Arithmetic Mean vs Geometric Mean

Arithmetic mean and geometric mean are both averages, but they answer different questions.

Arithmetic mean asks: what single value would add up to the same total?

Geometric mean asks: what single factor would multiply to the same overall effect?

That difference is small for ordinary values and important for growth rates, ratios, investment returns, index values, and normalized scores.

Quick rule

Use arithmetic mean when values add together. Use geometric mean when values multiply or compound.

| Data type | Better average | Why | | --- | --- | --- | | Test scores | Arithmetic mean | Scores add on the same scale | | Daily temperatures | Arithmetic mean | Values are ordinary measurements | | Investment returns | Geometric mean | Returns compound across periods | | Growth factors | Geometric mean | Factors multiply | | Index relatives | Geometric mean | Values represent relative change |

For a deeper side-by-side guide, see Geometric Mean vs Arithmetic Mean.

Arithmetic mean formula

Arithmetic mean adds the values and divides by the count:

AM = (x1 + x2 + ... + xn) / n

For scores 72, 80, 88, 96, the arithmetic mean is:

(72 + 80 + 88 + 96) / 4 = 84

This is useful because each score contributes directly to the total.

Geometric mean formula

Geometric mean multiplies positive values and takes the nth root:

GM = (x1 * x2 * ... * xn)^(1 / n)

For factors 1.06, 1.07, 1.08, the geometric mean is the representative factor that produces the same compound effect.

Example: ordinary scores

| Score | Meaning | | ---: | --- | | 72 | Test result | | 80 | Test result | | 88 | Test result | | 96 | Test result |

Arithmetic mean is the right summary because the scores are additive. The result, 84, is easy to explain: it is the average score across the tests.

The geometric mean is possible because the values are positive, but it does not answer the natural question for this dataset.

Example: investment returns

Now compare returns:

| Year | Return | Factor | | --- | ---: | ---: | | 1 | +6% | 1.06 | | 2 | +7% | 1.07 | | 3 | +8% | 1.08 | | 4 | -35% | 0.65 | | 5 | +10% | 1.10 |

The arithmetic average return is:

(6 + 7 + 8 - 35 + 10) / 5 = -0.8%

But the compound path is based on multiplication:

1.06 x 1.07 x 1.08 x 0.65 x 1.10

That is why geometric mean is the better return summary. It follows the money path instead of averaging the percentages as if they were independent additions.

Why geometric mean is usually lower

For positive values, geometric mean is less than or equal to arithmetic mean. They are equal only when every value is the same.

That gap is useful information. A large gap can signal volatility, skew, or one extreme value.

| Dataset | Arithmetic mean | Geometric mean | What it suggests | | --- | ---: | ---: | --- | | 10, 12, 14, 16 | 13 | 12.81 | Values are fairly even | | 4, 9, 16, 100 | 32.25 | 15.46 | One value is pulling the arithmetic mean upward |

When not to use geometric mean

Do not use ordinary geometric mean when:

The dataset includes zero
Values are negative and are not return percentages converted to factors
The question is about a simple total or score average
Units are mixed, such as dollars, percentages, and counts together

If the values are rates, the Harmonic Mean Calculator may be more appropriate. If you are deciding among several average types, use Which Average Should I Use?.

How to choose the mean in practice

The easiest way to choose between arithmetic mean and geometric mean is to ask what one value in the dataset does to the next value.

If the values are amounts on the same scale, they usually add. Test scores, temperatures, survey ratings, page counts, and daily ticket totals are common examples. A score of 80 and a score of 90 make a combined total of 170 points. In that situation, the arithmetic mean answers a clear question: what single score would give the same total if every observation were equal?

If the values are factors, ratios, index relatives, or percentage changes that compound, they multiply. A return factor of 1.10 followed by a return factor of 0.90 does not leave you with the arithmetic average of the two factors. It leaves you with 1.10 x 0.90 = 0.99, which means the final value is 99% of the starting value. In that situation, the geometric mean answers a different question: what single repeated factor would create the same final product?

This difference matters most when the values are uneven. When values are very close together, arithmetic and geometric means can look similar. That similarity can make the choice feel unimportant. But the moment one value is much larger, one period has a large loss, or a ratio changes the base for the next period, arithmetic mean can tell a story that the actual process does not follow.

Worked comparison: same numbers, different question

Consider the values 4, 9, 16, 25. You can calculate both means, but they do not answer the same question.

The arithmetic mean is:

(4 + 9 + 16 + 25) / 4 = 13.5

That is the correct summary if the values are additive measurements, such as four quiz scores out of 25, four daily counts, or four ordinary observations on the same scale.

The geometric mean is:

(4 x 9 x 16 x 25)^(1 / 4) = 10.9545

That is the correct summary if the values represent multiplying factors, normalized ratios, or scale changes. The geometric mean is lower because it respects the product, not the sum.

Now change only the question. If the four values are test scores, 13.5 is easier to defend. If they are four relative performance factors against a baseline, 10.9545 better represents the common multiplying level. Same numbers, different meaning, different average.

Decision table

| Question you are answering | Better mean | Reason | | --- | --- | --- | | What is the typical score or count? | Arithmetic mean | Values add on one scale | | What is the typical growth factor? | Geometric mean | Values multiply over time | | What is the average investment return path? | Geometric mean | Returns change the next period’s base | | What is the average speed over equal distances? | Harmonic mean | Rates need reciprocal weighting | | What is the middle value in skewed data? | Median | The middle observation resists outliers |

This table is useful because it starts with the user question rather than the formula. A formula is only correct when it matches the data-generating process.

How to explain the result clearly

When you report an arithmetic mean, use language about totals:

The average score was 84, meaning the four scores have the same total as four scores of 84.

When you report a geometric mean, use language about repeated factors:

The geometric mean return factor was 0.975, meaning the same final value would be reached by applying a 0.975 factor each period.

That wording prevents a common mistake: treating geometric mean as if it were just a lower arithmetic mean. It is not a discount. It is a different kind of average for a different kind of process.

If you are explaining investment returns, convert the factor back to a percentage:

0.975 - 1 = -0.025 = -2.5%

Then say it plainly:

The compound average return was about -2.5% per period.

If you are explaining normalized ratios, keep the ratio language:

The typical relative performance was 1.12x the baseline.

The unit of the explanation should match the unit of the input.

Using the calculators together

Use the Geometric Mean Calculator when your values are positive factors, ratios, or percentage returns converted into factors. Use the Arithmetic Mean Calculator when your values are ordinary additive measurements. Use the Variance and Standard Deviation Calculator when you need to understand how spread out the values are before trusting any average.

For returns, enter factors such as 1.06, 1.07, 0.65, 1.10, not raw percentages like 6, 7, -35, 10. For ordinary scores, enter the scores themselves. The result will only be as meaningful as the interpretation of the inputs.

A simple review checklist

Before choosing the average, review these five points:

Are all values measuring the same thing?
Do the values add to create a meaningful total?
Do the values multiply or compound from one period to the next?
Are there zeros, negative values, or mixed units?
Will the reader understand what the final number means?

If the answer is “the values add,” arithmetic mean is usually the first candidate. If the answer is “the values multiply,” geometric mean is usually the first candidate. If neither sentence sounds right, another summary may be better.

What the gap between the means tells you

The distance between arithmetic mean and geometric mean can be useful on its own. If the two values are very close, the dataset is probably fairly even or the compounding path is smooth. If the arithmetic mean is much higher than the geometric mean, the data may include volatility, skew, or a large value that makes the simple average look better than the product-based reality.

For investment returns, that gap often points to volatility drag. A portfolio can have a positive arithmetic average return while still ending below the starting value if losses are large enough. For normalized performance ratios, the gap can show that one component is much weaker than the others. For classroom examples, the gap helps students see why “middle” is not one single idea.

Use the gap as a prompt, not as a final verdict. Ask what caused it. Was there one extreme value? Was there a large negative return? Were the values ratios rather than amounts? The answer tells you whether to report both means, switch to another statistic, or explain the limitation in plain language.

FAQ

Is geometric mean better than arithmetic mean?

It is better only when the data is multiplicative. Arithmetic mean is better for ordinary additive data.

Can arithmetic mean be used for returns?

It can describe the simple average of returns, but it does not show the actual compounded return path. For investment performance, geometric mean or CAGR is usually more useful.

Why do I need to convert percentages?

Geometric mean multiplies values. Returns need to become factors first, so 8% becomes 1.08 and -10% becomes 0.90.

Use the Arithmetic Mean Calculator when your values add together. Use the Geometric Mean Calculator when your values are growth factors, ratios, or returns. Use Geometric Mean vs Arithmetic Mean for a full decision guide.