Fig 1.
Geographic locations of prehistoric settlement systems examined in this study.
Background image courtesy of NASA Earth Observatory [39].
Fig 2.
Examples of temporally diagnostic projectile points.
Temporally diagnostic artifacts are used to (a) assign archaeological sites to settlement systems and (b) measure site size in terms of artifact counts. Top row: Titicaca Basin Late Archaic Period. 2nd row: Titicaca Basin Middle Archaic Period. 3rd row: Titicaca Basin Early Archaic Period. 4th row: Gila River Middle Archaic (images reproduced with permission of author [43]). Bottom row: Jequetepeque Paijan.
Fig 3.
Cumulative mass function plots for artifact-per-site counts (discrete data).
Axes are logarithmic. The log-linear structure is consistent with power-law structure.
Table 1.
MLE parameters and goodness-of-fit results for artifact count data.
Fig 4.
Summary of model-selection results for empirical datasets.
Power-law models are favored in the artifact-count (discrete) data but not in the site-area (continuous) data. Bar labels indicate number of datasets found to fit the given statistical model. Proportions are within the discrete and continuous categories. Error bars indicate 5–95% quantile range derived by bootstrapping with 10,000 iterations.
Fig 5.
Results of power analysis for artifact-count (discrete) data.
The analysis shows that given the sample sizes and MLE model-parameter values (a) the probability of failing to identify power-law structure when power-law structure is present (type II error) is highly unlikely and (b) the probability of spuriously identifying power-law structure given Poisson or geometric data (type I error) is also highly unlikely. See also S1 Table.
Fig 6.
Cumulative density function plots for site-area (continuous) data.
Axes are logarithmic. The upwardly convex data structure suggests an absence of power-law structure over the full range of data. Only the 50% threshold data are displayed for the Altiplano datasets.
Table 2.
MLE parameters and goodness-of-fit results for site area data.
Table 3.
Results of AIC and AIC weight analysis.
Fig 7.
Results of power analysis for site-area (continuous) data.
The analysis shows that given the sample sizes and MLE model-parameter values, (a) the probability of failing to identify power-law structure when power-law structure is present (type II error) is unlikely and (b) the probability of spuriously identifying power-law structure given normal, exponential, or lognormal data (type I error) is also unlikely. See also S2 Table.