listw

Listw is a class in the R package spdep that represents a spatial weights list. It is a data structure used to encode the strength and pattern of spatial relationships among observations, typically derived from a defined neighborhood structure.

A listw object stores weights as a list, with one numeric vector per observation. Each vector contains the weights assigned to that observation’s neighbors, in the order of the corresponding neighbor list (nb). The object also carries metadata, including the original neighbor structure and a weighting style that determines how the weights are normalized or scaled. Common style options include row-standardized weights (weight rows so sums equal one), binary weights (0/1 without normalization), and unstandardized weights (weights kept as provided). The handling of observations with no neighbors can be controlled via a policy parameter, often referred to as zero.policy.

Listw objects are typically created from a neighbor list using functions such as nb2listw, or from a weight matrix using appropriate converters. They are central to many spatial analyses, because they define how nearby observations influence each other in computations. They are used to compute spatially lagged variables (via lag.listw), perform spatial autocorrelation tests (such as Moran’s I), and support various spatial regression and diagnostic procedures.

In practice, listw objects interact with related structures in spdep, including nb (neighbor lists) and mat or other representations of weights. They can be inspected, subsetted, and passed to analysis functions, with methods designed to respect the underlying neighbor relationships and the chosen weighting style. This makes listw a foundational component for modeling and testing spatial processes in R.