Unequal class intervals. The small number of observations greater than 7 in Exercise 3.119 may cause some difficulty in constructing a frequency distribution. To keep class intervals equal, one is faced with the dilemma of either creating too many classes for only 30 observations or using a small number of classes with excessive loss of information in the first few classes. In such cases, one is tempted to drop the rule of equal- size classes, using a larger interval for the last class.

(a) If that were done, what would the resulting frequency distribution become?

(b) Is there a unique class interval?