Unfortunately, the paper (http://home.medewerker.uva.nl/s.shalvi/bestanden/Shalvi%20et...) doesn't say whether the lower mean for the people in the low time pressure group is because they lied less frequently, or because they lied with lower numbers.

I wish they'd included graphs of the actual distributions of reported numbers. That would have been more useful than just the mean, standard deviation, and chi-square results.