ISSN:0162-1459    

DOI:10.1080/01621459.1992.10475168

出版商:Taylor & Francis Group    

年代:1992

数据来源: Taylor

摘要:

Identification of ice floes and their outlines in satellite images is important for understanding physical processes in the polar regions, for transportation in ice-covered seas, and for the design of offshore structures intended to survive in the presence of ice. At present this is done manually, a long and tedious process that precludes full use of the great volume of relevant images now available. We describe an accurate and almost fully automatic method for identifying ice floes and their outlines. Floe outlines are modeled as closed principal curves, a flexible class of smooth nonparametric curves. We propose a robust method of estimating closed principal curves that reduces both bias and variance. Initial estimates of floe outlines come from the erosion-propagation (EP) algorithm, which combines erosion from mathematical morphology with local propagation of information about floe edges. The edge pixels from the EP algorithm are grouped into floe outlines using a new clustering algorithm. This extends existing clustering methods by allowing groups to be centered about principal curves rather than points or lines. This may open the way to efficient feature extraction using cluster analysis in images more generally. The method is implemented in an object-oriented programming environment, for which it is well suited, and is quite computationally efficient.

ISSN:0162-1459    

DOI:10.1080/01621459.1992.10475169

出版商:Taylor & Francis Group    

年代:1992

数据来源: Taylor

摘要:

The survival of like-sex twins born between 1881 and 1930 in Denmark, of which both were alive by the age of 15, is studied by means of models for bivariate survival data, with special reference to the degree of dependence. The 8,985 pairs were followed until 1980. The dependence is assumed to be generated by a common unobserved risk level. Several families of distributions for this level, including positive stable and gamma, are compared. General aspects of such data are discussed, including how to measure the degree of dependence. A simple choice is Kendall's coefficient of concordance, which for monozygotic males and females is about .17 and .15 and for dizygotic .09 and .10. Adjusting for cohort effects by a covariate describing year of birth reduces the dependence measure only slightly. The dependence is so small that a prediction for one twin is only slightly improved if the survival status of the other is known.

ISSN:0162-1459    

DOI:10.1080/01621459.1992.10475170

出版商:Taylor & Francis Group    

年代:1992

数据来源: Taylor

摘要:

Many studies have reported significant empirical associations between family structure during childhood and children's outcomes later in life. It may be that living in a nonintact family has adverse consequences for children. On the other hand, it may be that some unobserved process jointly determines family structure and children's outcomes. How then should one interpret the empirical evidence on the relationship between family structure and children's outcomes? The answer depends on the question asked and on the prior information available to the researcher. We seek to interpret the association between family structure and high school graduation found among respondents in the National Longitudinal Study of Youth. We seek to answer the traditional question of the literature on treatment effects: How would the probability of high school graduation vary with family structure if family structure were not selected by parents but were, instead, an exogenously assigned “treatment,” as in a clinical trial or other controlled experiment? The inferential problem is that the data alone do not suffice to identify the treatment effect. Hence any attempt to estimate a treatment effect depends critically on the prior information available to the researcher. We develop alternative estimates of the effect of family structure on high school graduation, obtained under differing assumptions about the actual process generating family structure and high school outcomes. We first assume strong prior information and present estimates of a set of parametric latent-variable models explaining family structure and children's outcomes. We then assume no prior information at all and report estimates of nonparametric bounds on the graduation probabilities. Finally, we give non-parametric estimates obtained under the assumption that family structure is exogenous with respect to high school graduation. Our empirical analysis strengthens the evidence that living in an intact family increases the probability that a child will graduate from high school. We also report that the probability of high school graduation increases markedly with both parents’ education, regardless of family structure. At the same time, we stress that prior information is necessary if one is to do more than bound the effect of family structure on children's outcomes. Any point estimate embodies prior information about the process generating family structure and children's outcomes. As long as social scientists are heterogenous in their beliefs about this process, their estimates of family-structure effects may vary.

ISSN:0162-1459    

DOI:10.1080/01621459.1992.10475171

出版商:Taylor & Francis Group    

年代:1992

数据来源: Taylor

摘要:

This article proposes new methods for modeling household fertility decisions. Much of the demographic literature on this subject suggests that decisions relating to fertility are influenced by the orientations (attitudes, desires, intentions) of both husbands and wives, but the methods used in previous work do not indicate how wife's (husband's) orientation influences husband's (wife's) orientation, nor how these separate phenomena are combined to produce a joint decision. As such, these methods cannot be used to model the process by which husbands and wives come to have similar or dissimilar orientations or to assess one of the key issues in the literature, namely, the relative influence of the husband (wife) on the subsequent decision. To address these issues, we construct and estimate a nonlinear simultaneous equation probit that has not been considered in previous statistical work. Specifically, we model the trivariate distribution of wife's stated desire for additional children, husband's stated desire for additional children, and subsequent fertility. In the model, the stated desire of the husband (wife) is viewed as an indicator of the husband's (wife's) latent disposition toward subsequent fertility. The husband's (wife's) disposition is allowed to depend on the wife's (husband's) disposition. The two dispositions are then combined to generate the couple's propensity for subsequent fertility. We show how such models can be estimated and tested and how the parameters can be used to assess the relative influence of each partner on the propensity. To illustrate the approach, we reanalyze data on post–World War II fertility from the Princeton Fertility Study. We find that recursive models do not fit the data, that both partners’ dispositions influence subsequent fertility and (under additional assumptions) that the relative influence of each partner on the couple's propensity is .5. Next, we show how to extend the basic framework to treat other types of decision making. Specifically, we take up the case where the observed variables are not necessarily binary, as in the example, and we discuss differences between the binary case and the more general cases. For such cases we also propose a number of new models that have not been considered in previous work. We also show how the framework can be extended to the case where the latent variables have multiple indicators, the case where multiple decisions are jointly made, and the case where the decision(s) depends on three or more decision makers.

ISSN:0162-1459    

DOI:10.1080/01621459.1992.10475172

出版商:Taylor & Francis Group    

年代:1992

数据来源: Taylor

摘要:

In 1976 the U.S. Congress removed the threat of withdrawal of certain highway funds from states that failed to enact motorcycle helmet laws. Since then over half the states have either repealed or weakened these laws. Most researchers in the field agree that this has lead to a significant increase in injuries and fatalities among motorcyclists involved in accidents. Potential limitations of many of the studies on which these conclusions are based include the facts that fatalities can result from injuries to parts of the body not protected by helmets and that other factors, such as speed and alcohol use, are not taken into account, usually due to lack of data. The former will result in a loss of power and the latter in the introduction of bias. In this article I model the level of head injury rather than the fatality rate and build a multivariate model that includes the other factors. The basic model is an ordered probit model with heteroscedasticity in the errors. The adequacy of the model is tested by Lagrange multiplier and goodness-of-fit tests. The former include tests for the normality of the errors and the specification of the regressors. Predictions from the model include that helmets lead to a 42% increase in the number of riders with no head injury and a $1,700 per rider decrease in the direct medical cost of treating the riders.

ISSN:0162-1459    

DOI:10.1080/01621459.1992.10475173

出版商:Taylor & Francis Group    

年代:1992

数据来源: Taylor

ISSN:0162-1459    

DOI:10.1080/01621459.1992.10475174

出版商:Taylor & Francis Group    

年代:1992

数据来源: Taylor

摘要:

Methods for estimating nonparametric models for conditional quantiles are suggested based on the regression quantile methods of Koenker and Bassett. Spline parameterizations of the conditional quantile functions are used. The methods are illustrated by estimating hierarchical models for household electricity demand using data from the Chicago metropolitan area. The empirical results show that lower quantiles of demand (“base-load”) vary only slightly across residential households. This variability is difficult to explain using household characteristics. However, upper quantiles of the demand distribution vary considerably and are systematically related to household characteristics and appliance ownership. The implications of analyzing mean demand behavior rather than various quantiles of the distribution of demand are also discussed.

ISSN:0162-1459    

DOI:10.1080/01621459.1992.10475175

出版商:Taylor & Francis Group    

年代:1992

数据来源: Taylor

摘要:

Human brain evoked potentials are elicited by a stimulus and can be recorded by scalp electrodes. Many researchers have proposed models in which evoked potentials are generated by equivalent electrical dipoles in the brain. Each dipole is defined by a set of parameters that specify its location, orientation, and magnitude. Existing approaches to estimation of dipole parameters do not realistically account for errors resulting from background brain electrical activity (“noise”) and thus lead to inefficient estimators and incorrect confidence sets. As an alternative, we derive frequency domain maximum likelihood estimators of the dipole parameters. The frequency domain approach simplifies the representation of the noise process and leads to substantial data reduction. The Fourier coefficients of the noise are approximately complex normal and independent across frequencies. This leads to a multivariate complex normal likelihood with a mean vector that is a nonlinear function of the dipole parameters. We compute the maximum likelihood estimates using iterative Fisher scoring. The results of a simulation study demonstrate that the parameter and standard error estimators are approximately unbiased when the model is correctly specified. An application to data from four subjects indicates that electrical activity approximately 50 milliseconds following an auditory click stimulus can be represented by an equivalent dipole in midline subcortical structures. We discuss the problem of model misspecification in applications to real data and describe possible approaches to improving the model and reducing bias due to misspecification.

ISSN:0162-1459    

DOI:10.1080/01621459.1992.10475176

出版商:Taylor & Francis Group    

年代:1992

数据来源: Taylor

摘要:

Radioactively labeled gene probes are observed as “grains” that identify specific sites of hybridization within chromosomes, suggesting potential sites for a target gene. Based on the resolution of the assay, chromosomes are subdivided into bins of approximately equal length, and the number of grains in each bin is tallied. To distinguish true hybridization sites from nonspecific binding, we propose a test statistic,Zmax, based on the number of grains in any two adjacent bins. An efficient algorithm for the exact null distribution ofZmaxis provided, as well as power calculations and a large-sample approximation. A step-down method that protects the experiment-wide error rate is given to identify multiple hybridization sites.

ISSN:0162-1459    

DOI:10.1080/01621459.1992.10475177

出版商:Taylor & Francis Group    

年代:1992

数据来源: Taylor