Anybody else bothered by the lack of p-values on these tables? Kudos to the authors for doing some research to see whether there are differences between the performance of the various landing page and banner designs... but the way they're reported their results leaves a lot to be desired.
The the stated differences (in terms of mean donation as well as conversion rate) between banners and landing pages strike me as pretty small (e.g., $26.92 vs. $27.07)- confidence intervals would be helpful here, as I suspect that the differences aren't significant.
Also, is mean really the right metric to be reporting? I suspect that these data aren't normally distributed, or, at the very least, have some outliers on either side of the spectrum (some people who donated hardly anything, plus a few "high rollers"), so it seems to me that the median would be a more informative statistic.
The the stated differences (in terms of mean donation as well as conversion rate) between banners and landing pages strike me as pretty small (e.g., $26.92 vs. $27.07)- confidence intervals would be helpful here, as I suspect that the differences aren't significant.
Also, is mean really the right metric to be reporting? I suspect that these data aren't normally distributed, or, at the very least, have some outliers on either side of the spectrum (some people who donated hardly anything, plus a few "high rollers"), so it seems to me that the median would be a more informative statistic.