|
1. |
Modeling the Labeling Index Distribution: An Application of Functional Data Analysis |
|
Journal of the American Statistical Association,
Volume 90,
Issue 431,
1995,
Page 813-821
PatriciaM. Grambsch,
BryanL. Randall,
RoberdM. Bostick,
JohnD. Potter,
ThomasA. Louis,
Preview
|
PDF (837KB)
|
|
摘要:
This article presents exploratory data analytic methodology for visualizing and summarizing data that can be represented as individual-specific curves. We propose a simplified form of functional data analysis. A nonparametric scatterplot smooth is applied to each individual's data, followed by a principal components analysis of the smoothed data. We then display the individual smooth curves in an array organized by principal component scores. The display suggests interpretable summary measures. The methodology is applied to the measurement of proliferative activity, a biomarker for colon cancer risk. We use the summary measures in the analysis of a pilot study clinical trial.
ISSN:0162-1459
DOI:10.1080/01621459.1995.10476579
出版商:Taylor & Francis Group
年代:1995
数据来源: Taylor
|
2. |
Handling “Don't Know” Survey Responses: The Case of the Slovenian Plebiscite |
|
Journal of the American Statistical Association,
Volume 90,
Issue 431,
1995,
Page 822-828
DonaldB. Rubin,
HalS. Stern,
Vasja Vehovar,
Preview
|
PDF (804KB)
|
|
摘要:
The critical step in the drive toward an independent Slovenia was the plebiscite held in December 1990, at which the citizens of Slovenia voted overwhelmingly in favor of a sovereign and independent state. The Slovenian Public Opinion (SPO) survey of November/December 1990 was used by the government of Slovenia to prepare for the plebiscite. Because the plebiscite counted as “YES voters” only those voters who attended and voted for independence (nonvoters counted as “NO voters”), “Don't Know” survey responses can be thought of as missing data—the true intention of the voter is unknown but must be either “YES” or “NO.” An analysis of the survey data under the missing-at-random assumption for the missing responses provides remarkably accurate estimates of the eventual plebiscite outcome, substantially better than ad hoc methods and a nonignorable model that allows nonresponse to depend on the intended vote.
ISSN:0162-1459
DOI:10.1080/01621459.1995.10476580
出版商:Taylor & Francis Group
年代:1995
数据来源: Taylor
|
3. |
Modeling Dose and Local Control in Radiotherapy |
|
Journal of the American Statistical Association,
Volume 90,
Issue 431,
1995,
Page 829-838
Rick Chappell,
DavidM. Nondahl,
JohnF. Fowler,
Preview
|
PDF (1214KB)
|
|
摘要:
We discuss models for predicting local control (prevention of tumor recurrence) after therapeutic radiation in cancer patients. The probability of control is first formulated from theoretical precepts. Biophysical principles dictate that the three factors in therapy that most universally influence outcome are total dose, number of sessions in which the dose is administered, and total time under treatment. We show that these principles also suggest the scale, or link function, on which local control probability for a tumor of given size is a linear function of these predictors. The probabilities are given clinical relevance by assigning a mixing distribution to tumor size; effective size, the number of actively dividing cells in a tumor, is an unmeasurable but of course quite influential quantity. We show in this case that a gamma distribution on tumor size induces linearity on a subset of the class of links first proposed by Burr. Next, we discuss methods of modeling control by a finite follow-up time. We demonstrate a new result, that minor assumptions on the effects of size on recurrence allow models developed for permanent control to be applied directly to recurrence by a finite time. We also describe adjustments for accommodating losses to follow-up before that time. Finally, we develop inference on the mixing distribution of tumor size along with results of the effect of misspecifying the distribution. We illustrate the methods with a new analysis of a radiotherapy study. Though developed for a specific type of failure data, many of the results also apply to any time-dependent binary outcome.
ISSN:0162-1459
DOI:10.1080/01621459.1995.10476581
出版商:Taylor & Francis Group
年代:1995
数据来源: Taylor
|
4. |
Estimating Products in Forensic Identification Using DNA Profiles |
|
Journal of the American Statistical Association,
Volume 90,
Issue 431,
1995,
Page 839-844
DavidJ. Balding,
Preview
|
PDF (658KB)
|
|
摘要:
In many areas, such as reliability and forensic identification, it is of interest to estimate a product of unknown parameters, each of which is estimated from test data. The maximum likelihood estimator usually used in both applications is approximately unbiased but has an asymmetric sampling distribution, particularly when the sample size is not large. Consequently, most estimates understate the parameter value, often substantially, and this may be a serious drawback if “overoptimism” is highly undesirable. Further, widely used methods of interval estimation based on confidence limits suffer from a range of theoretical and computational problems. Alternative methods of estimation are developed and discussed, based on frequentist, Bayesian, and likelihood approaches.
ISSN:0162-1459
DOI:10.1080/01621459.1995.10476582
出版商:Taylor & Francis Group
年代:1995
数据来源: Taylor
|
5. |
Regression Models for a Bivariate Discrete and Continuous Outcome with Clustering |
|
Journal of the American Statistical Association,
Volume 90,
Issue 431,
1995,
Page 845-852
GarrettM. Fitzmaurice,
NanM. Laird,
Preview
|
PDF (764KB)
|
|
摘要:
Developmental toxicity studies of laboratory animals play a crucial role in the testing and regulation of chemicals and pharmaceutical compounds. Exposure to developmental toxicants typically causes a variety of adverse effects, such as fetal malformations and reduced fetal weight at term. In this article, we discuss regression methods for jointly analyzing bivariate discrete and continuous outcomes that are motivated by the statistical problems that arise in analyzing data from developmental toxicity studies. We focus on marginal regression models; that is, models in which the marginal expectation of the bivariate response vector is related to a set of covariates by some known link functions. In these models the regression parameters for the marginal expectation are of primary scientific interest, whereas the association between the binary and continuous response is considered to be a nuisance characteristic of the data. We describe a likelihood-based approach, based on the general location model of Olkin and Tate, that yields maximum likelihood estimates of the marginal mean parameters that are robust to misspecification of distributional assumptions. Finally, we describe an extension of this model to allow for clustering, using generalized estimating equations, a multivariate analog of quasi-likelihood. A motivating example, using fetal weight and malformation data from a developmental toxicity study of ethylene glycol in mice, illustrates this methodology.
ISSN:0162-1459
DOI:10.1080/01621459.1995.10476583
出版商:Taylor & Francis Group
年代:1995
数据来源: Taylor
|
6. |
Spatial Interpolation Errors for Monitoring Data |
|
Journal of the American Statistical Association,
Volume 90,
Issue 431,
1995,
Page 853-861
Gudmund Høst,
Henning Omre,
Paul Switzer,
Preview
|
PDF (803KB)
|
|
摘要:
Separate modeling of the spatial mean field, the spatial variance field, and the space-time residual fields can give a more detailed and possibly more accurate representation of spatial interpolation errors when we have repeated observations on a fixed monitoring network. This article gives expressions for the spatial interpolation errors in terms of the statistics of the component fields, which enable us to assess the relative importance of different kinds of uncertainty. This modeling approach is applied to data of sulfur dioxide concentrations in Europe, and a comparison with neighborhood kriging is done by means of cross-validation.
ISSN:0162-1459
DOI:10.1080/01621459.1995.10476584
出版商:Taylor & Francis Group
年代:1995
数据来源: Taylor
|
7. |
Threshold Models for Combination Data from Reproductive and Developmental Experiments |
|
Journal of the American Statistical Association,
Volume 90,
Issue 431,
1995,
Page 862-870
PamelaF. Schwartz,
Chris Gennings,
VernonM. Chinchilli,
Preview
|
PDF (1018KB)
|
|
摘要:
In risk assessment, thresholds are generally believed to exist for toxic reproductive and developmental agents. Therefore, dose combinations exist below the threshold where the response is not distinguishable from background and above the threshold where a dose-response trend results. Because data from combination experiments are more reflective of human exposure to environmental agents, threshold models that incorporate the effect of interactions among these agents are developed. Forsagents acting in combination, the threshold is ans-dimensional surface. More specifically, for a combination of two agents, the threshold is a two-dimensional contour of dose combinations, and for a single agent, the threshold is represented by a single dose value, not necessarily a dose level used in the experiment. In toxicological experiments designed to investigate the reproductive or developmental effects of a chemical on laboratory animals, litter effects typically are present, yielding overdispersion. Instead of assuming a particular distributional form for the binary responses, quasi-likelihood methods often are used where parameters are introduced to measure the overdispersion. This article describes threshold models for combination reproductive and developmental experiments and develops parameter estimation techniques using quasi-likelihood methods. Once parameters are estimated, a hypothesis test for an overall dose-response relationship is performed using a quasi-likelihood ratio statistic. Next, a confidence interval about the threshold parameter is constructed using a quasi-likelihood ratio statistic, and finally, a confidence region is constructed about the threshold surface. Illustrative examples are presented that use single-agent reproductive data from the National Toxicology Program and three-agent combination developmental data from the Environmental Protection Agency and ManTech Environmental Technology, Inc.
ISSN:0162-1459
DOI:10.1080/01621459.1995.10476585
出版商:Taylor & Francis Group
年代:1995
数据来源: Taylor
|
8. |
A Saturated Model for Analyzing Exchangeable Binary Data: Applications to Clinical and Developmental Toxicity Studies |
|
Journal of the American Statistical Association,
Volume 90,
Issue 431,
1995,
Page 871-879
Dale Bowman,
E.Olusegun George,
Preview
|
PDF (817KB)
|
|
摘要:
Correlated binary data occur very frequently in statistical practice. In many applications, it is reasonable to assume that data from the same cluster are exchangeable. Such data are commonly encountered in cluster sample surveys, teratological experiments, ophthalmologic and otolaryngologic studies, and other clinical trials. The standard methods of analyzing these data include the use of beta-binomial models and generalized estimating equations with third and fourth moments specified by “working matrices.” The focus of these procedures is an estimation of the mean and variance parameters. More information can be obtained when data are exchangeable. By expressing the joint distribution of a set of exchangeable binary random variables in terms of the probability of similar response within cluster, this article introduces a procedure for obtaining maximum likelihood estimates of population parameters such as the marginal means, moments, and correlations of orders two and higher. Applications are made to data sets from a clinical trial and a developmental toxicity study.
ISSN:0162-1459
DOI:10.1080/01621459.1995.10476586
出版商:Taylor & Francis Group
年代:1995
数据来源: Taylor
|
9. |
Optimal Confidence Sets, Bioequivalence, and the Limaçon of Pascal |
|
Journal of the American Statistical Association,
Volume 90,
Issue 431,
1995,
Page 880-889
LawrenceD. Brown,
George Casella,
J.T. Gene Hwang,
Preview
|
PDF (897KB)
|
|
摘要:
We begin with a decision-theoretic investigation into confidence sets that minimize expected volume at a given parameter value. Such sets are constructed by inverting a family of uniformly most powerful tests, and hence they also enjoy the optimality property of being uniformly most accurate. In addition, these sets possess Bayesian optimal volume properties and represent the first case (to our knowledge) of a frequentist 1 – α confidence set that possesses a Bayesian optimality property. The hypothesis testing problem that generates these sets is similar to that encountered in bioequivalence testing. Our sets are optimal for testing bioequivalence in certain settings; in the case of the normal distribution, the optimal set is a curve known as the limaçon of Pascal. We illustrate the use of these curves with a biopharmaceutical example.
ISSN:0162-1459
DOI:10.1080/01621459.1995.10476587
出版商:Taylor & Francis Group
年代:1995
数据来源: Taylor
|
10. |
Estimation of the Number of True Gray Levels, Their Values, and Relative Frequencies in a Noisy Image |
|
Journal of the American Statistical Association,
Volume 90,
Issue 431,
1995,
Page 890-899
Fred Godtliebsen,
Chih-Kang Chu,
Preview
|
PDF (1675KB)
|
|
摘要:
In some applications information is presented as a two-dimensional image corrupted by random noise. Due to the precision of the equipment that forms the image, we can typically have a large number,v, of observed gray levels. But in many situations we know that the number of true gray levels,p, corresponding to, for example, the number of tissue types in a brain slice, is much less thanv. In this article we propose a method based on the kernel density estimator for estimating thepunderlying true gray levels and their relative frequencies. The strong convergence rates for estimators of these quantities are established. The method is successfully applied to artificial and magnetic resonance images.
ISSN:0162-1459
DOI:10.1080/01621459.1995.10476588
出版商:Taylor & Francis Group
年代:1995
数据来源: Taylor
|
|