O'Reilly logo
  • Nathan Lomeli thinks this is interesting:

Normal probability plot for random samples from normal distributions


Cover of Think Stats, 2nd Edition


Take your original sample values. Sort them. Draw an equal number of values from the standard normal distribution. Sort them. Graph the original values on the standard normal distribution values as y-values on x-values. The mean (mu) will be the y-intercept at the line's midpoint and the standard deviation (sigma) will be the slope of the line. So lines with higher midpoints will have higher means and lines with steeper slope will have greater variance. If a line is not produced from the plot, the original data does not come from a normal distribution.