In response to a survey offaculty regarding their commercial software preferences for advanced analysis of variance courses, it was found that the most frequently used packages were SAS, spssx, and BMDP, all originally mainframe packages. The fourth choice, SYSTAT, was written for microcomputers but is currently also available for other host computers. Difficulty for students was the most frequently cited reason for not using one of these four packages. The most important criteria for choosing software packages were variety of designs, accuracy, and ease of use. Cost was a factor only for microcomputer licenses. Packages using the general linear model approach were paramount to some, while others would not consider such packages. Instructors decried the lack of the "ideal program," but the diversity oftheir desires makes it clear that no program could be ideal for all of them. Many faculty seemed unaware of newer software packages or of the extent to which older packages (particularly Minitab) have been modified; some had their choices constrained by departmental limitations. Better dissemination of information about statisti· cal software is needed, whether it be from software publishers or through the professional literature.Each year new statistical software packages are added to the already abundant number available, and existing packages are constantly being modified, corrected, and extended. All of the original mainframe packages are still available, and all have been (at least partially) rewritten for IBM-eompatible microcomputers. New packages have been developed specifically for IBM-compatible microcomputers and some of these and others are rapidly becoming available in the Macintosh world. Most of us cannot hope to keep up with all of the latest developments in statistical software.Although the packages offer a wide variety of statistical techniques, this paper focuses on the choice of packages for a senior-or graduate-level course in analysis of variance (ANDYA). Advanced ANDYA is the most widely taught statistical technique in graduate training programs (Aiken, West, Sechrest, & Reno, 1990). Among the 222 PhD programs in psychology surveyed, 65 % offer a full course in ANDYA; 88 % offer at least a partial course.We were interested in learning which software packages were currently being used, which had been used in the past, and which had been evaluated and rejected, and why. We also sought to determine the importance of such dimensions as coverage of a wide variety of ANDYA designs, cost, ease of use, and whether or not the general linear model approach was available. To obtain this information, we surveyed many faculty regarding their choices among packages for this type of course.
We appreciate the time and attention devoted to our book, Using Multivariate Statistics, by 1~. F.Widaman in his review that appeared in Volume 8, Number 1, pages 119-120. Although much of his description of the book is accurate, we believe that some of the problems with it that he mentions were unfounded.His first complaint is that the order of presentation of the analytical techniques is haphazard. The rationale for our arrangement is presented in chapter 3 in which topics are grouped by major research question. Following that logic, regression (predicting one variable from many) is followed by canonical correlation (predicting several variables from many) instead of by ANCOVA with which it is more closely related mathematically. For those who prefer the mathematical connection, rather than the organization by major research question, we recommend that chapters be read in another order, with little or no loss in comprehension.Similarly, a case could be made for an earlier overview of the multivariate general linear model, left to the end of the book because it was seen to be the most intimidating topic. Nothing prevents reading of that chapter directly after the 6 'G~ide to Techniques&dquo; (chap. 3) nor, for that matter, before it.A second complaint is that more pages were allocated to less frequently used topics. The point is certainly correct; however, pages were determined by difficulty of explaining the material, not by extent of use of the technique.Minor problems cited are that the definition of r would be more consistent with tradition using N rather than N -1 in the denominator-tme-and that the degrees of freedom in the test of the second variable on page 142 are wrong-untrue. Widaman might wish to disagree with the equation from which the degrees of freedom were derived (pp. 112-113), but the degrees of freedom used are consistent with that equation. Widaman accurately claims that there are features of SPSS and BMDP programs that are tabulated, but never explained. Indeed, the programs are much more flexible than we have indicated, a flexibility that requires in many cases great sophistication on the part of the user. We chose to include features of the programs in the tabulation that sophisticated users would understand, but that our book did not explain.We do not understand Widaman's complaint that we failed to explain the rationale for the number of factors in the example in chapter 10 because the rationale is explained on page 425.The last major problem for which Widaman chides us is the failure to include SAS in the book. Our excuse-called ' 'lame&dquo; -is that SAS was unavailable to us at the time the book was written (and is still unavailable, for that matter). We plead guilty and eagerly look forward to an offer from Widaman for access to SAS through his computing facility.
Preparatory data analyses (data screening) are conducted before a main analysis to assess the fit between the data and the assumptions of that main analysis. Different main analyses have different assumptions that vary in importance; violation of some assumptions can lead to the wrong inferential conclusion (and a potential failure of replication) while violation of others yields an analysis that is correct as far as it goes, but misses certain additional relationships in the data. Assumptions that are often relevant for continuous variables are normality of sampling distributions, pairwise linearity, absence of outliers and collinearity, independence of errors, and homoscedasticity; these are evaluated by both graphical and statistical methods. When violation is detected, variables are often transformed or an alternative analytic strategy is employed. Relevant issues in the choice of when and how to screen are the level of measurement of the variables, whether the design produces grouped or ungrouped data, whether cases provide a single response or more than one response, and whether the variables themselves or the residuals of analysis are screened.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.