lots of ppl seem very sure that my high-sample-size survey is bunk because the responses are anon and how do u verify?
Well, u can't verify for sure, but there's some ways to check/calibrate how accurate the responses are. 1/
10
4
135
Conversation
Replying to
They're bunk because there's a major sample bias. Not criticising just a reality of the data collection process you chose
2
5
Replying to
major sample bias does not mean bunk, this is a super common misconception
also: i can see the differences between how ppl answer questions when they come from different sources, so i can *check* the level of bias
1
20
Bias does actually mean bunk. Statisticians work hard to remove bias from studies. Internet polls are notoriously biased and usually not representative. Consider this: do you often see men and women represented equally in your polls? Are your results representative of population?
2
1
In other words, do your followers accurately represent the population you're trying to study? If so, then bias may be OK, but you have to understand you're studying your followers and not the population at large.
1
1
1
in my tweet thread, if u read, u can see that i found that my data is consistent with the actual populations im surveying
The actual population you’re surveying is the population of people who would possibly stumble upon your survey. Not sure you’re claiming it’s broader, but if so, it’s 100% bunk.
Any consistency with other studies just means those two different populations are similar.
1
5
Just because they are consistent along one dimension, it doesn't mean they are consistent along all dimensions. Implicit in your assumption is that a population can be identified by (e.g.) the correlation between being assaulted as a child and as an adult.
2
I read, but correlation != causation. Using this as evidence your surveys aren't biased is tantamount to using an incorrect method to come up with the right answer. Sure, you got the right answer, but how did you arrive there?