Random numbers generation

You mean truncated normal random deviates?

Pardon ?

       ---A---- --B--- -----------------------------C------------------------------
   1       Mean    0.5 B1: Input                                                   
   2         SD    1.2 B2: Input                                                   
   3      Lower  -1.65 B3: Input                                                   
   4      Upper    2.2 B4: Input                                                   
   5                                                                               
   6    Prob(L) 0.0366 B6: =NORMDIST(B3, B1, B2, TRUE)                             
   7    Prob(U) 0.9217 B7: =NORMDIST(B4, B1, B2, TRUE)                             
   8                                                                               
   9   Deviates  1.935 B8 and down: =NORMINV($B$5 + RAND()*($B$6-$B$5), $B$1, $B$2)
  10             1.320                                                             
  11            -0.681                                                             
  12             1.100                                                             
  13             2.193                                                             
  14             0.108

I have a problem

The calculated Mean and SD is not as the input values. Please see attachment

If you truncate a normal distribution, of course the SD of the surviving values changes -- you chopped off the tails.

If you truncate it asymmetrically, of course the mean of the surviving values changes -- you chopped more off one side than the other. Quantitatively, the mean changes to

= mean + SD^2 * (NORMDIST(lower, mean, SD, FALSE) - NORMDIST(upper, mean, SD, FALSE)) / Z

where Z = NORMDIST(upper, mean, SD, TRUE) - NORMDIST(lower, mean, SD, TRUE)

If you had a bunch of random values that had a certain mean and standard deviation, but a truncated distribution, those would not be normal deviates.