A frequency distribution is a table of data detailing the frequency with which certain characteristics appear in a sample population. For example, you could make a frequency distribution of the heights of major league basketball players. After collecting heights for each member of the sample population (the number of players), you would construct the table, which would include the class width. The class width is the range of data values in each section of your chart. In this example, you might have one class representing heights of 60 to 69 inches, the next of 70 to 79 inches, and so on for as many classes as you want in your frequency distribution. There is a mathematical method for determining the range of values for your class widths.
Determine the largest data value in your sample data set. For the basketball player height example, this would be the height of the tallest basketball player.
Determine the smallest data value in your set. In this example, use the height of the shortest basketball player.
Subtract the smallest data value from the largest data value. In this example, subtract the shortest player's height from the tallest player's height.
Divide the difference between the shortest and tallest players' heights by the number of classes that you wish to have in your frequency distribution. For example, if you want to make a frequency distribution with five classes, divide the difference by five. The wider the range of data values you have accumulated, the more classes you should select.
Round up the dividend to the next whole number. If your dividend is 11.4, round it up to 12. Note that this is not the same as the normal rules of rounding. This number is the class width.
TL;DR (Too Long; Didn't Read)
If you are determining the class width from a frequency table that has already been constructed, simply subtract the bottom value of one class from the bottom value of the next-highest class.