SPSS Categories 11.0

Win new business The data are a 2x5x6 table containing information on two genders, ve age groups and six products. This plot shows the results of a two-dimensional multiple correspondence analysis of the table. Notice that products such as A and B are chosen at younger ages and by males, while products such as G and C are preferred at older ages.
Predict deal closings, based on geographic region and sales reps, using categorical regression.
Increase customer loyalty
Learn how your customers perceive your products and services using PROXSCAL multidimensional scaling.

Increase cross-selling

Produce perceptual maps that enable you to discover product afnities by using optimal scaling techniques.

Detect fraud

Understand groupings using perceptual maps to predict categorical outcomes
Enhance your research with SPSS Categories 11.0. Youll go beyond unwieldy tables to clearly see relationships in your data using revealing biplots and perceptual maps.
You can analyze your data more completely and easily using SPSS Categories best-of-class correspondence analysis procedure. And, to ensure you have all the tools you need, SPSS Categories has a procedure for categorical regression analysis, as well as four others, so you get the most from your multivariate data analysis. based on variables of mixed measurement levels (nominal, ordinal and numerical variables). Youll also be able to incorporate variables of different measurement levels into sets and analyze them by using nonlinear canonical correlation analysis.
Discover fraud patterns across product sets using optimal scaling techniques.
Rely on SPSS Categories whenever you need to: s Understand information in large, unwieldy two-way and multi-way tables s Work with and understand ordinal and nominal data in procedures similar to conventional regression, principal components and canonical correlation s Visualize and explore multivariate categorical data Learn more about your categorical data Whatever types of categories you study market segments, subcultures, political parties or biological species SPSS Categories optimal scaling procedures free you from the restrictions of two-way tables, placing the relationships among your variables in a larger frame of reference. See a map of your data not just a statistical report. Youll understand your data better when you use nonlinear principal components analysis. This type of analysis summarizes your data by important components
Use charts to bring relationships into sharper focus High-resolution summary charts give you unique insight into relationships among more than two variables. For example, when you perform multiple correspondence (homogeneity) analysis, you see the relationships among categorical variables summarized in a single, easy-to-read, high-resolution chart. Present and share dynamic, interactive results When you use the correspondence analysis and categorical regression analysis procedures, you can display your results in tables, graphs or report cubes that feature unique pivoting technology. This award-winning technology empowers you to discover new insights into your data with a few mouse clicks. You can swap rows, columns and layers of report cubes or quickly change information and statistics in graphs for new levels of understanding. Then, you can share dynamic results using the optional SmartViewer Web Server, SPSS product for Web-based reporting. You give clients and colleagues who dont have SPSS installed on their machines the power to manipulate information and immediately interact with results.
Look whos using SPSS Categories:
Market researchers study what attributes characterize products Survey researchers relate questionnaire responses to the demographics of the respondents Health scientists rate the effectiveness of drugs Behavioral scientists relate the principle worries of a sample of subjects to the subjects residence and parents
Researchers studied the images of six brands of iced coffee sold in South Australia. Brands are denoted AA to FF and are characterized by various attributes. SPSS Correspondence procedure produced the correspondence map shown in this figure. Brand AA, the market leader, is near the popular attribute. CC and DD target consumers interested in health and low-fat products. FF is perceived as a rich, sweet premium brand.*

How you can use SPSS Categories Use correspondence analysis to analyze two-way contingency tables or data that can be expressed as a two-way table, such as brand preference or sociometric choice data. Correspondence analysis describes the relationship between two nominal variables in a low-dimensional space, while simultaneously describing the relationship between categories for each variable. For example, you can use correspondence analysis to graphically display the relationship between staff category and smoking habits. You might nd, with regard to smoking, junior managers differ from secretaries, but secretaries do not differ from senior managers. You might also nd that heavy smoking is associated with junior managers, whereas light smoking is associated with secretaries. Categorical regression predicts the values of a categorical dependent variable from a combination of categorical independent variables. Regression with optimal scaling quanties categorical variables using optimal scaling, resulting in an optimal linear regression equation for the transformed variables. You can use regression with optimal scaling to describe, for example, how job satisfaction depends on job category, geographic region and amount of workrelated travel. You may nd high levels of satisfaction correspond to managers and low travel. You could use the resulting regression equation to predict job satisfaction for any combination of the three independent variables.
You can use homogeneity analysis, also known as multiple correspondence analysis, to analyze a categorical multivariate data matrix when you are willing to make no stronger assumption than that all variables are analyzed at the nominal level. Homogeneity analysis is similar to correspondence analysis but doesnt limit you to two variables. Homogeneity analysis quanties nominal data by assigning numerical values to the cases and categories. Its goal is to describe the relationships between two or more nominal variables in a low-dimensional space containing the variable categories as well as the objects in those categories. For example, you can use homogeneity analysis to graphically display the relationship between job category, minority classication and gender. You might nd that minority classication and gender discriminate between people, but that job category does not. You might also nd the Latino and African-American categories are similar to each other. PROXSCAL helps you assign observations to specic locations in conceptual space. You can represent similarity and dissimilarity data in a low-dimensional space to gain spatial understanding of how objects relate. CATPCA, OVERALS and CATREG enable you to specify a transformation type of nominal, ordinal or numeric on a variable-by-variable basis. All three procedures use an alternating least-squares algorithm. CATPCA generalizes principal components analysis to accommodate variables of mixed measurement levels. OVERALS generalizes canonical correlation analysis, while CATREG does the same thing for multiple regression.

* Source for data and example: Kennedy, R., Riquier, C., and Sharp, Byron. 1996. Practical Applications of Correspondence Analysis to Categorical Data in Market Research, Journal of Targeting, Measurement and Analysis for Marketing, Vol. 5, No. 1, pp. 56-70.
SPSS Categories 11.0 in action a real-life case study
With SPSS Categories 11.0, you can easily incorporate outside data by using supplementary points. To illustrate this, we will use an example presented in Michael Greenacres book, Theory and Applications of Correspondence Analysis. Greenacre presents and analyzes the two-variable table relating staff group to smoking category in a particular workplace (see Figure 1).
Figure 1. With SPSS Categories, you can use correspondence analysis to easily display the relationship between staff category and smoking habits.
Greenacre also presents supplementary row-and-column information, which can be represented in the space of staff group and smoking category. First, suppose you know the levels of smoking nationwide. You can show the nationwide average in the space of staff group by smoking category. Second, suppose you have additional information on staff group versus whether a person consumes alcoholic beverages or not. You can see this information in Figure 2. These additional columns can be projected into the staff group by smoking category space. Figure 3 shows the biplot from SPSS Correspondence procedure.
Figure 3. With SPSS Categories, you can use symmetrical normalization if you are primarily interested in the differences or similarities between two variables. This is usually the preferred method for making biplots.
Regarding the nationwide estimate point in Figure 3, note that the point lies midway between none and light smoking. This indicates the sample consists of generally high proportions of smokers compared to the nationwide average. When you look at the consumption of alcohol, note that consume alcohol beverages yes and consume alcohol beverages no are aligned with the second (vertical) axis, while no smoking versus various levels of smoking are aligned with the rst (horizontal) axis. From this information, Greenacre concludes that a strong association does not exist in the sample between non-drinking and non-smoking, and there is a suggested association between drinking and level of smoking with relatively more drinkers in the high-smoking group.

Figure 2. Easily incorporate supplementary information on additional variables in SPSS Categories 11.0. Here, look at alcohol consumption by staff group.
Greenacre, M. 1984. Theory and Applications of Correspondence Analysis. New York: Academic Press.

SPSS Categories features

PROXSCAL Multi-dimensional scaling of proximity data to nd a least-squares representation of the objects in low dimension space s Read one or more square matrices of proximities, either symmetrical or asymmetrical, or provide specications for matrices with proximities in a stacked format s Read weights, initial congurations, xed coordinates and independent variables s Treat proximities as ordinal (nonmetric) or numeric (metric); ordinal transformations can treat tied observations as discrete or continuous s Specify multi-dimensional scaling with three individual differences models, as well as the identity model s Specify xed coordinated or independent variables to restrict the conguration or specify the transformations (numerical, nominal, ordinal and splines) for independent variables s Specify output that includes the original and transformed proximities, history of iterations, individual space weights, distances and decomposition of the stress CATPCA Principal components analysis via alternating least squares s Specify the optimal scaling level at which you want to analyze each variable: spline ordinal, spline nominal, ordinal, nominal, multiple nominal or numerical s Discretize fractional value variables or to recode categorical variables s Specify how you want missing data handled: pairwise exclusion, listwise exclusion or impute missing data s Specify objects and variables to be treated as supplementary s Specify the number of dimensions s Five options for normalizing the object scores and variables: association between variables, distances between objects, symmetrical, independent, user specied s Specify the maximum number of iterations s Specify the convergence criteria s Print results: model summary; history statistics; descriptive statistics; variance accounted for per variable and per dimension; component loadings for variables with optimal scaling level that results in vector quantication; category quantications and category coordinates for each dimension; history of iterations; correlations of the transformed variables and the eigenvalues of this correlation matrix; correlations of the original variables and the eigenvalues of this correlation matrix; object (component) scores Plot results: plots of the object points and category points; plot of component loadings, transformation points, residuals per variable, objects and variables; object points and component loadings; object points and centroids; plot of object points, component loadings for variables with an optimal scaling level that results in vector quantication; centroids for variables with optimal scaling level MNOM; joint plot of category points for variables in varlist; plot of centroids of variable varname projected on each variable in varlist; plots for one-dimensional solutions Create maximum exibility in the biplots using two variable lists for the LOADING keyword in the PLOT subcommand (one for loadings and one for centroids)

SPSS Categories features continued
Add variables to the working data le or write variable to external data le; transformed variables, the object scores and the approximation for variables that do not have optimal scaling level Full output is included for categories that occur only for supplementary objects, especially for centroids that classify supplementary objects into groups

New New

CORRESPONDENCE Correspondence analysis, related singular value decomposition of a two-way array of nonnegative numbers s Input data as a case le or directly as table input s Specify the number of dimensions of the solution s Choice of distance measure Chi-square distances for correspondence analysis Euclidean distances for more biplot analysis types s Five types of standardization: remove row means; remove column means; remove row-and-column means; equalize row totals, remove row means; equalize column totals; remove column means s Five types of normalization: symmetrical; row principal; column principal; customized s Print results Table Summary table (singular values, inertia, proportion of inertia accounted for by the dimension, cumulative proportion of inertia accounted for by the dimensions, condence statistics for the maximum number of dimensions; row proles; column proles) Overview of row-and-column points: mass, scores, inertia, contribution of the points to the inertia of the dimension, contribution of the dimensions to the inertia of the points Row-and-column condence statistics: standard deviations and correlations for the active row points Permutated table: table with rows and columns ordered by the row-and-column scores for a given dimension s Plot results: row scores; column scores; biplot (joint plot of the row-and-column scores) s Output les Row-and-column scores, variances, covariances to a matrix data le CATREG Categorical regression analysis via optimal scaling s Indicate transformation type for each variable: nominal; ordinal; numerical; monotonic; nonmonotonic spline

New New New

Specify the method used to compute the initial conguration Control the number of iterations Specify the convergence criterion Specify missing value treatment: listwise exclusion or replace with mode, creating an additional category Choose treatment for each variable separately Discretization options for continuous variables. Get the ability to group original scores into a pre-selected number of categories according to an optimal distribution (normal, uniform), a xed interval or multiplying. Print results Multiple R, R2, adjusted R2 Standardized regression coefcients, standard errors, zero-order correlation, part correlation, partial correlation, Pratts relative importance measure for the transformed predictors, tolerance before and after transformation, F statistics Table of descriptive statistics including marginal frequencies, transformation type, number of missing values and mode Iteration history Tables for t and model parameters: ANOVA table with degrees of freedom according to optimal scaling level; Model Summary table with Adjusted RSQ for Optimal Scaling; t-values; signicance levels; a separate table with the zero-order, part and partial correlation, the importance and tolerance before and after transformation Correlations of the transformed predictors q Eigenvalues Correlations of the original predictors q Eigenvalues Category quantications Transformed data list Plot of the optimal quantications versus the original category values Plots that display the optimal, nonlinear transformations with respect to the residuals in the regression model Write the transformed variables to an external data le Fit predicted values for supplementary objects by using regression and optimal scaling results for the active objects Write discretized data to an external le Save to the working le: transformed variables, predicted variables, residuals Number of categories no longer to be declared Use long strings as input Added functionality by allowing user/system missing

HOMALS Homogeneity analysis via alternating least squares, also known as multiple correspondence analysis s Specify the number of dimensions s Specify maximum number of iterations s Specify convergence criterion s Print results: marginal frequencies for variables in the analysis; iteration history; eigenvalues; discrimination measures; object scores; category quantications s Plot results: discrimination measures; object scores; category quantications s Write category quantications to a matrix system le OVERALS Canonical correlation analysis of two or more sets of variables via alternating least squares s Indicate transformation type for each variable: multiple nominal; single nominal; ordinal; numerical s Specify the method used to compute the initial conguration s Specify number of sets, and which variables belong in each set s Specify the number of dimensions s Specify maximum number of iterations s Specify convergence criterion s Print results: marginal frequencies for variables in the analysis; iteration history; multiple t, single t and single loss per variable; category quantication scores, projected centroids and centroids; object scores; category quantications and single, multiple coordinates; weights, component loadings s Plot results: category quantications; category coordinates; category centroids s Write category quantications, coordinates, centroids, weights and component loadings to a matrix system le SYSTEM REQUIREMENTS s SPSS Base 11.0 s 1MB hard drive space s Other system requirements vary according to platform

Learn more about SPSS

Contact your nearest SPSS ofce or visit us at www.spss.com.
SPSS Inc. Toll-free SPSS Argentina SPSS Asia Pacic SPSS Australasia Toll-free SPSS Belgium SPSS Benelux SPSS Brasil SPSS Czech Republic SPSS Danmark +1.312.651.3000 +1.800.543.2185 +5411.4814.5030 +65.245.9110 +61.2.9954.5660 +1.800.024.836 +32.16.317070 +31.183.651.777 +55.11.5505.3644 +420.2.24813839 +45.45.46.02.00 SPSS East Africa +254.2.577.262 SPSS Federal Systems +1.703.527.6777 Toll-free +1.800.860.5762 SPSS Finland +358.9.4355.920 SPSS France +01.55.35.27.00 SPSS Germany +49.89.4890740 SPSS Hellas +30.1.72.51.925 SPSS Hispanoportuguesa +34.91.447.37.00 SPSS Hong Kong +852.28119662 SPSS Ireland +353.1.415.0234 SPSS Israel +972.3.6166616 SPSS Italia SPSS Japan SPSS Korea SPSS Latin America SPSS Malaysia SPSS Mexico SPSS Miami SPSS Norway SPSS Polska SPSS Russia SPSS San Bruno +800.437300 +81.3.5466.5511 +82.2.3446.7651 +1.312.651.3539 +603.6203.2300 +52.5.682.87.68 +1.305.627.5700 +47.22.40.20.60 +48.12.6369680 +7.095.125.0069 +1.650.794.2692 SPSS Schweiz SPSS Singapore SPSS South Africa SPSS South Asia SPSS Sweden SPSS Taiwan SPSS Thailand SPSS UK +41.1.266.90.30 +65.324.5150 +27.21.7120929 +91.80.2088069 +46.8.506.105.50 +886.2.25771100 +66.2.260.7070 +44.1483.719200

In addition to these ofces, SPSS has a worldwide network of distributors. Contact the SPSS ofce nearest you for assistance.

About SPSS Inc.

SPSS Inc., (Nasdaq: SPSS) headquartered in Chicago, IL, USA, is a worldwide provider of analytical technology for business, government, and higher education. The company's solutions and products enable customers to make better decisions by learning from the past, understanding the present and anticipating the future. With this insight, organizations gain a true competitive advantage: the ability to manage the future. SPSS analytical technology is brought to the market through ve divisions: CustomerCentric Solutions (for integrated analytical CRM solutions); SPSS BI (for data mining and statistical products and services used to solve business problems); ShowCase (for analytical products operating on IBM iSeries/AS400 platform); SPSS MR (for analytical solutions in the market research industry); and SPSS Enabling Technologies (for licensing SPSS technologies for use in other analytical applications). For more information, visit www.spss.com.
SPSS is a registered trademark and the other SPSS products named are trademarks of SPSS Inc. All other names are trademarks of their respective owners. Printed in the U.S.A. Copyright 2001 SPSS Inc. SCT11SPC-0701