Mean, median and typical rent: do not swap the labels

Calculate both on the same sample

Take illustrative rents of 600, 700, 800, 900 and 2,000. Their sum is 5,000, so the arithmetic mean is 1,000. The ordered middle value, and therefore the unweighted sample median, is 800. These are invented teaching numbers, not observations from a city. The expensive final item affects the mean differently from the median.

Preserve the provider’s statistic

If a source publishes an average, check whether it defines an arithmetic mean, a weighted mean or another measure. Do not rename it a median because that sounds more robust. A phrase such as typical rent needs a method to be meaningful. A model prediction and a median calculated from observations also require different labels, even when their numeric values happen to agree.

Check the sample before choosing a summary

Neither statistic fixes mixed property types, outdated periods or a biased sample. Keep the number of observations, geographic scope and rental cohort with the result. Without compatible observations, a true sample median cannot be reconstructed from a broad national range. Use an honestly labelled estimate when evidence is incomplete rather than manufacturing the appearance of a measured city statistic.

Sources and further reading

• Lifovra — methodology and limitations