It’s alleged that the number of people donating zero is large, and more generally I would expect people to round off their donation amounts when reporting. Ages are clearly also quantized. So there may be lots of points on top of one another. Is it easy to jitter them in the plot, or something like that, to avoid this source of visual confusion?
Just eyeballing the charts without any jitter, it kinda looks like Effective Altruists more often report precise donation quantities, while non-EAs round things off in standard ways, producing dramatic orange lines perpendicular to the Y axis and more of a cloud for the blues. Not sure what to make of this, even if true.
Possibly EAs think about their donating more and are thus more likely to be able to look up the exact sums they’ve donated. While non-EAs might be more likely to just donate to some random thing that seems like a nice enough cause, and then forget the details.
If EAs are reporting actual numbers they’ve looked up and non-EAs are reporting their vague estimates of how much they might have given, I would expect non-EAs’ figures to be worse overestimates than EAs’.
As Caplan would say, social desirability bias seems like an issue here. One would not expect people to penalize themselves in such a self-report rather than nudge estimates upwards and make themselves look better in their own mind.
It’s alleged that the number of people donating zero is large, and more generally I would expect people to round off their donation amounts when reporting. Ages are clearly also quantized. So there may be lots of points on top of one another. Is it easy to jitter them in the plot, or something like that, to avoid this source of visual confusion?
Just eyeballing the charts without any jitter, it kinda looks like Effective Altruists more often report precise donation quantities, while non-EAs round things off in standard ways, producing dramatic orange lines perpendicular to the Y axis and more of a cloud for the blues. Not sure what to make of this, even if true.
Possibly EAs think about their donating more and are thus more likely to be able to look up the exact sums they’ve donated. While non-EAs might be more likely to just donate to some random thing that seems like a nice enough cause, and then forget the details.
If EAs are reporting actual numbers they’ve looked up and non-EAs are reporting their vague estimates of how much they might have given, I would expect non-EAs’ figures to be worse overestimates than EAs’.
As Caplan would say, social desirability bias seems like an issue here. One would not expect people to penalize themselves in such a self-report rather than nudge estimates upwards and make themselves look better in their own mind.
Apply
jitter()
inside ofqplot
. Also, transparency:qplot(jitter(Age), jitter(CharityLog,a=0.1), color=EffectiveAltruism, data=survey,alpha=I(0.5))
Default
jitter()
worked great onAge
, but was negligible onCharityLog
.