Showing papers on "Matching (statistics) published in 1980"

Bias Reduction Using Mahalanobis-Metric Matching

Abstract: SUMMARY Monte Carlo methods are used to study the ability of nearest available Mahalanobis metric matching to make the means of matching variables more similar in matched samples than in random samples.

Statistical Methods for Comparative Studies Techniques for Bias Reduction

Abstract: Introduction. Confounding Factors. Expressing the Treatment Effect. Randomized and Nonrandomized Studies. Some General Considerations in Controlling Bias. Matching. Standardization and Stratification. Analysis of Covariance. Logit Analysis. Log--Linear Analysis. Survival Analysis. Analyzing Data From Premeasure/Postmeasure Designs. Choice of Procedure. Considerations in Assessing. Association and Causality. Index.

Intensity perception XI. Experimental results on the relation of intensity resolution to loudness matching.

Abstract: According to a recent extension of our theory of intensity perception [Lim et al., J. Acoust. Soc. Am 62, 1256-1267 (1977)], two stimuli are matched in loudness if and only if their intensities divide the respective dynamic ranges proportionally in terms of just noticeable differences. This study reports results of intensity discrimination and loudness matching experiments designed to test this prediction. Data were obtained over most of the dynamic range for three different types of sounds: a 1000-Hz tone in quiet, a 1000-Hz tone partially masked by a 2-octave band of noise, and spectrally flat wide-band noise. Of the five subjects tested, three produced results that had sufficient internal consistency to be useful for testing the predictions. For these subjects, the data and the theory were found to be reasonably consistent. Comparison with other studies, however, reveals that loudness matching results vary with matching paradigm by an amount that is significant with respect to the task of evaluating the theory. Hence, a rigorous test of the theory requires an improved understanding of the effects of matching paradigm.

Bias and Efficiency in Logistic Analyses of Stratified Case-Control Studies

Abstract: Two approaches to the estimation of the parameters in the logistic model applied to the analysis of stratum matched case-control studies are compared. It is shown that the "unconditional likelihood method' gives estimates of odds ratios which may be severely exaggerated if the stratum sizes are not large; however, if stratum sizes are not small the "conditional likelihood method' is computationally prohibitive. The use of small stratum sizes (e.g., 1-to-1 matching) leads to some loss of efficiency if the matching is irrelevant but it is shown that the loss is small for odds ratios up to at least 3. There would seem to be strong arguments for the routine use of matched designs as the reduction in confounding bias produced by such matching may be considerable whereas there is only a small loss in efficiency compared with unmatched designs.

A Study of Disjunctive Reasoning

Abstract: Whilst it is currently fashionable to construct sequential models of reasoning, the results of recent experiments on conditional reasoning tasks are apparently incompatible with this notion. There is evidence of two orthogonal statistical components, one logical and the other non-logical. It has been suggested that non-logical tendencies are ‘response biases’ independent of the linguistic structure of the rule. However, certain results suggest that the ‘matching bias’ observed in conditional tasks may not occur with disjunctives. Two experiments are reported in which subjects generated ‘subjective truth tables’ for disjunctive rules, in which the presence and absence of negative components was varied. Experiment I provided a powerful test of the matching bias hypothesis and found no evidence of its presence. The presence of negative components did, however, systematically affect performance. Experiment II, with the aid of latency measurements, provided support for the hypothesis that negatives were affecting the ease of interpretation of the disjunctive sentences, with one negative component causing at least as much difficulty as two. In discussion, it is pointed out that the present results can be reconciled with those on conditional tasks, on the assumption that ‘matching bias’ is a special case of associational biases which are dependent on the linguistic form of the rule. It still appears necessary to retain the distinction between logical and non-logical components of the performance. However, this does not necessarily imply parallel processes underlying the data. A single muddled thought process, in which a rational sequential strategy fails to be consistently applied, could give rise to the type of data obtained.

Industrial Pass/Fail Colour Matching Part 1‐Preparation of Visual Colour‐matching Data

Abstract: The preparation of a comprehensive set of visual colour–difference pairs covering 640 samples grouped around 55 colour centres uniformly distributed throughout colour space is described 25, 590 pass/fail decisions of a panel of eight observers are analysed to provide acceptability ratings for each of the sample pairs The data also provide information on the within– and between–observer errors and the reproducibility of the panel The experiment provides a useful data set for the evaluation of colour–difference formulae

Person-Specific Information in the Labor Market

Abstract: Heterogeneity on both sides of the labor market implies that the correct matching of individuals to firms is of importance. If there is uncertainty about individual productive characteristics, there are both private and social returns to activities that generate information facilitating the assortative matching process. A model of individual investment in information is presented and analyzed. A key assumption is that there are no individuals who have an absolute advantage in all jobs. Under this assumption, all individuals view more accurate information as beneficial and therefore invest in its production. Comparative statics are derived and the model is compared to human capital and signaling-screening models.

Number of observations required for the comparison of two correlated proportions

Abstract: Previous sample size tables for studies with pairwise matching of experimental units and all-or-none responses have relied upon asymptotic results for the power function given the effective sample size of the experiment. In this paper the exact power function is employed to construct detailed sample size tables for such studies.

On the Analysis of Tree-Matching Algorithms

Abstract: 0 INTRODUCTION : This paper deals wi th the average case performance analysis of tree-m~tc~Lng algorithms. The trees we consider here are planar label led trees as occurs in programming experience under the form of syntax t rees, expression trees or tree-representat ion of structured objects (records). More spec i f i ca l l y , we are interested in methods of estimating the average time of matching algorithms as a funct ion of the pattern size and the tree s ize, under a wide class of s t a t i s t i c s on the imputs. Unt i l recent ly , analysis of (planar) tree algorithms has received rather l i t t l e a t tent ion. The basic resul ts appear in [KNUTH ; 1968] and other problems are considered, for example in [de BRUIJN, KNUTH ; RICE ; 1972] and [FLAJOLET, ODLYZKO ; 1980] (stack size in explorat ion of t rees) , [FLAJOLET ; 1979] (binary tree matching) or [KEMP ; 1979] and [FLAJOLET, RAOULT, VUILLEMIN ; 1979] ( reg is ter a l locat ion problems). These works deal almost exc lus ive ly wi th the C~taJZanst~Y~L~ • where the average performance of an algorithm is determined over the set of a l l possible shapes o f trees of a given size. We consider here more general s t a t i s t i c s corresponding to various classes of label led trees, the purpose of which is to c losely model pa r t i cu la r appl icat ions. Following [MEIR, MOON ; 1978], a s~mply generated fami ly of trees is in formal ly defined by speci fy ing a f i n i t e set of admissible labels for each node-degree. Instances of simply generated fami l ies of trees include : expression trees where each node has degree O, I or 2 ; nodes of degree 0 are label led var ( " va r iab le " ) , nodes of degree i are label led e.g. +, , log, exp, s in , nodes of degree 2 are label led +, , x, ÷. trees where the nodes of degree 0 are label led var , nodes of degree I are label led +, , log, s in , nodes of degree 2 are label led +, , ×, ÷, ÷ ; nodes of degree 3 are label led cond, i t e r . . . ; th is is a simply generated family as can be defined to approximate syntax trees in a structured programming language. unlabel led trees (binary or general) that appear in th is framework as trees wi th at most one possible label for each degree. Thus, simply generated fami l ies of trees include the classical fami l ies of binary and general trees as subcases. We can now formulate our i n i t i a l problem as fol lows : given a tree-matching A, l e t

Industrial Pass/Fail Colour Matching. Part II – Methods of Fitting Tolerance Ellipsoids

Abstract: Methods of fitting tolerance ellipsoids to the J & P Coats and to the Davidson and Friede matching data are described. The fitted ellipsoids vary in size according to their position in colour space. Various colour matching equations are derived to describe these variations and these are found to give better correlation with visual observation than other colour matching formulae.

Dynamic compensation for exact model-matching of two-dimensional systems

Abstract: The paper considers the problem of matching exactly the transfer function matrix of a given 2-D (two-dimensional) system to that of a desired 2-D model using output feedback, where in the feedback loop proportional (P), integral (I) and derivative (D) terms are present. The approach followed reduces the problem to that of solving a linear system of algebraic equations. Furthermore, necessary and sufficient conditions are established for exact matching. An example is included to illustrate the proposed technique.

A New Role for Psychographics in Media Selection

Abstract: Most media selection problems involve some kind of matching between media audience and target market membership. As a rule, this matching is done indirectly, using demographics as mediating variables. Recent studies have suggested that psychographics might also be used for this purpose. This study evaluates the performance of a standard set of psychographic variables as mediators of the media-market relationship. It then discusses the implications of this performance for media selection theory.

Empathy revisited: the effect of representational system matching on certain counseling process and outcome variables

Abstract: Empathy Revisited: The Effect of Representational System Matching on Certain Counseling Process and Outcome Variables William Philip Brockman, Ed.D. The College of William and Mary in Virginia, May 1980 Chairman: Charles 0. Matthews, Ph.D. Therapist-offered empathy has been shown to be an important ingredient in the counseling relationship. Many operational definitions of empathy and tools for measurement of this elusive quality exist. Most empathy measures have been criticized on methodological grounds and their construct validity is suspect. Yet there is little argument with the trend which emerges from the data; the overall relationship between empathy, or those dimensions tapped by empathy measures and effective therapy appears positive. The nature of empathy however remains enigmatic and it is evident that all of the variables which account for the empathic process have not been explicated. This study defined and investigated the validty and effect on counseling of a new dimension of empathy. From their linguistic analysis of effective therapy Bandler and Grinder have formulated the construct of representational systems or internal maps used by individuals to organize reality. Such maps are visual, auditory or kinesthetic and are reflected in natural language. Do you see what I mean? Empathy, then, is operationally defined as the counselor's matching language with the representational system used by the client. It was hypothesized that counselors who use representa­ tional system matching would: 1 ) be perceived by subjects as more empathic than counselors who do not (accepted, £ = • OOI4S ) ; 2 ) be perceived by judges as more empathic than counselors who do not (accepted, £ = .0165); elicit a greater willingness to self-disclose than counselors who do not (rejected) and i|) be preferred by clients over counselors who do not use representational matching (accepted £ < . 05). Reproduced with perm ission o f the copyright owner. Further reproduction prohib ited w ithout permission. Subjects (N=20) were undergraduates at The College of William and Mary who met with each of two counselors, in counterbalanced order, for an analogue of a beginning counseling interview. One counselor used representational system matching; the other counselor took a more generic, human relations, approach to empathy. After each interview subjects completed Barrett-Lennard’s Relationship Inventory (RI) and Jourard1s Willingness-to-Disclose Questionnaire (WDQ). Following their second interview subjects indicated their preferred counselor. Covariates were: 1) Carkhuff's Empathic Understanding Scale (EU) which also served as a dependent measure; 2) The Counseling Readiness Scale (Crs) of Gough and Heilbrun’s Adjective Check List and 3) Rotter’s I-E scale. The Latin-square design produced data analyzed by: repeated measures analysis of covariance (Hypotheses 1-3); stepwise regression (Hypotheses 1 & 2) and Chi square (Hypothesis ). Results indicate that both subjects and judges per­ ceived the representational system matching counselor as more empathic than the generic empathy counselor. While EU accounted for 11.76% of the variance on Rl-empathy scale scores, representational system matching accounted for 11.9i+% of the variance beyond that accounted for by EU. Clients preferred the representational system matching counselor by a ratio of 3 to 1. It was concluded that representational system matching is an important dimension of empathy and the recommendation was made that beginning courses in counseling techniques and human relations training include a section on identify ing and responding to client’s representational systems. Recommendations were made for further study. Reproduced with perm ission o f the copyright owner. Further reproduction prohib ited w ithout permission.

A danger in matched-control studies.

Abstract: IN READING the medical research literature, one often sees references to or use of matched controls. This procedure may be included, for instance, in comparative drug trials or studies of the putative etiologic effect of certain environmental factors. Actually, there are a variety of research settings that may lend themselves to this approach. What do we mean by matching? Briefly, it ordinarily consists of the selection, for each test subject in a study (eg, one with a certain disease or receiving a certain therapy), of a comparison subject who is as similar to the test subject as possible except for the presence of the disease or treatment under investigation. One of the ways in which matching differs from stratification or blocking is that the comparison subjects in matching are selected individually to match individual test subjects. For example, in a case-control study of the possible effect of fiber consumption in

Improved method for scene matching of dissimilar imagery

Abstract: The problem of matching two digitized images of a scene obtained from dissimilar sensors is considered. A new method for improving the scene matching performance is given. Simulation results are given which illustrate the improved performance.