Where does centralization collide?

The degree to which the authority to make decisions in an organization is concentrated at the top rather than dispersed among lower levels and units: a highly centralized organization reserves choices about resources, policy, and work for a few senior roles, while a decentralized one pushes them down to those who do the work. Hage and Aiken's 1967 Relationship of centralization to other structural properties treated it as a measurable feature of structure, indexed by how much staff take part in decisions and how far their discretion reaches. The referent is a property of one organization's design, scored across its members, and it is not a property of a network graph, a division of power between governments, or where a group lives in a city.[1]

Unit

One organization's decision structure, scaled from centralized to decentralized

Recovery means

Not applicable

Measured by

Operationalised through surveys of who takes part in decisions and how much discretion staff hold over their own work

The concentration of political and fiscal authority in a national government rather than in regional or local governments: a centralized state decides and funds from the centre, while a decentralized or federal one grants subnational governments their own powers to tax, spend, and legislate. Treisman's 2007 The architecture of government pulled apart the many things this word bundles together, separating decision, fiscal, and appointment authority across the tiers of a country's government. The referent is the vertical division of power among governments in a country, and it has nothing to do with an organization's internal design, the shape of a graph, or where a population lives.[2]

Unit

One country's allocation of authority across tiers of government

Recovery means

Not applicable

Measured by

Assessed through indicators such as the subnational share of public spending and revenue and formal grants of authority to lower tiers

A single number describing a whole network, capturing the extent to which its connections revolve around one or a few central points rather than spreading evenly: a star graph, where every link runs through one hub, is maximally centralized, while a ring, where all points are equivalent, is not. Freeman's 1978 Centrality in social networks conceptual clarification defined this graph-level index as the variation in how central the individual points are, and kept it apart from the centrality of any single point. The referent is a property of an entire graph, not of one node within it, and it is unrelated to an organization's chain of command, a nation's tiers of government, or the map location of a group.[3]

Unit

One network's graph-level index, from a ring near zero to a star at one

Recovery means

Not applicable

Measured by

Computed from the gaps between the most central point's score and every other point's, summed and scaled against a star graph's maximum

In the study of residential segregation, the degree to which a population group settles near the physical centre of an urban area rather than in its outskirts: a group is centralized when its members cluster in the older, inner districts around downtown. Massey and Denton's 1988 The dimensions of residential segregation named centralization as one of five distinct spatial dimensions along which groups can be segregated, gauged against the metropolitan centre. The referent is a pattern of where people live in a city, measured by distance from the centre, and it says nothing about who makes decisions in an organization, how power is split among governments, or the shape of a network.[4]

Unit

One group's position relative to a metro centre, scaled from central to peripheral

Recovery means

Not applicable

Measured by

Calculated from the distribution of a group's residents across census tracts by their distance from the central business district