SUGI 1978 Papers - Index

58 papers. Source: SAS Users Group International conference proceedings.

#	Title ×	Authors ×	Abstract ×	Keywords ×	Extras	Pages
1	"Adjusted Means" from SAS76	James A. Bolognese	Users of SAS76 have found that the "adjusted means" available in SAS72 from `PROC` REGR, in particular from the ADJMEANS statement, are not directly available in SAS76. This paper will show how to obtain "adjusted means" by writing the appropriate linear full-rank model, coding its design matrix, and choosing the appropriate parameter estimates. A simple method for coding this model using `PROC` MATRIX (SAS76) will be presented. It will then be explained how `PROC GLM` can be run to obtain the desired "adjusted means".	anova, proc genmod, proc step	source	4
2	A More 'Intelligent' Merge Capability in SAS	Max E. Ellis	Existing data manipulation functions of SAS do not provide the user with a complete set of functions required for research data management. Certain logical, data-set combining operations using the SORT and MERGE functions require excessive programming, and in some cases, extra passes of a data set(s) must be made to produce the resultant composite file. A hierarchical logical file structure is used to illustrate how a user would combine physical SAS data sets to form output data sets corresponding to various analysis levels of the hierarchy. Household, family, person and work data sets are combined via 1) existing SAS program statements and 2) proposed new statements. By enhancing the data manipulation language of SAS, the utility of SAS can be increased and the programming now required to combine logically complex, related data sets is reduced.	dataset, keep statement	source	6
3	A Ridge Regression Algorithm Versus the Classical Least Squares Regression Model	Bradford T. Greene, Kareem A. Nakkash	The unbiased least squares (L.S.) estimation procedure for deriving the coefficients of the linear regression model is compared with a biased estimation procedure - the ridge regression technique. In the case of ill-conditioned data (i.e., non-orthogonal correlation matrix), a ridge solution provides a more stable set of regression coefficients than the classical least squares solution. Examples are presented to compare these two methods. This paper also utilizes the matrix procedures in the SAS program and presents a computer program for the implementation of the ridge solution.	regression analysis, correlation, ridge regression, covariance, proc step	source	6
4	A SAS Macro for Calculating Coefficients of Ridge Regression	Ernest Hilderbrand, Robert Rogers	A SAS Macro (Version 5) is used to produce ridge regression coefficients which predict and extrapolate better than least squares when predictor variables are highly correlated. The method of selecting coefficient values and variables using this macro are illustrated with an example from forestry. Forest site productivity is related to a selected subset of soil factors.	regression analysis, ridge regression, predictor variable, correlation, macro definition	source	7
5	A SAS Macro for Maximum Likelihood Probit Analysis with Multiple Predictors	Carolyn L. Bean, James E. Dunn	A SAS macro statement for maximum likelihood probit analysis with multiple predictors is presented along with derivations of the equations used in the algorithm. An attempt is made to arrive at a mathematical model which will predict the presence of certain species of zooplankton.	likelihood function, proc genmod, proc step, macro definition	source	3
6	A SAS-Based Small-Area Profile System: Its Use in Primary Care Resource Development	Edward M. Bosanac	The development of rural primary care clinics has been a dominant activity on the West Virginia health scene for the last several years. Although much of this development occurred without an adequate initial examination, the situation was such that nearly all these efforts were successful. This proliferation of clinic development has closed off many of the choice sites, leading to increasingly marginal site location. Faced with these pressures, while maintaining a commitment to increasing the availability of care to the State's residents, major payors and primary care developers formed a study group to review development activities in portions of the State. The group needed a "study" methodology and supporting data system which could provide an adequate examination of the area in question on short notice at a reasonable cost. A two-stage decision process included secondary source data in an initial stage with primary source or "shoe leather" data in the latter phase. The secondary data from a variety of sources consisted of machine readable files including: travel time, small area demographic profiles, manpower and facilities files, service demand data, hospital patient flow information, third party coverage data, and a geo-conversion file. The initial profile attempt (principally by hand) cost 25 man-days, while the latest was done in less than three man-days. An initial questioning of the utility of secondary source data on a small area basis has been replaced by a questioning of the necessity of primary data collection. This paper details the ongoing decision process and the data system that supports these reviews.	raw data, proc means	source	6
7	Adherence Monitoring at the Clinic Level	Lynn B. Eggering	One of the significant problems in the management of long-term intervention trials is insuring patient compliance to the prescribed treatment. Basic to any program to enhance adherence is a monitoring system which tracks the adherence record of each study participant. With the use of SAS at the clinic level, monitoring and data management are performed rapidly and economically.	proc step, proc freq	source	5
8	Analysis of a Repeated Measurement Experiment with Incomplete Data	William L. Sanders	In many biological experiments (well designed and others) repeated measures over time on each experimental unit occur in patterns often uncontrollable by the researcher. However, various analytical strategies may be employed when this problem type is encountered by chaining several features readily available in SAS. An example using data from a ruminant nutrition study is presented. Diurnal curves approximated by fitting two harmonics of a Fourier series were estimated separate for each animal in the study via `PROC NLIN`. The vectors of coefficients were then inputted to GLM as 'data' and a multivariate analysis performed to test hypothesis of no treatment effects on the diurnal curves.	proc step, proc genmod, correlation	source	4
9	Applications in Econometrics: Problems, Programs, and Procedures	Kenneth J. White	This paper surveys some recent examples of statistical procedures in Economics. Studies with autoregressive errors, ARIMA models, distributed lag coefficients, random coefficients, and nonlinear systems of equations are used to demonstrate the procedure. A description of capabilities of computer programs in Econometrics is included.	regression analysis, residual, interest rate, likelihood function	source	6
10	Capabilities of SAS as a Clinical Research Data Management and Analysis System	Frank E. Harrell Jr.	SAS has many features which make it a flexible tool for the management and analysis of data from clinical research projects. However, there are many problems which at first glance do not appear to be reasonably handled by SAS. Some examples are dealing with hierarchical files and encrypted data, and editing and correcting data. Because of SAS's flexibility, methods for tackling such problems exist. Some general solutions will be discussed and SAS will be compared from an analysis viewpoint with some packages designed specifically for handling hierarchical files.	error message, output	source	4
11	Conversational Instruction in the Statistical Analysis System	Ione H. Cockrell, Dorsey A. Glenn	Teaching relatively unsophisticated users the Statistical Analysis System poses special difficulties in a large geographically dispersed computer network. Am interactive SAS course has been developed at the Computer Services Division of the University of South Carolina which overcomes many of these difficulties. Students may sign on to the computer assisted course at any one of approximately 300 interactive terminals throughout the network. The course, which introduces the potential SAS user to the USC computer system and to SAS, consists of textual material interspersed with questions. Each student is given instant feedback to his responses, may see as many or as few of the sessions as he desires, and may completely determine the. pace of the course.	by group processing, proc step, data science, proc sort	source	4
12	Data Base and Analysis of Power Plant Cooling Systems	Sharon A. Bucci	This paper discusses a nationwide, computerized data base on cooling systems and related water quality characteristics for steam-electric power plants in the U.S. with a generating capacity greater than 100 MWe. In the course of assembling the data base, information was reviewed and extracted from major environmental data bases, from existing power plant data bases and from document sources such as Environmental Reports and 316 Variance applications. The SAS data base management system was used to organize the information in the computer in a manner that facilitates the storage, modification and retrieval of data and is compatible with anticipated analysis programs. Using the BMDP statistical routines in conjunction with SAS's analytical procedures, the data was subjected to various quantitative analyses in order to determine the degree of association among the variables.	dataset, proc step, regression analysis, proc freq	source	5
13	Generating Multivariate Normal Data in SAS	Daniel M. Chilko, E. James Harner	Computer random number generators offer a solution to the problem of finding suitable examples of data for classroom use. A SAS procedure for generating multivariate normal data has been developed. This provides the instructor a single system to generate data, analyze then, and display results. Its use in teaching multivariate statistics is illustrated.	covariance, sample, proc step, regression analysis, proc princomp	source	4
14	History, Problems and Practical Applications of SAS in a Large Industrial Organization	Maxine B. Potter	The purpose of this paper is to provide some insight into the wide acceptance SAS has enjoyed as a problem-solving tool in a large industrial complex and to discuss certain problems which can develop due to such acceptance. Various applications, which exploit both the statistical and data management capabilities of SAS, are discussed. The problem areas addressed are the possible pitfalls encountered upon implementation of a generalized, easy-to-use statistical data management system; specifically, the chance for application misuse by "instant" Statisticians and Computer Scientists. Armco's approach to the prevention or at least the reduction of SAS misuse is presented.	data wrangling, data governance, proc step, proc freq	source	4
15	Improving SAS PROCS without Rewriting SAS	Nat H. Wooding Jr.	Printed output from SAS PROCS might sometimes be more useful in reports if additional information could be added to the standard output. Conversely, information not included in output data sets may be useful for further work. Heretofore, the information had to be typed on the output or copied and repunched for subsequent processing. The new features, `PROC PRINTTO` and substring handling, allow considerable capability in modifying and using SAS printout. Portions of output may also be used as input for further processing where there is now no output capability. Simple examples show the augmentation of `PROC FREQ` and `PROC ANOVA` and the retrieval of information from `PROC GLM`.	proc step, anova, proc genmod, proc freq	source	7
16	Interfacing a Mumps-Based Data Entry System to SAS	Joel F. Achtenberg, J. Philip Miller	A discussion of the rationale for providing mini-computer based data entry facilities interfaced to established statistical analysis packages is presented along with the description of one such system currently in use.	raw data, proc step	source	7
17	IPFPHC: A SAS Procedure for Hierarchical Clustering Using Transaction Flow Tables	Daniel M. Chilko, P. B. Slater	Transaction flow tables, such as internal migration, input-output, trip distribution, trade, and occupational mobility matrices, are frequently available to social scientists. They often assume large dimensions. A two-stage procedure for effectively summarizing this form of data structure has been implemented in SAS. In the first stage, the table is adjusted so that all its row and column sums are equal. In the second, the units of the adjusted table, which is viewed as an asymmetric similarity matrix, are hierarchically clustered using the directed graph analogue of the single linkage method.	hierarchical clustering, correlation	source	5
18	K.W.I.C. Indexes with SAS	Edward L. Spitznagel	Specialized indexes such as the keyword-in-context index are useful tools in information retrieval and display. For example, the author has developed a KWIC index to the routines available in major statistics packages with references to the package, the routine name, and a benchmark CPU time for the routine. Although any number of KWIC indexing programs already exist, the author chose to write a new one in SAS to obtain precisely the form of index desired. The program is short, simple, and easily adapted to other uses.	cpu time, proc step	source	4
19	KRUWALC, A Macro for Categorical Data	Donald J. Henderson	A general purpose macro which computes the Kruskal-Wallis H-test, including the ties correction factor, and Scheffe type contrasts for ordered categories in a one-way layout is presented. An example of its use on a set of ground beef taste panel data is given.	sample, proc step, raw data, proc freq	source	5
20	Macro Commands for Tukey Type Smoothing	Donald J. Henderson	A set of macros is presented which enable the user to easily perform most of the smoothing techniques presented by Tukey (1977). The results are stored as two SAS data sets. One contains the trend and the other contains the residuals. Their use on a set of insect count data is presented.	if-then-else, proc step	source	5
21	Maintenance and Analysis of Anesthesia/Surgery Data with SAS	William D. Owens, Edward L. Spitznagel	The Washington University Medical Center requires a form detailing pre-operative and operative information to be filled out for every patient undergoing anesthesia and surgery. Data from this form is later merged with date-of-discharge and death information from hospital records. The result is a seventy-variable, 19000-case-per-year data set yielding much valuable information on the relationship between pre-operative and operative factors and final outcome. SAS has been an invaluable tool in both the maintenance and analysis of this data set.	regression analysis, class variable	source	3
22	Management of Diverse Environmental Data with SAS	Richard J. Olson, Rodney H. Strand	Environmental assessment and planning for energy development ongoing at Oak Ridge National Laboratory requires rapid access to data at appropriate spatial and temporal scales. In the Environmental Sciences Division we have developed, and are extending as needed, a spatial data base at the county-subcounty level of resolution. The data base contains information on terrain, water resources, forestry, agriculture, land use, wildlife, air quality, climate, critical natural areas, human population, health, and energy. Each of these subject sectors is stored as a SAS data set. The SAS directory is used extensively to describe the 800 variables contained in the present 30 data sets. The file currently occupies 10 million bytes of storage for a 16 state area in the southeastern United States. The data base is unique in that it was compiled from many existing data sources into a single, standardized file for a large geographic area. The data base is being used, for example, to study habitat relations and population dynamics of wildlife, potential changes in vegetation patterns related to predicted temperature shifts from increased atmospheric CO2, and regional energy facility impact assessment and siting.	proc step, dataset	source	7
23	Maximum Likelihood Estimation of the Multivariate Logistic	Martha McCrate, Michael A. Province, Reimut Wette	A frequent problem occurring in the analysis of clinical studies is the construction of a risk function which allows the calculation of the probability of some outcome such as death on the basis of a training set of observations. In this training set, the observations are identified as belonging to either population P0, or P1. Associated with each observation is a vector X of length p which is composed of observations on various symptoms or laboratory measurements which are to be used to calculate the probability of membership in Pi.	sample, odds ratio, likelihood function, logistic distribution, proc step, maximum likelihood estimation, sample size	source	3
24	Multivariate Prediction of Neonatal Mortality with Emphasis on Health Care Inputs	Merwyn R. Greenlick, Jonathan D. Quick	Using the SAS multiple regression procedure, Oregon vital statistics data from 1972 to 1974 were used to predict the likelihood of neonatal mortality. Maternal sociodemographic characteristics, medical-obstetric risks, and health care inputs were used to predict outcome. After selection of a ten percent random sample of all viable births and a 100 percent sample of neonatal deaths and appropriate transformations of nonlinear variables, the SAS `GLM` procedure was used to create a multivariate function predictive of neonatal mortality. The results of the analysis are presented and discussed in light of recent national guidelines for perinatal services. The ability of SAS data manipulation capabilities to facilitate the multivariate analysis of rare events in large sample populations is demonstrated.	regression analysis, multiple regression, sample, main effect, dataset, predictor variable, proc freq	source	6
25	Nonlinear Model for Pesticide Accumulation in Selected Estuarine Species	Lowell H. Bahner, Jerry L. Oglesby	Extensive testing has shown that pesticides, such as Kepone and endrin, are rapidly accumulated by estuarine animals from water, food, or sediment. After initial uptake, a pesticide can depurate rapidly or slowly, depending on its chemical composition, the test species, and temperature. Chemicals that depurate slowly can threaten aquatic animals and man in their movement through aquatic food webs. Simulated aquatic exposures of Kepone and endrin to estuarine shrimp and crabs were used to test our nonlinear statistical models for pesticide uptake, equilibrium, and depuration. The models describe biological data as a single equation, thus allowing variations due to many physical, chemical, biological, and random error factors to be analyzed simultaneously.	proc nlin, sample	source	5
26	Orthogonal Polynomial Fitting with Arbitrary Spacing and Correlated Means	Walter R. Harvey, L. A. Swiger	Computing procedures for obtaining the sums of squares and tests of significance for single degree orthogonal polynomials which is applicable to either balanced or unbalanced data and to a single factor or to two factors are presented. The method presented utilizes weighted least squares procedures rather than an orthogonalizing procedure, such as the Gram-Schmidt method, which could then lead to the use of orthogonal polynomial coefficients. The non-orthogonal partial regression coefficients are computed in order to plot the curve or the three dimensional surface.	heteroscedasticity, proc step	source	5
27	Pinpointing Data with "DARTS "	Mark R. Schultz	Modeling complex analytic systems using real data often involves large amounts of data reduction and processing to identify system parameters. The Analytic Sciences Corporation (TASC) has developed a procedure using the Statistical Analysis System (SAS) to "track" data through a processing cycle, with the intent of reducing duplication of effort and eliminating the pitfalls of manual recordkeeping. Three requirements were considered of major importance in the system design: automation of the capturing of the tracking information to the greatest extend possible; design of the system to be user (rather than programmer) oriented, and sufficient flexibility to allow the system to be used on other projects with a minimum of redesign and reprogramming. The Data Analysis and Retrieval Tracking System (DARTS) was implemented to meet these needs. In addition to SAS, several other components were also incorporated in this system: routines callable from user programs to extract dataset information, interactive conversational procedures for data retrieval and updating, and tracking of remote processing on another System/370. The choice of SAS for the data management aspects of DARTS had the added advantage of an easily accessible statistical and report generating capability.	data science, data wrangling	source	5
28	Principal Component Regression with PROC CORR and PROC MATRIX	M. E. McBride, Robert P. Parks	The paper discusses a SAS program that calculates principal component regression estimates. The program uses the SAS procedures CORR and MATRIX. An application of the program to a housing price equation is given.	proc step, proc princomp	source	3
29	PROC CODIN, A Procedure for Coded Data Entry	Paul D. Hopkins, Donna P. Woodworth	`PROC` CODIN is a procedure to input coded data directly into SAS. Coded data consists of a pair of fixed fields. The first field is a code designating the item, and the second field is the value of the item designated. The user specifies the correspondence between the ID codes and SAS variable names. Within a given input record, the coded pairs may be specified in any order, and for every code not referenced, the corresponding variable will be assigned the missing value for that observation.	format statement, proc step	source	3
30	Repeated Measures in SAS	John A. Courtright	One of the most common research designs in the analysis of empirical data is the repeated measures ANOVA. Yet despite the frequency with which this type of design is encountered, SAS offers no "automatic" procedure for analyzing such data. Consequently, the purpose of the present paper is to show that repeated measures designs are indeed capable of being analyzed by SAS. With a few relatively simple assignment statements, a user can obtain not only an "overall" repeated measures ANOVA, but can also test specified contrasts from among the repeated measures. In terms of the overall analysis, the paper details how the user can (1) employ the OUTPUT statement to appropriately modify the data array for entry into `PROC ANOVA` or `PROC GLM`; (2) utilize an artificial classification variable (SUBJECTS) as a nested variable to create the error term for all completely crossed factors; and (3) use the residual error term to test the repeated measures main effects and interactions. The paper goes on to discuss a method for further decomposing the design into single df contrasts. It discusses not only how to construct such contrasts but also how to normalize them so as to obtain numerically correct sums-of-squares. In addition, it explains how the NOINT option in GLM should be employed to test these contrasts. Data from two recent empirical studies are employed to illustrate these analyses.	repeated measures, anova, residual, main effect, output, response variable, proc genmod, sample	source	5
31	Response Surface Contour Plotting in SAS	Daniel M. Chilko	Response surface equations are frequently used to analyze data consisting of a dependent variable and a set of independent variables, all measured on continuous scales. The SAS `PROCEDURE GLM` facilitates the development of such equations. A SAS procedure has been developed to facilitate the visual exploration of response surfaces by contour plotting.	contour plot, predictor variable, proc step, proc genmod	source	4
32	SAS - The Necessary Utility for Computer Performance Evaluation and Control (CPEC)	Mario M. Morino	The role of Computer Performance and Evaluation (CPE) has increasingly gained attention in more organizations as CPE tools and techniques are called upon to solve critical problems. Unfortunately, this traditional role of CPE has seldom been extended in use beyond the "solve-this-crisis" mode. The typical CPE process is more oriented to after-the-fact problem resolution rather than ongoing preventative control. The traditional CPE cycle is shown in the following series of events: Installation crisis; Management reaction; Performance evaluation of a situation; Management decision (or indecision?); Watt for next crisis. Consequently, organizations continue to be managed in a reactive fashion and the potential benefits of CPE have not been fully realized. These benefits could be attained, however, if state-of-the-art CPE know how is used to define and implement an information system which would allow continual control over an installation's operational performance.	batch job, data wrangling	source	6
33	SAS and the Analysis of Clinical Data at the Upjohn Company	Patricia A. Kemp, Michael S. Lajiness, Timon P. Tesar	Clinical information is retrieved from a tape by an SAS program which generates several SAS data sets on disk. These files are backed up on tape and the data analyzed by other SAS programs. Part of the analysis is performed by two in-house written SAS procedures. One of these procedures, AOVMEAN, is a one-way analysis of variance that allows user inputted contrasts. Another SAS procedure, CCPLOT produces Calcomp plots. AOVMEAN is used to compare treatment effects over several variables, adjusted for initial differences. CCPLOT is normally used to plot the means of several treatments across time. In addition to the above analysis, descriptive statistics, such as medians, are obtained and displayed in tabular form with the aid of `PROC PRINTTO` and PUT statements.	proc step, proc printto	source	7
34	SAS and the Evaluation of Drug Reinforcement	Philip I. Good, Michael S. Lajiness	SAS proved to be an important tool in the evaluation of the reinforcing properties of drugs. The versatility of the tabulation and display functions of SAS facilitated the development of a drug score. This score was based in part on a bivariate confidence region and on ranking across several variables within an SAS observation. Several measurements of reinforcement were made on each experimental subject. Transformationa of the raw data were necessary to obtain a meaningful scale of measurement and to insure the normality of the data. The dependence of the day to day responses were discovered with the aid of SAS. This led to the development of a bivariate confidence region. Details of the development may be found in the design and development section. Generation of the confidence region is discussed in the section on tabulation and display. The drug score is based on three aspects of the response rates: the relationship of the high-low bivariate response to a confidence region, the relative number of key presses on the high and low dose, and the trend of the response rates with changes in the number of key presses required for injection. This trend is determined by ranking across several variables within a single SAS observation. The method used to generate the ranking is described in detail in the section concerning the drug score.	geomean, proc step	source	6
35	SAS Applications in a Three-Course Sequence in Statistical Computing	Thomas A. Bubolz, James E. Gentle	The role of SAS in a sequence of courses in statistical computing is discussed. The three courses, sponsored jointly by the statistics and computer science departments, focus on current topics in statistical computing: programming principles and program development; statistical numerical analysis, particularly with regard to linear algebra; computing systems design; data management; and random number generation and applications. Although the emphasis in the latter courses in on computing methods rather than data analysis, SAS is a useful pedagogic tool. `PROC` MATRTX is used to introduce topics in numerical linear algebra as well as nonstandard techniques like ridge regression. In the third course in the sequence the students work in teams to install a SAs procedure.	regression analysis, proc step, sample	source	7
36	SAS as a Management System for Routine Echocardiograms	Glenn A. Augustine, Emilio R. Giuliani, Kenneth P. Offord, Vernon P. Weber	This paper will discuss how SAS is used in a batch processing mode as a data management and retrieval system to handle the collecting of data from echocardiograms done at the Mayo Clinic. A discussion of the edit/update program written in SAS will be presented, as well as why SAS was selected, and the benefits it has provided. Echocardiography is a non-invasive technique for examination of the heart by use of ultrasound. It was used sparingly in the late 1960's at Mayo, but in 1970 when the ultrasound unit was interfaced with a strip chart recorder, and new techniques were mastered, it gained in popularity and usefulness. In 1971, only 188 echoes were done, but in 1976, 4254 were done. The monthly average in 1977 through November was slightly over 400.	dataset, clinical research	source	2
37	SAS as a User Maintained Data Base System	Frank J. Potter	The rationale, design, and implementation of SAS 76 as a data base management system for the processing of moderate scale clinical trials is presented in two phases. The first phase demonstrates SAS 76 flexibility of design and ease of use which permits efficient user implementation with little reliance on a computer systems department. These two properties are the primary rationale for SAS's use by a medical research team. A multiple record type input was used in the construction of SAS files which comprise the subdivisions of the data base. Through the use of embedded key identification fields for each observation, all SAS files are available for merging. This maximizes the information available for analysis of each observation, while minimizing input and storage requirements. A flexible system of data input, updating and report generation is also demonstrated. The second phase involves modifications of the data base system to enhance data manipulation and to minimize input errors through the use of processing dictionaries for adverse experience teams, laboratory normal range standards, and drug therapy. A simple method of storage space optimization is also presented along with diagrams demonstrating the benefits that result from this modification.	clinical trial, length statement, proc step	source	5
38	SAS Procedure for the Analysis of Covariance with Intra-Class Regression Model in One- and Two-Way Classifications	Harji I. Patel	This paper deals with a SAS procedure to analyze the covariance model in one- and two-way classifications with a single covariate. No meaningful comparison of all levels of an effect can be made if the regression slopes in the classes of that effect are heterogeneous. Based upon the invariance property studied by Reeder and Carter (presented at the 1976 ASA annual meeting at Boston}, the main effects and interactions that are invariant to the covariate are detected and corresponding tests of significance performed. Orthogonal contrasts are defined for the main effects and interactions. When the comparison of all levels of a given effect is dependent on the value of the covariate, orthogonal contrasts that are invariant to the covariate are detected.	main effect, proc genmod, regression analysis, proc glm, proc step, ancova, covariance	source	5
39	SAS Supplemental Procedure SPSS	William C. Reynolds	Unquestionably, the two most important and widely used general purpose statistical computer programs are SAS (Statistical Analysis System) and SPSS (Statistical Package for the Social Sciences). As the two Programs have their own unique capabilities and limitations, clearly an interface between the two would enhance the utility of each. SAS procedure SPSS is such an interface. The procedure provides for the automatic conversion of SAS data sets to SPSS system files, the execution of SPSS statements and procedures and the conversion of SPSS system files to SAS data sets. This paper describes this interface and its use, illustrates common applications and indicates those areas where it may be profitably employed by the data analyst.		source	3
40	SAS's Vital Role in a Statewide Fuel Conversion and Energy Conservation Study	E. N. Caldwell	SAS was used extensively for the data management and analysis of the four separate databases in the first comprehensive coal conversion study for residential, commercial, and institutional buildings in the United States. SAS was very instrumental in the success of the project despite formidable handicaps.	proc step, proc freq	source	5
41	Simulating a Hierarchical Master File Structure Using SAS	Helene Enid Cavior	RICHIS, an integrated data management system, combines the objectives of program monitoring with an historical research data base. The conceptual design is that of a hierarchical or tree structure with one observation per person containing unlimited repetitions of segments representing various events. RICHIS is in actuality organized as separate fixed length files, for the root and each segment, loosely linked by an identifying variable which is unique for each prisoner. This organization minimizes processing costs, avoids the problems of widely varying repetitions of events, and provides the flexibility necessary for monitoring a dynamic program. SAS was selected because it provides a unique and powerful combination of data management, report writing, and statistical analysis capabilities in addition to permitting this conceptual simulation of a single hierarchical master file.		source	5
42	Some Applications of the Discriminant Function Procedure	Emilio A. Icaza, Kenneth L. Koonce	Essentially a multivariate statistical technique, discriminant analysis has been utilized primarily for deriving a linear function of p variables which maximizes the distance between centroids or mid-points of the multivariate distributions of k groups. Such a function is frequently used as a classification criterion for deciding into which group a random observation from one of the k groups (with measures on the p variables) should be classed. Information on the relative frequency of occurrence of the k groups, when available, is also useful in minimizing the degree of misclassification. `PROCEDURE DISCRIM` in SAS has some techniques which are generally not available in other statistical packages, the main one being a test for equality of within group covariance matrices. This feature allows the use of inequalities of covariances to further reduce misclassification. Examples of `PROC DISCRIM` applications from three research areas are presented; education, geology and zoology. The zoology example, species classification, illustrates one of the first applied uses of discriminant analysis. Two non-discriminant analysis uses of `PROC DISCRIM`, within group correlation and univariate equality of variances test, are presented. Some deficiencies of the SAS `DISCRIM` procedure are highlighted: no method for testing the statistical significance of either the overall discriminant function or specific coefficients is given and no variable selection technique analogous to stepwise regression is available.	proc step, discriminant analysis, proc discrim, covariance, anova, correlation, sample, response variable, predictor variable, regression analysis	source	4
43	Stability Scheduling Using SAS	Robert J. Bronstein	Federal regulations will soon be published defining Good Manufacturing Practices (GMP's). A portion of these regulations will specify procedures that must be performed when determining the shelf life or stability of a product. At USV Pharmaceutical Corporation¹ a programming system is being developed to handle these newly defined requirements imposed by the Federal Drug Administration.	sample, proc step	source	5
44	The 1978 Interface Between SAS and BMDP and Some BMDP Features of Interest to SAS Users	James W. Frane	Recent changes in SAS-76 and BMDP-77 enable RMDP programs to obtain data from SAS files without the SAS file first being converted to a BMDP file. This new interface makes it possible for SAS users to economically make use of many features in BMDP that are not available in SAS. The new interface also makes SAS an extremely attractive complement to BMDP.	regression analysis, proc step, anova, proc freq	source	4
45	The Data Base Approach at the Bureau of Labor Statistics	Rudolph C. Mendelssohn	The Bureau of Labor Statistics (BLS) has Pained economy and speedier results through the use of a data base manager as the central feature for most of its computer uses, Associated generalized systems are embedded in very high-level, nonprocedural languages that may be used by personnel not knowledgeable in the computer sciences,	time series analysis, sample	source	4
46	The KWIKPLOT Procedure	Paul D. Hopkins, Peggy A. Pearsall	The KWIKPLOT procedure is used as an interface between SAS and the software package for Houston Instrument's COMPLOT incremental plotter. It allows users to produce one or more plots with a minimum of control statements. This procedure is designed for interactive use only. It will produce either a simple scatter diagram or a line plot as output. Options and parameters can be used to vary the line type, plot size, pen color, pen type, and the plot symbol to be used.	by group processing, proc step	source	3
47	The LIFETEST Procedure for Analyzing Reliability Data	Hoke S. Hill, Frederick W. Morgan, Richard Nelson	The `PROCEDURE LIFETEST` has been developed for estimating and testing hypotheses concerning the parameters of distributions commonly encountered in evaluating the reliability of system components. Currently, the procedure contains point and interval estimates, and "p" values for testing the parameters of the exponential distribution for a variety of sampling schemes. The sampling includes two types of censoring where each censoring scheme can be conducted with or without the replacement of the items that have failed. The complete `LIFETEST` procedure will include the Weibull, gamma, and log-normal distributions.	sample, confidence interval	source	3
48	The Use of SAS for the Analysis of Family Studies	C. Robert Cloninger, J. Philip Miller, Deborah Powell	The analysis of family studies present significant problems with respect to the use of standard statistical packages as the data set does not conform to the simple rectangular observations by variables structure. The data management facilities of SAS, however, provide the necessary tools to produce the desired analyses. The principal challenge of family studies is the variety of ways in which the data needs to be organized for the production of the required analyses. Some analyses require only the selection of an appropriate subset of the database, e.g. all probands, all fathers, all first degree relatives of probands. Others require the linking of related individuals, e.g. mother-daughter, father-offspring, husband-wife. Since the amount and/or type of information often depends upon the familial relationship of the individual, we have found that an effective strategy is to store the data in multiple SAS datasets according to the type of information obtained, but all within a single SAS library. Analysis data sets are then straightforwardly developed with the use of sorting and merging operations within SAS. These analysis data sets can either be created as temporary, work data sets or saved in the library if they are to be frequently utilized.	order by, proc sort	source	6
49	Unsymmetrical Parallel Line Bioassay Using SAS	Chiao Yeh	Even in the most carefully conducted bio-assay experiment, an accident circumstance or unequal litter size may cause the loss of observations and so destroy the symmetry of the design. In this paper, testing for non-validity, point and interval estimation procedures of relative potency in an unsymmetrical parallel line assay will be discussed using SAS. A numerical example will be given.	regression analysis, confidence interval, anova, covariance	source	5
50	Use of MATRIX for Single Degree of Freedom Contrasts in Factorial Experiments.	Samuel G. Carmer	A SAS-76 MACRO has been prepared to compute single degree of freedom linear contrasts on data from balanced factorial experiments. The MACRO utilizes MATRIX to perform a tabular analysis on the treatment totals. An example involving a 3 x 2 factorial is given.	degrees of freedom, response variable, main effect, anova, regression analysis, predictor variable	source	4
51	Use of SAS at Oklahoma State University During the Fall Semester 1977	Robert D. Morrison	The tape records for all jobs logged in the Oklahoma State University computing center from August 1 to December 31, 1977 were searched for SAS usage. CPU time, elapsed time and frequency of use were summarized by type of user.	proc freq, cpu time	source	5
52	Use of SAS in a Statistical Methodology Course	William J. Wilson	The use of SAS in teaching an intermediate level Statistics Methods course is proposed as a viable alternative to the more classic approach. Problems encountered in the three years this approach has been used at the University of North Florida are discussed and possible solutions suggested. Extensions of this approach to other courses are also considered.		source	5
53	Use of SAS Procedures in Estimating Survival Curves	Ernest Hilderbrand, Melvin Moeschberger	If a particular parametric family of distributions is assumed (as is often the case in human or animal survival studies) then maximum likelihood techniques are appropriate to estimate the parameters of the life distributions of interest. From estimates of these parameters one can estimate the average life expectancy along with other pertinent quantities of interest. This paper will present an application of SAS procedures (viz., `PROC` NONLINEAR and `PROC` MATRIX) towards finding such maximum likelihood estimates along with their estimated asymptotic variances. An example dealing with breast cancer will be used to illustrate the technique.	proc step, likelihood function, maximum likelihood estimation	source	7
54	Use of the Absorb Statement in SAS	Michael Conlon, Jean B. Holzer	Use of the ABSORS statement in PROCS GLM and ANOVA in SAS 76 is illustrated for three repeated measures design examples. Coding procedures and limitations are discussed, and actual savings in GPU time and memory required are documented.		source	4
55	Using SAS - GLM in Kinetic Modeling	Irving K. Hwang	A kinetic model is generally expressed in differential equations. These equations may not be solved analytically. To establish these equations the parameters or coefficients muse be estimated directly from experimental data. Innumerable procedures, such as nonlinear least squares, absolute least deviation, quasilinearization, and invariant embedding have been proposed and used to solve the nonlinear estimation problems. The purpose of this paper is to demonstrate that SAS - GLM can also be used to provide good parameter estimates through linear least squares, based on the estimation procedure of Hwang and Yeh. Numerical example in a two-compartment open model in clinical pharmacokinetics is given and the results of a comparison with the SAS - NLIN are presented.	proc genmod, proc step	source	4
56	Using SAS to Tune MVS	H. W. Barry Merrill	Using the Statistical Analysis System (SAS) and SMF records (including RMF) can provide a comprehensive source of information for the measurement of MVS. This paper describes a specific implementation of the data collection and reporting portions of such a system.	dataset, cpu time, proc freq	source	24
57	Using Software Physics to Size Workloads with SAS	Kenneth W. Kolence	This paper discusses the use of software work vectors as the fundamental form for workload characterization purposes. Software work is one of the basic properties of software physics, and has the property of being independent of processor speed, configuration design, and is unaffected by other workload elements. The general vector form of characterization is defined. Equations for calculating work are given, and examples of its uses considered. The relation of this form to the data used for synthetic benchmarks and workload analyses is briefly reviewed.	cluster analysis, sample	source	5
58	What Does GLM Do? The Case of the Missing Cell	Rudolf J. Freund	The `GLM` procedure was developed for the purpose of providing information on the structure and meaning of results of linear models analyses of incomplete data. The brevity of descriptions of GLM in the Users' Guide may make it difficult for the average user to properly understand the given results. A simple example is used to illustrate 1) what information GLM provides and how to use it, 2) the non-uniqueness of the Type IV functions, and 3) what the REGR procedure of SAS-72 did with the same data.	interaction effect, proc genmod, proc step, proc glm	source	6

Back to Conf go 1979 go 1977

This site serves as a convenient hosting location and does not work as a publisher. All copyrights belong to their respective owners.

The information on this site is provided for general reference purposes only. Whilst every effort is made to ensure that the site is up to date and accurate, no warrant, nor any responsibility or liability for, the accuracy or completeness of the content or for any loss which may arise from reliance on information contained in this site is provided. Links to other sites are provided for your convenience but no responsibility or liability for the content of those sites or of any external site which links to this site is accepted. References to any specific commercial product, process, service, or corporation are also for information only, and do not constitute endorsement, recommendation or favoring.

SAS is a registered trademark of SAS Institute Inc. SAS, Statistical Analysis System, and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.