from my escort survey, self-reported age and estimated monthly earnings.
sample size kinda small, each bin is n=~40
9
5
134
Conversation
Replying to
As someone with a degree in economics and data science, 40 is TOO small a sample size for the data you are trying to measure considering how large of potential sample pool there actually is (ie- millions of SWers)
5
7
Since I have a tremor disorder and not enough characters to explain on Twitter, here’s an article that easily explains you don’t just do a survey on social media and call it a day.:
The following media includes potentially sensitive content. Change settings
View
1
well, while i agree with you in theory, i think i have a few issues with you saying all this. (1) the bins have an n=40, but according to the survey the overall n=283. second, SWer N is difficult to determine so an appropriate SWer n is also then difficult to determine 1/?
1
1
there are important methodological issues happening here (e.g., selection bias and undercoverage), but imo the author isn’t claiming causality or generalizability (tho i could be wrong); rather it seems like they’re explaining associations which is different than causality 2/?
2
1
but a lot of these problems could be solved with a more robust demographics section to calculate appropriate weights. but again, the author seems more concerned with something like “there is a trend for SWers on twitter” rather than “this is a causal factor in outcomes”
1
1
Hahaha so yeah… we are essentially saying the same thing (in real time) to each other 😁
1
you're not, though? you're acting like i'm making claims far beyond what i actually am. i'm sharing results from a survey i did, i'm not saying this is perfectly representative of all escorts.
also i did estimate of total escort # here
2
2
Yea but it’s an irrelevant survey because you have an extremely skewed sample. I’d like to see the percentages of where your answers came from, because I bet they mostly came from Twitter which isn’t where most SWers advertise.
1
why do people just *assume* my data came from the worst possible places instead of asking me?
I have over a million followers across six different platforms and recruited participants from all of them.
and I collected data on where my respondents advertised so i can see!