Trimean — Definition, Formula & Example
The trimean is a measure of center that blends the median with the first and third quartiles. The median gets twice the weight of each quartile. Because it uses only these three values, a few extreme data points barely move it.
For a data set with first quartile , median , and third quartile , Tukey's trimean is the weighted average . Equivalently, it is the average of the median and the midhinge .
Key Formula
Where:
- = the trimean
- = first quartile (25th percentile)
- = median (50th percentile)
- = third quartile (75th percentile)
Worked Example
Problem: Nine students record how many texts they sent in one day: 4, 8, 10, 12, 14, 16, 18, 22, 76. Find the trimean and compare it with the mean.
Step 1: Find the three quartiles. The data are already sorted. With 9 values, the median is the 5th value. Q1 is the median of the lower half (4, 8, 10, 12), and Q3 is the median of the upper half (16, 18, 22, 76).
Step 2: Substitute into the trimean formula. Count the median twice.
Step 3: For comparison, compute the mean. The single heavy texter (76) pulls it upward.
Answer: The trimean is 14.25 texts, close to the median of 14. The mean is 20, which is pulled up by the outlier 76. Unlike the median alone, the trimean also reflects where the middle 50% of the data sits. Here Q3 lies farther above the median than Q1 lies below it, so the trimean lands slightly above the median.
Why It Matters
In AP Statistics you choose between the mean and the median based on skew and outliers. The trimean is a middle option. It resists outliers like the median, but it also uses information from the quartiles. Exploratory data analysis, the approach John Tukey developed, uses it to summarize skewed data such as incomes or response times.
Common Mistakes
Mistake: Averaging Q1, the median, and Q3 equally by dividing their sum by 3.
Correction: The median counts twice, so you divide by 4: (Q1 + 2Q2 + Q3)/4. Also note that software packages compute quartiles in different ways, so your trimean may differ slightly from a calculator's result. State which quartile method you used.
