Loss of Information

Grouping raw data into a frequency distribution has one built-in drawback: a loss of information. The table is concise and easy to read, but the individual observations disappear. Once data are grouped, a single observation no longer counts in further calculations — every value in a class is treated as equal to that class's class mark.

For example, the class 20–30 holds six observations: 25, 25, 20, 22, 25 and 28. Grouped, the table records only that the frequency is 6; the actual six values are lost, and all of them are taken to equal the mid-value 25. Every later statistical calculation then rests on the class mark, not on the true observations — so some information is genuinely s …