Z-scores poorly estimated
Given the amount of people (>50?) doing n>9 picture n-backs on this site in the past, and the amount of people who visit this site (>1 million?), the z-score for average n>7 is too overinflated by at least (3.5-2.5?) standard deviations. Does anyone know of a site providing precise or genuine statistics for dual n-back and n-back?
That is merely your conjecture based on your assumption of 50 or so persons visiting the sight. I don't think the statistics are inflated, because it is clear that the stats do not track people, but completed sessions. Hence, for each person, hundreds of completed sessions result. And since most will gravitated to the lower n-levels, that makes the standard score for completed higher n-levels "artificially" higher, that is, makes them seem rare (when they in fact are rarer). Hence, it is important that you do not interpret the z-scores as locating yourself on a continuum of other participants, but rather you should interpret it w.r.t. the typical performance outcome on the task.