Nearest Neighbors, Boundary Effect, and the Old Flag Trick: A General Solution

Abstract

This paper presents a method for quantifying "boundary effect" in square, rectangular, triangular, and circular study areas. It is based upon a very different, theoretically oriented modification of the classic computation methods, is applicable to the variously shaped study areas described, and produces results which approximate simulation data quite closely.

Refining the definition of cultural levels at Karabi Tamchin: a quantitative approach to vertical intra-site spatial analysis

“…Whallon (1973) (Carr 1984 145 Whallon (1973: 268) (Clark and Evans 1954) however, this type of analysis has also been used in archaeology for the study patterning within distributions of archaeological materials (Whallon 1974;Pinder et al I979;McNutt 1981;Durand and Pippin 1992). Nearest neighbour analysis is designed to interpret spatial relationships between distributions of objects (i.e., artefacts) to create an overall description of the spatial characteristics of the sample.…”

Locational structure of archeological underground storage pits in northwest Córdoba, Argentina.

“…For the spatial analysis we selected the technique of the nearest-neighbor analysis (NNA) (Carr, 1984(Carr, , 1985Earle, 1976;Flannery, 1976;Hodder 1972;McNutt, 1981;Pinder, Shimada and Gregory, 1979;Stark and Young, 1981;Washburn, 1974;Whallon, 1974;Zubrow, 1975), which provides an objective description of the randomness of elements in a finite space. Working with this type of analysis we could provide a thorough and u n b iased descrip tio n of the observable patterning of the structures within the site location.…”

“…But, in spite of our scale, we also found disturbances in the measures introduced by features lying outside the delimited area, known as the boundary effect. To avoid this problem we chose McNutt's proposal (McNutt, 1981), which includes a series of solutions for different area shapes. This way, selecting the formula for the best suitable geometric shape in concordance with the study area, the effect introduced by items with their actual first nearest neighbor outside the chosen area boundaries can be calculated.…”

Spatial analysis on a dugong consumption site at Princess Charlotte Bay, North Queensland

“…This approach has already been employed in a number of studies of intersite settlement patterns in the Americas (Earle 1976), Europe (Hodder and Orton 1976) and recently in Australia (Lilley 1985). The value of nearest neighbour studies in intra-site applications is somewhat limited by the commonly discussed problems of arbitrary study area size and the attendant edge effects (see Donnelly 1978, Hodder and Orton 1976, McNutt 1981, Pinder et al 1979. It should further be stressed that, while nearest neighbour analysis may inform on the degree of clumping, randomness or regularity in a distribution, it tells us nothing about its shape or form.…”