News for hackers. Stuff that matters.
Re:Looks vs. Results: My ugly ad got 150% more clicks than a... by clircle on Sunday January 03, 2021 @08:22PM (#25627002)
The effect may be statistically significant depending on your tolerance:

  R> prop.test(c(12, 24), c(5563, 4403))

   2-sample test for equality of proportions with continuity correction

  data:  c(12, 24) out of c(5563, 4403)
  X-squared = 6.5211, df = 1, p-value = 0.01066
  alternative hypothesis: two.sided
  95 percent confidence interval:
   -0.0059903650 -0.0005970741
  sample estimates:
       prop 1      prop 2 
  0.002157109 0.005450829
Re:Looks vs. Results: My ugly ad got 150% more clicks than a... by jldugger on Sunday January 03, 2021 @10:31PM (#25627678)
Not that it harms the argument, but do you really need it two sided here?
Re:Looks vs. Results: My ugly ad got 150% more clicks than a... by clircle on Monday January 04, 2021 @09:10AM (#25631511)
No, a one sided test is also appropriate. I tend to default to two sided tests because the acceptance of one sided tests varies by field.