Skip to content

Instantly share code, notes, and snippets.

@robertfeldt
Created April 26, 2017 06:12
Show Gist options
  • Save robertfeldt/867aaa283806fb2149501fa142df9a02 to your computer and use it in GitHub Desktop.
Save robertfeldt/867aaa283806fb2149501fa142df9a02 to your computer and use it in GitHub Desktop.
using HypothesisTests
# Checked num downloads from each papers page in SpringerLink 2017-04-26 08:05 CET
OpenAccessPaperDownloads = [2400, 589] # says 2.4k downloads, approximates with 2400
NonOAPaperDownloads = [134, 126, 85, 104, 56, 228, 112]
println(pvalue(MannWhitneyUTest(OpenAccessPaperDownloads, NonOAPaperDownloads)))
# p-value is 0.055
# Executed 2017-04-26 08:10 CET on MacBook Pro 13" 2015 running Julia 0.5.1 with HypothesisTests version 0.5.1
@robertfeldt
Copy link
Author

Note that since Mann Whitney U is based on ranks the actual difference in medians will make no difference here and p-value 0.055 is the lowest one can get with a 2vs7 comparison. So a bit tongue-in-cheek... ;)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment