Original scientific paper
A multivariate methodology for modelling regional development in Croatia
Dario Cziraky
Jakša Puljiz
Krešimir Jurlin
Sanja Maleković
Mario Polić
Abstract
This paper proposes a multivariate statistical approach based on covariance structure analysis for assessment of the regional development level with an application to development ranking of 545 Croatian municipalities. Municipality-level data ware collected on economic, structural, and demographic dimensions and preliminary factor and principal component analysis were computed to analyse empirical groupings of the variables. Next, confirmatory factor analytic models were estimated with the maximum likelihood technique and subsequently their implied structure was formally tested. Testing was extended to a joint model including all three dimensions (economic, structural and demographic) and their covariance structure was modelled with a recursive structural equation model. Finally, scores were estimated for latent variables thereby allowing (i) estimation of the latent development level of the territorial units, (ii) ranking of all units on an interval scale in respect to their latent development level, and (iii) selection of a given percentage of units for inclusion into special state-care subsidy programme.
Keywords
Croatia; regional development; statistics; methodology; multivariate methodology; covariance structure analysis; factor analysis; structural equation modelling; latent variables
Hrčak ID:
7081
URI
Publication date:
1.1.2002.
Visits: 1.506 *