Multiscale methods for data on graphs and irregular multidimensional situations

Summary.  For regularly spaced one-dimensional data, wavelet shrinkage has proven to be a compelling method for non-parametric function estimation. We create three new multiscale methods that provide wavelet-like transforms both for data arising on graphs and for irregularly spaced spatial data in more than one dimension. The concept of scale still exists within these transforms, but as a continuous quantity rather than dyadic levels. Further, we adapt recent empirical Bayesian shrinkage techniques to enable us to perform multiscale shrinkage for function estimation both on graphs and for irregular spatial data. We demonstrate that our methods perform very well when compared with several other methods for spatial regression for both real and simulated data. Although we concentrate on multiscale shrinkage (regression) we present our new 'wavelet transforms' as generic tools intended to be the basis of methods that might benefit from a multiscale representation of data either on graphs or for irregular spatial data.     

Lesbian struggle against a pillow or a wall: a Dutch-American dialogue

A dialogue between two lesbian women is conducted in an effort to demonstrate the differences in attitude and level of tolerance towards lesbians in the United States as compared to those in The Netherlands. Their observations suggest there are positive and negative aspects to being a lesbian in both countries and they hope that the positive benefits in both countries will combine.     

Evolution and population dynamics in stochastic environments

Inter-generational temporal variability of the environment is important in the evolution and adaptation of phenotypic traits. We discuss a population-dynamic approach which plays a central role in the analysis of evolutionary processes. The basic principle is that the phenotypes with the greatest long-term average growth rate will dominate the entire population. The calculation of longterm average growth rates for populations under temporal stochasticity can be highly cumbersome. However, for a discrete non-overlapping population, it is identical to the geometric mean of the growth rates (geometric mean fitness), which is usually different from the simple arithmetic mean of growth rates. Evolutionary outcomes based on geometric mean fitness are often very different from the predictions based on the usual arithmetic mean fitness. In this paper we illustrate the concept of geometric mean fitness in a few simple models. We discuss its implications for the adaptive evolution of phenotypes, e.g. foraging under predation risks and clutch size. Next, we present an application: the risk-spreading egg-laying behaviour of the cabbage white butterfly, and develop a two-patch population dynamic model to show how the optimal solution diverges from the ssual arithmetic mean approach. The dynamics of these stochastic models cannot be predicted from the dynamics of simple deterministic models. Thus the inclusion of stochastic factors in the analyses of populations is essential to the understanding of not only population dynamics, but also their evolutionary dynamics.     

Market transition or path dependency?: Changing effects of income determinants in the Czech Republic,Hungary, Poland,Russia, and Slovakia, 1991–2002

A major issue in the market transition debate on transitional societies is the extent to which changing economic institutions are responsible for the changing stratification order in post-Communist societies, as assumed by the Market Transition Theory (MTT). Advocates of path dependent transformation processes assume varying transformation processes across post-Communist societies. They have criticized the MTT of being insensitive to the different paths of transformation post-Communist societies go through and the way this will influence changes in the effects of income determinants. This study will test the extent to which trends in the effects of income determinants converge across post-Communist societies. A cross-national and across time comparative research design will be employed, analyzing 61 cross-sectional, standardized surveys for five Central and Eastern countries covering the period 1991–2002. Results from modified weighted least squares regression analyses show an increasing trend in the effect of years of education on standardized personal income for all five countries as predicted by the MTT. Income effects of years of experience, private sector employment and gender do not show increasing trends for all countries, contradicting the MTT. The empirical trends in effects of income determinants are to a great extent consistent with Stark's typology of privatization strategies, giving some support to the theoretical notion of path dependent transformation processes.