OVERVIEW: Normal distributions are quite common inreal life settings. Any set of normally-distributed observations canbe examined efficiently by converting the data to standardizedobservations know as z-scores. A z-transformation changes a normalrandom variable with mean m and standarddeviation s into a standard normal randomvariable with mean 0 and standard deviation 1.
Any set of numbers has a mean and standard deviation. The set W ={5,10,20,65,80} has m = 36 and
The standardized value (sometimes called a z-score)of an observation, x, is
The following table displays the standardized values forobservations from the set W.
| z-score |
| -1.0140 |
| -0.8507 |
| -0.5235 |
| 0.9489 |
| 1.4397 |
If a variable x has a normal distribution with mean
Suppose a set of observations is approximately normalwith mean = 50 and standard deviation = 4.
We know that approximately 68% of the observationswill be within one standard deviation of the mean. Note that
normalcdf(46,54,50,4) = 0.6826894809
normalcdf(-1,1,0,1) = 0.6826894809If we want to know what percent of the scores are above 56, we cannote that z56 = (56-50)/4 = 1.5.
normalcdf(56,1E99,50,4) = 0.0668072287
normalcdf(1.5,1E99,0,1) = 0.0668072287
Important to note: You canstandardize any set of numerical observations and obtain z-scores.The z-scores simply reflect how many standard deviations anobservation is from the mean. The z-scores will form a normaldistribution only if the original data set is normal. You cancorrectly use the normal distribution table to interpret z-scoresonly if the original data set is normal. Standardizing scores doesnot magically convert non-normal data into normal data.
If you have a data set in a TI-83 list, you can use the calculatorto construct a normal probability plot. This is one of theoptions when you do a stat plot on the calculator.
RETURN TO TEXTBOOK HOME PAGE /Back to the top of this page