Διδακτορικές διατριβές

Μόνιμο URI για αυτήν τη συλλογήhttps://pyxida.aueb.gr/handle/123456789/14

Περιήγηση

A B C D E F G H I J K L M N O P Q R S T U V W X Y Z

Α Β Γ Δ Ε Ζ Η Θ Ι Κ Λ Μ Ν Ξ Ο Π Ρ Σ Τ Υ Φ Χ Ψ Ω

Τώρα δείχνει 1 - 1 από 1

Statistical methodology for estimating patterns of demographic phenomena
Peristera, Paraskevi M.; Περιστερά, Παρασκευή Μ.; Athens University of Economics and Business, Department of Statistics; Kostaki, Anastasia
The main objective of this work is to develop statistical methodology for estimating patterns of demographic phenomena. The main topics are briefly described below. Several techniques have been proposed in the literature for graduating mortality data. An innovative topic for graduating mortality data is the use of kernel regression estimators. In this work we provide an analytical evaluation of the alternative kernel estimators as tools for graduating mortality data. In order to evaluate the efficiency and accuracy of kernel techniques for graduation purposes, we apply these to several empirical data sets and compare the results with those of classical graduation techniques. The interesting finding is that the Gasser-Muller estimator proves more adequate compared to other kernel-type regression estimators. The study of sex mortality differences is a topic that has been widely studied. Particular emphasis is given to the causes that provoke these differences as well as the evolution of the sex mortality gap through time. The gap between the mortality patterns of the two sexes significantly varies through ages exhibiting a pattern that became typical since the middle of the 20th century. This pattern consists of two humps, the first around age 20 and the second around age 60. However the literature lacks a model that captures this pattern. Therefore in this work a parametric model is developed for estimating the age-specific pattern of sex mortality differences over the whole age range. This model is useful for many purposes. In fact, such a model, eliminating the random variations of empirical data can serve for simplifying comparisons of the mortality patterns between sexes. Moreover, capturing the pattern into a limited set of parameter values, it can simplify comparisons of the shape and the severity of the mortality deviations of the two sexes, through space and time.In order to evaluate the performance of such a parametric representation, we fit the model to a variety of empirical data sets. The results suggest that a widening of the sex mortality differential is reflected through time. In addition in recent years, the modes of the two humps tend to move to higher ages, except of some Baltic and East-European countries. It is of course interesting to further investigate the reasons for which this sex mortality widening occurs in recent years.Furthermore, a parametric model is developed in order to graduate the age-specific patterns of fertility and nuptiality. Considering fertility, it is known that the age-specific fertility pattern has a typical shape common in all human populations through years. In recent years, a distorted fertility pattern is observed for the UK, the USA and Ireland characterised by a second hump at earlier ages. As expected, the existing models are unable to estimate the new shape of the fertility pattern and therefore the use of more appropriate representations is required. The model proposed here is able to capture both the old and the new distorted fertility pattern. In this work, this model is utilised for estimating the age-specific fertility pattern the age-specific parity rates of several European populations. The results reveal that the patterns of early-age fertility, previously confined to a few mostly English-speaking countries, are now more widely distributed in Europe. Another finding is that for countries with enhanced early-age fertility the pattern of first births also exhibits a strongly intense hump in younger ages and even stronger than the pattern of total fertility. Furthermore the fertility pattern of the USA when differentiated by the ethnicity of the mother is quite heterogeneous. These facts provide a strong evidence of heterogeneity in the female populations associated not only to the marital status, race and birth order but also to the educational level, social and economic status as well as the religiosity of the mothers. The model proposed is also adequate for estimating the age-specific pattern of nuptiality. To evaluate its adequacy, comparisons with existing models are provided, for several populations through time. The results indicate that these models generally show deviations between the empirical and estimated rates at the tails of the first-marriage distribution for the majority of populations. This might be an indication of heterogeneity in these populations and therefore, as shown here, mixture modes are more adequate for fitting the nuptiality pattern of modern populations.

Περιήγηση

Πλοήγηση Διδακτορικές διατριβές ανά Επιβλέπων "Kostaki, Anastasia"