Search

Experimental design considerations have been virtually ignored by economists who have conducted computer simulation experiments with models of economic systems. The objective of this paper is to spell out in detail the relationship between existing experimental design techniques and techniques of data analysis and the design of simulation experiments with economic systems. We begin by defining the problem of experimental design as applied to computer simulation experiments. With the aid of an example model, we explore several techniques of data analysis and a number of specific experimental design problems. Although this paper is oriented towards the design of computer simulation experiments in economics, the techniques which are discussed are of a general nature and should be applicable to the design of simulation experiments in other disciplines. [ABSTRACT FROM AUTHOR]

Some optimum nonparametric procedures for estimating and testing contrasts in two-way layouts are proposed and studied. These procedures are based on the Chernoff-Savage [4] type of rank order statistics which include the Wilcoxon and normal scores statistics among others. In the first three sections, the properties of the proposed point estimators of contrasts, such as symmetry, invariance and asymptotic normality, are studied and their asymptotic relative efficiencies with respect to the corresponding least-squares estimators are obtained. In particular, it is shown that the procedures based on the normal scores statistics are asymptotically at least as efficient as the corresponding procedures based on the method of least squares, whatever be parent cumulative distribution functions. In sections 4 and 5, the corresponding problems of testing and confidence intervals are discussed, and generalized to two-way layouts with several observations per cell. [ABSTRACT FROM AUTHOR]

A simple derivation of asymptotic efficiency for the two sample Kolmogorov-Smirnov statistic is given and evaluated for normal location and normal scale alternatives. Using equal samples to simplify the derivation, the limiting efficiency is obtained by letting the type I error go to zero while the type II error goes to beta, 0 < beta < 1. For unimodal symmetric location alternatives, the efficiency is the same as that obtained for the Mood and Brown median test. Limits of relative efficiencies for alternatives which approach the null hypothesis are 2/pi for normal location alternatives and (pi e)[sup -1] for normal scale alternatives. [ABSTRACT FROM AUTHOR]

The article presents a table of the values of the power of the F-test corresponding to the degrees of freedom. The use of F-test while using a classical analysis of variance model, rejects the null hypothesis at the significance level. The values of &b.beta;, the Type II error, are tabulated to three decimal places. In this article the values of &b.beta; are tabulated to four decimal places and to obtain the values of &b.beta; for odd values, a linear interpolation may be used which generally gives three decimal place accuracy. The computation was terminated when the values of &b.beta; no longer changed in the sixth decimal figure. These values agree with the values of the power-function of the non-central chi-square distribution.

The problem of iterated tests of hypotheses has been treated from the standpoint of multiple decision theory. Under the assumption that there is a UMP test (not necessarily similar) for each component hypothesis, a procedure has been developed which ensures an upper bound for various probabilities of misclassification and which maximizes the probabilities of correct classification in some class. [ABSTRACT FROM AUTHOR]

The article discusses a class of modified distribution free measures of goodness of fit. A random variable with the continuous probability distribution function is defined. There are various measures of the difference between a specified continuous distribution function and the sample distribution function. The discrepancies between the continuous and the empirical distribution function are jointly distribution free. They can also be expressed as any function and may be regarded as a distribution free statistic of goodness of fit. Even the sum of all deviations can be regarded as a one-sided test statistic. Short-cut methods can be defined using the statistics for one-sided tests and for two-sided tests. To get the exact null distributions, one can make use of the fact that the discrepancies are unchanged by any move of the sample point within an interval. Thus one can reduce the problems to the classical urn model. By means of a computer the exact null distribution and critical values for small number is calculated.

The standard tables used for the Kolmogorov-Smirnov test are valid when testing whether a set of observations are from a completely specified continuous distribution. If one or more parameters must be estimated from the sample then the tables are no longer valid. A table is given in this note for use with the Kolmogorov-Smirnov statistic for testing whether a set of observations is from a normal population when the mean and variance are not specified but must be estimated from the sample. The table is obtained from a Monte Carlo calculation. A brief Monte Carlo investigation is made of the power of the test. [ABSTRACT FROM AUTHOR]

Abstract The procedures of Scheffe [4] and Welch [5] for testing equality of means of two normal populations are compared according to the expected lengths of the confidence intervals they yield, and a criterion is developed for deciding which of the two methods is better in any given situation. An expression for the expected length of the confidence interval for the Scheffe procedure was developed by Scheffe. Because of the asymptotic series nature of the Welch procedure, in this case, the analogous expression is more involved. It is developed below. A numerical computation was carried out for the case of a two-sided 90 per cent confidence interval. The results are presented in graphical form. Finally, it is shown that in large samples, the two procedures are equivalent. [ABSTRACT FROM AUTHOR]

Abstract The expected significance level (ESL) of a test, proposed by Dempster and Schatzoff (4) as a criterion for comparing sensitivities of competing test statistics, is defined as the expected value of the observed significance level under a simple alternative hypothesis. For example, the ESL of a lower tailed test of the hypothesis H[sub 0]: X is similar to F(x) against H[sub 1]: X is similar to G(x) is given by Integral of F(x)dG(x). In this paper, the sensitivities of several proposed tests of the general linear hypothesis are compared by simulating their ESL's under a wide variety of parameterizations. The results indicate that Wilks's likelihood ratio criterion A and Hotelling's T[sup 2, sub 0] provide about the same levels of protection over a wide spectrum of alternatives, whereas Roy's largest characteristic root is shown to be rather insensitive to alternatives involving more than one non-zero root. Other proposed test statistics are shown to be less sensitive as judged by the ESL criterion. [ABSTRACT FROM AUTHOR]

An investigation of stratification is reported for a universe with strong positive skewness. The choice of stratification variable, the method of constructing strata, the sample allocation, and the number of strata are considered. Comparisons are made among four types of allocation in combination with the corresponding optimum stratifications. For four simpler methods of stratification, equal allocation and allocation proportional to stratum aggregates are contrasted. Gain from stratification are examined for several estimation variables. Illustrations utilize data for the 1961 universe of United States nonfederal general medical hospitals for short-term care. However, the finding apply to the stratification of other populations having similar distribution functions. [ABSTRACT FROM AUTHOR]

The critical regions for the uniformly most powerful unbiased tests of ... and ..., where σ² is the variance of a normal distribution with unknown mean, can be expressed in terms of equations which require numerical methods for solution. Graphs are included from which these critical regions may be read for significance levels α = .05, .10. The graphs may also be used to obtain critical regions for the corresponding hypotheses concerning the parameters of two other distributions. Calculation of power is discussed. [ABSTRACT FROM AUTHOR]

A number of examples which involve restricted alternatives are given; each of these, except one, deals with normal distributions. The examples demonstrate the following: (a) one way to construct a simple test when the alternative hypothesis is restricted, (b) a method by which the additional restrictions can be justified, and (c) a procedure that can be used to defend the original model. An interesting relationship between Bartholomew's chi[sup 2] and a test statistic, which is used in this paper, is pointed out. Finally, a certain distribution-free problem and a solution of it are proposed. [ABSTRACT FROM AUTHOR]

This paper compares the Pearson Chi-Square and Kolmogorov goodness-of-fit tests with respect to validity under the following conditions: (1) the N independent observations are tabulated and arranged into k mutually exclusive groups that are equally probable under the hypothesis to be tested; and (2) both N and k are "small"; i.e., not greater than 50. A random sampling experiment was performed, and the results show that in general for the conditions considered, the Pearson test is more valid than the Kolmogorov test. [ABSTRACT FROM AUTHOR]

The Kolmogorov-Smirnov test is important in nonparametric statistical inference. The limiting distribution of the Kolmogorov-Smirnov statistic under the null hypothesis is well known and has been derived by several different methods. The limiting distribution of this statistic under the alternative hypothesis is unknown so that it is not possible, in general, to compute the power of the Kolmogorov-Smirnov test. It is possible, however, to compute a lower bound for the power of this test. This result is used to calculate a lower bound for the asymptotic efficiency of the Kolmogorov-Smirnov test relative to the optimum likelihood ratio test. Applications are given to Cauchy, exponential and normal populations. In addition, a lower bound for the asymptotic efficiency of the Kolmogorov-Smirnov test relative to the t-test is computed for translation alternatives. [ABSTRACT FROM AUTHOR]

This paper presents the results of a limited investigation which brings into focus the difficulties encountered in developing exact distribution free methods for stratified simple random sampling. It is shown that exact distribution-free confidence intervals for a population median can be obtained from a stratified sample under very special circumstances. The principal result concerns the order statistics of the combined separate strata samples, where proportional allocation is employed. It is proved that a pair of symmetric order statistics provides a confidence interval for the population median whose confidence coefficient is not less than the confidence coefficient obtained from the corresponding order statistics of a random sample of the same total size drawn from the entire population. Furthermore, it is shown by means of a numerical example that this result necessarily holds only when proportional allocation is employed. [ABSTRACT FROM AUTHOR]

After a brief motivating discussion of approaches to significance tests in Section 1, the concept of expected significance level (ESL) is defined and its basic properties are presented in Section 2. Section 3 discusses the relationship between ESL and the familiar criterion of power. While a complete description of power for each size alpha is more informative than ESL, it is argued that ESL is often a reasonable compromise between high ideals and practicability. Section 4 gives two estimation methods for Monte Carlo simulation of ESL and shows how to estimate variances and covariances for the second of these methods. The first method leads to the familiar Wilcoxon two-sample statistic which is clearly a sample analogue of ESL. The second and more apposite method leads to a modification of the Wilcoxon statistic. [ABSTRACT FROM AUTHOR]

Two exact tests are presented for testing the hypothesis that the residuals from a least squares regression are homoscedastic. The results can be used to test the hypothesis that a linear [ratio] model explains the relationship between variables as opposed to the alternative that the ratio [linear ] specification is correct. The first test is parametric and uses the F-statistic. The second test is nonparametric and uses the number of peaks in the ordered sequence of unsigned residuals. In conclusion, the results of some experimental calculations of the powers of the tests are discussed. [ABSTRACT FROM AUTHOR]

This paper deals with the asymptotic relative efficiency (A.R.E.) and the asymptotic power of certain tests. These tests are distribution-free goodness-of-fit tests based on the empirical distribution function. They are the one and two sided Kolmogorov-Smirnov tests and a new one based on the absolute values of the order statistics. The method of obtaining the A.R.E. is due to Quade [5] and is outlined in Section 1. In this paper we are concerned with the case of scalar alternatives. The asymptotic power for the Kolmogorov-Smirnov tests is obtained in Section 2. Section 3 deals with a new test which allows a more facile use of a one-sided statistic while utilizing the same information normally included in a two-sided statistic. In Section 4, three examples are considered to illustrate the superiority of this new test. [ABSTRACT FROM AUTHOR]

In this article, four ratio estimators designated as simple, Quenouille's, Beale's and modified ratio estimators are compared with respect to bias, efficiency, approach to normality and computational convenience. They are shown to be asymptotically minimum variance bound estimators. Some additional ratio estimators are discussed briefly and compared with these. Quenouille's, Beale's, and modified ratio estimators are found to be more attractive than the alternatives compared. [ABSTRACT FROM AUTHOR]

Confidence intervals for the population mean of a normal distribution can be determined from the distribution of the variate H = Square root of n(x - mu)/d, analogue to Student's t-distribution but based on the mean absolute deviation d instead of the standard deviation. The H-distribution is derived: the frequency function is symmetric about zero, with central ordinate (1/pi) Square root of (1-n[sup -1]); asymptotically it is normal, N(0, Square root of (pi/2)). An approximate formula for the calculation of the percent values is developed and numerical factors tabulated by which the mean absolute deviation of a normal sample of size n is to be multiplied in order to obtain 95 percent and 50 percent confidence limits of the mean, for n = 2(1)15(5)30, 40, 60,120. It is shown that the increase in length against confidence intervals from the standard deviation and Student's t is negligible. The usefulness of confidence intervals from the mean absolute deviation for estimating the precision of measurements in Physics is discussed. [ABSTRACT FROM AUTHOR]

In the case of moving average schemes or what are termed the m-dependent stochastic processes, the successive observations in the series are not stochastically independent. As a result, the usual sign test, the test based on the rank correlation tau (for randomness against trend alternatives), and the test proposed by Moore and Wallis [9] and further developed by Goodman and Grunfeld [2] (for the comovement between two time series) are not valid and require some modifications. Accordingly, the modified forms of these non-parametric tests are studied here in detail, and their applications considered. [ABSTRACT FROM AUTHOR]

The need for data compression, a consequence of the demands made on the telemetry system of a space vehicle, prompts consideration of the use of sample quantiles in estimating population parameters and obtaining tests of goodness of fit for large samples. In this paper optimal unbiased estimators of the mean and standard deviation are given using up to twenty quantiles when the parent population is normal. Moreover, the estimators are relatively insensitive to deviations from normality. A distribution-free goodness-of-fit test is presented based on the sum of the squares of four quantiles after an orthogonal transformation to independent normal deviates. If a frequency function is of the form f(x; p) = pf[sub 1](x) + (1 - p) f[sub 2](x), 0 < p < 1, where f, and f[sub 2] are normal frequency functions, the distribution is likely to be bimodal. Another goodness-of-fit test is obtained using four quantiles, which is likely to have considerable power with a null hypothesis of normality and the alternative hypothesis of bimodality. The "data compression ratios" obtained with the use of a quantile system can be on the order of 100 to 1. [ABSTRACT FROM AUTHOR]

Explores the use of statistical tests for decision-making in regional geographical construction in the United States. Chi square; Variance analysis; Relationship between regional constructs and the areal dispersion of selected traits within regional frames.

The article focuses on prestige patterns in scholarship and science. To throw some light on this question a questionnaire was recently sent to 1,275 individual chosen at random from the membership rosters of twelve different learned societies. The homogeneity of the returns was determined by the Chi square test. The sample was checked for homogeneity with respect to academic rank and section where the respondent was employed at the time he made out his questionnaire and at the time he was listed in the roster of his society. It is Interesting to note that history and economics are the only two fields in which one individual achieved a ranking frequency considerably above that of any other man listed. In every field individuals were listed prominently for their administrative ability, personal leadership in establishing schools and stimulating followers, for important offices held and for other characteristics not directly relevant to research in the limited sense.

The article presents information on a possibility theorem on dynamic organizational research. This discussion of the problems of rigorous research in the dynamic organizational context is specifically methodological, addressing itself to the demonstration of possibility by indirect proofs. For illustrative purposes, the authors concern themselves with organization seeking to effect resolution of social problems. Although research has more frequently addressed itself to these particular organizations in the guise of action research, it is not to be assumed that this restricts the general applicability of the findings. While experimental designs are frequently used in research within an organization, for example, in industrial research applications, designs are rarely applied to the study of an organization and its management. Scholar James L. Price has pointed out that rigorous research designs are virtually unknown in the study of organizations. Scholar Stanely E. Seashore, in his discussion of problems and prospects of the utilization of rigorous designs, points out that the five or ten attempts to use designs reported in the literature are primitive, pioneering ventures. It is not difficult to explain why the organizational researcher has tended to case studies and anecdotal evidence. The universe of organizations is ill defined; the number of units accessible to an investigator is small.

Methodological presentation of the panel, drawing attention to the interest of this method for thorough analysis of change, its nature and its elements. Short history of its application to electoral decisions insisting on the value of certain results and the interest of the perspectives thus opened. Lastly, the A. endeavors to show that the particularities of the French political system do not rule out the use of the panel and that its application could lead to a realistic study of the French citizen, and the process of influence to which he is subjected. [ABSTRACT FROM AUTHOR]

This article presents information regarding parole prediction variables and the time factor in violations by burglars. Parole prediction studies have shown that it is possible to determine, before an inmate is released from prison under parole supervision, what are his relative chances for making a successful extramural adjustment. The original data were set up in 23 contingency tables with the number of months before violation as the independent variable in each case and one of the 23 parole prediction factors as the dependent variable. The Chi-square test was undertaken to test the probability that differences between the actual and the theoretical frequencies would occur by chance. In these instances where the probability that the differences were due to chance was very small the coefficient of mean square contingency was determined in order to measure the amount of the relationship that exists between the variables. Whether the relationship is a positive or a negative, one was determined by inspection of the data and the signs of the differences.

A sample of 183 South Dakota farm operators has been interviewed in order to test the hypothesis that motivational factors such as social status, net worth, and money invested in livestock and machinery influence in a similar fashion (a) the seeking of contacts with agricultural agents and (b) the direct learning of agricultural technology. Two supporting factors, education and age, have been included in the testing of this hypothesis. The first part of this hypothesis is tested by the use of partial correlation analysis and the second part by tile use of three-way tables. Social status is the only factor which is related in a similar fashion to both the seeking of contacts with agricultural agents and to the direct learning of agricultural technology. In other words, high-status farmers have more contacts, and they also get more out of the same contacts. The factors of net worth and money invested in livestock and machinery only partially support this hypothesis. This is the case because, although all the relationships which are involved in the testing of this hypothesis are positive as expected, most of them are not significant at the required level. Age is related to the seeking of contacts negatively and to the direct learning of agricultural technology positively. Education is not related to these two aspects of behavior in any way whatsoever. [ABSTRACT FROM AUTHOR]

Copyright of Rural Sociology is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

In this article the authors described a method of testing relative merits of area and quota sampling by drawing various samples from a known universe, which is made up of records based on a house-to-house survey in Iowa City. Stratification simply means dividing the universe into groups which are presumed to be more homogeneous than the entire universe. This is merely a technique for the reduction of the variance of a sample estimate. The differentiation between quota and area samples is made by the way the respondents are selected within strata, not through the presence of strata or their choice. A statistical method, chi-square test was developed by Karl Pearson to test the hypothesis that an observed distribution is a random sample from a proposed universe. The hypothesis is rejected when the probability of securing a value of chi-square as large as, or larger than, the value obtained is small, usually less than 5 percent. It can be seen from the given formula that chi-square is defined in absolute frequencies, not relative frequencies.

The article investigates the efficiency of an alternative set of multivariate hierarchical methods used in plant ecology, called similarity analyses. These methods, also called agglomerative-polythetic methods, of hierarchically classifying elements into set can take a large number of different forms, based on the sorting method or type of fusion strategy and the coefficient employed in measuring similarity. The authors compare the strategy and results of the most effective methods of similarity analyses with those of association-analysis. Findings show that "centroid" sorting provides better results than "nearest neighbor" sorting, but the combination of centroid sorting and information-statistic coefficient is much superior in producing clear and ecologically acceptable hierarchies.

The article presents a study which examined alternative tests of independence between the stochastic regressors and disturbances. Researchers have noted that in testing hypothesis on the coefficients of a linear regression model with stochastic regressors, the usual tests are applicable only if the stochastic regressors are statistically independent of the disturbances. Therefore, it is desirable to develop a procedure that can be used to test the hypothesis that the stochastic regressors and disturbances are independent.

The problem to be considered in this paper is that in a linear regression model, y = Xβ + ϵ (where X is n × k of rank r ⩽ k), the disturbance vectors ϵ' = (ϵ1,ϵ2.. . ϵn) is distributed according to the null hypothesis, H0, as multivariate normal with mean vector 0 and variance-covariance matrix proportional to Σ0, against the alternative hypothesis, H1, that it is distributed as multivariate normal with mean vector 0 and variance-covariance matrix proportional to Σ1. Three test statistics, s1, s2 and s3, all functions of estimated disturbances from the fitted regression are proposed to test the hypothesis H0. It is shown (in Section 3) that all three tests based on s1, s2, and s3 are unbiased and that the test T(l) based on s1 is most powerful. In Section 4, ϵ is assumed to have a special covariance structure, namely, a first order stationary Markov process, uniform covariance structure, and moving average of order one, and the general results obtained in Section 3 are simplified. It is also shown that the hypothesis H0, in general, cannot be tested and that only an implication of it can be tested. Section 5 contains three numerical illustrations comparing the results pro- posed in this study with the Durbin-Watson procedure. [ABSTRACT FROM AUTHOR]

The article examines the asymptotic efficacies of the exact tests for autocorrelation, introduced in an equation where the Gaussian variable has a mean zero and constant variance and the zeros lie outside the unit circle. The exact tests for autocorrelation were provided by the classical regression procedures. The efficiency of the procedures were considered after the results were extended to a vector case. The individual identifiable variable in the linear combination becomes the constant term. The modification of the standard test procedures are considered based upon the joint distribution of ratios and the associated confidence interval procedures.

This article discusses problems in the interpretation of F ratios for hypotheses of differences between means. It has been reported that some sociologists may be making errors of confusing significance with strength of association and misinterpreting form of relationship. Errors of a similar elementary nature appear to occur in the interpretation of F ratios and, given the popularity of analysis of variance techniques, are indicated to be of professional interest. This paper is designed to bring to more general attention possible sources of misinterpretation of F ratios for hypotheses of differences between two means. Each procedure discussed is illustrated by testing a hypothesis using a fictitious data. The reader can assume that the subject were randomly assigned to the two groups, and that the subjects in group E were individually subjected to a newly proposed treatment for increasing self-acceptance, while those assigned to group C were individually administered a placebo treatment. Procedures for conducting tests of differences between variances is briefly reviewed. To facilitate comparisons, all tests was run at the .05 level of significance.

The article describes the procedures followed for studying the reliability of sampling plans in 100 percent auditing and a summary of its results. The author explains the methods followed in the study which required the selection of an actual case consisting of a large group of accounting data. The summary of classification of errors and an application of the indicated ratio for other miscellaneous errors has been presented in Exhibit I. Errors were classified as minor and major errors which involve errors in footing and computations, among others and the misclassification of cost respectively. He enumerates the objectives of auditing and a description of the selected samples. He highlights the concept of using opportunities of error in sample selection rather than the absolute number of items actually examined and the direct consequences of using the concept. He enumerates the criterion for acceptance in auditing situations and illustrates the determination of acceptable error by calculating the cost of a complete audit and setting the maximum amount of error acceptable before resorting to a 100 percent audit. He then enumerates the general observations that can be made while determining the relative reliability of the samples studied.

Describes different models for testing competing hypotheses. Utility maximization model; Profit maximization model; Characteristics of Paul Johnson's traditional model.

A technique previously developed for interpreting measures of association for fourfold tables based on markovian assumptions is here extended to derive measures of second order association. Transition rates between the states represented by [ABSTRACT FROM AUTHOR]

This article reviews the salient points of criticism in statistical methodology of hypothesis testing and the problems in statistical inference which have been associated with the widespread use of significance test. It also summarizes the applications of interval estimation contrasted with significance tests in the investigation of hypotheses and models and in the development of a body of empirical data. The primary objective of hypothesis testing methodology is the formulation of rules to use in determining whether the data should be construed as rejecting or accepting the hypothesis under investigation.

This article focuses on various applications of variance-covariance designs in educational research. It is noted that although analysis of variance is most commonly used to compare means, it may, when certain assumptions are met, be used to estimate components of variance ascribable to the different factors being investigated. To provide a rationale for dealing primarily with variances, the author asserted that the goal of the course was increase in variance from pretest to post-test, under the assumption that greater variance in a group indicates greater expression of individual differences.

This article investigates the aviation psychology program of the U.S. Navy. The development of the naval aviation psychology program up to and following the time of the establishment of the central office in October 1942 may be read in several descriptive summaries. Psychologists were originally commissioned to administer, score, and interpret tests for the selection of naval aviation cadets; however, the program soon broadened to include the development of experimental designs for research projects, statistical analyses, methods for selecting flight instructors and aircraft gunners, investigation of attrition, development of training aids, advisory aid to other bureaus, and research on vision and communication. About a year and one half before the Pearl Harbor attack, work had begun on the validation of a group of tests for the selection of naval aviators. From the forty different tests investigated, three were selected. Each of these three tests was validated on groups of over 3000 cadets.

This article examines the application of newer statistical techniques into research procedures in education in the U.S. The article attempts a general appraisal of the applications from the technician's point of view. The chapter has been organized with reference to techniques. Only as many studies have been cited with reference to each technique as were needed to yield illustrations of important prevalent errors and misconceptions, or, occasionally, of valid applications. In some of the studies cited the error may be of relatively minor consequence and in all other respects the study may be very competently conducted. The treatment is divided into three general, somewhat overlapping areas; problems involving correlation, significance of differences between means, and analysis of variance and covariance.

In hypothesis testing, the level of significance, α, often performs a screening as well as an evaluating function. Results which are not "significant" at some preset value of α are simply not reported. Unfortunately, this value of α is usually arbitrarily set at .05 or .01 and bears no relation to the needs of potential users of the study. This paper suggests an approach to making the selection of α more responsive to these potential applications. [ABSTRACT FROM AUTHOR]

Controversies over the value of statistical significance tests in assessing substantive significance in basic sociological research have persisted because the methodological issues involved have not been clarified. The issues are (1) the meanings or definitions assigned to the terms statistical ant substantive significance ; (2) the nature of relations between the definitions, and (3) the grounds for choosing definitions and their relations. An argument is made for : (a) defining subs'antive and statistical significance independently of each other and (b) forming methodological propositions relating these definitions in which statistical significance tests play no role in assessing substantive significance. The concept of methodological significance is introduced as a basis for choosing between alternative definitions and methodological propositions. [ABSTRACT FROM AUTHOR]

Experimental designs have previously been treated unsystematically. This article presents all possible experimental designs that use from one to four groups, the groups being differentiated by whether they receive a pretest, a stimulus, neither or both. Pretesting a group on the dependent variable may change the experiment's outcome, as may uncontrolled events. Both factors are worked into the analysis, as well as all possible interactions among these and the stimulus. Contrary to general belief, it develops that the classic two-group design is not optimal, nor are the three-group and four-group designs now in the literature. For every design the investigator must assume a value for one or more quantities, and tables are presented with solutions for unknown quantities under each combination of assumptions. Statistical procedures for data analysis are included. [ABSTRACT FROM AUTHOR]

This paper applies the exponential distribution to stock price reactions to determine, at three confidence levels, the critical percentage price reaction beyond which a reaction constitutes a strong likelihood of a major reversal or halt in the stock's present general price trend. We show that these critical values can be used to determine stop losses, so that a trader's position is closed when the probability of a major reversal or halt against his position is large, and an open position is kept open when the probability that the reaction is a major one is small or moderate. The distribution fit on ten stocks on daily and weekly bases for over a year's duration each, tested using the chi-squared statistic, was found to be good. Further, three stops, corresponding to three confidence levels of the distribution were tested for each stock against a subsequent six-month period of price action. The percentage of successful tests of these stops for each of the stocks corresponded very closely to expected results ascertained by using data from the previous period. [ABSTRACT FROM AUTHOR]

The article discusses the problem of estimating an optimal interval containing the location of the maximum of a unimodal function by simultaneous experimentation for the case when an arbitrary probability density function for the distribution of the maximum is given. A particular type of problem is examined. Estimate the location x* of the maximum of an unknown unimodal function z(x). The estimate is taken as an interval D in which the maximum is located. Simultaneously observations of the function values at N points are performed before making the estimate. The objective is to distribute the observations in an optimal way. The resultant optimal strategy has the following properties: When the number of experiments is even, the optimal strategy selects an interval D, containing the maximum x*, for which the maximal a priori probability of containing x* is minimized. When the probability density function for the distribution of x* is uniform, the optimal strategy coincides with the mini-max search. When the number of experiments is odd, a search pattern results that always yields more information than an even search with one less experiment.

This paper presents derivations of expressions for the probability of killing a randomly located point target and for the expected coverage of an area target, of variable density. Particular emphasis is placed upon a general situation in which the phase distribution is not necessarily uniform. General distributions are employed to describe the phase, radial, and component behavior of the target and from these functions the distribution of a quadratic form is derived that provides the expected coverage or kill probability. The targeting conditions include a fixed lethal radius (kill-no-kill) when the center of gravity of the weapon impact points coincides with the center of the target density and a variable lethal radius when a weapon bias is present. [ABSTRACT FROM AUTHOR]

Let F(x, theta) denote the distribution function of a vector variate x, whose range does not depend on the parameter theta. Then assuming that theta admits a complete and sufficient estimator theta, we derive minimum variance unbiased estimate of P{x *(This character cannot be converted in ASCII text) a} = F(a, theta), where a is a known vector. The result is applied to a coverage problem in a normal set up. [ABSTRACT FROM AUTHOR]

Experimental optimization methods are used to determine optimal decision policies in the cases where the functional form or parameter values are unknown. The optimal design of such experiments is developed in this paper for a variety of methods of experimentation. Parameter values are estimated, an optimal decision policy is obtained, and estimates of the expected gain from experimentation are derived. Expected gains are compared for alternative experimental designs. In particular, methods are derived for obtaining the optimal size, organization structure, and duration time of the experiment. The general methods are applied to the problem of design of a maintenance experiment to obtain the scheduled maintenance frequency that achieves maximum cost savings. [ABSTRACT FROM AUTHOR]