, but you can't discount the results of on account of that unless you can meaningfully show a non-zero covariance between the response variable and likelihood of non-response.
By all means it's a caveat but it doesn't make these results entirely non-informative.
In any case I cited iid conditions to explain why asking all their friends is certain to produce a useless estimate of the population proportion.
They know that asking their friends is useless. They were trying to make a point about sampling bias.
To me they were saying this numbers are totally made up.
If it's not representative, it basically is
A biased estimator is not a meaningless or non-informative estimator.
Look at it this way. When the US wants to prosecute a suspect in federal court, where do they generally hold the trials? In the whitest, most affluent part of Virginia. So the jury, while random, is still taken from a specific pool of people who lean a certain way politically.
If you think polling data isn't politically motivated and influenced by sample location, age, and the way the questions are formulated, you're deceiving yourself. What's that saying? There are small lies, big lies, and statistics.