for the m and sd ... simple ... for skewness and kurtosis ... i am not sure ... of course, if you already have a distribution with the amount of skewness and kurtosis you want ... again, no problem


in minitab, changing a given distribution (any one) to one with different m and sd ... is easy

say we have the following data ... 1 to 10 ...

MTB > set c50
DATA> 1 2 3 4 5 6 7 8 9 10 <<<< just input the data ... you could make this ANY data set you want
DATA> end
MTB > name c50='X'
MTB > desc c50


Descriptive Statistics: X


Variable N Mean Median TrMean StDev SE Mean X 10 5.500 5.500 5.500 3.028 0.957

Variable       Minimum    Maximum         Q1         Q3
X                1.000     10.000      2.750      8.250

m=5.5 and sd=3.028

say you want to have a set of data where m=32 and sd=4.6?

MTB > cent c50 c51 <<< convert to z scores
MTB > name c51='zX'
MTB > let c52=c51*4.6 + 32 <<< use linear transformation to new scale ... z * new sd + new m
MTB > name c52='newX'
MTB > prin c50-c52 <<< here are the data columns


Data Display


Row X zX newX


   1      1  -1.48630   25.1630
   2      2  -1.15601   26.6823
   3      3  -0.82572   28.2017
   4      4  -0.49543   29.7210
   5      5  -0.16514   31.2403
   6      6   0.16514   32.7597
   7      7   0.49543   34.2790
   8      8   0.82572   35.7983
   9      9   1.15601   37.3177
  10     10   1.48630   38.8370

MTB > desc c50-c52

Descriptive Statistics: X, zX, newX


Variable N Mean Median TrMean StDev SE Mean X 10 5.500 5.500 5.500 3.028 0.957 zX 10 0.000 0.000 0.000 1.000 0.316 newX 10 32.00 32.00 32.00 4.60 1.45

note newX has m=32 and sd=4.6

now, IF the original data had the desired skewness ... and kurtosis ... the newX set would too


. . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================

Reply via email to