With the exception of specific regions, the predictive performance of population size as a guide to spacing was poor, and King went on to test five other hypotheses related to the characteristics of the region in which the town was located and to its occupational structure. He argued that towns of a given size were likely to be more widely spaced where (i) rural population density is low, (ii) farming is extensive, (iii) agricultural production is low, (iv) where the overall population density is low, and (v) where the town itself has a low proportion of workers in manufacturing. Regression analysis showed that while all five were slightly more valuable than town size in predicting spacing, only one, overall population density, could explain more than 10 per cent of the variation. Indeed, all six hypotheses working together could only explain one quarter of the variation in spacing, though here again performance improved with certain agricultural zones (shown in the final column in Table 4.8). Clearly there is a very considerable problem in building accurate predictive models (cf. Chapter 16) for spacing for an area as large as the United States where differences in the historical development of settlements in the
With the exception of specific regions, the predictive performance of population size as a guide to spacing was poor, and King went on to test five other hypotheses related to the characteristics of the region in which the town was located and to its occupational structure. He argued that towns of a given size were likely to be more widely spaced where (i) rural population density is low, (ii) farming is extensive, (iii) agricultural production is low, (iv) where the overall population density is low, and (v) where the town itself has a low proportion of workers in manufacturing. Regression analysis showed that while all five were slightly more valuable than town size in predicting spacing, only one, overall population density, could explain more than 10 per cent of the variation. Indeed, all six hypotheses working together could only explain one quarter of the variation in spacing, though here again performance improved with certain agricultural zones (shown in the final column in Table 4.8). Clearly there is a very considerable problem in building accurate predictive models (cf. Chapter 16) for spacing for an area as large as the United States where differences in the historical development of settlements in the
การแปล กรุณารอสักครู่..
