|
1. |
Assumptions for Statistical Inference |
|
The American Statistician,
Volume 47,
Issue 1,
1993,
Page 1-11
GeraldJ. Hahn,
WilliamQ. Meeker,
Preview
|
PDF (1280KB)
|
|
摘要:
In this article we overview and discuss some of the important practical assumptions underlying statistical inference. What we say, though not new, is stressed insufficiently in teaching statistical methods and applications. We build on the important conceptual difference between enumerative and analytic studies, emphasized by W. Edwards Deming. Our comments, however, go beyond the published views of Deming. We emphasize the assumptions needed for both types of studies and illustrate the concepts with examples.
ISSN:0003-1305
DOI:10.1080/00031305.1993.10475924
出版商:Taylor & Francis Group
年代:1993
数据来源: Taylor
|
2. |
A Remarkable Scatterplot |
|
The American Statistician,
Volume 47,
Issue 1,
1993,
Page 12-19
Ian Spence,
RobertF. Garrison,
Preview
|
PDF (845KB)
|
|
摘要:
The Hertzsprung-Russell diagram, invented by astronomers just after the turn of the century, has had a profound influence on almost every aspect of modern astronomy, but especially on theories of stellar evolution. The diagram is an excellent pedagogical example of a scatterplot and is remarkable on many counts: it is an impressive illustration of the power of graphical display; it may be used to show the strengths and limitations of smoothing, or the fitting of parametric functions; and variations of the plot can be used to represent change over time, large numbers of data points, or differing relationships in distinct strata.
ISSN:0003-1305
DOI:10.1080/00031305.1993.10475925
出版商:Taylor & Francis Group
年代:1993
数据来源: Taylor
|
3. |
A Note on Obtaining Confidence Intervals for Discrete Parameters |
|
The American Statistician,
Volume 47,
Issue 1,
1993,
Page 20-23
Maureen Tingley,
Chunhui Li,
Preview
|
PDF (366KB)
|
|
摘要:
Peskun considers the problem of constructing a closed confidence interval for a single parameter, using a discrete test statistic. The interval is constructed by generalizing the method of Konijn. Peskun gives an example where the interval fails to be well defined. In this article a confidence interval is constructed by generalizing the method of Cochran, in a manner similar to the way in which Peskun generalizes Konijn's technique. Buonaccorsi compares the Konijn and Cochran type intervals in the hypergeometric setting. This article shows that the Cochran interval is always well defined and contains the Peskun interval, and that the Cochran interval contains, at most, two points more than the Peskun interval. Discrepancies between the two kinds of intervals, and ill definition of the Peskun interval, follow from discreteness of the parameter space, rather than discreteness of the test statistic.
ISSN:0003-1305
DOI:10.1080/00031305.1993.10475926
出版商:Taylor & Francis Group
年代:1993
数据来源: Taylor
|
4. |
An Improved Modification of a Regression-Toward-the-Mean Demonstration |
|
The American Statistician,
Volume 47,
Issue 1,
1993,
Page 24-26
JoelR. Levin,
Preview
|
PDF (351KB)
|
|
摘要:
Earlier simulations illustrating regression toward the mean have been improved so that they bear a closer resemblance to the actual phenomenon. In addition to the improved correspondence, the modified simulation permits a clear understanding of regression artifacts in two quasi-experimental designs, theextreme groupsdesign and thematching-from-different-populationsdesign. Highlighted in the simulation are: (1) the inverse relationship between measurement reliability (or between-measure correlations) and the magnitude of regression effects; (2) the stability of means and variances in the presence of regression effects; and (3) the time-reversed character of regression effects.
ISSN:0003-1305
DOI:10.1080/00031305.1993.10475927
出版商:Taylor & Francis Group
年代:1993
数据来源: Taylor
|
5. |
The Relationship between the Magnitudes of SSR(x2) and SSR(x2|x1): A Geometric Description |
|
The American Statistician,
Volume 47,
Issue 1,
1993,
Page 26-30
HarryM. Schey,
Preview
|
PDF (438KB)
|
|
摘要:
We use geometric methods to investigate the relative magnitudes of SSR(x2), the sum of squares for regression onx2alone, and SSR(x2|x1), the increase in the regression sum of squares resulting from the addition ofx2to a model that already containsx1. We examine a variety of cases, emphasizing those in which SSR(x2|x1) > SSR(x2). We also point out that SSR(x2) and SSR(x2|x1) can be equal even whenx1andx2are correlated. We present contrived data sets illustrating these points, and examine the relative magnitudes of SSR(x2) and SSR(x2|x1) for two real data sets.
ISSN:0003-1305
DOI:10.1080/00031305.1993.10475928
出版商:Taylor & Francis Group
年代:1993
数据来源: Taylor
|
6. |
A Geometric Interpretation of Partial Correlation Using Spherical Triangles |
|
The American Statistician,
Volume 47,
Issue 1,
1993,
Page 30-32
Guy Thomas,
John O'quigley,
Preview
|
PDF (247KB)
|
|
摘要:
This article shows how spherical triangles may be helpful in interpreting and visualizing the relations between partial and simple correlations. The formula giving the partial correlation coefficient in terms of the pairwise simple correlations is seen to be identical with the fundamental formula of spherical trigonometry. The spherical representation is applied to illustrate the masking effect of one variable on another in multiple linear regression.
ISSN:0003-1305
DOI:10.1080/00031305.1993.10475929
出版商:Taylor & Francis Group
年代:1993
数据来源: Taylor
|
7. |
Characteristic Functions and Their Empirical Counterparts: Geometrical Interpretations and Applications to Statistical Inference |
|
The American Statistician,
Volume 47,
Issue 1,
1993,
Page 33-38
T.W. Epps,
Preview
|
PDF (617KB)
|
|
摘要:
The value of a characteristic function of a random variableXat some real numbertis the center of mass of the distribution oftXwrapped around the unit circle in the complex plane. This geometrical representation is used to illustrate how various properties of frequency functions and characteristic functions correspond and to illuminate the role of empirical characteristic functions in statistical inference.
ISSN:0003-1305
DOI:10.1080/00031305.1993.10475930
出版商:Taylor & Francis Group
年代:1993
数据来源: Taylor
|
8. |
A Note on the Kaplan—Meier Estimator |
|
The American Statistician,
Volume 47,
Issue 1,
1993,
Page 39-40
David Oakes,
Preview
|
PDF (242KB)
|
|
摘要:
An anomalous feature of the Kaplan-Meier estimator is that certain estimated survival probabilities can be decreased when the data are perturbed in a way that improves the overall group survival. An alternative estimator based on the so-called reduced-sample method does not have this disadvantage. However, its construction requires knowledge of the potential censoring times of subjects who were actually observed to fail. If these censoring times are unknown, it is natural to proceed by imputing them. A self-consistency argument shows that this leads back to the Kaplan-Meier estimator. Although the method does not yield a new estimator, it gives some insight into the anomalous behavior of the original one.
ISSN:0003-1305
DOI:10.1080/00031305.1993.10475931
出版商:Taylor & Francis Group
年代:1993
数据来源: Taylor
|
9. |
Accent on Teaching Materials |
|
The American Statistician,
Volume 47,
Issue 1,
1993,
Page 40-42
HarryO. Posten,
Preview
|
PDF (358KB)
|
|
摘要:
A Computer Package for Teaching and Learning StatisticsUnderstanding Statistics, a suite of six programs forIBMcompatible computers equipped with EGA (256K video memory required) or VGA displays, is available from the Centre for Statistics Education, University of Sheffield, Sheffield S3 7RH, England, on either 3.5- or 5.25-inch disks: U.S. $63 by airmail or U.S. $58 by sea. Reviewed by Rosemary A. Roberts
ISSN:0003-1305
DOI:10.1080/00031305.1993.10475932
出版商:Taylor & Francis Group
年代:1993
数据来源: Taylor
|
10. |
Confidence Regions for Parameter Pairs |
|
The American Statistician,
Volume 47,
Issue 1,
1993,
Page 43-45
J.B. Douglas,
Preview
|
PDF (229KB)
|
|
摘要:
Marginal confidence intervals for pairs of parameters are easily misinterpreted if the estimators are not nearly uncorrelated. But transforming to (nearly) orthogonal parameters may complicate physical interpretations, and may not be necessary if confidence regions are used directly.
ISSN:0003-1305
DOI:10.1080/00031305.1993.10475933
出版商:Taylor & Francis Group
年代:1993
数据来源: Taylor
|
|