part 1 of 2 The differences are of course statistically significant, they are not necessarily real. There are of course the big sources of uncertainty in the numbers even if none of the uncertainty is attributable to sampling. Some would say that even though differences are not attributable to sampling they are still subject to uncertainty (not completely real). Different statistical subcultures have different attitudes toward uncertainty. Some would say "uncertainty is made up of sampling and oh yes nonsampling error". Others would say that "uncertainty is made up of the big sources of uncertainty such as intrinsic variation in the phenomenon, errors is applying the measurement, lack of correspondence between the measurement and the construct, unreliability of the measurement method, memory error, coding error, and oh yes uncertainty due to sampling".
. . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
