Direct answer
Use the arithmetic mean for additive observations on one scale, the geometric mean for multiplicative factors, the harmonic mean for equal-quantity rates, and RMS when squared magnitude is the meaningful quantity.
What this calculation tells you
A mean compresses several values into one representative quantity, but different means preserve different mathematical relationships.
No single average describes spread, skew, subgroups, or uncertainty; interpretation belongs beside the result.
Where it is used
Education and reporting
Summarise comparable observations while retaining sample size and spread.
Growth analysis
Combine successive multiplicative factors with a geometric mean.
Travel and production
Combine rates over equal distances or equal work quantities with a harmonic mean.
Engineering and signals
Summarise signed varying magnitude through a root-mean-square measure.
Common situations
- Averaging ordinary measurements.
- Combining multi-period growth factors.
- Averaging speeds over equal distances.
- Comparing arithmetic and RMS values for a varying signal.
Match the relationship
Arithmetic averaging assumes equal additive contribution unless explicit weights are supplied. Multiplicative processes and reciprocal rates preserve different structures.
Keep the denominator honest
Average speed depends on total distance divided by total time; averaging displayed speeds directly is valid only under particular weighting conditions.
Inspect what the mean hides
Report sample size, range or another spread measure when variation matters. A stable mean can coexist with changing subgroups or extreme values.
Common mistakes
Frequent errors include averaging percentages with unequal bases, using a geometric mean with unsupported signs, and treating a mean as a typical observation.
- Name the quantity being preserved.
- Inspect the distribution.
- State weighting and exclusions.
Practical questions
Frequently asked questions
Is the arithmetic mean always the average?
It is the most common meaning of average, but it is not the appropriate summary for every relationship.
Why use a geometric mean for growth?
Successive growth acts through multiplication, and the geometric mean preserves the equivalent constant factor.
Can a mean be outside the data range?
The ordinary arithmetic mean with nonnegative weights lies within the range, while transformed or signed weighting conventions can behave differently.