Spatial Autocorrelation of a Demographic Phenomenon

Jaroslav Kraus

Spatial Autocorrelation of a Demographic Phenomenon

Číslo: 4/2019
Periodikum: Statistika

Klíčová slova: Population and housing census, households, spatial autocorrelation, Czech Republic

Pro získání musíte mít účet v Citace PRO.

Přečíst po přihlášení

Anotace: The paper aims to examine spatial distribution and to subsequently analyse a demographic phenomenon; share of one-family households and one-person households of all households by municipality in the Czech Republic (CR), because these two types of households are in a way contradictory. Although the Czech Republic is rather small and homogeneous with respect to demographic processes, it is questionable whether this also applies to the spatial distribution of households. Methods of local and global spatial autocorrelation represented by Moran’s index were used; however, the challenge of normalisation of both variables using Box-Cox transformation had to be addressed before. For comparison classical measures of association on NUTS3 level were used with respect to the type of data being examined – e.g. the Pearson chi-square, the uncertainty coefficient, the lambda coefficient, as well as the Gini index (Gini ratio), which is a measure of statistical dispersion and the most commonly used measurement of inequality.

From the results a small statistically significant global autocorrelation was ascertained. Local results show that shares of both types of households are not a complementary phenomenon in terms of space. From the relationship of local and global Moran’s index, it can be concluded that the global rate does not sufficiently depict detected local differences.