URBP7005 Chap.4 Composite Indices, Weighting and City Rankings
Composite Indices, Weighting and City Rankings
Three decisions turn indicators into a number
A composite indicator combines two or more individual indicators, and its appeal is that it reduces several numbers to one a reader can hold. The cost is three contestable decisions. Normalisation puts components on a common scale and destroys the units, so nobody can any longer read the result in the measure a regulator uses.
Weighting decides how much each component contributes and destroys the separate identity of the components. Aggregation decides how they combine and destroys the profile, so two cities with opposite strengths can share a score.
A composite is therefore useful for attention and useless for diagnosis.
What practice actually chooses
A survey of composite indicator studies reports equal weighting in just under half of them, and additive aggregation in a far larger share, with principal components, geometric aggregation, the analytic hierarchy process, public opinion, budget allocation and others sharing what remains.
Two readings of that distribution are available and the second is the one worth writing down: rather more than half of the studies chose something other than equal weights, which means most practitioners who considered the question concluded that their components were not equally important.
One index, read as decisions
An urban sustainability index built for major Chinese cities has three sub indices.
The economic one scales output per head between an explicit floor and ceiling on a logarithmic scale, so cities outside the bounds become indistinguishable and equal proportional gains count equally. The environmental one combines air quality and noise in fixed proportions, and the equity one combines a rural to urban income ratio with an urbanisation ratio in the same proportions.
Nothing in the construction explains why those proportions, which is the point: a weighting has to be argued for in the text.
Ranking places without ranking their residents
A green city ranking of the same cities measures annual carbon dioxide emissions per household, and controls for household characteristics by estimating emissions for one standardised household in every city, defined as three members with a head aged forty five and a stated annual income.
Without that control a city of small, poor, older households would outrank a city of large, affluent, young ones on nothing but demography. A per capita denominator by itself controls for size and for nothing else.
What this chapter covers
- 01
Normalise, weight, aggregate, and what each step destroys
- 02
Why a composite is for attention rather than for diagnosis
- 03
The published distribution of weighting and aggregation methods
- 04
Equal weighting as a substantive claim rather than a default
- 05
Explicit bounds and the cities an index can no longer separate
- 06
Two thirds to one third splits and where they have to be argued
- 07
The standardised household as a control card
- 08
Compensation, and when adding is the wrong rule
- 09
The disclosure set a usable composite has to publish
Tell a client which of two rankings is correct
- 3Locate the disagreement in the right step.
- 3Say what the published distribution of practice does and does not settle.
- 4State what the client is actually owed.
Key terms
- Normalisation
- The step that places each component of a composite on a common scale, usually zero to one, so that different measurement units drop out of the comparison.
- Compensatory Aggregation
- A combination rule, typically additive, under which a strong component is permitted to offset a weak one so that a mid table score can hide an extreme profile.
- Standardised Household
- A fixed household specification, such as three members with a head of a stated age and income, used so that a cross city emissions comparison measures places rather than populations.
- Scale Bounds
- The floor and ceiling chosen for a normalised component, which decide which cases the index can still discriminate between and which are censored at the limits.
- Sensitivity
- How far a ranking moves when a defensible alternative weighting is applied. Reporting it is what turns a composite from an assertion into an argument.
Composite Indices, Weighting and City Rankings FAQ
Is equal weighting the neutral option?
No. Giving every component the same weight asserts that every component matters equally to the thing being measured, which for an index of urban sustainability means claiming that a unit of air quality improvement is worth exactly a unit of income improvement.
That is sometimes defensible, usually because nobody has better information, but it is never neutral, and describing an equally weighted index as unweighted is the phrasing to avoid in a paper.
Why does a per capita figure not solve the comparison problem?
Because dividing by population controls for size and for nothing else. Household composition, age structure, income and climate all survive the division, and in a cross city comparison at least one of them is usually the real driver of the difference you are about to attribute to planning.
Where you cannot standardise properly, name the uncontrolled variable that would most plausibly explain your result and say how large it would have to be.
What has to be disclosed for a composite to be usable?
Four things. The bounds used for each normalised component, because they decide who is still distinguishable. The weights and the basis on which they were set. The aggregation rule, because adding permits compensation and other rules do not. And the sensitivity of the ranking to a defensible alternative weighting. A composite missing any of the four cannot be argued with, which also means it cannot be defended.
Are composite indicators a bad idea then?
Not at all, and the reading treats them as one useful way of building an integrating indicator, since combining across dimensions is what integration means. The simple example given is the cost of recycling per ton recycled, which holds an economic and an environmental consideration together.
For some purposes a composite is the only device that satisfies the first of the four properties a sustainability indicator is supposed to have.
Assessment move
Build a two component index for two cities you know, once with equal weights and once with a two to one split, and see whether the order changes. Ten minutes of arithmetic will teach you more about why disclosure matters than any amount of reading, and the resulting table is exactly the kind of appendix material the term paper allows beyond the word count.
Working through Composite Indices, Weighting and City Rankings in URBP7005? Sia is AskSia’s AI Environmental Science tutor — ask any URBP7005 Composite Indices, Weighting and City Rankings question and get a clear, step-by-step explanation grounded in how URBP7005 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.
