Page 48 - IJPS-8-2

International Journal of
Population Studies Used versus Offered densities of human population

A B

Figure 7. Approximations of (a) offered density PDF and (b) used density PDF
Abscissas in log-scale.
Source: Author’s calculations based on INSEE data (2021a; 2021b).

Table 1. Deciles of offered density versus used density The RID and RIQ of x are quite high for their kinds of
O
(France, 2019), in persons per km² indicators. As for x , the RID and RIQ values are still much
U
higher: they reveal the very large heterogeneity of used
α 10% 20% 30% 40% 50% 60% 70% 80% 90% density, that is, of human density as lived by the people.
x (α) 7 12 17 24 33 46 64 96 186
O
We utilized the dataset to calculate the alternative
x (α) 46 92 167 306 600 1,146 2,213 3,872 6,999 indicators of average density recalled in §2.5. Craig’s
U
Source: Author’s calculations based on INSEE data (2021a; 2021b). geometric mean of the used density has a value of
 x = 590 p/km²: Thus, it is close to the geometric midpoint
high density land are associated to 20 – 80% shares of low between x and x , since  x ≈ 445. x and x ≈ 493. x  .
versus high density people. Such striking contrast calls O U O 2 U
for quantitative metrics to complement density maps in Both ratios are close to value 1 O 4 69. that was
spatial analysis. expected using a lognormal approximation of x .
o
r
The deciles pave the way to the qualitative assessment The Stair’s generalized indicators x in (22) were
k
of low to high levels of density. With respect to people computed for indices k from 0 to 9 and r from 1 to 5. Figure 8
living in France, the median used density, that is, 600 exhibits the reduced indicators ln(x r / M )/(k + 1 ) r as
persons per square kilometer may be taken as “medium k O 2
level of density,” low densities for the bottom 20%, that functions of k depending on r. The salient values are those
is, below 92 p/km², high densities for the top 20%, that associated to pairs kr, ( ,)01 and (1,1), that is, to x
O
is, above 4000 p/km². These people-based values are close and x , while the other pairs yield an overall pattern about
U
to the values selected by the Regional and Urban General
Directorate of the European Commission (Eurostat, 2019). value 1 that corresponds to lognormal distributions.
The land-based deciles have little relevance to depict urban 4. Discussion
conditions. The average offered density is just a ratio to
summarize the intensity of human occupation over a given 4.1. On the statistical populations and random
stretch of land – nothing less, nothing more, especially not variables of human density
about the used density of population. The history of statistics began with early censuses of the

The average densities are meaningful metrics. The human population in a couple of countries, that is, with
standard deviation of offered density makes little sense human populations as statistical populations – hence,
to people: and not much more for land, in fact. The Gini the very name of the latter concept. In such a historical
index is much more meaningful and so are the relative perspective, the used density lends itself to be modeled
dispersions of used and offered density. as a random variable in the human population. However,
as geography and cartographic methods have been well
As for interquartile ratios to measure distribution
heterogeneity, the outcomes are: developed long before the advent of computers, the plain,
offered density has been introduced long before the used,
• For offered density, inter-decile ratio (RID) of 186/7 = 27 lived density. In the pioneering contributions of Craig
and inter-quartile ratio (RIQ) of 80/15 = 5.3. (1975; 1979; 1980; 1984) and Stairs (1977a), it has been
• For used density, inter-decile ratio (RID) of 7000/46= 152 clearly stated that the average plain density is area weighted,
and inter-quartile ratio (RIQ) of 3000/120 = 25. whereas the average lived density is a population-weighted

Volume 8 Issue 2 (2022) 42 https://doi.org/10.36922/ijps.v8i2.297