Conversation

Aella
@Aella_Girl
The comments here are full of poor understanding of statistics, in case ppl wanted to go help and e.g. explain what a "bin" means
Quote Tweet
Aella
@Aella_Girl
Escort monthly income by BMI, from my escort survey I've had open for a while. Sample size is still kinda low, like n=~30 per bin, and this is sampling escorts online enough to take my survey, but there still looks like a trend I'd expect. Interesting to see the degree of impact!
Show this thread
Image
28
120
LysanderSpoonedher
@spoonedher
Replying to
@Aella_Girl
More valuable information would be the standard deviation of how much is earned. If you know how to use seaborn you can easily graph correlation matrices to identify which elements contribute more to earnings
2
2
Show replies
Bob English (self-parody)
@uneasilybemused
Replying to
@Aella_Girl
I’m not sure what randomness you’re trying to measure here. If these were random samples of a particular population, error bars would reflect the range of likely results from the full population. In a self-selected sample, it’s not clear what the “full”population actually is.
1
mikko
@MikkoHaavisto_
Replying to
@Aella_Girl
Sample size of 30 per bin is more than enough, i don't think you are limited by that at all. More sample would be wasted at this simple level of analysis.
Murray Gothbard
@MurrayGothbard
Replying to
@Aella_Girl
You have to specify what level of confidence the CI error bars are. When the sample is high enough (30 is good) I prefer to use standard error for the error bars because it allows you to eyeball significance (non overlapping bars means p < 0.05)
Erd Tird Mans
@theErdTirdMans
Replying to
@Aella_Girl
I'm still just surprised the peak isn't 21-22. I'd assume if you're paying for it, you'd be almost exclusively seeking the most optimally sexually desirable options (i.e. big boobs\butt) which should tip the BMIs a bit, but maybe I don't understand why\who solicits these services