For AI agents: use /llms.txt for the Nakafa content index.
How do we measure the spread of data presented in a grouped frequency table? For example, data on phone battery duration grouped into hour intervals (, , etc.).
Since we don't know the exact value of each data point within a class interval (e.g., in the class, we don't know if the duration was exactly , , or something else), we need to make an assumption.
The most common assumption is that all data within a class interval are evenly distributed. Therefore, we can represent all data in that class using the midpoint () of that class.
Using the midpoint () and frequency () of each class, the formulas are slightly different:
Variance () The commonly used (and easier to compute) formula is the computational formula adapted for grouped data:
This formula essentially calculates the average of the squared midpoints weighted by frequency, minus the square of the average midpoint weighted by frequency (the mean of the grouped data).
Standard Deviation () Just like with ungrouped data, the standard deviation is the square root of the variance:
Suppose a study on phone battery duration yielded the following data:
| Battery duration (hours) | Frequency () |
|---|---|
Let's determine the variance and standard deviation for this battery duration data.
We need to calculate the midpoint () for each class, then compute and .
| Battery duration (hours) | Midpoint, | Frequency, | ||
|---|---|---|---|---|
Plug the total values from the table into the variance formula:
So, the variance of the battery duration data is approximately (in units of hours squared).
Take the square root of the variance:
The standard deviation of the phone battery duration is approximately . This gives us an idea that the average deviation of battery duration from the mean (which can be calculated as ) is about .
The smaller the standard deviation, the more uniform the phone battery durations were in the study.
| Total |
