When two cities report the same average household income, it is tempting to assume their residents are equally well off. But averages hide a crucial detail: how spread out the values really are. One city might have incomes clustered tightly around the mean, while the other has extreme poverty sitting next to extreme wealth. This is exactly what measures of dispersion capture. They tell us how much the data points in a dataset vary from each other and from the central value, turning a flat single number into a fuller picture of reality.
Table of Contents
Defining dispersion and why it matters
Dispersion, also called variability or spread, describes the extent to which the values of a variable differ from one another. Measures of dispersion describe how different the values of a variable are from the centre and from each other. While measures of central tendency like the mean and median give us a single representative value, they say nothing about consistency. Dispersion fills that gap.
Consider two neighbourhoods with the same average monthly rent of ₹15,000. In the first, almost every flat rents for between ₹14,000 and ₹16,000. In the second, rents swing from ₹6,000 to ₹30,000. The average is identical, but the lived experience is completely different. A planner who relies only on the mean would miss the affordability crisis in the second neighbourhood. While central tendency finds a representative value, dispersion reveals how much the data points scatter around it.
Statisticians group dispersion into two broad families. Absolute measures are expressed in the same units as the original data, such as rupees, kilometres, or years. Relative measures, also called coefficients of dispersion, are unitless ratios that let you compare datasets measured in different units. Range, variance, standard deviation, and mean deviation are absolute measures, while coefficients of dispersion are relative and dimensionless.
Key measures of dispersion
Let us walk through the main tools one by one, starting from the simplest and moving towards the most powerful. Each has its own strengths, weaknesses, and ideal use cases.
Range
The range is the easiest measure to grasp. It is simply the difference between the largest and smallest values in a dataset. If the highest property price in a locality is ₹80 lakh and the lowest is ₹20 lakh, the range is ₹60 lakh.
The strength of the range is its simplicity. It gives a quick sense of the overall spread. Its weakness is equally obvious: it depends only on the two extreme values and ignores everything in between. Because the range relies only on the outermost values, two datasets with the same range can still have drastically different distributions. A single outlier, such as one luxury penthouse, can inflate the range and mislead the analysis.
Percentiles and quartiles
Percentiles divide an ordered dataset into 100 equal parts. The 25th percentile is the value below which 25% of observations fall. The most commonly used percentiles are the quartiles: the first quartile (Q1) at the 25th percentile, the second quartile (the median) at the 50th, and the third quartile (Q3) at the 75th.
From these we get the interquartile range (IQR), which is Q3 minus Q1. The interquartile range is the difference between the upper quartile (Q3) and the lower quartile (Q1). The IQR captures the spread of the middle 50% of the data and is far more robust against outliers than the full range. Half of the IQR gives the quartile deviation, also known as the semi-interquartile range.
Percentiles are widely used in policy work because they describe position rather than raw value. A census tract in the 95th percentile of a risk index has a value higher than 95 percent of tracts in its state. This makes percentiles ideal for ranking neighbourhoods, schools, or income groups in a way that is easy to communicate.
Mean deviation
The mean deviation, sometimes called the average absolute deviation, measures how far each value sits from a central point on average. The mean deviation gives the average of the absolute deviations of the data about a central point, which can be the mean, median, or mode.
To calculate it, you first find the central value, then measure the distance of every observation from it, take the absolute value of each distance so negatives do not cancel positives, and finally average these distances. Unlike the range, the mean deviation uses every data point, giving a more complete view of spread. Its main drawback is the use of absolute values, which makes it awkward for further algebraic manipulation. Mean deviation considers all data points but sees limited use in advanced statistical work compared to standard deviation.
Variance and standard deviation
The standard deviation is the most widely used and respected measure of dispersion. It is built on the idea of variance. To compute the variance, you subtract the mean from each value, square the result, add up all the squared differences, and divide by the number of observations. Standard deviation is the square root of the arithmetic average of the squared deviations measured from the mean.
Squaring the deviations serves two purposes: it removes negative signs and it gives greater weight to values that lie far from the mean. The downside is that variance is expressed in squared units, which are hard to interpret. Taking the square root brings the measure back into the original units, which is why the standard deviation is preferred. A low standard deviation means data points cluster close to the mean, while a high standard deviation means they are spread far from it.
One technical point worth noting: when working with a sample rather than an entire population, statisticians divide by the sample size minus one rather than the sample size. Dividing by n minus one rather than n provides an unbiased estimate of the population parameter. This small correction matters in research where conclusions about a whole population are drawn from a limited sample, which is common in urban surveys.
The coefficient of variation
The measures above are powerful, but they share a limitation. You cannot meaningfully compare the standard deviations of two datasets that use different units or have very different averages. Comparing the spread of household incomes in rupees with the spread of commute times in minutes makes no sense. This is where the coefficient of variation (CV) comes in.
The coefficient of variation is the ratio of the standard deviation to the mean, usually expressed as a percentage. It is calculated by dividing the standard deviation by the mean and multiplying by 100. Because both the standard deviation and the mean carry the same units, those units cancel out, leaving a pure number. This is what makes the CV so useful for comparison across scales.
A lower CV indicates more consistency, while a higher CV signals greater relative variability. The coefficient of variation indicates the size of a standard deviation relative to its mean, allowing comparison between disparate groups. For instance, if commute times in City A have a CV of 12% and City B has a CV of 35%, City A offers far more predictable travel, regardless of how long the average commute actually is.
One caution: the CV becomes unreliable when the mean is close to zero, because a tiny mean in the denominator produces wild swings in the result. The coefficient of variation is most useful when comparing variation for one variable among groups with different means or among different measurement variables. It also requires data on a ratio scale with a true zero, so it does not work for things like temperature in Celsius.
Real-world applications in urban planning and research
Measures of dispersion are not abstract classroom exercises. They sit at the heart of how cities are studied, planned, and governed.
Income inequality and housing. Urban economists routinely build income distributions by splitting a city’s population into percentile bins. Researchers studying Indian cities construct percentile income distributions and divide them into equal-sized bins to analyse inequality. The spread between the bottom and top percentiles reveals how unequal a city is, which directly informs affordable housing targets and welfare schemes.
Targeting public funds. Housing programmes often define eligibility using percentages of the area median income. Housing need data are split by income limits set at 30, 50, and 80 percent of median income to show the extent of housing problems. Local governments use these thresholds to decide where to direct grants, build social housing, or expand transport links. Without measures of spread, a city would only see its average and miss the households at the margins.
Service delivery and consistency. Suppose a municipal body measures the time taken to clear garbage across different wards. Two wards may have the same average collection time, but the ward with the lower standard deviation is delivering a more reliable service. The coefficient of variation lets planners compare consistency across wards of very different sizes, helping identify where systems break down.
Infrastructure and amenities. Researchers study how amenities such as parks, schools, and hospitals are distributed across a city. Studies of urban amenities establish statistical patterns in the quantity and spatial distribution of facilities like restaurants, parks, and universities. High dispersion in amenity access points to inequity, where some neighbourhoods are well served while others are deprived.
In each case, the logic is the same. The average tells you the typical situation, but dispersion tells you about fairness, reliability, and risk. For anyone monitoring or evaluating an urban programme, ignoring spread means ignoring the very inequalities that programmes are usually designed to fix.
Choosing the right measure
No single measure is best for every situation. The range gives a fast snapshot but breaks down with outliers. The interquartile range resists outliers and suits skewed data such as house prices. Mean deviation uses all values but is hard to extend mathematically. Standard deviation is the workhorse for symmetric data and feeds directly into advanced techniques. The coefficient of variation steps in whenever you need to compare datasets with different units or vastly different averages. For symmetric data without outliers, standard deviation is most informative, while the interquartile range is a more robust choice for skewed distributions. A frequent mistake is reporting an average with no measure of spread at all, which leaves readers unable to judge how reliable that average really is.
What do you think? If you were evaluating two housing schemes in different cities, would the coefficient of variation tell you something the standard deviation alone could not? And in your own city, which neighbourhoods do you suspect hide the widest gaps behind a comfortable-looking average?
References
- https://libguides.lib.miamioh.edu/data_analysis/dispersion
- https://flexbooks.ck12.org/cbook/ck-12-cbse-maths-class-11/section/13.1/primary/lesson/measures-of-dispersion/
- https://www.cuemath.com/data/measures-of-dispersion/
- https://www.geeksforgeeks.org/maths/measures-of-dispersion/
- https://www.urban.org/data-tools/mapping-neighborhoods-highest-risk-housing-instability-and-homelessness
- https://foodsafety.institute/research-methodology/measures-of-dispersion-range-mean-deviation-standard-deviation/
- https://medium.com/@aw_marcell/measure-of-dispersion-range-variance-and-standard-deviation-b7eaa7b4a90b
- https://www.cuemath.com/coefficient-of-variation-formula/
- https://statisticsbyjim.com/basics/coefficient-variation/
- https://stats.libretexts.org/Courses/Las_Positas_College/Math_40%3A_Statistics_and_Probability/03%3A_Data_Description/3.03%3A_Measures_of_Variation/3.3.01%3A_Coefficient_of_Variation
- https://arxiv.org/pdf/2410.04737
- https://www.huduser.gov/portal/datasets/cp.html
- https://epjdatascience.springeropen.com/articles/10.1140/epjds/s13688-022-00362-6
Leave a Reply