3 ms·

Ok, comments on fig. 2. The 95% confidence intervals are plotted, but clearly the distribution of their subjects seems to sample heavily near where they find t

• joveian 10y ago
  Also, if you look at the right side of figure 1 there are lines on the bottom of the complexity chart for each of the tasks, presumably representing people who essentially did not do the task (although they still seem to have spent a variable amount of time not doing the task). It looks like this skews older, and just by eyeballing the chart some of the 75+ would be upward trends if not for that bottom line. I skimmed the article and did not see a discussion of this.
  • BlackFly 10y ago
    Furthermore, the poor sampling near the edges of the plots leads me to suspect that the trend away from a flat line is due to Runge's phenomenon. The peak may be nothing other than the oscillations from this phenomenon leaking into the interior of their data set.