Hacker News Reader

Comment by its_so_on | Hacker News Reader

Why Perl, you lying sack of shit!!

  C:\>perl -we "for(1..76){$sum += (int rand 6) + 1;} print $sum / 76
  3.71052631578947

If that should have been 3.5, then 0.21 * 76 = 15.9 extra points over average.

Perl, you lying oaf, you owe me 159 shekels. pay up.

I guess I'm being ironic, but with n=76, it would be great if "Those who were not under any time pressure reported a mean roll of 3.9. Both groups lied, then." came with a disclaimer about how confident we could be about that.

View on HN

In the paper in question (http://home.medewerker.uva.nl/s.shalvi/bestanden/Shalvi%20et...), it does. Chi-square test shows p<0.01 and p=0.02 for significance for the high and low time pressure groups, respectively.

After reading it I was going to say the same thing. What I don't understand is why they compared it to the statistical average rather than just recording what they rolled. It would have been fairly trivial to eliminate that particular statistical variability from the experiment.

The experimenters want the test subjects to believe that the die roll is truly private and therefore that the subjects can, individually, get away with a lie. An easy way to do that is to actually make the die roll private.

Maybe you could devise equipment that would convince subjects that roll was private even though it really wasn't. Ideas on how to do that? Hidden wireless/infrared camera in the cup?

I thought a hidden camera scenario would be fairly easy. Two way mirror, hidden camera, etc.

If Justin Bieber is allowed hidden cameras, surely science is, too.

Also interesting: the experimenters are testing lying but refuse to lie themselves in order to properly carry out the experiment. I feel like there's some special irony in there somewhere.

despite my statistical uncertainty, I think it is a lot easier to show a subject that they're not being tricked if they aren't. People are REALLY good at picking up being tricked or not from very slight nonverbal communication and environmental clues.

Hmmm... maybe don't tell the administrator who runs it? I'm sure there is a way to actually observe the whole thing without giving it away. There is no cake, after all.

A die equipped with a gyroscope that records the result of each roll.

n=76 also seems low to me for this type of experiment. There was a great New Yorker article titled "THE TRUTH WEARS OFF", which if I remember correctly, discussed the difficulties of reproducing scientific experiments because we underestimate the occurrences of statistical anomalies. Good time to read that article again.

http://www.newyorker.com/reporting/2010/12/13/101213fa_fact_...

Yea I felt the same way. I found this app to visualize how long it takes the law of large numbers to take affect with a die:

http://www.math.usu.edu/~schneit/CTIS/LawOfAverages/

76 is plenty for this. Read the paper before complaining.

The first experiment reported 76 participants, the second recruited 74.

Reply on news.ycombinator.com