Blog Archives
WHY CHOOSE STATISTICA DATA MINER?
WHY CHOOSE STATISTICA DATA MINER?
STATISTICA Data Miner is the most powerful and comprehensive selection of data mining solutions on the market, with an iconbased, extremely easytouse user interface and deployment engine. It features a selection of automated and readytodeploy data mining.
At the same time, STATISTICA Data Miner is highly customizable and can be tailored to meet very specific and demanding analysis requirements through its open architecture. It also features a straightforward, wizardstyle Data Miner Recipes user interface that guides the user stepbystep with plain English prompts to create a desired model or a workflow, or alternatively creates a model in one step by bypassing all prompts and using intelligent defaults selected by the program.
A flexible and powerful solution:
• Highly scalable, leadingedge algorithms, trees, boosting, ensembles
• All standard analytic modeling tools, logistic regression, linear models…
• Industry standard interfaces
• Ability to connect to all industry standard database and file formats
• Comprehensive API allows quick and straightfoward connection to
existing IT assets, including scheduling engines
Preferred tools and solutions recognized by industry experts, experienced
users, and novices alike:
• Rated highest in customer satisfaction in largest independent surveys
of users for three consecutive years1
• DoubleVictor in Predictive Analytics Industry Quadrant2
• Reviewers’ Choice: Statistica software offers superior price to
performance ratio3
• Used by hundreds of thousands of users worldwide; supported by
30 offices on all continents
Comprehensive BIG DATA support
Enterprise versions of Data Miner offer:
• Access to Hadoop and other distributed file system formats
• Proprietary algorithms to optimize calculation of analytics on extremely
large data sets (with mapreduce analytics options)
• Multithreading optimizations to leverage multicore CPU architecture
and provide high performance even on relatively inexpensive hardware
• Managed automation to build, recalibrate, and deploy models, so you
can work with thousands of models to extract big insights from big data
To read more click here to see the full Data Miner Brochure.
StatSoft Southern Africa – Research (Pty) Ltd.
1st Floor
Petervale Centre
Cnr. Frans Hals & Cambridge Streets
Petervale, Sandton 2196
SOUTH AFRICA
Phone: +27112346148
Fax: +27112347868
email: info@statsoft.co.za, sales@statsoft.co.za, support@statsoft.co.za, accounts@statsoft.co.za
Sales, Training, Consulting Services
Website: StatSoft Southern AfricaResearch (Pty) Ltd.
Power Plant Performance Reporter
 Review historical performance data
 Conduct performance analyses
 Identify trends and run comparisons
All with no effort or help from your IT department!
Power Plant Performance Reporter is an inexpensive and simpletouse system for summarizing and reporting on all aspects of your power plant’s performance; installs in minutes right from your desktop!
Power Plant Performance Reporter is a simpletoinstall, simpletouse, and inexpensive software solution, to combine all available data from different sources for summary analyses, adhoc graphing, and summary reporting
Do you currently have any tools to:
 Create summary and trend reports of how your plant performed last week, month, year?
 Create scatterplots, trend plots, box plots, etc. of key power plant performance indicators, to summarize and explore validated, aggregated historical data?
 Create reports automatically on your desktop every morning (as PDF or Word documents) to summarize what happened last night, last week, etc.?
If not, you need Power Plant Performance Reporter.
Power Plant Performance Reporter enables you to know what is happening with your power plant, instead of “guessing” what may have happened!
Power Plant Performance Reporter will be the most significant and least expensive tool you ever buy to keep you informed and in control!
For example, with Power Plant Performance Reporter, you can create informative reports on key power plant performance indicators, quickly and without effort:
 Create summary plots and statistical reports of flame temperature trends, emissions, heat transfer efficiencies by dayoftheweek, hour, month, etc.
 Automatically create default reports every morning about operations over the past 24 hours
 Quickly create graphs of primary, secondary, and tertiary air flows, stochoimetric ratios, OFA vs. CO, NOx, for hourly summaries, for the entire last month, based on validated data
What it does:
Power Plant Performance Reporter(PPPR) enables you to create quickly summary reports, graphs, statistics, or adhoc analyses of the data you are already collecting.
How it works:
Power Plant Performance Reporter will automatically connect to your OSI Pi and other auxiliary databases (e.g., Access databases), clean and validate the data, aggregate the data (e.g., to hourly averages, minimum/maximum values, etc.), and align the data so that you can quickly examine meaningful graphs, summary reports, and trend reports.
Call us today to discuss a project to implement the most costeffective solution for power plant optimization.
Data Mining & StatSoft Power Solutions
Analytics supported
by third party report
The nonprofit Electric Power Research Institute (EPRI) recently conducted a study of the StatSoft technology to determine its suitability for optimizing the performance (heatrate, emissions, LOI) in an older coalfired power plant. EPRI ordered from StatSoft an optimization project to be conducted under scrutiny of their inspectors.
Using nine months worth of detailed 6minute interval data describing more than 140 parameters of the process, EPRI found that process data analysis using STATISTICA is a costeffective solution for optimizing the use of current process hardware to save cost and reduce emissions.
Overview of the Approach
StatSoft Power Solutions offer solution packages designed for utility companies, for optimizing power plant performance, increasing the efficiency, and reducing emissions. Based on over 20 years of experience in applying advanced datadriven, data mining optimization technologies for process optimization in various industries, these solutions will allow power plants to get the most out of their equipment and control systems, by leveraging all data collected at your site to identify opportunities for improvement, even for older designs such as coalfired Cyclone furnaces (as well as wallfired or Tfired designs).
Opportunities for Data Driven Strategies to Improve Powerplant Performance
Many (most) power generation facilities are collecting “lots of data” into dedicated historical process data bases. (such as OSI PI) However, in most cases, only simple charts and “afterthefact” adhoc analyses are performed on a small subset of those data; most information is simply not used.
For example, for coal fired power plants, our solutions can help you identify optimum settings for stoichiometric ratio, primary/tertiary air flows, secondary air biases, distribution of OFA (overfired air), burner tilts and yaw positions, and other controllable parameters to reduce NOx, CO, and LOI, without requiring any reengineering of existing hardware.
What is Data Mining? Why Data Mining?
Data Mining is the term used to describe the application of various machine learning and/or pattern recognition algorithms and techniques, to identify complex relationships among observed variables. These techniques can reveal invaluable insights when the data contain meaningful information which is “hidden” deep inside your data set, and cannot be identified with simple methods. Advanced data mining can reveal those insights by processing many variables and complex interrelations between them, all at the same time.
Unlike CFD, data mining allows you to model the “real world” from “real data,” describing your specific plant. Using this approach, you can:
 Identify from among hundreds or even thousands of input parameters those that are critical for lowemissions efficient operations
 Determine the ranges for those parameters, and combinations of parameter ranges that will result in robust and stable lowemissions operations, without costly excursions (highemissions events, unscheduled maintenance and expensive generation rollbacks).
These results can be implemented using your existing closedoropen loop control system to achieve sustained improvements in power plant performance, or you can use StatSoft MultiStream to create a stateoftheart advanced process monitoring system to achieve permanent improvements.
How is this Different from “Neural Nets” for Closed Loop Control?
One frequently asked question is: How do these solutions differ from neural networks based computer programs that can control critical power plant operations in a closed loop system (an approach used at some plants, often with less than expected success)?
The answer is that, because those systems are based on relatively simple, traditional neural networks technology which typically can only simultaneously process relatively few parameters, they are not capable of identifying the important parameters from among hundreds of possible candidates, and they will not identify specific combinations of parameter ranges (“sweet spots”) that make overall power plant operations more robust.
The cuttingedge technologies developed by StatSoft Power Solutions will not simply implement a cookiecutter approach to use a few parameters common to all power plants to achieve some (usually only very modest) overall process performance improvements. Instead, our approach allows you to take a fresh look at all your data and operations, to optimize them for best performance. This will allow you to focus your process monitoring efforts, operator training, or automation initiatives only on those parameters that actually drive boiler efficiency, emissions, and so on at your plant and for your equipment.
What we are offering is not simply another neural net for closed loop control; instead, it provides flexible tools based on cuttingedge data processing technologies to optimize all systems, and also provides smart monitoring and advisory options capable of predicting problems, such as emissions related to combustion optimization or maintenance issues.
Contact StatSoft Southern Africa for more information about our services, software solutions, and recent success stories. lorraine@statsoft.co.za
Considering Alternatives to SAS?
Do you use SAS for predictive modeling, advanced analytics, business intelligence, insurance or financial applications, or data visualization?
Why Choose STATISTICA?
SAS software is expensive and carries high, unpredictable annual licensing costs. SAS software is difficult to use, requiring specific SAS programming expertise, and it drives users toward dependency on only SASspecific solutions (e.g., their proprietary data warehouses). Data visualization is integral for analytics, but SAS’s graphics have major shortcomings.
STATISTICA has consistently been ranked the highest in ease of use and customer satisfaction in independent surveys of analytics professionals. Click here to see the results of the most recent Rexer survey (2010), the largest survey of data mining professionals in the industry.
We offer the breadth of analytics capabilities and performance, including the most comprehensive data mining solution on the market, using more open, modern technologies. StatSoft software is designed to facilitate interfacing with all industry standard components of your computer infrastructure (e.g., ultrafast integration with Oracle, MS SQL Server, and other databases) instead of locking you into proprietary standards and total dependence on one vendor.
STATISTICA is significantly faster than SAS. StatSoft is an Intel® Software Premiere Elite Partner and has developed technologies that leverage Intel CPU architecture to deliver unmatched parallel processing performance (press release with Intel) and rapidly process terabytes of data. StatSoft’s robust, cuttingedge enterprise system technology drives the analytics and analytic data management at some of the largest computer infrastructures in the world at Fortune 100 and Fortune 500 companies.
Quotes from SAS Customers
“We acquired our SAS license seven years ago and quickly learned that with SAS, you do not pay just an annual renewal and support fee – you practically have to “buy” the software again every year. Our first year renewal fee was already 60% of the initial purchase price, and it increased steadily and every year. Two years ago, our annual fee exceeded the initial purchase price we paid, and it keeps going up much faster than the inflation. This is not sustainable.” – CEO, Technology Company
“It took 8 weeks to install SAS Enterprise Miner. The installer just didn’t work. And we’re a midsize company, so we were a low priority for SAS’s technical support.” – Engineer, Chemical Company
“Early in our evaluation, we eliminated SAS from our consideration of fraud detection solutions primarily due to the exorbitant cost.” – Chief Actuary, Insurance Company
“We had used SAS ondemand for my data mining class. A few days before finals, all of our students’ project files were corrupted. Our SAS technical support representative confirmed there was nothing that could be done to restore the files. We’re switching to STATISTICA.” – University Professor
“Now, all graduate students use R. It is getting more difficult to find SAS programmers.” – Head of Statistics, Pharmaceutical Company
“We used SAS until May 2009 when we converted to WPS. The conversion went remarkably smoothly and was completed on time. Not only did we save a substantial amount in licensing fees, we also regained functionality such as Graphs that we had previously removed because of the cost.” – Survey respondent on KDNuggets.com
How to Proceed
StatSoft makes it easy to transition your current SAS environment to STATISTICA, either gradually or all at once. STATISTICA offers:
Direct import/export to SAS files
Deployment of predictive models to SAS code to score against SAS data sets
Native integration to run R program
For more information and for specific recommendations to suit your needs, please contact one of our representatives using the form below:
Featured Article: Full Rexer Report Shows StatSoft STATISTICA’s a Big Winner Among Users!
Featured Article: Full Rexer Report Shows STATISTICA’s a Big Winner Among Users
Rexer Analytics has just released the full summary report of its 5th Annual Data Miner Survey, and we are excited to share that StatSoft’s STATISTICA blew away the competition.
Not only was STATISTICA the primary data mining tool chosen most often by users, but STATISTICA received the highest user satisfaction in 16 out of 20 categories as shown here:
Download A STATISTICA Advanced v10 Demo
Has my university/institution got a license?
For more information Contact us…..
LORRAINE EDEL
StatSoft Southern Africa Research
Tel: 011 2346148
Fax: 086 5441172
Cell: 082 5678 330
Mail: lorraine@statsoft.co.za
Web: www.statsoft.co.za
Powerful solutions for:
* Business intelligence
* Data mining
* Research Analytics
* Quality control
Generalized Additive Models (GAM)
The methods available in Generalized Additive Models are implementations of techniques developed and popularized by Hastie and Tibshirani (1990). A detailed description of these and related techniques, the algorithms used to fit these models, and discussions of recent research in this area of statistical modeling can also be found in Schimek (2000).
Additive Models
The methods described in this section represent a generalization of multiple regression (which is a special case of general linear models). Specifically, in linear regression, a linear leastsquares fit is computed for a set of predictor or X variables, to predict a dependent Y variable. The well known linear regression equation with m predictors, to predict a dependent variable Y, can be stated as:
Y = b0 + b1*X1 + … + bm*Xm
Where Y stands for the (predicted values of the) dependent variable, X1through Xm represent the m values for the predictor variables, and b0, and b1 through bm are the regression coefficients estimated by multiple regression. A generalization of the multiple regression model would be to maintain the additive nature of the model, but to replace the simple terms of the linear equation bi*Xi with fi(Xi) where fi is a nonparametric function of the predictor Xi. In other words, instead of a single coefficient for each variable (additive term) in the model, in additive models an unspecified (nonparametric) function is estimated for each predictor, to achieve the best prediction of the dependent variable values.
Generalized Linear Models
To summarize the basic idea, the generalized linear model differs from the general linear model (of which multiple regression is a special case) in two major respects: First, the distribution of the dependent or response variable can be (explicitly) nonnormal, and does not have to be continuous, e.g., it can be binomial; second, the dependent variable values are predicted from a linear combination of predictor variables, which are “connected” to the dependent variable via a link function. The general linear model for a single dependent variable can be considered a special case of the generalized linear model: In the general linear model the dependent variable values are expected to follow the normal distribution, and the link function is a simple identity function (i.e., the linear combination of values for the predictor variables is not transformed).
To illustrate, in the general linear model a response variable Y is linearly associated with values on the X variables while the relationship in the generalized linear model is assumed to be
Y = g(b0 + b1*X1 + … + bm*Xm)
where g(…) is a function. Formally, the inverse function of g(…), say gi(…), is called the link function; so that:
gi(muY) = b0 + b1*X1 + … + bm*Xm
where muY stands for the expected value of Y.
Distributions and Link Functions
Generalized Additive Models allows you to choose from a wide variety of distributions for the dependent variable, and link functions for the effects of the predictor variables on the dependent variable (see McCullagh and Nelder, 1989; Hastie and Tibshirani, 1990; see also GLZ Introductory Overview – Computational Approach for a discussion of link functions and distributions):
Normal, Gamma, and Poisson distributions:
Log link: f(z) = log(z)
Inverse link: f(z) = 1/z
Identity link: f(z) = z
Binomial distributions:
Logit link: f(z)=log(z/(1z))
Generalized Additive Models
We can combine the notion of additive models with generalized linear models, to derive the notion of generalized additive models, as:
gi(muY) = Si(fi(Xi))
In other words, the purpose of generalized additive models is to maximize the quality of prediction of a dependent variable Y from various distributions, by estimating unspecific (nonparametric) functions of the predictor variables which are “connected” to the dependent variable via a link function.
Estimating the Nonparametric Function of Predictors via Scatterplot Smoothers
A unique aspect of generalized additive models are the nonparametric functions fi of the predictor variables Xi. Specifically, instead of some kind of simple or complex parametric functions, Hastie and Tibshirani (1990) discuss various general scatterplot smoothers that can be applied to the X variable values, with the target criterion to maximize the quality of prediction of the (transformed) Y variable values. One such scatterplot smoother is the cubic smoothing splines smoother, which generally produces a smooth generalization of the relationship between the two variables in the scatterplot. Computational details regarding this smoother can be found in Hastie and Tibshirani (1990; see also Schimek, 2000).
To summarize, instead of estimating single parameters (like the regression weights in multiple regression), in generalized additive models, we find a general unspecific (nonparametric) function that relates the predicted (transformed) Y values to the predictor values.
A Specific Example: The Generalized Additive Logistic Model
Let us consider a specific example of the generalized additive models: A generalization of the logistic (logit) model for binary dependent variable values. As also described in detail in the context of Nonlinear Estimation and Generalized Linear/Nonlinear Models, the logistic regression model for binary responses can be written as follows:
y=exp(b0+b1*x1+…+bm*xm)/{1+exp(b0+b1*x1+…+bm*xm)}
Note that the distribution of the dependent variable is assumed to be binomial, i.e., the response variable can only assume the values 0 or 1 (e.g., in a market research study, the purchasing decision would be binomial: The customer either did or did not make a particular purchase). We can apply the logistic link function to the probability p (ranging between 0 and 1) so that:
p’ = log {p/(1p)}
By applying the logistic link function, we can now rewrite the model as:
p’ = b0 + b1*X1 + … + bm*Xm
Finally, we substitute the simple singleparameter additive terms to derive the generalized additive logistic model:
p’ = b0 + f1(X1) + … + fm(Xm)
An example application of the this model can be found in Hastie and Tibshirani (1990).
Fitting Generalized Additive Models
Detailed descriptions of how generalized additive models are fit to data can be found in Hastie and Tibshirani (1990), as well as Schimek (2000, p. 300). In general there are two separate iterative operations involved in the algorithm, which are usually labeled the outer and inner loop. The purpose of the outer loop is to maximize the overall fit of the model, by minimizing the overall likelihood of the data given the model (similar to the maximum likelihood estimation procedures as described in, for example, the context of Nonlinear Estimation). The purpose of the inner loop is to refine the scatterplot smoother, which is the cubic splines smoother. The smoothing is performed with respect to the partial residuals; i.e., for every predictor k, the weighted cubic spline fit is found that best represents the relationship between variable k and the (partial) residuals computed by removing the effect of all other j predictors (j ¹ k). The iterative estimation procedure will terminate, when the likelihood of the data given the model can not be improved.
Interpreting the Results
Many of the standard results statistics computed by Generalized Additive Models are similar to those customarily reported by linear or nonlinear model fitting procedures. For example, predicted and residual values for the final model can be computed, and various graphs of the residuals can be displayed to help the user identify possible outliers, etc. Refer also to the description of the residual statistics computed by Generalized Linear/Nonlinear Models for details.
The main result of interest, of course, is how the predictors are related to the dependent variable. Scatterplots can be computed showing the smoothed predictor variable values plotted against the partial residuals, i.e., the residuals after removing the effect of all other predictor variables.
This plot allows you to evaluate the nature of the relationship between the predictor with the residualized (adjusted) dependent variable values (see Hastie & Tibshirani, 1990; in particular formula 6.3), and hence the nature of the influence of the respective predictor in the overall model.
Degrees of Freedom
To reiterate, the generalized additive models approach replaces the simple products of (estimated) parameter values times the predictor values with a cubic spline smoother for each predictor. When estimating a single parameter value, we lose one degree of freedom, i.e., we add one degree of freedom to the overall model. It is not clear how many degrees of freedom are lost due to estimating the cubic spline smoother for each variable. Intuitively, a smoother can either be very smooth, not following the pattern of data in the scatterplot very closely, or it can be less smooth, following the pattern of the data more closely. In the most extreme case, a simple line would be very smooth, and require us to estimate a single slope parameter, i.e., we would use one degree of freedom to fit the smoother (simple straight line); on the other hand, we could force a very “nonsmooth” line to connect each actual data point, in which case we could “useup” approximately as many degrees of freedom as there are points in the plot. Generalized Additive Models allows you to specify the degrees of freedom for the cubic spline smoother; the fewer degrees of freedom you specify, the smoother is the cubic spline fit to the partial residuals, and typically, the worse is the overall fit of the model. The issue of degrees of freedom for smoothers is discussed in detail in Hastie and Tibshirani (1990).
A word of Caution
Generalized additive models are very flexible, and can provide an excellent fit in the presence of nonlinear relationships and significant noise in the predictor variables. However, note that because of this flexibility, you must be extra cautious not to overfit the data, i.e., apply an overly complex model (with many degrees of freedom) to data so as to produce a good fit that likely will not replicate in subsequent validation studies. Also, compare the quality of the fit obtained from Generalized Additive Models to the fit obtained via Generalized Linear/Nonlinear Models. In other words, evaluate whether the added complexity (generality) of generalized additive models (regression smoothers) is necessary in order to obtain a satisfactory fit to the data. Often, this is not the case, and given a comparable fit of the models, the simpler generalized linear model is preferable to the more complex generalized additive model. These issues are discussed in greater detail in Hastie and Tibshirani (1990).
Another issue to keep in mind pertains to the interpretability of results obtained from (generalized) linear models vs. generalized additive models. Linear models are easily understood, summarized, and communicated to others (e.g., in technical reports). Moreover, parameter estimates can be used to predict or classify new cases in a simple and straightforward manner. Generalized additive models are not easily interpreted, in particular when they involve complex nonlinear effects of some or all of the predictor variables (and, of course, it is in those instances where generalized additive models may yield a better fit than generalized linear models). To reiterate, it is usually preferable to rely on a simple well understood model for predicting future cases, than on a complex model that is difficult to interpret and summarize.
General Linear Models (GLM)
This topic describes the use of the general linear model in a wide variety of statistical analyses. If you are unfamiliar with the basic methods of ANOVA and regression in linear models, it may be useful to first review the basic information on these topics in Elementary Concepts. A detailed discussion of univariate and multivariate ANOVA techniques can also be found in the ANOVA/MANOVA topic.
Basic Ideas: The General Linear Model
The following topics summarize the historical, mathematical, and computational foundations for the general linear model. For a basic introduction to ANOVA (MANOVA, ANCOVA) techniques, refer to ANOVA/MANOVA; for an introduction to multiple regression, see Multiple Regression; for an introduction to the design an analysis of experiments in applied (industrial) settings, see Experimental Design.
Historical Background
The roots of the general linear model surely go back to the origins of mathematical thought, but it is the emergence of the theory of algebraic invariants in the 1800’s that made the general linear model, as we know it today, possible. The theory of algebraic invariants developed from the groundbreaking work of 19th century mathematicians such as Gauss, Boole, Cayley, and Sylvester. The theory seeks to identify those quantities in systems of equations which remain unchanged under linear transformations of the variables in the system. Stated more imaginatively (but in a way in which the originators of the theory would not consider an overstatement), the theory of algebraic invariants searches for the eternal and unchanging amongst the chaos of the transitory and the illusory. That is no small goal for any theory, mathematical or otherwise.
The wonder of it all is the theory of algebraic invariants was successful far beyond the hopes of its originators. Eigenvalues, eigenvectors, determinants, matrix decomposition methods; all derive from the theory of algebraic invariants. The contributions of the theory of algebraic invariants to the development of statistical theory and methods are numerous, but a simple example familiar to even the most casual student of statistics is illustrative. The correlation between two variables is unchanged by linear transformations of either or both variables. We probably take this property of correlation coefficients for granted, but what would data analysis be like if we did not have statistics that are invariant to the scaling of the variables involved? Some thought on this question should convince you that without the theory of algebraic invariants, the development of useful statistical techniques would be nigh impossible.
The development of the linear regression model in the late 19th century, and the development of correlational methods shortly thereafter, are clearly direct outgrowths of the theory of algebraic invariants. Regression and correlational methods, in turn, serve as the basis for the general linear model. Indeed, the general linear model can be seen as an extension of linear multiple regression for a single dependent variable. Understanding the multiple regression model is fundamental to understanding the general linear model, so we will look at the purpose of multiple regression, the computational algorithms used to solve regression problems, and how the regression model is extended in the case of the general linear model. A basic introduction to multiple regression methods and the analytic problems to which they are applied is provided in the Multiple Regression.
The Purpose of Multiple Regression
The general linear model can be seen as an extension of linear multiple regression for a single dependent variable, and understanding the multiple regression model is fundamental to understanding the general linear model. The general purpose of multiple regression (the term was first used by Pearson, 1908) is to quantify the relationship between several independent or predictor variables and a dependent or criterion variable. For a detailed introduction to multiple regression, also refer to the Multiple Regression section. For example, a real estate agent might record for each listing the size of the house (in square feet), the number of bedrooms, the average income in the respective neighborhood according to census data, and a subjective rating of appeal of the house. Once this information has been compiled for various houses it would be interesting to see whether and how these measures relate to the price for which a house is sold. For example, we might learn that the number of bedrooms is a better predictor of the price for which a house sells in a particular neighborhood than how “pretty” the house is (subjective rating). We may also detect “outliers,” for example, houses that should really sell for more, given their location and characteristics.
Personnel professionals customarily use multiple regression procedures to determine equitable compensation. We can determine a number of factors or dimensions such as “amount of responsibility” (Resp) or “number of people to supervise” (No_Super) that we believe to contribute to the value of a job. The personnel analyst then usually conducts a salary survey among comparable companies in the market, recording the salaries and respective characteristics (i.e., values on dimensions) for different positions. This information can be used in a multiple regression analysis to build a regression equation of the form:
Salary = .5*Resp + .8*No_Super
Once this socalled regression equation has been determined, the analyst can now easily construct a graph of the expected (predicted) salaries and the actual salaries of job incumbents in his or her company. Thus, the analyst is able to determine which position is underpaid (below the regression line) or overpaid (above the regression line), or paid equitably.
In the social and natural sciences multiple regression procedures are very widely used in research. In general, multiple regression allows the researcher to ask (and hopefully answer) the general question “what is the best predictor of …”. For example, educational researchers might want to learn what are the best predictors of success in highschool. Psychologists may want to determine which personality variable best predicts social adjustment. Sociologists may want to find out which of the multiple social indicators best predict whether or not a new immigrant group will adapt and be absorbed into society.
Computations for Solving the Multiple Regression Equation
A onedimensional surface in a twodimensional or twovariable space is a line defined by the equation Y = b_{0} + b_{1}X. According to this equation, the Y variable can be expressed in terms of or as a function of a constant (b_{0}) and a slope (b_{1}) times the X variable. The constant is also referred to as the intercept, and the slope as the regression coefficient. For example, GPA may best be predicted as 1+.02*IQ. Thus, knowing that a student has an IQ of 130 would lead us to predict that her GPA would be 3.6 (since, 1+.02*130=3.6). In the multiple regression case, when there are multiple predictor variables, the regression surface usually cannot be visualized in a two dimensional space, but the computations are a straightforward extension of the computations in the single predictor case. For example, if in addition to IQ we had additional predictors of achievement (e.g., Motivation, Selfdiscipline) we could construct a linear equation containing all those variables. In general then, multiple regression procedures will estimate a linear equation of the form:
Y = b_{0} + b_{1}X_{1} + b_{2}X_{2} + … + b_{k}X_{k}
where k is the number of predictors. Note that in this equation, the regression coefficients (or b_{1} … b_{k} coefficients) represent the independent contributions of each in dependent variable to the prediction of the dependent variable. Another way to express this fact is to say that, for example, variable X_{1} is correlated with the Y variable, after controlling for all other independent variables. This type of correlation is also referred to as a partial correlation (this term was first used by Yule, 1907). Perhaps the following example will clarify this issue. We would probably find a significant negative correlation between hair length and height in the population (i.e., short people have longer hair). At first this may seem odd; however, if we were to add the variable Gender into the multiple regression equation, this correlation would probably disappear. This is because women, on the average, have longer hair than men; they also are shorter on the average than men. Thus, after we remove this gender difference by entering Gender into the equation, the relationship between hair length and height disappears because hair length does not make any unique contribution to the prediction of height, above and beyond what it shares in the prediction with variable Gender. Put another way, after controlling for the variable Gender, the partial correlation between hair length and height is zero.
The regression surface (a line in simple regression, a plane or higherdimensional surface in multiple regression) expresses the best prediction of the dependent variable (Y), given the independent variables (X‘s). However, nature is rarely (if ever) perfectly predictable, and usually there is substantial variation of the observed points from the fitted regression surface. The deviation of a particular point from the nearest corresponding point on the predicted regression surface (its predicted value) is called the residual value. Since the goal of linear regression procedures is to fit a surface, which is a linear function of the X variables, as closely as possible to the observed Y variable, the residual values for the observed points can be used to devise a criterion for the “best fit.” Specifically, in regression problems the surface is computed for which the sum of the squared deviations of the observed points from that surface are minimized. Thus, this general procedure is sometimes also referred to as least squares estimation. (see also the description of weighted least squares estimation).
The actual computations involved in solving regression problems can be expressed compactly and conveniently using matrix notation. Suppose that there are n observed values of Y and n associated observed values for each of k different X variables. Then Y_{i}, X_{ik}, and e_{i }can represent the ith observation of the Y variable, the ith observation of each of the X variables, and the ith unknown residual value, respectively. Collecting these terms into matrices we have
The multiple regression model in matrix notation then can be expressed as
Y = Xb + e
where b is a column vector of 1 (for the intercept) + k unknown regression coefficients. Recall that the goal of multiple regression is to minimize the sum of the squared residuals. Regression coefficients that satisfy this criterion are found by solving the set of normal equations
X’Xb = X’Y
When the X variables are linearly independent (i.e., they are nonredundant, yielding an X’X matrix which is of full rank) there is a unique solution to the normal equations. Premultiplying both sides of the matrix formula for the normal equations by the inverse of X’X gives
(X’X)^{1}X’Xb = (X’X)^{1}X’Y
or
b = (X’X)^{1}X’Y
This last result is very satisfying in view of its simplicity and its generality. With regard to its simplicity, it expresses the solution for the regression equation in terms just 2 matrices (X and Y) and 3 basic matrix operations, (1) matrix transposition, which involves interchanging the elements in the rows and columns of a matrix, (2) matrix multiplication, which involves finding the sum of the products of the elements for each row and column combination of two conformable (i.e., multipliable) matrices, and (3) matrix inversion, which involves finding the matrix equivalent of a numeric reciprocal, that is, the matrix that satisfies
A^{1}AA=A
for a matrix A.
It took literally centuries for the ablest mathematicians and statisticians to find a satisfactory method for solving the linear least square regression problem. But their efforts have paid off, for it is hard to imagine a simpler solution.
With regard to the generality of the multiple regression model, its only notable limitations are that (1) it can be used to analyze only a single dependent variable, (2) it cannot provide a solution for the regression coefficients when the X variables are not linearly independent and the inverse of X’X therefore does not exist. These restrictions, however, can be overcome, and in doing so the multiple regression model is transformed into the general linear model.
Extension of Multiple Regression to the General Linear Model
One way in which the general linear model differs from the multiple regression model is in terms of the number of dependent variables that can be analyzed. The Y vector of n observations of a single Y variable can be replaced by a Y matrix of n observations of m different Y variables. Similarly, the b vector of regression coefficients for a single Y variable can be replaced by a b matrix of regression coefficients, with one vector of b coefficients for each of the m dependent variables. These substitutions yield what is sometimes called the multivariate regression model, but it should be emphasized that the matrix formulations of the multiple and multivariate regression models are identical, except for the number of columns in the Y and b matrices. The method for solving for the b coefficients is also identical, that is, m different sets of regression coefficients are separately found for the m different dependent variables in the multivariate regression model.
The general linear model goes a step beyond the multivariate regression model by allowing for linear transformations or linear combinations of multiple dependent variables. This extension gives the general linear model important advantages over the multiple and the socalled multivariate regression models, both of which are inherently univariate (single dependent variable) methods. One advantage is that multivariate tests of significance can be employed when responses on multiple dependent variables are correlated. Separate univariate tests of significance for correlated dependent variables are not independent and may not be appropriate. Multivariate tests of significance of independent linear combinations of multiple dependent variables also can give insight into which dimensions of the response variables are, and are not, related to the predictor variables. Another advantage is the ability to analyze effects of repeated measure factors. Repeated measure designs, or withinsubject designs, have traditionally been analyzed using ANOVA techniques. Linear combinations of responses reflecting a repeated measure effect (for example, the difference of responses on a measure under differing conditions) can be constructed and tested for significance using either the univariate or multivariate approach to analyzing repeated measures in the general linear model.
A second important way in which the general linear model differs from the multiple regression model is in its ability to provide a solution for the normal equations when the X variables are not linearly independent and the inverse of X’X does not exist. Redundancy of the X variables may be incidental (e.g., two predictor variables might happen to be perfectly correlated in a small data set), accidental (e.g., two copies of the same variable might unintentionally be used in an analysis) or designed (e.g., indicator variables with exactly opposite values might be used in the analysis, as when both Male and Female predictor variables are used in representing Gender). Finding the regular inverse of a nonfullrank matrix is reminiscent of the problem of finding the reciprocal of 0 in ordinary arithmetic. No such inverse or reciprocal exists because division by 0 is not permitted. This problem is solved in the general linear model by using a generalized inverse of the X’X matrix in solving the normal equations. A generalized inverse is any matrix that satisfies
AA^{–}A = A
for a matrix A. A generalized inverse is unique and is the same as the regular inverse only if the matrix A is full rank. A generalized inverse for a nonfullrank matrix can be computed by the simple expedient of zeroing the elements in redundant rows and columns of the matrix. Suppose that an X’X matrix with r nonredundant columns is partitioned as
where A_{11} is an r by r matrix of rank r. Then the regular inverse of A_{11} exists and a generalized inverse of X’X is
where each 0 (null) matrix is a matrix of 0’s (zeroes) and has the same dimensions as the corresponding A matrix.
In practice, however, a particular generalized inverse of X’X for finding a solution to the normal equations is usually computed using the sweep operator (Dempster, 1960). This generalized inverse, called a g2 inverse, has two important properties. One is that zeroing of the elements in redundant rows is unnecessary. Another is that partitioning or reordering of the columns of X’X is unnecessary, so that the matrix can be inverted “in place.”
There are infinitely many generalized inverses of a nonfullrank X’X matrix, and thus, infinitely many solutions to the normal equations. This can make it difficult to understand the nature of the relationships of the predictor variables to responses on the dependent variables, because the regression coefficients can change depending on the particular generalized inverse chosen for solving the normal equations. It is not cause for dismay, however, because of the invariance properties of many results obtained using the general linear model.
A simple example may be useful for illustrating one of the most important invariance properties of the use of generalized inverses in the general linear model. If both Male and Female predictor variables with exactly opposite values are used in an analysis to represent Gender, it is essentially arbitrary as to which predictor variable is considered to be redundant (e.g., Male can be considered to be redundant with Female, or vice versa). No matter which predictor variable is considered to be redundant, no matter which corresponding generalized inverse is used in solving the normal equations, and no matter which resulting regression equation is used for computing predicted values on the dependent variables, the predicted values and the corresponding residuals for males and females will be unchanged. In using the general linear model, we must keep in mind that finding a particular arbitrary solution to the normal equations is primarily a means to the end of accounting for responses on the dependent variables, and not necessarily an end in itself.
SigmaRestricted and Overparameterized Model
Unlike the multiple regression model, which is usually applied to cases where the X variables are continuous, the general linear model is frequently applied to analyze any ANOVA or MANOVA design with categorical predictor variables, any ANCOVA or MANCOVA design with both categorical and continuous predictor variables, as well as any multiple or multivariate regression design with continuous predictor variables. To illustrate, Gender is clearly a nominal level variable (anyone who attempts to rank order the sexes on any dimension does so at his or her own peril in today’s world). There are two basic methods by which Gender can be coded into one or more (nonoffensive) predictor variables, and analyzed using the general linear model.
Sigmarestricted model (coding of categorical predictors). Using the first method, males and females can be assigned any two arbitrary, but distinct values on a single predictor variable. The values on the resulting predictor variable will represent a quantitative contrast between males and females. Typically, the values corresponding to group membership are chosen not arbitrarily but rather to facilitate interpretation of the regression coefficient associated with the predictor variable. In one widely used strategy, cases in the two groups are assigned values of 1 and 1 on the predictor variable, so that if the regression coefficient for the variable is positive, the group coded as 1 on the predictor variable will have a higher predicted value (i.e., a higher group mean) on the dependent variable, and if the regression coefficient is negative, the group coded as 1 on the predictor variable will have a higher predicted value on the dependent variable. An additional advantage is that since each group is coded with a value one unit from zero, this helps in interpreting the magnitude of differences in predicted values between groups, because regression coefficients reflect the units of change in the dependent variable for each unit change in the predictor variable. This coding strategy is aptly called the sigmarestricted parameterization, because the values used to represent group membership (1 and 1) sum to zero.
Note that the sigmarestricted parameterization of categorical predictor variables usually leads to X’X matrices which do not require a generalized inverse for solving the normal equations. Potentially redundant information, such as the characteristics of maleness and femaleness, is literally reduced to fullrank by creating quantitative contrast variables representing differences in characteristics.
Overparameterized model (coding of categorical predictors). The second basic method for recoding categorical predictors is the indicator variable approach. In this method a separate predictor variable is coded for each group identified by a categorical predictor variable. To illustrate, females might be assigned a value of 1 and males a value of 0 on a first predictor variable identifying membership in the female Gender group, and males would then be assigned a value of 1 and females a value of 0 on a second predictor variable identifying membership in the male Gender group. Note that this method of recoding categorical predictor variables will almost always lead to X’X matrices with redundant columns, and thus require a generalized inverse for solving the normal equations. As such, this method is often called the overparameterized model for representing categorical predictor variables, because it results in more columns in the X’X than are necessary for determining the relationships of categorical predictor variables to responses on the dependent variables.
True to its description as general, the general linear model can be used to perform analyses with categorical predictor variables which are coded using either of the two basic methods that have been described.
Summary of Computations
To conclude this discussion of the ways in which the general linear model extends and generalizes regression methods, the general linear model can be expressed as
YM = Xb + e
Here Y, X, b, and e are as described for the multivariate regression model and M is an m x s matrix of coefficients defining s linear transformation of the dependent variables. The normal equations are
X’Xb = X’YM
and a solution for the normal equations is given by
b = (X’X)^{–}X’YM Here the inverse of X’X is a generalized inverse if X’X contains redundant columns.
Add a provision for analyzing linear combinations of multiple dependent variables, add a method for dealing with redundant predictor variables and recoded categorical predictor variables, and the major limitations of multiple regression are overcome by the general linear model.
Types of Analyses
A wide variety of types of designs can be analyzed using the general linear model. In fact, the flexibility of the general linear model allows it to handle so many different types of designs that it is difficult to develop simple typologies of the ways in which these designs might differ. Some general ways in which designs might differ can be suggested, but keep in mind that any particular design can be a “hybrid” in the sense that it could have combinations of features of a number of different types of designs.
In the following discussion, references will be made to the design matrix X, as well as sigmarestricted and overparameterized model coding. For an explanation of this terminology, refer to the section entitled Basic Ideas: The General Linear Model, or, for a brief summary, to the Summary of computations section.
A basic discussion to univariate and multivariate ANOVA techniques can also be found in the ANOVA/MANOVA topic; a discussion of multiple regression methods is also provided in the Multiple Regression topic.
BetweenSubject Designs
 Overview
 Oneway ANOVA
 Main effect ANOVA
 Factorial ANOVA
 Nested designs
 Balanced ANOVA
 Simple regression
 Multiple regression
 Factorial regression
 Polynomial regression
 Response surface regression
 Mixture surface regression
 Analysis of covariance (ANCOVA)
 Separate slopes designs
 Homogeneity of slopes
 Mixedmodel ANOVA and ANCOVA
Overview. The levels or values of the predictor variables in an analysis describe the differences between the n subjects or the n valid cases that are analyzed. Thus, when we speak of the between subject design (or simply the between design) for an analysis, we are referring to the nature, number, and arrangement of the predictor variables.
Concerning the nature or type of predictor variables, between designs which contain only categorical predictor variables can be called ANOVA (analysis of variance) designs, between designs which contain only continuous predictor variables can be called regression designs, and between designs which contain both categorical and continuous predictor variables can be called ANCOVA (analysis of covariance) designs. Further, continuous predictors are always considered to have fixed values, but the levels of categorical predictors can be considered to be fixed or to vary randomly. Designs which contain random categorical factors are called mixedmodel designs (see the Variance Components and Mixed Model ANOVA/ANCOVA section).
Between designs may involve only a single predictor variable and therefore be described as simple (e.g., simple regression) or may employ numerous predictor variables (e.g., multiple regression).
Concerning the arrangement of predictor variables, some between designs employ only “main effect” or firstorder terms for predictors, that is, the values for different predictor variables are independent and raised only to the first power. Other between designs may employ higherorder terms for predictors by raising the values for the original predictor variables to a power greater than 1 (e.g., in polynomial regression designs), or by forming products of different predictor variables (i.e., interaction terms). A common arrangement for ANOVA designs is the fullfactorial design, in which every combination of levels for each of the categorical predictor variables is represented in the design. Designs with some but not all combinations of levels for each of the categorical predictor variables are aptly called fractional factorial designs. Designs with a hierarchy of combinations of levels for the different categorical predictor variables are called nested designs.
These basic distinctions about the nature, number, and arrangement of predictor variables can be used in describing a variety of different types of between designs. Some of the more common between designs can now be described.
OneWay ANOVA. A design with a single categorical predictor variable is called a oneway ANOVA design. For example, a study of 4 different fertilizers used on different individual plants could be analyzed via oneway ANOVA, with four levels for the factor Fertilizer.
In genera, consider a single categorical predictor variable A with 1 case in each of its 3 categories. Using the sigmarestricted coding of A into 2 quantitative contrast variables, the matrix X defining the between design is
That is, cases in groups A_{1}, A_{2}, and A_{3} are all assigned values of 1 on X_{0} (the intercept), the case in group A_{1} is assigned a value of 1 on X_{1} and a value 0 on X_{2}, the case in group A_{2} is assigned a value of 0 on X_{1} and a value 1 on X_{2}, and the case in group A_{3} is assigned a value of 1 on X_{1} and a value 1 on X_{2}. Of course, any additional cases in any of the 3 groups would be coded similarly. If there were 1 case in group A_{1}, 2 cases in group A_{2}, and 1 case in group A_{3}, the X matrix would be
where the first subscript for A gives the replicate number for the cases in each group. For brevity, replicates usually are not shown when describing ANOVA design matrices.
Note that in oneway designs with an equal number of cases in each group, sigmarestricted coding yields X_{1} … X_{k} variables all of which have means of 0.
Using the overparameterized model to represent A, the X matrix defining the between design is simply
These simple examples show that the X matrix actually serves two purposes. It specifies (1) the coding for the levels of the original predictor variables on the X variables used in the analysis as well as (2) the nature, number, and arrangement of the X variables, that is, the between design.
Main Effect ANOVA. Main effect ANOVA designs contain separate oneway ANOVA designs for 2 or more categorical predictors. A good example of main effect ANOVA would be the typical analysis performed on screening designs as described in the context of the Experimental Design section.
Consider 2 categorical predictor variables A and B each with 2 categories. Using the sigmarestricted coding, the X matrix defining the between design is
Note that if there are equal numbers of cases in each group, the sum of the crossproducts of values for the X_{1} and X_{2} columns is 0, for example, with 1 case in each group (1*1)+(1*1)+(1*1)+(1*1)=0. Using the overparameterized model, the matrix X defining the between design is
Comparing the two types of coding, it can be seen that the overparameterized coding takes almost twice as many values as the sigmarestricted coding to convey the same information.
Factorial ANOVA. Factorial ANOVA designs contain X variables representing combinations of the levels of 2 or more categorical predictors (e.g., a study of boys and girls in four age groups, resulting in a 2 (Gender) x 4 (Age Group) design). In particular, fullfactorial designs represent all possible combinations of the levels of the categorical predictors. A fullfactorial design with 2 categorical predictor variables A and B each with 2 levels each would be called a 2 x 2 fullfactorial design. Using the sigmarestricted coding, the X matrix for this design would be
Several features of this X matrix deserve comment. Note that the X_{1} and X_{2} columns represent main effect contrasts for one variable, (i.e., A and B, respectively) collapsing across the levels of the other variable. The X_{3 }column instead represents a contrast between different combinations of the levels of A and B. Note also that the values for X_{3} are products of the corresponding values for X_{1} and X_{2}. Product variables such as X_{3 }represent the multiplicative or interaction effects of their factors, so X_{3} would be said to represent the 2way interaction of A and B. The relationship of such product variables to the dependent variables indicate the interactive influences of the factors on responses above and beyond their independent (i.e., main effect) influences on responses. Thus, factorial designs provide more information about the relationships between categorical predictor variables and responses on the dependent variables than is provided by corresponding oneway or main effect designs.
When many factors are being investigated, however, fullfactorial designs sometimes require more data than reasonably can be collected to represent all possible combinations of levels of the factors, and highorder interactions between many factors can become difficult to interpret. With many factors, a useful alternative to the fullfactorial design is the fractional factorial design. As an example, consider a 2 x 2 x 2 fractional factorial design to degree 2 with 3 categorical predictor variables each with 2 levels. The design would include the main effects for each variable, and all 2way interactions between the three variables, but would not include the 3way interaction between all three variables. Using the overparameterized model, the X matrix for this design is
The 2way interactions are the highest degree effects included in the design. These types of designs are discussed in detail the 2**(kp) Fractional Factorial Designs section of the Experimental Design topic.
Nested ANOVA Designs. Nested designs are similar to fractional factorial designs in that all possible combinations of the levels of the categorical predictor variables are not represented in the design. In nested designs, however, the omitted effects are lowerorder effects. Nested effects are effects in which the nested variables never appear as main effects. Suppose that for 2 variables A and B with 3 and 2 levels, respectively, the design includes the main effect for A and the effect of B nested within the levels of A. The X matrix for this design using the overparameterized model is
Note that if the sigmarestricted coding were used, there would be only 2 columns in the X matrix for the B nested within A effect instead of the 6 columns in the X matrix for this effect when the overparameterized model coding is used (i.e., columns X_{4} through X_{9}). The sigmarestricted coding method is overlyrestrictive for nested designs, so only the overparameterized model is used to represent nested designs.
Balanced ANOVA. Most of the between designs discussed in this section can be analyzed much more efficiently, when they are balanced, i.e., when all cells in the ANOVA design have equal n, when there are no missing cells in the design, and, if nesting is present, when the nesting is balanced so that equal numbers of levels of the factors that are nested appear in the levels of the factor(s) that they are nested in. In that case, the X’X matrix (where X stands for the design matrix) is a diagonal matrix, and many of the computations necessary to compute the ANOVA results (such as matrix inversion) are greatly simplified.
Simple Regression. Simple regression designs involve a single continuous predictor variable. If there were 3 cases with values on a predictor variable P of, say, 7, 4, and 9, and the design is for the firstorder effect of P, the X matrix would be
and using P for X_{1} the regression equation would be
Y = b_{0} + b_{1}P
If the simple regression design is for a higherorder effect of P, say the quadratic effect, the values in the X_{1} column of the design matrix would be raised to the 2nd power, that is, squared
and using P^{2} for X_{1} the regression equation would be
Y = b_{0} + b_{1}P^{2}
The sigmarestricted and overparameterized coding methods do not apply to simple regression designs and any other design containing only continuous predictors (since there are no categorical predictors to code). Regardless of which coding method is chosen, values on the continuous predictor variables are raised to the desired power and used as the values for the X variables. No recoding is performed. It is therefore sufficient, in describing regression designs, to simply describe the regression equation without explicitly describing the design matrix X.
Multiple Regression. Multiple regression designs are to continuous predictor variables as main effect ANOVA designs are to categorical predictor variables, that is, multiple regression designs contain the separate simple regression designs for 2 or more continuous predictor variables. The regression equation for a multiple regression design for the firstorder effects of 3 continuous predictor variables P, Q, and R would be
Y = b_{0} + b_{1}P + b_{2}Q + b_{3}R
Factorial Regression. Factorial regression designs are similar to factorial ANOVA designs, in which combinations of the levels of the factors are represented in the design. In factorial regression designs, however, there may be many more such possible combinations of distinct levels for the continuous predictor variables than there are cases in the data set. To simplify matters, fullfactorial regression designs are defined as designs in which all possible products of the continuous predictor variables are represented in the design. For example, the fullfactorial regression design for two continuous predictor variables P and Q would include the main effects (i.e., the firstorder effects) of P and Q and their 2way P by Q interaction effect, which is represented by the product of P and Q scores for each case. The regression equation would be
Y = b_{0} + b_{1}P + b_{2}Q + b_{3}P*Q
Factorial regression designs can also be fractional, that is, higherorder effects can be omitted from the design. A fractional factorial design to degree 2 for 3 continuous predictor variables P, Q, and R would include the main effects and all 2way interactions between the predictor variables
Y = b_{0} + b_{1}P + b_{2}Q + b_{3}R + b_{4}P*Q + b_{5}P*R + b_{6}Q*R
Polynomial Regression. Polynomial regression designs are designs which contain main effects and higherorder effects for the continuous predictor variables but do not include interaction effects between predictor variables. For example, the polynomial regression design to degree 2 for three continuous predictor variables P, Q, and R would include the main effects (i.e., the firstorder effects) of P, Q, and R and their quadratic (i.e., secondorder) effects, but not the 2way interaction effects or the P by Q by R 3way interaction effect.
Y = b_{0} + b_{1}P + b_{2}P^{2} + b_{3}Q + b_{4}Q^{2} + b_{5}R + b_{6}R^{2}
Polynomial regression designs do not have to contain all effects up to the same degree for every predictor variable. For example, main, quadratic, and cubic effects could be included in the design for some predictor variables, and effects up the fourth degree could be included in the design for other predictor variables.
Response Surface Regression. Quadratic response surface regression designs are a hybrid type of design with characteristics of both polynomial regression designs and fractional factorial regression designs. Quadratic response surface regression designs contain all the same effects of polynomial regression designs to degree 2 and additionally the 2way interaction effects of the predictor variables. The regression equation for a quadratic response surface regression design for 3 continuous predictor variables P, Q, and R would be
Y = b_{0} + b_{1}P + b_{2}P^{2} + b_{3}Q + b_{4}Q^{2} + b_{5}R + b_{6}R^{2} + b_{7}P*Q + b_{8}P*R + b_{9}Q*R
These types of designs are commonly employed in applied research (e.g., in industrial experimentation), and a detailed discussion of these types of designs is also presented in the Experimental Design topic (see Central composite designs).
Mixture Surface Regression. Mixture surface regression designs are identical to factorial regression designs to degree 2 except for the omission of the intercept. Mixtures, as the name implies, add up to a constant value; the sum of the proportions of ingredients in different recipes for some material all must add up 100%. Thus, the proportion of one ingredient in a material is redundant with the remaining ingredients. Mixture surface regression designs deal with this redundancy by omitting the intercept from the design. The design matrix for a mixture surface regression design for 3 continuous predictor variables P, Q, and R would be
Y = b_{1}P + b_{2}Q + b_{3}R + b_{4}P*Q + b_{5}P*R + b_{6}Q*R
These types of designs are commonly employed in applied research (e.g., in industrial experimentation), and a detailed discussion of these types of designs is also presented in the Experimental Design topic (see Mixture designs and triangular surfaces).
Analysis of Covariance. In general, between designs which contain both categorical and continuous predictor variables can be called ANCOVA designs. Traditionally, however, ANCOVA designs have referred more specifically to designs in which the firstorder effects of one or more continuous predictor variables are taken into account when assessing the effects of one or more categorical predictor variables. A basic introduction to analysis of covariance can also be found in the Analysis of covariance (ANCOVA) section of the ANOVA/MANOVA topic.
To illustrate, suppose a researcher wants to assess the influences of a categorical predictor variable A with 3 levels on some outcome, and that measurements on a continuous predictor variable P, known to covary with the outcome, are available. If the data for the analysis are
then the sigmarestricted X matrix for the design that includes the separate firstorder effects of P and A would be
The b_{2} and b_{3} coefficients in the regression equation
Y = b_{0} + b_{1}X_{1} + b_{2}X_{2} + b_{3}X_{3}
represent the influences of group membership on the A categorical predictor variable, controlling for the influence of scores on the P continuous predictor variable. Similarly, the b_{1} coefficient represents the influence of scores on P controlling for the influences of group membership on A. This traditional ANCOVA analysis gives a more sensitive test of the influence of A to the extent that P reduces the prediction error, that is, the residuals for the outcome variable.
The X matrix for the same design using the overparameterized model would be
The interpretation is unchanged except that the influences of group membership on the A categorical predictor variables are represented by the b_{2}, b_{3} and b_{4} coefficients in the regression equation
Y = b_{0} + b_{1}X_{1} + b_{2}X_{2} + b_{3}X_{3} + b_{4}X_{4}
Separate Slope Designs. The traditional analysis of covariance (ANCOVA) design for categorical and continuous predictor variables is inappropriate when the categorical and continuous predictors interact in influencing responses on the outcome. The appropriate design for modeling the influences of the predictors in this situation is called the separate slope design. For the same example data used to illustrate traditional ANCOVA, the overparameterized X matrix for the design that includes the main effect of the threelevel categorical predictor A and the 2way interaction of P by A would be
The b_{4}, b_{5}, and b_{6 coefficients in the regression equation }
Y = b_{0} + b_{1}X_{1} + b_{2}X_{2} + b_{3}X_{3} + b_{4}X_{4} + b_{5}X_{5} + b_{6}X_{6}
give the separate slopes for the regression of the outcome on P within each group on A, controlling for the main effect of A.
As with nested ANOVA designs, the sigmarestricted coding of effects for separate slope designs is overly restrictive, so only the overparameterized model is used to represent separate slope designs. In fact, separate slope designs are identical in form to nested ANOVA designs, since the main effects for continuous predictors are omitted in separate slope designs.
Homogeneity of Slopes. The appropriate design for modeling the influences of continuous and categorical predictor variables depends on whether the continuous and categorical predictors interact in influencing the outcome. The traditional analysis of covariance (ANCOVA) design for continuous and categorical predictor variables is appropriate when the continuous and categorical predictors do not interact in influencing responses on the outcome, and the separate slope design is appropriate when the continuous and categorical predictors do interact in influencing responses. The homogeneity of slopes designs can be used to test whether the continuous and categorical predictors interact in influencing responses, and thus, whether the traditional ANCOVA design or the separate slope design is appropriate for modeling the effects of the predictors. For the same example data used to illustrate the traditional ANCOVA and separate slope designs, the overparameterized X matrix for the design that includes the main effect of P, the main effect of the threelevel categorical predictor A, and the 2way interaction of P by A would be
If the b_{5}, b_{6}, or b_{7} coefficient in the regression equation
Y = b_{0} + b_{1}X_{1} + b_{2}X_{2} + b_{3}X_{3} + b_{4}X_{4} + b_{5}X_{5} + b_{6}X_{6} + b_{7}X_{7}
is nonzero, the separate slope model should be used. If instead all 3 of these regression coefficients are zero the traditional ANCOVA design should be used.
The sigmarestricted X matrix for the homogeneity of slopes design would be
Using this X matrix, if the b_{4}, or b_{5} coefficient in the regression equation
Y = b_{0} + b_{1}X_{1} + b_{2}X_{2} + b_{3}X_{3} + b_{4}X_{4} + b_{5}X_{5}
is nonzero, the separate slope model should be used. If instead both of these regression coefficients are zero the traditional ANCOVA design should be used.
Mixed Model ANOVA and ANCOVA. Designs that contain random effects for one or more categorical predictor variables are called mixedmodel designs. Random effects are classification effects where the levels of the effects are assumed to be randomly selected from an infinite population of possible levels. The solution for the normal equations in mixedmodel designs is identical to the solution for fixedeffect designs (i.e., designs which do not contain Random effects. Mixedmodel designs differ from fixedeffect designs only in the way in which effects are tested for significance. In fixedeffect designs, between effects are always tested using the mean squared residual as the error term. In mixedmodel designs, between effects are tested using relevant error terms based on the covariation of random sources of variation in the design. Specifically, this is done using Satterthwaite’s method of denominator synthesis (Satterthwaite, 1946), which finds the linear combinations of sources of random variation that serve as appropriate error terms for testing the significance of the respective effect of interest. A basic discussion of these types of designs, and methods for estimating variance components for the random effects can also be found in the Variance Components and Mixed Model ANOVA/ANCOVA topic.
Mixedmodel designs, like nested designs and separate slope designs, are designs in which the sigmarestricted coding of categorical predictors is overly restrictive. Mixedmodel designs require estimation of the covariation between the levels of categorical predictor variables, and the sigmarestricted coding of categorical predictors suppresses this covariation. Thus, only the overparameterized model is used to represent mixedmodel designs (some programs will use the sigmarestricted approach and a socalled “restricted model” for random effects; however, only the overparameterized model as described in General Linear Models applies to both balanced and unbalanced designs, as well as designs with missing cells; see Searle, Casella, & McCullock, 1992, p. 127). It is important to recognize, however, that sigmarestricted coding can be used to represent any between design, with the exceptions of mixedmodel, nested, and separate slope designs. Furthermore, some types of hypotheses can only be tested using the sigmarestricted coding (i.e., the effective hypothesis, Hocking, 1996), thus the greater generality of the overparameterized model for representing between designs does not justify it being used exclusively for representing categorical predictors in the general linear model.
WithinSubject (Repeated Measures) Designs
 Overview
 Oneway withinsubject designs
 Multiway withinsubject designs
 The multivariate approach to Repeated Measures
 Doubly multivariate withinsubject designs
Overview. It is quite common for researchers to administer the same test to the same subjects repeatedly over a period of time or under varying circumstances. In essence, we are interested in examining differences within each subject, for example, subjects’ improvement over time. Such designs are referred to as withinsubject designs or repeated measures designs. A basic introduction to repeated measures designs is also provided in the Betweengroups and repeated measures section of the ANOVA/MANOVA topic.
For example, imagine that we want to monitor the improvement of students’ algebra skills over two months of instruction. A standardized algebra test is administered after one month (level 1 of the repeated measures factor), and a comparable test is administered after two months (level 2 of the repeated measures factor). Thus, the repeated measures factor (Time) has 2 levels. Now, suppose that scores for the 2 algebra tests (i.e., values on the Y_{1} and Y_{2} variables at Time 1 and Time 2, respectively) are transformed into scores on a new composite variable (i.e., values on the T_{1}), using the linear transformation
T = YM
where M is an orthonormal contrast matrix. Specifically, if
then the difference of the mean score on T_{1} from 0 indicates the improvement (or deterioration) of scores across the 2 levels of Time.
OneWay WithinSubject Designs. The example algebra skills study with the Time repeated measures factor (see also withinsubjects design Overview) illustrates a oneway withinsubject design. In such designs, orthonormal contrast transformations of the scores on the original dependent Y variables are performed via the M transformation (orthonormal transformations correspond to orthogonal rotations of the original variable axes). If any b_{0} coefficient in the regression of a transformed T variable on the intercept is nonzero, this indicates a change in responses across the levels of the repeated measures factor, that is, the presence of a main effect for the repeated measure factor on responses.
What if the between design includes effects other than the intercept? If any of the b_{1} through b_{k} coefficients in the regression of a transformed T variable on X are nonzero, this indicates a different change in responses across the levels of the repeated measures factor for different levels of the corresponding between effect, i.e., the presence of a within by between interaction effect on responses.
The same betweensubject effects that can be tested in designs with no repeatedmeasures factors can also be tested in designs that do include repeatedmeasures factors. This is accomplished by creating a transformed dependent variable which is the sum of the original dependent variables divided by the square root of the number of original dependent variables. The same tests of betweensubject effects that are performed in designs with no repeatedmeasures factors (including tests of the between intercept) are performed on this transformed dependent variable.
MultiWay WithinSubject Designs. Suppose that in the example algebra skills study with the Time repeated measures factor (see the withinsubject designs Overview), students were given a number problem test and then a word problem test on each testing occasion. Test could then be considered as a second repeated measures factor, with scores on the number problem tests representing responses at level 1 of the Test repeated measure factor, and scores on the word problem tests representing responses at level 2 of the Test repeated measure factor. The within subject design for the study would be a 2 (Time) by 2 (Test) fullfactorial design, with effects for Time, Test, and the Time by Test interaction.
To construct transformed dependent variables representing the effects of Time, Test, and the Time by Test interaction, three respective M transformations of the original dependent Y variables are performed. Assuming that the original Y variables are in the order Time 1 – Test 1, Time 1 – Test 2, Time 2 – Test 1, and Time 2 – Test 2, the M matrices for the Time, Test, and the Time by Test interaction would be
The differences of the mean scores on the transformed T variables from 0 are then used to interpret the corresponding withinsubject effects. If the b_{0} coefficient in the regression of a transformed T variable on the intercept is nonzero, this indicates a change in responses across the levels of a repeated measures effect, that is, the presence of the corresponding main or interaction effect for the repeated measure factors on responses.
Interpretation of within by between interaction effects follow the same procedures as for oneway within designs, except that now within by between interactions are examined for each within effect by between effect combination.
Multivariate Approach to Repeated Measures. When the repeated measures factor has more than 2 levels, then the M matrix will have more than a single column. For example, for a repeated measures factor with 3 levels (e.g., Time 1, Time 2, Time 3), the M matrix will have 2 columns (e.g., the two transformations of the dependent variables could be (1) Time 1 vs. Time 2 and Time 3 combined, and (2) Time 2 vs. Time 3). Consequently, the nature of the design is really multivariate, that is, there are two simultaneous dependent variables, which are transformations of the original dependent variables. Therefore, when testing repeated measures effects involving more than a single degree of freedom (e.g., a repeated measures main effect with more than 2 levels), you can compute multivariate test statistics to test the respective hypotheses. This is a different (and usually the preferred) approach than the univariate method that is still widely used. For a further discussion of the multivariate approach to testing repeated measures effects, and a comparison to the traditional univariate approach, see the Sphericity and compound symmetry section of the ANOVA/MANOVA topic.
Doubly Multivariate Designs. If the product of the number of levels for each withinsubject factor is equal to the number of original dependent variables, the withinsubject design is called a univariate repeated measures design. The within design is univariate because there is one dependent variable representing each combination of levels of the withinsubject factors. Note that this use of the term univariate design is not to be confused with the univariate and multivariate approach to the analysis of repeated measures designs, both of which can be used to analyze such univariate (singledependentvariableonly) designs. When there are two or more dependent variables for each combination of levels of the withinsubject factors, the withinsubject design is called a multivariate repeated measures design, or more commonly, a doubly multivariate withinsubject design. This term is used because the analysis for each dependent measure can be done via the multivariate approach; so when there is more than one dependent measure, the design can be considered doublymultivariate.
Doubly multivariate design are analyzed using a combination of univariate repeated measures and multivariate analysis techniques. To illustrate, suppose in an algebra skills study, tests are administered three times (repeated measures factor Time with 3 levels). Two test scores are recorded at each level of Time: a Number Problem score and a Word Problem score. Thus, scores on the two types of tests could be treated as multiple measures on which improvement (or deterioration) across Time could be assessed. M transformed variables could be computed for each set of test measures, and multivariate tests of significance could be performed on the multiple transformed measures, as well as on the each individual test measure.
Multivariate Designs
Overview. When there are multiple dependent variables in a design, the design is said to be multivariate. Multivariate measures of association are by nature more complex than their univariate counterparts (such as the correlation coefficient, for example). This is because multivariate measures of association must take into account not only the relationships of the predictor variables with responses on the dependent variables, but also the relationships among the multiple dependent variables. By doing so, however, these measures of association provide information about the strength of the relationships between predictor and dependent variables independent of the dependent variable interrelationships. A basic discussion of multivariate designs is also presented in the Multivariate Designs section in the ANOVA/MANOVA topic.
The most commonly used multivariate measures of association all can be expressed as functions of the eigenvalues of the product matrix
E^{1}H
where E is the error SSCP matrix (i.e., the matrix of sums of squares and crossproducts for the dependent variables that are not accounted for by the predictors in the between design), and H is a hypothesis SSCP matrix (i.e., the matrix of sums of squares and crossproducts for the dependent variables that are accounted for by all the predictors in the between design, or the sums of squares and crossproducts for the dependent variables that are accounted for by a particular effect). If
l_{i} = the ordered eigenvalues of E^{1}H, if E^{1} exists
then the 4 commonly used multivariate measures of association are
Wilks’ lambda = P[1/(1+l_{i})]
Pillai’s trace = Sl_{i}/(1+l_{i})
HotellingLawley trace = Sl_{i}
Roy’s largest root = l_{1}
These 4 measures have different upper and lower bounds, with Wilks’ lambda perhaps being the most easily interpretable of the 4 measures. Wilks’ lambda can range from 0 to 1, with 1 indicating no relationship of predictors to responses and 0 indicating a perfect relationship of predictors to responses. 1 – Wilks’ lambda can be interpreted as the multivariate counterpart of a univariate Rsquared, that is, it indicates the proportion of generalized variance in the dependent variables that is accounted for by the predictors.
The 4 measures of association are also used to construct multivariate tests of significance. These multivariate tests are covered in detail in a number of sources (e.g., Finn, 1974; Tatsuoka, 1971).
Estimation and Hypothesis Testing
The following sections discuss details concerning hypothesis testing in the context of STATISTICA‘s GLM module, for example, how the test for the overall model fit is computed, the options for computing tests for categorical effects in unbalanced or incomplete designs, how and when customerror terms can be chosen, and the logic of testing customhypotheses in factorial or regression designs.
Whole Model Tests
Partitioning Sums of Squares. A fundamental principle of least squares methods is that variation on a dependent variable can be partitioned, or divided into parts, according to the sources of the variation. Suppose that a dependent variable is regressed on one or more predictor variables, and that for convenience the dependent variable is scaled so that its mean is 0. Then a basic least squares identity is that the total sum of squared values on the dependent variable equals the sum of squared predicted values plus the sum of squared residual values. Stated more generally,
S(y – ybar)^{2} = S(yhat – ybar)^{2} + S(y – yhat)^{2}
where the term on the left is the total sum of squared deviations of the observed values on the dependent variable from the dependent variable mean, and the respective terms on the right are (1) the sum of squared deviations of the predicted values for the dependent variable from the dependent variable mean and (2) the sum of the squared deviations of the observed values on the dependent variable from the predicted values, that is, the sum of the squared residuals. Stated yet another way,
Total SS = Model SS + Error SS
Note that the Total SS is always the same for any particular data set, but that the Model SS and the Error SS depend on the regression equation. Assuming again that the dependent variable is scaled so that its mean is 0, the Model SS and the Error SS can be computed using
Model SS = b’X’Y
Error SS = Y’Y – b’X’Y
Testing the Whole Model. Given the Model SS and the Error SS, we can perform a test that all the regression coefficients for the X variables (b1 through bk) are zero. This test is equivalent to a comparison of the fit of the regression surface defined by the predicted values (computed from the whole model regression equation) to the fit of the regression surface defined solely by the dependent variable mean (computed from the reduced regression equation containing only the intercept). Assuming that X’X is fullrank, the whole model hypothesis mean square
MSH = (Model SS)/k
is an estimate of the variance of the predicted values. The error mean square
s^{2} = MSE = (Error SS)/(nk1)
is an unbiased estimate of the residual or error variance. The test statistic is
F = MSH/MSE
where F has (k, n – k – 1) degrees of freedom.
If X’X is not full rank, r + 1 is substituted for k, where r is the rank or the number of nonredundant columns of X’X.
Note that in the case of nonintercept models, some multiple regression programs will compute the full model test based on the proportion of variance around 0 (zero) accounted for by the predictors; for more information (see Kvålseth, 1985; Okunade, Chang, and Evans, 1993), while other will actually compute both values (i.e., based on the residual variance around 0, and around the respective dependent variable means.
Limitations of Whole Model Tests. For designs such as oneway ANOVA or simple regression designs, the whole model test by itself may be sufficient for testing general hypotheses about whether or not the single predictor variable is related to the outcome. In more complex designs, however, hypotheses about specific X variables or subsets of X variables are usually of interest. For example, you might want to make inferences about whether a subset of regression coefficients are 0, or you might want to test whether subpopulation means corresponding to combinations of specific X variables differ. The whole model test is usually insufficient for such purposes.
A variety of methods have been developed for testing specific hypotheses. Like whole model tests, many of these methods rely on comparisons of the fit of different models (e.g., Type I, Type II, and the effective hypothesis sums of squares). Other methods construct tests of linear combinations of regression coefficients in order to test mean differences (e.g., Type III, Type IV, and Type V sums of squares). For designs that contain only firstorder effects of continuous predictor variables (i.e., multiple regression designs), many of these methods are equivalent (i.e., Type II through Type V sums of squares all test the significance of partial regression coefficients). However, there are important distinctions between the different hypothesis testing techniques for certain types of ANOVA designs (i.e., designs with unequal cell n‘s and/or missing cells).
All methods for testing hypotheses, however, involve the same hypothesis testing strategy employed in whole model tests, that is, the sums of squares attributable to an effect (using a given criterion) is computed, and then the mean square for the effect is tested using an appropriate error term.
When there are categorical predictors in the model, arranged in a factorial ANOVA design, then we are typically interested in the main effects for and interaction effects between the categorical predictors. However, when the design is not balanced (has unequal cell n’s, and consequently, the coded effects for the categorical factors are usually correlated), or when there are missing cells in a full factorial ANOVA design, then there is ambiguity regarding the specific comparisons between the (population, or leastsquares) cell means that constitute the main effects and interactions of interest. These issues are discussed in great detail in Milliken and Johnson (1986), and if you routinely analyze incomplete factorial designs, you should consult their discussion of various problems and approaches to solving them.
In addition to the widely used methods that are commonly labeled Type I, II, III, and IV sums of squares (see Goodnight, 1980), we also offer different methods for testing effects in incomplete designs, that are widely used in other areas (and traditions) of research.
Type V sums of squares. Specifically, we propose the term Type V sums of squares to denote the approach that is widely used in industrial experimentation, to analyze fractional factorial designs; these types of designs are discussed in detail in the 2**(kp) Fractional Factorial Designs section of the Experimental Design topic. In effect, for those effects for which tests are performed all population marginal means (least squares means) are estimable.
Type VI sums of squares. Second, in keeping with the Type i labeling convention, we propose the term Type VI sums of squares to denote the approach that is often used in programs that only implement the sigmarestricted model (which is not well suited for certain types of designs; we offer a choice between the sigmarestricted and overparameterized model models). This approach is identical to what is described as the effective hypothesis method in Hocking (1996).
Contained Effects. The following descriptions will use the term contained effect. An effect E1 (e.g., A * B interaction) is contained in another effect E2 if:
 Both effects involve the same continuous predictor variable (if included in the model; e.g., A * B * X would be contained in A * C * X, where A, B, and C are categorical predictors, and X is a continuous predictor); or
 E2 has more categorical predictors than does E1, and, if E1 includes any categorical predictors, they also appear in E2 (e.g., A * B would be contained in the A * B * C interaction).
Type I Sums of Squares. Type I sums of squares involve a sequential partitioning of the whole model sums of squares. A hierarchical series of regression equations are estimated, at each step adding an additional effect into the model. In Type I sums of squares, the sums of squares for each effect are determined by subtracting the predicted sums of squares with the effect in the model from the predicted sums of squares for the preceding model not including the effect. Tests of significance for each effect are then performed on the increment in the predicted sums of squares accounted for by the effect. Type I sums of squares are therefore sometimes called sequential or hierarchical sums of squares.
Type I sums of squares are appropriate to use in balanced (equal n) ANOVA designs in which effects are entered into the model in their natural order (i.e., any main effects are entered before any twoway interaction effects, any twoway interaction effects are entered before any threeway interaction effects, and so on). Type I sums of squares are also useful in polynomial regression designs in which any lowerorder effects are entered before any higherorder effects. A third use of Type I sums of squares is to test hypotheses for hierarchically nested designs, in which the first effect in the design is nested within the second effect, the second effect is nested within the third, and so on.
One important property of Type I sums of squares is that the sums of squares attributable to each effect add up to the whole model sums of squares. Thus, Type I sums of squares provide a complete decomposition of the predicted sums of squares for the whole model. This is not generally true for any other type of sums of squares. An important limitation of Type I sums of squares, however, is that the sums of squares attributable to a specific effect will generally depend on the order in which the effects are entered into the model. This lack of invariance to order of entry into the model limits the usefulness of Type I sums of squares for testing hypotheses for certain designs (e.g., fractional factorial designs).
Type II Sums of Squares. Type II sums of squares are sometimes called partially sequential sums of squares. Like Type I sums of squares, Type II sums of squares for an effect controls for the influence of other effects. Which other effects to control for, however, is determined by a different criterion. In Type II sums of squares, the sums of squares for an effect is computed by controlling for the influence of all other effects of equal or lower degree. Thus, sums of squares for main effects control for all other main effects, sums of squares for twoway interactions control for all main effects and all other twoway interactions, and so on.
Unlike Type I sums of squares, Type II sums of squares are invariant to the order in which effects are entered into the model. This makes Type II sums of squares useful for testing hypotheses for multiple regression designs, for main effect ANOVA designs, for fullfactorial ANOVA designs with equal cell ns, and for hierarchically nested designs.
There is a drawback to the use of Type II sums of squares for factorial designs with unequal cell ns. In these situations, Type II sums of squares test hypotheses that are complex functions of the cell ns that ordinarily are not meaningful. Thus, a different method for testing hypotheses is usually preferred.
Type III Sums of Squares. Type I and Type II sums of squares usually are not appropriate for testing hypotheses for factorial ANOVA designs with unequal ns. For ANOVA designs with unequal ns, however, Type III sums of squares test the same hypothesis that would be tested if the cell ns were equal, provided that there is at least one observation in every cell. Specifically, in nomissingcell designs, Type III sums of squares test hypotheses about differences in subpopulation (or marginal) means. When there are no missing cells in the design, these subpopulation means are least squares means, which are the best linearunbiased estimates of the marginal means for the design (see, Milliken and Johnson, 1986).
Tests of differences in least squares means have the important property that they are invariant to the choice of the coding of effects for categorical predictor variables (e.g., the use of the sigmarestricted or overparameterized model) and to the choice of the particular g2 inverse of X’X used to solve the normal equations. Thus, tests of linear combinations of least squares means in general, including Type III tests of differences in least squares means, are said to not depend on the parameterization of the design. This makes Type III sums of squares useful for testing hypotheses for any design for which Type I or Type II sums of squares are appropriate, as well as for any unbalanced ANOVA design with no missing cells.
The Type III sums of squares attributable to an effect is computed as the sums of squares for the effect controlling for any effects of equal or lower degree and orthogonal to any higherorder interaction effects (if any) that contain it. The orthogonality to higherorder containing interactions is what gives Type III sums of squares the desirable properties associated with linear combinations of least squares means in ANOVA designs with no missing cells. But for ANOVA designs with missing cells, Type III sums of squares generally do not test hypotheses about least squares means, but instead test hypotheses that are complex functions of the patterns of missing cells in higherorder containing interactions and that are ordinarily not meaningful. In this situation Type V sums of squares or tests of the effective hypothesis (Type VI sums of squares) are preferred.
Type IV Sums of Squares. Type IV sums of squares were designed to test “balanced” hypotheses for lowerorder effects in ANOVA designs with missing cells. Type IV sums of squares are computed by equitably distributing cell contrast coefficients for lowerorder effects across the levels of higherorder containing interactions.
Type IV sums of squares are not recommended for testing hypotheses for lowerorder effects in ANOVA designs with missing cells, even though this is the purpose for which they were developed. This is because Type IV sumofsquares are invariant to some but not all g2 inverses of X’X that could be used to solve the normal equations. Specifically, Type IV sums of squares are invariant to the choice of a g2 inverse of X’X given a particular ordering of the levels of the categorical predictor variables, but are not invariant to different orderings of levels. Furthermore, as with Type III sums of squares, Type IV sums of squares test hypotheses that are complex functions of the patterns of missing cells in higherorder containing interactions and that are ordinarily not meaningful.
Statisticians who have examined the usefulness of Type IV sums of squares have concluded that Type IV sums of squares are not up to the task for which they were developed:
 Milliken & Johnson (1992, p. 204) write: “It seems likely that few, if any, of the hypotheses tested by the Type IV analysis of [some programs] will be of particular interest to the experimenter.”
 Searle (1987, p. 463464) writes: “In general, [Type IV] hypotheses determined in this nature are not necessarily of any interest.”; and (p. 465) “This characteristic of Type IV sums of squares for rows depending on the sequence of rows establishes their nonuniqueness, and this in turn emphasizes that the hypotheses they are testing are by no means necessarily of any general interest.”
 Hocking (1985, p. 152), in an otherwise comprehensive introduction to general linear models, writes: “For the missing cell problem, [some programs] offers a fourth analysis, Type IV, which we shall not discuss.”
So, we recommend that you use the Type IV sums of squares solution with caution, and that you understand fully the nature of the (often nonunique) hypotheses that are being testing, before attempting interpretations of the results. Furthermore, in ANOVA designs with no missing cells, Type IV sums of squares are always equal to Type III sums of squares, so the use of Type IV sums of squares is either (potentially) inappropriate, or unnecessary, depending on the presence of missing cells in the design.
Type V Sums of Squares. Type V sums of squares were developed as an alternative to Type IV sums of squares for testing hypotheses in ANOVA designs in missing cells. Also, this approach is widely used in industrial experimentation, to analyze fractional factorial designs; these types of designs are discussed in detail in the 2**(kp) Fractional Factorial Designs section of the Experimental Design topic. In effect, for effects for which tests are performed all population marginal means (least squares means) are estimable.
Type V sums of squares involve a combination of the methods employed in computing Type I and Type III sums of squares. Specifically, whether or not an effect is eligible to be dropped from the model is determined using Type I procedures, and then hypotheses are tested for effects not dropped from the model using Type III procedures. Type V sums of squares can be illustrated by using a simple example. Suppose that the effects considered are A, B, and A by B, in that order, and that A and B are both categorical predictors with, say, 3 and 2 levels, respectively. The intercept is first entered into the model. Then A is entered into the model, and its degrees of freedom are determined (i.e., the number of nonredundant columns for A in X’X, given the intercept). If A‘s degrees of freedom are less than 2 (i.e., its number of levels minus 1), it is eligible to be dropped. Then B is entered into the model, and its degrees of freedom are determined (i.e., the number of nonredundant columns for B in X’X, given the intercept and A). If B‘s degrees of freedom are less than 1 (i.e., its number of levels minus 1), it is eligible to be dropped. Finally, A by B is entered into the model, and its degrees of freedom are determined (i.e., the number of nonredundant columns for A by B in X’X, given the intercept, A, and B). If B‘s degrees of freedom are less than 2 (i.e., the product of the degrees of freedom for its factors if there were no missing cells), it is eligible to be dropped. Type III sums of squares are then computed for the effects that were not found to be eligible to be dropped, using the reduced model in which any eligible effects are dropped. Tests of significance, however, use the error term for the whole model prior to dropping any eligible effects.
Note that Type V sums of squares involve determining a reduced model for which all effects remaining in the model have at least as many degrees of freedom as they would have if there were no missing cells. This is equivalent to finding a subdesign with no missing cells such that the Type III sums of squares for all effects in the subdesign reflect differences in least squares means.
Appropriate caution should be exercised when using Type V sums of squares. Dropping an effect from a model is the same as assuming that the effect is unrelated to the outcome (see, e.g., Hocking, 1996). The reasonableness of the assumption does not necessarily insure its validity, so when possible the relationships of dropped effects to the outcome should be inspected. It is also important to note that Type V sums of squares are not invariant to the order in which eligibility for dropping effects from the model is evaluated. Different orders of effects could produce different reduced models.
In spite of these limitations, Type V sums of squares for the reduced model have all the same properties of Type III sums of squares for ANOVA designs with no missing cells. Even in designs with many missing cells (such as fractional factorial designs, in which many highorder interaction effects are assumed to be zero), Type V sums of squares provide tests of meaningful hypotheses, and sometimes hypotheses that cannot be tested using any other method.
Type VI (Effective Hypothesis) Sums of Squares. Type I through Type V sums of squares can all be viewed as providing tests of hypotheses that subsets of partial regression coefficients (controlling for or orthogonal to appropriate additional effects) are zero. Effective hypothesis tests (developed by Hocking, 1996) are based on the philosophy that the only unambiguous estimate of an effect is the proportion of variability on the outcome that is uniquely attributable to the effect. The overparameterized coding of effects for categorical predictor variables generally cannot be used to provide such unique estimates for lowerorder effects. Effective hypothesis tests, which we propose to call Type VI sums of squares, use the sigmarestricted coding of effects for categorical predictor variables to provide unique effect estimates even for lowerorder effects.
The method for computing Type VI sums of squares is straightforward. The sigmarestricted coding of effects is used, and for each effect, its Type VI sums of squares is the difference of the model sums of squares for all other effects from the whole model sums of squares. As such, the Type VI sums of squares provide an unambiguous estimate of the variability of predicted values for the outcome uniquely attributable to each effect.
In ANOVA designs with missing cells, Type VI sums of squares for effects can have fewer degrees of freedom than they would have if there were no missing cells, and for some missing cell designs, can even have zero degrees of freedom. The philosophy of Type VI sums of squares is to test as much as possible of the original hypothesis given the observed cells. If the pattern of missing cells is such that no part of the original hypothesis can be tested, so be it. The inability to test hypotheses is simply the price we pay for having no observations at some combinations of the levels of the categorical predictor variables. The philosophy is that it is better to admit that a hypothesis cannot be tested than it is to test a distorted hypothesis that may not meaningfully reflect the original hypothesis.
Type VI sums of squares cannot generally be used to test hypotheses for nested ANOVA designs, separate slope designs, or mixedmodel designs, because the sigmarestricted coding of effects for categorical predictor variables is overly restrictive in such designs. This limitation, however, does not diminish the fact that Type VI sums of squares can b
Error Terms for Tests
LackofFit Tests using Pure Error. Whole model tests and tests based on the 6 types of sums of squares use the mean square residual as the error term for tests of significance. For certain types of designs, however, the residual sum of squares can be further partitioned into meaningful parts which are relevant for testing hypotheses. One such type of design is a simple regression design in which there are subsets of cases all having the same values on the predictor variable. For example, performance on a task could be measured for subjects who work on the task under several different room temperature conditions. The test of significance for the Temperature effect in the linear regression of Performance on Temperature would not necessarily provide complete information on how Temperature relates to Performance; the regression coefficient for Temperature only reflects its linear effect on the outcome.
One way to glean additional information from this type of design is to partition the residual sums of squares into lackoffit and pure error components. In the example just described, this would involve determining the difference between the sum of squares that cannot be predicted by Temperature levels, given the linear effect of Temperature (residual sums of squares) and the pure error; this difference would be the sums of squares associated with the lackoffit (in this example, of the linear model). The test of lackoffit, using the mean square pure error as the error term, would indicate whether nonlinear effects of Temperature are needed to adequately model Tempature’s influence on the outcome. Further, the linear effect could be tested using the pure error term, thus providing a more sensitive test of the linear effect independent of any possible nonlinear effect.
Designs with Zero Degrees of Freedom for Error. When the model degrees of freedom equal the number of cases or subjects, the residual sums of squares will have zero degrees of freedom and preclude the use of standard hypothesis tests. This sometimes occurs for overfitted designs (designs with many predictors, or designs with categorical predictors having many levels). However, in some designed experiments, such as experiments using splitplot designs or highly fractionalized factorial designs as commonly used in industrial experimentation, it is no accident that the residual sum of squares has zero degrees of freedom. In such experiments, mean squares for certain effects are planned to be used as error terms for testing other effects, and the experiment is designed with this in mind. It is entirely appropriate to use alternatives to the mean square residual as error terms for testing hypotheses in such designs.
Tests in Mixed Model Designs. Designs which contain random effects for one or more categorical predictor variables are called mixedmodel designs. These types of designs, and the analysis of those designs, is also described in detail in the Variance Components and Mixed Model ANOVA/ANCOVA topic. Random effects are classification effects where the levels of the effects are assumed to be randomly selected from an infinite population of possible levels. The solution for the normal equations in mixedmodel designs is identical to the solution for fixedeffect designs (i.e., designs which do not contain random effects). Mixedmodel designs differ from fixedeffect designs only in the way in which effects are tested for significance. In fixedeffect designs, between effects are always tested using the mean square residual as the error term. In mixedmodel designs, between effects are tested using relevant error terms based on the covariation of sources of variation in the design. Also, only the overparameterized model is used to code effects for categorical predictors in mixedmodels, because the sigmarestricted model is overly restrictive.
The covariation of sources of variation in the design is estimated by the elements of a matrix called the Expected Mean Squares (EMS) matrix. This nonsquare matrix contains elements for the covariation of each combination of pairs of sources of variation and for each source of variation with Error. Specifically, each element is the mean square for one effect (indicated by the column) that is expected to be accounted by another effect (indicated by the row), given the observed covariation in their levels. Note that expected mean squares can be computing using any type of sums of squares from Type I through Type V. Once the EMS matrix is computed, it is used to the solve for the linear combinations of sources of random variation that are appropriate to use as error terms for testing the significance of the respective effects. This is done using Satterthwaite’s method of denominator synthesis (Satterthwaite, 1946). Detailed discussions of methods for testing effects in mixedmodels, and related methods for estimating variance components for random effects, can be found in the Variance Components and Mixed Model ANOVA/ANCOVA topic.
Testing Specific Hypotheses
Whole model tests and tests based on sums of squares attributable to specific effects illustrate two general types of hypotheses that can be tested using the general linear model. Still, there may be other types of hypotheses the researcher wishes to test that do not fall into either of these categories. For example, hypotheses about subsets of effects may be of interest, or hypotheses involving comparisons of specific levels of categorical predictor variables may be of interest.
Estimability of Hypotheses. Before considering tests of specific hypotheses of this sort, it is important to address the issue of estimability. A test of a specific hypothesis using the general linear model must be framed in terms of the regression coefficients for the solution of the normal equations. If the X’X matrix is less than full rank, the regression coefficients depend on the particular g2 inverse used for solving the normal equations, and the regression coefficients will not be unique. When the regression coefficients are not unique, linear functions (f) of the regression coefficients having the form
f = Lb
where L is a vector of coefficients, will also in general not be unique. However, Lb for an L which satisfies
L = L(X’X)^{–}X’X
is invariant for all possible g2 inverses, and is therefore called an estimable function.
The theory of estimability of linear functions is an advanced topic in the theory of algebraic invariants (Searle, 1987, provides a comprehensive introduction), but its implications are clear enough. One instance of nonestimability of a hypothesis has been encountered in tests of the effective hypothesis which have zero degrees of freedom. On the other hand, Type III sums of squares for categorical predictor variable effects in ANOVA designs with no missing cells (and the least squares means in such designs) provide an example of estimable functions which do not depend on the model parameterization (i.e., the particular g2 inverse used to solve the normal equations). The general implication of the theory of estimability of linear functions is that hypotheses which cannot be expressed as linear combinations of the rows of X (i.e., the combinations of observed levels of the categorical predictor variables) are not estimable, and therefore cannot be tested. Stated another way, we simply cannot test specific hypotheses that are not represented in the data. The notion of estimability is valuable because the test for estimability makes explicit which specific hypotheses can be tested and which cannot.
Linear Combinations of Effects. In multiple regression designs, it is common for hypotheses of interest to involve subsets of effects. In mixture designs, for example, we might be interested in simultaneously testing whether the main effect and any of the twoway interactions involving a particular predictor variable are nonzero. It is also common in multiple regression designs for hypotheses of interest to involves comparison of slopes. For example, we might be interested in whether the regression coefficients for two predictor variables differ. In both factorial regression and factorial ANOVA designs with many factors, it is often of interest whether sets of effects, say, all threeway and higherorder interactions, are nonzero. Tests of these types of specific hypotheses involve (1) constructing one or more Ls reflecting the hypothesis, (2) testing the estimability of the hypothesis by determining whether
L = L(X’X)^{–}X’X
and if so, using (3)
(Lb)’L’)^{1}(Lb)
to estimate the sums of squares accounted for by the hypothesis. Finally, (4) the hypothesis is tested for significance using the usual mean square residual as the error term. To illustrate this 4step procedure, suppose that a test of the difference in the regression slopes is desired for the (intercept plus) 2 predictor variables in a firstorder multiple regression design. The coefficients for L would be
L = [0 1 1]
(note that the first coefficient 0 excludes the intercept from the comparison) for which Lb is estimable if the 2 predictor variables are not redundant with each other. The hypothesis sums of squares reflect the difference in the partial regression coefficients for the 2 predictor variables, which is tested for significance using the mean square residual as the error term.
Planned Comparisons of Least Square Means. Usually, experimental hypotheses are stated in terms that are more specific than simply main effects or interactions. We may have the specific hypothesis that a particular textbook will improve math skills in males, but not in females, while another book would be about equally effective for both genders, but less effective overall for males. Now generally, we are predicting an interaction here: the effectiveness of the book is modified (qualified) by the student’s gender. However, we have a particular prediction concerning the nature of the interaction: we expect a significant difference between genders for one book, but not the other. This type of specific prediction is usually tested by testing planned comparisons of least squares means (estimates of the population marginal means), or as it is sometimes called, contrast analysis.
Briefly, contrast analysis allows us to test the statistical significance of predicted specific differences in particular parts of our complex design. The 4step procedure for testing specific hypotheses is used to specify and test specific predictions. Contrast analysis is a major and indispensable component of the analysis of many complex experimental designs (see also for details).
To learn more about the logic and interpretation of contrast analysis refer to the ANOVA/MANOVA topic Overview section.
PostHoc Comparisons. Sometimes we find effects in an experiment that were not expected. Even though in most cases a creative experimenter will be able to explain almost any pattern of means, it would not be appropriate to analyze and evaluate that pattern as if we had predicted it all along. The problem here is one of capitalizing on chance when performing multiple tests posthoc, that is, without a priori hypotheses. To illustrate this point, let’s consider the following “experiment.” Imagine we were to write down a number between 1 and 10 on 100 pieces of paper. We then put all of those pieces into a hat and draw 20 samples (of pieces of paper) of 5 observations each, and compute the means (from the numbers written on the pieces of paper) for each group. How likely do you think it is that we will find two sample means that are significantly different from each other? It is very likely! Selecting the extreme means obtained from 20 samples is very different from taking only 2 samples from the hat in the first place, which is what the test via the contrast analysis implies. Without going into further detail, there are several socalled posthoc tests that are explicitly based on the first scenario (taking the extremes from 20 samples), that is, they are based on the assumption that we have chosen for our comparison the most extreme (different) means out of k total means in the design. Those tests apply “corrections” that are designed to offset the advantage of posthoc selection of the most extreme comparisons. Whenever we find unexpected results in an experiment, we should use those posthoc procedures to test their statistical significance.
Testing Hypotheses for Repeated Measures and Dependent Variables
In the discussion of different hypotheses that can be tested using the general linear model, the tests have been described as tests for “the dependent variable” or “the outcome.” This has been done solely to simplify the discussion. When there are multiple dependent variables reflecting the levels of repeated measure factors, the general linear model performs tests using orthonormalized Mtransformations of the dependent variables. When there are multiple dependent variables but no repeated measure factors, the general linear model performs tests using the hypothesis sums of squares and crossproducts for the multiple dependent variables, which are tested against the residual sums of squares and crossproducts for the multiple dependent variables. Thus, the same hypothesis testing procedures which apply to univariate designs with a single dependent variable also apply to repeated measure and multivariate designs.
Healthcare Intelligence Network
New Chart: Top 5 Tools to Identify Patients at Risk for Readmissions
SUMMARY: Anxious to avoid looming financial penalties for excessive hospital readmissions, healthcare organizations have tightened coordination of care and management of care transitions for Medicare beneficiaries at risk of rehospitalization. We wanted to see how healthcare organizations identify patients most at risk for returning to the hospital.
Click here to see a larger, printable version of this chart.
HIN’s third annual survey on Reducing Hospital Readmissions, conducted in February 2012, documented the highest rates of targeted programs to reduce readmissions in the survey’s threeyear history. The survey captured details on these programs to curb readmission rates, along with the conditions most likely to trigger readmissions and much more. According to 119 healthcare companies who responded to the survey, the top 5 tools used to identify patients most at risk for readmissions are:
 Risk stratification: 55.2 percent
 Chart review: 41.4 percent
 Health claims: 39.7 percent
 Predictive modeling: 36.2 percent
 EHRs: 32.8 percent
For additional research data and insights on this topic:
Download the executive summary of 2012 Healthcare Benchmarks: Reducing Hospital Readmissions.
Principal Components and Factor Analysis
How to Reduce Number of Variables and Detect Relationships, Principal Components and Factor Analysis
General Purpose
The main applications of factor analytic techniques are: (1) to reduce the number of variables and (2) to detect structure in the relationships between variables, that is to classify variables. Therefore, factor analysis is applied as a data reduction or structure detection method (the term factor analysis was first introduced by Thurstone, 1931). The topics listed below will describe the principles of factor analysis, and how it can be applied towards these two purposes. We will assume that you are familiar with the basic logic of statistical reasoning as described in Elementary Concepts. Moreover, we will also assume that you are familiar with the concepts of variance and correlation; if not, we advise that you read the Basic Statistics topic at this point.
There are many excellent books on factor analysis. For example, a handson howto approach can be found in Stevens (1986); more detailed technical descriptions are provided in Cooley and Lohnes (1971); Harman (1976); Kim and Mueller, (1978a, 1978b); Lawley and Maxwell (1971); Lindeman, Merenda, and Gold (1980); Morrison (1967); or Mulaik (1972). The interpretation of secondary factors in hierarchical factor analysis, as an alternative to traditional oblique rotational strategies, is explained in detail by Wherry (1984).
Confirmatory factor analysis. Structural Equation Modeling (SEPATH) allows you to test specific hypotheses about the factor structure for a set of variables, in one or several samples (e.g., you can compare factor structures across samples).
Correspondence analysis. Correspondence analysis is a descriptive/exploratory technique designed to analyze twoway and multiway tables containing some measure of correspondence between the rows and columns. The results provide information which is similar in nature to those produced by factor analysis techniques, and they allow you to explore the structure of categorical variables included in the table. For more information regarding these methods, refer to Correspondence Analysis.
Basic Idea of Factor Analysis as a Data Reduction Method
Suppose we conducted a (rather “silly”) study in which we measure 100 people’s height in inches and centimeters. Thus, we would have two variables that measure height. If in future studies, we want to research, for example, the effect of different nutritional food supplements on height, would we continue to use both measures? Probably not; height is one characteristic of a person, regardless of how it is measured.
Let’s now extrapolate from this “silly” study to something that you might actually do as a researcher. Suppose we want to measure people’s satisfaction with their lives. We design a satisfaction questionnaire with various items; among other things we ask our subjects how satisfied they are with their hobbies (item 1) and how intensely they are pursuing a hobby (item 2). Most likely, the responses to the two items are highly correlated with each other. (If you are not familiar with the correlation coefficient, we recommend that you read the description in Basic Statistics – Correlations) Given a high correlation between the two items, we can conclude that they are quite redundant.
Combining Two Variables into a Single Factor. You can summarize the correlation between two variables in a scatterplot. A regression line can then be fitted that represents the “best” summary of the linear relationship between the variables. If we could define a variable that would approximate the regression line in such a plot, then that variable would capture most of the “essence” of the two items. Subjects’ single scores on that new factor, represented by the regression line, could then be used in future data analyses to represent that essence of the two items. In a sense we have reduced the two variables to one factor. Note that the new factor is actually a linear combination of the two variables.
Principal Components Analysis. The example described above, combining two correlated variables into one factor, illustrates the basic idea of factor analysis, or of principal components analysis to be precise (we will return to this later). If we extend the twovariable example to multiple variables, then the computations become more involved, but the basic principle of expressing two or more variables by a single factor remains the same.
Extracting Principal Components. We do not want to go into the details about the computational aspects of principal components analysis here, which can be found elsewhere (references were provided at the beginning of this section). However, basically, the extraction of principal components amounts to a variance maximizing (varimax) rotation of the original variable space. For example, in a scatterplot we can think of the regression line as the original X axis, rotated so that it approximates the regression line. This type of rotation is called variance maximizing because the criterion for (goal of) the rotation is to maximize the variance (variability) of the “new” variable (factor), while minimizing the variance around the new variable (see Rotational Strategies).
Generalizing to the Case of Multiple Variables. When there are more than two variables, we can think of them as defining a “space,” just as two variables defined a plane. Thus, when we have three variables, we could plot a three dimensional scatterplot, and, again we could fit a plane through the data.
With more than three variables it becomes impossible to illustrate the points in a scatterplot, however, the logic of rotating the axes so as to maximize the variance of the new factor remains the same.
Multiple orthogonal factors. After we have found the line on which the variance is maximal, there remains some variability around this line. In principal components analysis, after the first factor has been extracted, that is, after the first line has been drawn through the data, we continue and define another line that maximizes the remaining variability, and so on. In this manner, consecutive factors are extracted. Because each consecutive factor is defined to maximize the variability that is not captured by the preceding factor, consecutive factors are independent of each other. Put another way, consecutive factors are uncorrelated or orthogonal to each other.
How many Factors to Extract? Remember that, so far, we are considering principal components analysis as a data reduction method, that is, as a method for reducing the number of variables. The question then is, how many factors do we want to extract? Note that as we extract consecutive factors, they account for less and less variability. The decision of when to stop extracting factors basically depends on when there is only very little “random” variability left. The nature of this decision is arbitrary; however, various guidelines have been developed, and they are reviewed in Reviewing the Results of a Principal Components Analysis under Eigenvalues and the Numberof Factors Problem.
Reviewing the Results of a Principal Components Analysis. Without further ado, let us now look at some of the standard results from a principal components analysis. To reiterate, we are extracting factors that account for less and less variance. To simplify matters, you usually start with the correlation matrix, where the variances of all variables are equal to 1.0. Therefore, the total variance in that matrix is equal to the number of variables. For example, if we have 10 variables each with a variance of 1 then the total variability that can potentially be extracted is equal to 10 times 1. Suppose that in the satisfaction study introduced earlier we included 10 items to measure different aspects of satisfaction at home and at work. The variance accounted for by successive factors would be summarized as follows:
STATISTICA FACTOR ANALYSIS 
Eigenvalues (factor.sta) Extraction: Principal components 


Value 
Eigenval 
% total Variance 
Cumul. Eigenval 
Cumul. % 
1 2 3 4 5 6 7 8 9 10 
6.118369 1.800682 .472888 .407996 .317222 .293300 .195808 .170431 .137970 .085334 
61.18369 18.00682 4.72888 4.07996 3.17222 2.93300 1.95808 1.70431 1.37970 .85334 
6.11837 7.91905 8.39194 8.79993 9.11716 9.41046 9.60626 9.77670 9.91467 10.00000 
61.1837 79.1905 83.9194 87.9993 91.1716 94.1046 96.0626 97.7670 99.1467 100.0000 
Eigenvalues
In the second column (Eigenvalue) above, we find the variance on the new factors that were successively extracted. In the third column, these values are expressed as a percent of the total variance (in this example, 10). As we can see, factor 1 accounts for 61 percent of the variance, factor 2 for 18 percent, and so on. As expected, the sum of the eigenvalues is equal to the number of variables. The third column contains the cumulative variance extracted. The variances extracted by the factors are called the eigenvalues. This name derives from the computational issues involved.
Eigenvalues and the NumberofFactors Problem
Now that we have a measure of how much variance each successive factor extracts, we can return to the question of how many factors to retain. As mentioned earlier, by its nature this is an arbitrary decision. However, there are some guidelines that are commonly used, and that, in practice, seem to yield the best results.
The Kaiser criterion. First, we can retain only factors with eigenvalues greater than 1. In essence this is like saying that, unless a factor extracts at least as much as the equivalent of one original variable, we drop it. This criterion was proposed by Kaiser (1960), and is probably the one most widely used. In our example above, using this criterion, we would retain 2 factors (principal components).
The scree test. A graphical method is the scree test first proposed by Cattell (1966). We can plot the eigenvalues shown above in a simple line plot.
Cattell suggests to find the place where the smooth decrease of eigenvalues appears to level off to the right of the plot. To the right of this point, presumably, you find only “factorial scree” – “scree” is the geological term referring to the debris which collects on the lower part of a rocky slope. According to this criterion, we would probably retain 2 or 3 factors in our example.
Which criterion to use. Both criteria have been studied in detail (Browne, 1968; Cattell & Jaspers, 1967; Hakstian, Rogers, & Cattell, 1982; Linn, 1968; Tucker, Koopman & Linn, 1969). Theoretically, you can evaluate those criteria by generating random data based on a particular number of factors. You can then see whether the number of factors is accurately detected by those criteria. Using this general technique, the first method (Kaiser criterion) sometimes retains too many factors, while the second technique (scree test) sometimes retains too few; however, both do quite well under normal conditions, that is, when there are relatively few factors and many cases. In practice, an additional important aspect is the extent to which a solution is interpretable. Therefore, you usually examines several solutions with more or fewer factors, and chooses the one that makes the best “sense.” We will discuss this issue in the context of factor rotations below.
Principal Factors Analysis
Before we continue to examine the different aspects of the typical output from a principal components analysis, let us now introduce principal factors analysis. Let us return to our satisfaction questionnaire example to conceive of another “mental model” for factor analysis. We can think of subjects’ responses as being dependent on two components. First, there are some underlying common factors, such as the “satisfactionwithhobbies” factor we looked at before. Each item measures some part of this common aspect of satisfaction. Second, each item also captures a unique aspect of satisfaction that is not addressed by any other item.
Communalities. If this model is correct, then we should not expect that the factors will extract all variance from our items; rather, only that proportion that is due to the common factors and shared by several items. In the language of factor analysis, the proportion of variance of a particular item that is due to common factors (shared with other items) is called communality. Therefore, an additional task facing us when applying this model is to estimate the communalities for each variable, that is, the proportion of variance that each item has in common with other items. The proportion of variance that is unique to each item is then the respective item’s total variance minus the communality. A common starting point is to use the squared multiple correlation of an item with all other items as an estimate of the communality (refer to Multiple Regression for details about multiple regression). Some authors have suggested various iterative “postsolution improvements” to the initial multiple regression communality estimate; for example, the socalled MINRES method (minimum residual factor method; Harman & Jones, 1966) will try various modifications to the factor loadings with the goal to minimize the residual (unexplained) sums of squares.
Principal factors vs. principal components. The defining characteristic then that distinguishes between the two factor analytic models is that in principal components analysis we assume that all variability in an item should be used in the analysis, while in principal factors analysis we only use the variability in an item that it has in common with the other items. A detailed discussion of the pros and cons of each approach is beyond the scope of this introduction (refer to the general references provided in Principal components and Factor Analysis – Introductory Overview). In most cases, these two methods usually yield very similar results. However, principal components analysis is often preferred as a method for data reduction, while principal factors analysis is often preferred when the goal of the analysis is to detect structure (see Factor Analysis as a Classification Method).
Factor Analysis as a Classification Method
Let us now return to the interpretation of the standard results from a factor analysis. We will henceforth use the term factor analysis generically to encompass both principal components and principal factors analysis. Let us assume that we are at the point in our analysis where we basically know how many factors to extract. We may now want to know the meaning of the factors, that is, whether and how we can interpret them in a meaningful manner. To illustrate how this can be accomplished, let us work “backwards,” that is, begin with a meaningful structure and then see how it is reflected in the results of a factor analysis. Let us return to our satisfaction example; shown below is the correlation matrix for items pertaining to satisfaction at work and items pertaining to satisfaction at home.
STATISTICA FACTOR ANALYSIS 
Correlations (factor.sta) Casewise deletion of MD n=100 


Variable  WORK_1  WORK_2  WORK_3  HOME_1  HOME_2  HOME_3 
WORK_1 WORK_2 WORK_3 HOME_1 HOME_2 HOME_3 
1.00 .65 .65 .14 .15 .14 
.65 1.00 .73 .14 .18 .24 
.65 .73 1.00 .16 .24 .25 
.14 .14 .16 1.00 .66 .59 
.15 .18 .24 .66 1.00 .73 
.14 .24 .25 .59 .73 1.00 
The work satisfaction items are highly correlated amongst themselves, and the home satisfaction items are highly intercorrelated amongst themselves. The correlations across these two types of items (work satisfaction items with home satisfaction items) is comparatively small. It thus seems that there are two relatively independent factors reflected in the correlation matrix, one related to satisfaction at work, the other related to satisfaction at home.
Factor Loadings. Let us now perform a principal components analysis and look at the twofactor solution. Specifically, let us look at the correlations between the variables and the two factors (or “new” variables), as they are extracted by default; these correlations are also called factor loadings.
STATISTICA FACTOR ANALYSIS 
Factor Loadings (Unrotated) Principal components 


Variable  Factor 1  Factor 2 
WORK_1 WORK_2 WORK_3 HOME_1 HOME_2 HOME_3 
.654384 .715256 .741688 .634120 .706267 .707446 
.564143 .541444 .508212 .563123 .572658 .525602 
Expl.Var Prp.Totl 
2.891313 .481885 
1.791000 .298500 
Apparently, the first factor is generally more highly correlated with the variables than the second factor. This is to be expected because, as previously described, these factors are extracted successively and will account for less and less variance overall.
Rotating the Factor Structure. We could plot the factor loadings shown above in a scatterplot. In that plot, each variable is represented as a point. In this plot we could rotate the axes in any direction without changing the relative locations of the points to each other; however, the actual coordinates of the points, that is, the factor loadings would of course change. In this example, if you produce the plot it will be evident that if we were to rotate the axes by about 45 degrees we might attain a clear pattern of loadings identifying the work satisfaction items and the home satisfaction items.
Rotational strategies. There are various rotational strategies that have been proposed. The goal of all of these strategies is to obtain a clear pattern of loadings, that is, factors that are somehow clearly marked by high loadings for some variables and low loadings for others. This general pattern is also sometimes referred to as simple structure (a more formalized definition can be found in most standard textbooks). Typical rotational strategies are varimax, quartimax, and equamax.
We have described the idea of the varimax rotation before (see Extracting Principal Components), and it can be applied to this problem as well. As before, we want to find a rotation that maximizes the variance on the new axes; put another way, we want to obtain a pattern of loadings on each factor that is as diverse as possible, lending itself to easier interpretation. Below is the table of rotated factor loadings.
STATISTICA FACTOR ANALYSIS 
Factor Loadings (Varimax normalized) Extraction: Principal components 


Variable  Factor 1  Factor 2 
WORK_1 WORK_2 WORK_3 HOME_1 HOME_2 HOME_3 
.862443 .890267 .886055 .062145 .107230 .140876 
.051643 .110351 .152603 .845786 .902913 .869995 
Expl.Var Prp.Totl 
2.356684 .392781 
2.325629 .387605 
Interpreting the Factor Structure. Now the pattern is much clearer. As expected, the first factor is marked by high loadings on the work satisfaction items, the second factor is marked by high loadings on the home satisfaction items. We would thus conclude that satisfaction, as measured by our questionnaire, is composed of those two aspects; hence we have arrived at a classification of the variables.
Consider another example, this time with four additional Hobby/Misc variables added to our earlier example.
In the plot of factor loadings above, 10 variables were reduced to three specific factors, a work factor, a home factor and a hobby/misc. factor. Note that factor loadings for each factor are spread out over the values of the other two factors but are high for its own values. For example, the factor loadings for the hobby/misc variables (in green) have both high and low “work” and “home” values, but all four of these variables have high factor loadings on the “hobby/misc” factor.
Oblique Factors. Some authors (e.g., Cattell & Khanna; Harman, 1976; Jennrich & Sampson, 1966; Clarkson & Jennrich, 1988) have discussed in some detail the concept of oblique (nonorthogonal) factors, in order to achieve more interpretable simple structure. Specifically, computational strategies have been developed to rotate factors so as to best represent “clusters” of variables, without the constraint of orthogonality of factors. However, the oblique factors produced by such rotations are often not easily interpreted. To return to the example discussed above, suppose we would have included in the satisfaction questionnaire above four items that measured other, “miscellaneous” types of satisfaction. Let us assume that people’s responses to those items were affected about equally by their satisfaction at home (Factor 1) and at work (Factor 2). An oblique rotation will likely produce two correlated factors with lessthan obvious meaning, that is, with many crossloadings.
Hierarchical Factor Analysis. Instead of computing loadings for often difficult to interpret oblique factors, you can use a strategy first proposed by Thompson (1951) and Schmid and Leiman (1957), which has been elaborated and popularized in the detailed discussions by Wherry (1959, 1975, 1984). In this strategy, you first identify clusters of items and rotate axes through those clusters; next the correlations between those (oblique) factors is computed, and that correlation matrix of oblique factors is further factoranalyzed to yield a set of orthogonal factors that divide the variability in the items into that due to shared or common variance (secondary factors), and unique variance due to the clusters of similar variables (items) in the analysis (primary factors). To return to the example above, such a hierarchical analysis might yield the following factor loadings:
STATISTICA FACTOR ANALYSIS 
Secondary & Primary Factor Loadings  

Factor  Second. 1  Primary 1  Primary 2 
WORK_1 WORK_2 WORK_3 HOME_1 HOME_2 HOME_3 MISCEL_1 MISCEL_2 MISCEL_3 MISCEL_4 
.483178 .570953 .565624 .535812 .615403 .586405 .780488 .734854 .776013 .714183 
.649499 .687056 .656790 .117278 .079910 .065512 .466823 .464779 .439010 .455157 
.187074 .140627 .115461 .630076 .668880 .626730 .280141 .238512 .303672 .228351 
Careful examination of these loadings would lead to the following conclusions:
 There is a general (secondary) satisfaction factor that likely affects all types of satisfaction measured by the 10 items;
 There appear to be two primary unique areas of satisfaction that can best be described as satisfaction with work and satisfaction with home life.
Wherry (1984) discusses in great detail examples of such hierarchical analyses, and how meaningful and interpretable secondary factors can be derived.
Confirmatory Factor Analysis. Over the past 15 years, socalled confirmatory methods have become increasingly popular (e.g., see Jöreskog and Sörbom, 1979). In general, you can specify a priori, a pattern of factor loadings for a particular number of orthogonal or oblique factors, and then test whether the observed correlation matrix can be reproduced given these specifications. Confirmatory factor analyses can be performed via Structural Equation Modeling (SEPATH).
Miscellaneous Other Issues and Statistics
Factor Scores. We can estimate the actual values of individual cases (observations) for the factors. These factor scores are particularly useful when you want to perform further analyses involving the factors that you have identified in the factor analysis.
Reproduced and Residual Correlations. An additional check for the appropriateness of the respective number of factors that were extracted is to compute the correlation matrix that would result if those were indeed the only factors. That matrix is called the reproduced correlation matrix. To see how this matrix deviates from the observed correlation matrix, you can compute the difference between the two; that matrix is called the matrix of residual correlations. The residual matrix may point to “misfits,” that is, to particular correlation coefficients that cannot be reproduced appropriately by the current number of factors.
Matrix Illconditioning. If, in the correlation matrix there are variables that are 100% redundant, then the inverse of the matrix cannot be computed. For example, if a variable is the sum of two other variables selected for the analysis, then the correlation matrix of those variables cannot be inverted, and the factor analysis can basically not be performed. In practice this happens when you are attempting to factor analyze a set of highly intercorrelated variables, as it, for example, sometimes occurs in correlational research with questionnaires. Then you can artificially lower all correlations in the correlation matrix by adding a small constant to the diagonal of the matrix, and then restandardizing it. This procedure will usually yield a matrix that now can be inverted and thus factoranalyzed; moreover, the factor patterns should not be affected by this procedure. However, note that the resulting estimates are not exact.