1. |
Selection of Subsets of Regression Variables |
|
Journal of the Statistical Society of London,
Volume 147,
Issue 3,
2018,
Page 389-410
Alan J. Miller,
Preview
|
PDF (1559KB)
|
|
摘要:
SUMMARYComputational algorithms for selecting subsets of regression variables are discussed. Only linear models and the least‐squares criterion are considered. The use of planar‐rotation algorithms, instead of Gauss–Jordan methods, is advocated. The advantages and disadvantages of a number of “cheap” search methods are described for use when it is not feasible to carry out an exhaustive search for the best‐fitting subsets.Hypothesis testing for three purposes is considered, namely (i) testing for zero regression coefficients for remaining variables, (ii) comparing subsets and (iii) testing for any predictive value in a selected subset. Three small data sets are used to illustrate these tests. Spj⊘tvoll's (1972a) test is discussed in detail, though an extension to this test appears desirable.Estimation problems have largely been overlooked in the past. Three types of bias are identified, namely that due to the omission of variables, that due to competition for selection and that due to the stopping rule. The emphasis here is on competition bias, which can be of the order of two or more standard errors when coefficients are estimated from the same data as were used to select the subset. Five possible ways of handling this bias are listed. This is the area most urgently requiring further research.Mean squared errors of prediction and stopping rules are briefly discussed. Competition bias invalidates the use of existing stopping rules as they are commonly applied to try to produce optimal prediction equations.
ISSN:0959-5341
DOI:10.2307/2981576
出版商:Wiley
年代:2018
数据来源: WILEY
|
2. |
Discussion of Dr Miller's Paper |
|
Journal of the Statistical Society of London,
Volume 147,
Issue 3,
2018,
Page 410-425
Preview
|
PDF (1316KB)
|
|
ISSN:0959-5341
DOI:10.1111/j.2397-2327.1984.tb00231.x
出版商:Wiley
年代:2018
数据来源: WILEY
|
3. |
Tests of Significance for 2 × 2 Contingency Tables |
|
Journal of the Statistical Society of London,
Volume 147,
Issue 3,
2018,
Page 426-449
F. Yates,
Preview
|
PDF (1795KB)
|
|
摘要:
SUMMARYFisher's exact test, and the approximation to it by the continuity‐corrected χ2test, have repeatedly been attacked over the past 40 years, recently with the support of extensive computer exercises. The present paper argues, on commonsense grounds, supported by simple examples, that these attacks are misconceived, and are mainly due to uncritical acceptance of the Neyman–Pearson approach to tests of significance, the use of nominal levels, and refusal to accept the arguments for conditioning on the margins.Two‐sided tests have also added to the confusion; it is argued that the best definition of a two‐sided probability is twice the observed one‐tail probability.
ISSN:0959-5341
DOI:10.2307/2981577
出版商:Wiley
年代:2018
数据来源: WILEY
|
4. |
Discussion of Dr Yates's Paper |
|
Journal of the Statistical Society of London,
Volume 147,
Issue 3,
2018,
Page 449-463
Preview
|
PDF (1304KB)
|
|
ISSN:0959-5341
DOI:10.1111/j.2397-2327.1984.tb00233.x
出版商:Wiley
年代:2018
数据来源: WILEY
|
5. |
The Lognormal Distribution of the Duration of Strikes |
|
Journal of the Statistical Society of London,
Volume 147,
Issue 3,
2018,
Page 464-483
R. J. Lawrence,
Preview
|
PDF (1169KB)
|
|
摘要:
SUMMARYHow long strikes last is one measure of the state of industrial relations, used in year‐to‐year and international comparisons. Strike duration has been modelled as a stochastic process, and causal explanations have been drawn from the model used. But the lognormal distribution fits 1965 UK strike duration data, and possibly also US figures, previously analysed as inverse Gaussian or Weibull distributed. The log‐normal has the advantage of accounting for short strike numbers, and offers much greater economy of parameters. It suggests that strike durations were not affected by cause of dispute or by type of industry, except in the case of dockers and miners, in the 1960s. After 1975, data covering all UK strikes conform to the lognormal distribution except during the Winter of Discontent. Mean and variance change from year to year, and appear to be negatively correlated. Methods of fitting a left‐truncated lognormal distribution to discrete data, the use of the hazard function, and the shape of expected residual lifetime curves are discussed.
ISSN:0959-5341
DOI:10.2307/2981578
出版商:Wiley
年代:2018
数据来源: WILEY
|
6. |
Estimating Time Series Models Using the Relevant Forecast Evaluation Criterion |
|
Journal of the Statistical Society of London,
Volume 147,
Issue 3,
2018,
Page 484-487
A. A. Weiss,
A. P. Andersen,
Preview
|
PDF (265KB)
|
|
摘要:
SUMMARYThe usual procedure in time series analysis is for the parameters to be estimated using ordinary least squares. However, the forecasts are often evaluated using a totally different criterion, mean absolute percentage error, for example. Here, via the examination of a large number of data series, we examine the effects of using the forecast evaluation criterion as the estimation criterion.
ISSN:0959-5341
DOI:10.2307/2981579
出版商:Wiley
年代:2018
数据来源: WILEY
|
7. |
The Bivariate Generalized Waring Distribution and its Application to Accident Theory |
|
Journal of the Statistical Society of London,
Volume 147,
Issue 3,
2018,
Page 488-498
Evdokia Xekalaki,
Preview
|
PDF (666KB)
|
|
摘要:
SUMMARYThe univariate generalized Waring distribution was shown by Irwin (1968, 1975) to provide a useful accident model which enables one to split the variance into three additive components due to randomness, proneness and liability. The two non‐random variance components, however, cannot be separately estimated.In this paper a way of tackling this problem is suggested by defining a bivariate extension of the generalized Waring distribution. Using this it is possible to obtain distinguishable estimates for the variance components and hence inferences can be made about the role of the underlying accident factors. The technique is illustrated by two examples.
ISSN:0959-5341
DOI:10.2307/2981580
出版商:Wiley
年代:2018
数据来源: WILEY
|
8. |
Tests of Three Hypotheses Relating to the Leontief Input–Output Model |
|
Journal of the Statistical Society of London,
Volume 147,
Issue 3,
2018,
Page 499-509
Roger H. Bezdek,
Preview
|
PDF (834KB)
|
|
摘要:
SUMMARYThis paper examines three hypotheses central to the Leontief economic input–output model: (1) constant dollar coefficients are more stable than current dollar coefficients; (2) the intermediate coefficients are more stable than the final demand coefficients; and (3) aggregate coefficients are more stable than disaggregate coefficients. These hypotheses are first discussed and then tested empirically using data from the 1947, 1958, 1963, 1967 and 1972 US input–output studies. Our empirical analysis leads us to tentatively reject the first hypothesis and accept the second and third hypotheses. We conclude with a brief discussion of the implications of these findings and with some suggestions for further research.
ISSN:0959-5341
DOI:10.2307/2981581
出版商:Wiley
年代:2018
数据来源: WILEY
|
9. |
Statistical Methods for Forecasting |
|
Journal of the Statistical Society of London,
Volume 147,
Issue 3,
2018,
Page 510-511
W. D. Ray,
Preview
|
PDF (162KB)
|
|
摘要:
1. Statistical Methods for Forecasting.By Bovas Abraham and Johannes Ledolter. Chichester, Wiley, 1983, xv. 445 p. £28.65.
ISSN:0959-5341
DOI:10.2307/2981583
出版商:Wiley
年代:2018
数据来源: WILEY
|
10. |
The Biomathematics of Malaria |
|
Journal of the Statistical Society of London,
Volume 147,
Issue 3,
2018,
Page 511-512
J. Radcliffe,
Preview
|
PDF (180KB)
|
|
摘要:
2. The Biomathematics of Malaria.By Norman T. J. Bailey. London and High Wycombe, Charles Griffin, 1982. xii, 210 p. £16.50.
ISSN:0959-5341
DOI:10.2307/2981584
出版商:Wiley
年代:2018
数据来源: WILEY
|