Powerful modeling techniques for analyzing complex relationships

IBM SPSS Advanced Statistics provides univariate and multivariate modeling techniques to help users reach the most accurate conclusions when working with data describing complex relationships. These sophisticated analytical techniques are frequently applied to gain deeper insights from data used in disciplines such as medical research, manufacturing, pharmaceuticals and market research.

SPSS Advanced Statistics provides the following capabilities:

• General linear models (GLM) and mixed models procedures.

• Generalized linear models (GENLIN) including widely used statistical models, such as linear regression for normally distributed responses, logistic models for binary data and loglinear models for count data.

• Linear mixed models, also known as hierarchical linear models (HLM), which expands the general linear models used in the GLM procedure so that you can analyze data that exhibit correlation and non-constant variability.

• Generalized estimating equations (GEE) procedures that extend generalized linear models to accommodate correlated longitudinal data and clustered data.

• Generalized linear mixed models (GLMM) for use with hierarchical data and a wide range of outcomes, including ordinal values.

• Survival analysis procedures for examining lifetime or duration data.

General linear models (GLM)

• Describe the relationship between a dependent variable and a set of independent variables. Models include linear regression, analysis of variance (ANOVA), analysis of covariance (ANCOVA), multivariate analysis of variance (MANOVA) and multivariate analysis of covariance (MANCOVA).

• Use flexible design and contrast options to estimate means and variances and to test and predict means.

• Mix and match categorical and continuous predictors to build models, choosing from many model-building possibilities.

• Use linear mixed models for greater accuracy when predicting nonlinear outcomes, such as what a customer is likely to buy, by taking into account hierarchical and nested data structures.

• Formulate dozens of models, including split-plot design, multi-level models with fixed-effects covariance and randomized complete blocks design.

Generalized linear models (GENLIN)

• Provide a unifying framework that includes classical linear models with normally distributed dependent variables, logistic and probit models for binary data, and loglinear models for count data, as well as various other nonstandard regression-type models.

• Apply many useful general statistical models including ordinal regression, Tweedie regression, Poisson regression, Gamma regression and negative binomial regression

Linear mixed models/hierarchical linear models (HLM)

• Model means, variances and covariances in data that display correlation and non-constant variability, such as students nested within classrooms or consumers nested within families.

• Formulate dozens of models, including split-plot design, multi-level models with fixed-effects covariance, and randomized complete blocks design.

• Select from 11 non-spatial covariance types, including first-order ante-dependence, heterogeneous, and first-order autoregressive.

• Get more accurate results when working with repeated measures data, including situations in which there are different numbers of repeated measurements, different intervals for different cases, or both.

Generalized estimating equations (GEE) procedures

• Extend generalized linear models to accommodate correlated longitudinal data and clustered data.

• Model correlations within subjects.

Generalized linear mixed models (GLMM)

• Access, manage and analyze virtually any kind of data set including survey data, corporate databases or data downloaded from the web.

• Run the GLMM procedure with ordinal values to build more accurate models when predicting nonlinear outcomes such as whether a customer’s satisfaction level will fall under the low, medium or high category.

Survival analysis procedures

• Choose from a flexible and comprehensive set of techniques for understanding terminal events such as part failure, death or survival rates.

• Use Kaplan-Meier estimations to gauge the length of time to an event.

• Select Cox regression to perform proportional hazard regression with time-to-response or duration response as the dependent variable.

Create more reliable models and generate more accurate results

IBM SPSS Bootstrapping is an efficient way to ensure that analytical models are reliable and will produce accurate results. It can be used to test the stability of analytical models and procedures found throughout the SPSS Statistics product family, including descriptive, means, crosstabs, correlations, regression and many others.

SPSS Bootstrapping enables you to:

• Quickly and easily estimate the sampling distribution of an estimator by re-sampling with replacement from the original sample.

• Create thousands of alternate versions of a data set for a more accurate view of what is likely to exist in the population.

• Reduce the impact of outliers and anomalies, helping to ensure the stability and reliability of your models.

• Estimate the standard errors and confidence intervals of a population parameter such as the mean, median, proportion, odds ratio, correlation coefficient, regression coefficient and more.

Predict outcomes and reveal relationships in categorical data

IBM SPSS Categories makes it easy to visualize and explore relationships in your data and predict outcomes based on your findings. Using advanced techniques, such as predictive analysis, statistical learning, perceptual mapping and preference scaling, you can understand which characteristics consumers relate most closely to your product or brand, and learn how they perceive your products in relation to others.

SPSS Categories includes advanced analytical techniques to help you:

• Easily analyze and interpret multivariate data and its relationships more completely.

• Turn qualitative variables into quantitative ones by performing additional statistical operations on categorical data.

• Graphically display underlying relationships in whatever types of categories you study, including market segments, medical diagnoses, political parties or biological species.

Easily analyze and interpret multivariate data

• Use categorical regression procedures to predict the values of a nominal, ordinal or numerical outcome variable from a combination of numeric and (un)ordered categorical predictor variables.

• Quantify the variables to maximize the Multiple R with optimal scaling techniques.

• Clearly see relationships in your data using dimension reduction techniques such as perceptual maps and biplots.

• Gain insight into relationships among more than two variables with summary charts that display similar variables or categories.

Turn qualitative variables into quantitative ones

• Predict the values of a nominal, ordinal or numerical outcome variable from a combination of categorical predictor variables.

• Analyze two-way tables that contain some measurement of correspondence between rows and columns, as well as display rows and columns as points in a map. Also analyze multivariate categorical data by allowing the use of more than two variables in your analysis.

• Use optimal scaling to generalize the principal components analysis procedure so that it can accommodate variables of mixed measurement levels.

• Compare multiple sets of variables to one another in the same graph after removing the correlation within sets, and visually examine relationships between two sets of objects; for example, consumers and products.

• Perform multidimensional scaling of one or more matrices with similarities or dissimilarities (proximities).

Graphically display underlying relationships

• Place the relationships among your variables in a larger frame of reference with optical scaling.

• Create perceptual maps that graphically display similar variables or categories close to each other for unique insights into relationships between more than two categorical variables.

• Use biplots and triplots to look at the relationships among cases, variables and categories; for example, to define relationships between products, customers and demographic characteristics.

• Further visualize relationships among objects using preference scaling, which helps you perform non-metric analyses for ordinal data and obtain more meaningful results.

• Analyze similarities between objects and incorporate characteristics for objects in the same analysis

Analyze statistical data and interpret survey results from complex samples

IBM SPSS Complex Samples helps market researchers, public opinion researchers and social scientists make more statistically valid inferences by incorporating sample design into their survey analysis. SPSS Complex Samples provide the specialized planning tools and statistics you need when working with complex sample designs, such as stratified, clustered or multistage sampling.

• Incorporate sample design into survey analysis for more accurate results.

• Retain survey planning parameters for future use to speed analysis and increase efficiency.

• Manage complex survey data for thorough, detailed analysis.

• Use an intuitive interface and helpful wizards to analyze data and interpret survey results faster.

Incorporate sample design into survey analysis

• Increase the precision of your sample or ensure a representative sample from key groups.

• Select clusters or groups of sampling units to make your surveys more cost-effective.

• Employ multistage sampling to select a higher-stage sample.

Retain survey planning parameters for future use

• Publish public-use data sets that include your sampling and analysis plans.

• Use published plans as a template in order to save decisions made when creating the plan.

• Make plans available to others in the organization so they can replicate results or pick up where you left off.

Manage complex survey data

• Display one-way frequency tables or two-way cross-tabulations and associated standard errors, design effects, confidence intervals and hypothesis tests.

• Build linear regression, analysis of variance (ANOVA) and analysis of covariance (ANCOVA) models.

• Estimate means, sums and ratios, and compute standard errors, design effects confidence intervals and hypothesis tests for samples drawn by complex sampling methods.

• Perform binary logistic regression analysis and multiple logistic regression (MLR) analysis.

• Apply Cox proportional hazards regression to analysis of survival times.

Use an intuitive interface and helpful wizards

• Use the Analysis Preparation Wizard to specify how the samples are defined and how standard errors should be estimated.

• When creating your own samples, use the Sampling Plan Wizard to define the scheme and draw the sample.

• Use the IBM SPSS Complex Samples Selection (CSSELECT) procedure to select complex, probability-based samples from a population while mitigating the risk of over-representing or under-representing a subgroup.

Understand and measure purchasing decisions

IBM SPSS Conjoint helps market researchers increase their understanding of consumer preferences so they can more effectively design, price and market successful products. It enables them to model the consumer decision-making process so they can design products with the features and attributes most important to their target market.

SPSS Conjoint includes procedures that can help researchers:

• Design an orthogonal array of product attribute combinations using ORTHOPLAN, a design generator.

• Produce and print cards that study respondents can use to sort, rank or rate alternative products.

• Analyze research data using conjoint analysis, a specially tailored version of regression.

Design an orthogonal array of product attribute combinations

• Reduce the number of questions asked while ensuring enough information to perform a full analysis.

• Generate orthogonal main effects fractional factorial designs; ORTHOPLAN is not limited to two-level factors.

• Specify a variable list, optional variable labels, a list of values for each variable and optional value labels.

• Generate holdout cards to test the fitted conjoint model.

• Specify the desired number of cards for the plan.

Produce and print cards

• Use the PLANCARDS utility procedure to generate printed cards for use as stimuli by respondents.

• Specify the variables to be used as factors and the order in which their labels are to appear in the output.

• Choose listing-file formats and card formats.

• Display output in pivot tables.

Analyze research data

• Perform an ordinary least-squares analysis of preference or rating data with the conjoint procedure.

• Work with the plan file generated by PLANCARDS, or a plan file input by the user using a data list.

• Work with individual-level rank or rating data.

• Provide individual-level and aggregate results.

• Select from three conjoint simulation methods: max utility, Bradley-Terry-Luce (BTL) and logit.

Analyze your data with custom tables created in less time

IBM SPSS Custom Tables makes it easy to summarize IBM SPSS Statistics data in different styles for different audiences. It combines analytical capabilities to help you learn from your data with features that allow you to build tables people can easily read and interpret.

This software is useful for anyone who creates and updates reports on a regular basis, especially those who work in survey or market research, the social sciences, database or direct marketing and institutional research.

SPSS Custom Tables includes capabilities to help you:

• Get in-depth analyses so you can understand your data better and enhance reports for decision makers.

• Preview tables as you build them, ensuring that you create polished, accurate reports in less time.

• Customize table layout and appearance to communicate results clearly and accurately.

• Make results easily available by delivering information people can act on without further processing.

Get in-depth analyses

• Include inferential statistics, also known as significance testing, to highlight opportunities or problem areas.

• Compare means or proportions for demographic groups, customer segments, time periods or other categorical variables.

• Identify trends, changes or major differences in your data.

• Show the results of significance tests directly in SPSS Custom Tables output.

• Select from various summary statistics, including categorical variables, measures of dispersion, multiple response sets, scale variables and custom total summaries for categorical variables.

Preview tables as you build them

• Drag and drop variables onto the table builder and view them in a preview pane before adding them to your tables.

• Interact with the variables on your screen and know immediately how your data is structured.

• Move variables from row to column for precise positioning; add, swap and nest variables; or hide variable labels from within the table preview builder.

• Collapse large or complex tables for a more concise view and still see your variables.

• Preview the arrangement of variables, including dimensions, stacking or nesting, as well as the categories of each variable and requested statistics.

Customize table layout and appearance

• Create totals and subtotals without changing your data file and sort categories within your table without affecting the subtotal calculation.

• Change variable types, exclude categories, sort categories by any summary statistic and hide the categories that comprise subtotals.

• Add titles and captions, use table expressions in titles and specify minimum and maximum column widths during table creation.

• Select from pre-formatted styles found within IBM SPSS Statistics Base or create your own styles.

• Add scripts to automate formatting and other repetitive tasks through SPSS Statistics Base.

Make results easily available

• Create customized tabular reports suitable for a variety of audiences, including those without a technical background.

• Use syntax and automation to run frequently needed reports in production mode, or to create reports with the same structure.

• Produce all results as IBM SPSS pivot tables that can be easily exported to Microsoft® Word, Excel or HTML with formatting intact.

IBM SPSS Data Preparation performs advanced techniques that streamline the data preparation stage of the analytical process to deliver faster, more accurate data analysis results. Analysts can choose from a completely automated data preparation procedure for the fastest results, or select from several other methods to help prepare more challenging data sets.

With this software, you can easily identify suspicious or invalid cases, variables and data values. You can also view patterns of missing data, summarize variable distributions and more accurately work with algorithms designed for nominal attributes.

SPSS Data Preparation helps:

• Automate the data preparation process—to eliminate complex, time-consuming manual data preparation.

• Validate data without manual checks—to perform faster, more accurate data validation.

• Prevent outliers from skewing analyses—to automatically detect anomalies that could corrupt results.

Automate the data preparation process

• Prepare data in a single step.

• Detect and correct quality errors and impute missing values.

• Quickly determine which data to use in your analysis.

• View easy-to-understand reports with recommendations and visualizations.

Validate data without manual checks

• Ensure consistency of data validation from project to project.

• Apply validation rules based on each variable’s measure level (categorical or continuous).

• Receive reports of invalid cases, rule violation summaries and number of cases affected.

• Remove or correct suspicious cases at your discretion before analysis.

Prevent outliers from skewing analyses

• Search for unusual cases based upon deviations from similar cases.

• Flag outliers by creating a new variable.

• Examine unusual cases to determine if they should be included in analyses.

Easily identify groups and predict outcomes

IBM SPSS Decision Trees helps you better identify groups, discover relationships between them and predict future events. This module features highly visual classification and decision trees that enable you to present categorical results in an intuitive manner, so you can more clearly explain categorical analysis to non-technical audiences. It includes four tree-growing algorithms, giving you the ability to try different types and find the one that best fits your data.

The module provides specialized tree-building techniques for classification within the IBM SPSS Statistics environment. The four tree-growing algorithms include:

• CHAID - a fast, statistical, multi-way tree algorithm that explores data quickly and efficiently, and builds segments and profiles with respect to the desired outcome.

• Exhaustive CHAID - a modification of CHAID, which examines all possible splits for each predictor.

• Classification and regression trees (C&RT) - a complete binary tree algorithm that partitions data and produces accurate homogeneous subsets.

• QUEST - a statistical algorithm that selects variables without bias and builds accurate binary trees quickly and efficiently.

Easily identify the right customers and improve campaign results

IBM SPSS Direct Marketing helps you understand your customers in greater depth, improve your marketing campaigns and maximize the ROI of your marketing budget. Conduct sophisticated analyses of your customers or contacts easily – and with a high level of confidence in your results. Choose from recency, frequency and monetary value (RFM) analysis, cluster analysis, prospect profiling, postal code analysis, propensity scoring and control package testing.

SPSS Direct Marketing enables database and direct marketers to:

• Identify which customers are likely to respond to specific promotional offers.

• Develop a marketing strategy for each customer group.

• Compare the effectiveness of direct mail campaigns.

• Boost profits and reduce costs by mailing only to those customers most likely to respond.

• Identify by postal code the responses to your campaigns.

• Connect to Salesforce.com to extract customer information, collect details on opportunities and perform analyses.

Accurately analyze small data sets or those with rare occurrences

IBM SPSS Exact Tests enables you to use small samples and still feel confident about the results. If you have a small number of case variables with a high percentage of responses in one category, or have to subset your data into fine breakdowns, traditional tests could be incorrect. SPSS Exact Tests eliminates this risk.

With SPSS Exact Tests you can:

• Run a test at any time with just the click of a button.

• Choose from more than 30 exact tests, which cover the entire spectrum of nonparametric and categorical data problems for small or large data sets, contingency tables and on measures of association.

• Slice and dice your data into breakdowns. You aren't limited by required expected counts of five or more per cell for correct results.

• Search for rare occurrences within large data sets.

• Keep your original design or natural categories - for example, regions, income or age groups—and analyze what you intended to analyze.

Build expert forecasts—in a flash

IBM SPSS Forecasting enables analysts to predict trends and develop forecasts quickly and easily—without being an expert statistician. People new to forecasting can create sophisticated forecasts that take into account multiple variables, and experienced forecasters can use SPSS Forecasting to validate their models. You get the information you need faster because the software helps you every step of the way.

SPSS Forecasting offers:

• Advanced statistical techniques you need to work with time-series data regardless of your level of expertise.

• Procedures to help you get the most from your time-series analysis.

Advanced statistical techniques

• Analyze historical data, predict trends faster and deliver information in ways that your organization’s decision-makers can understand and use.

• Automatically determine the best-fitting ARIMA or exponential smoothing model to analyze your historic data.

• Model hundreds of different time series at once, rather than having to run the procedure for one variable at a time.

• Save models to a central file so forecasts can be updated when data changes, without having to reset parameters or re-estimate models.

• Write scripts so models can be updated with new data automatically.

Procedures

• TSMODEL—use the Expert Modeler to model a set of time-series variables, using either ARIMA or exponential smoothing techniques.

• TSAPPLY—apply saved models to new or updated data.

• SEASON—estimate multiplicative or additive seasonal factors for periodic time series.

• SPECTRA—decompose a time series into its harmonic components, which are sets of regular periodic functions at different wavelengths or periods.

Build better models when you estimate missing data

IBM SPSS Missing Values software is used by survey researchers, social scientists, data miners, market researchers and others to validate data. The software allows you to examine data to uncover missing data patterns, then estimate summary statistics and impute missing values using statistical algorithms.

With SPSS Missing Values software, you can impute your missing data, draw more valid conclusions and remove hidden bias.

• Quickly diagnose missing data imputation problems using diagnostic reports.

• Replace missing data values with estimates using a multiple imputation model.

• Display and analyze patterns to gain insight and improve data management.

Quickly diagnose missing data imputation problems

• Examine data from different angles using six diagnostic reports.

• Diagnose missing data using the data patterns report, which provides a case-by-case overview of your data.

• Determine the extent of missing data and any extreme values for each case.

Replace missing data values with estimates

• Understand missing patterns in your data set and replace missing values with plausible estimates.

• Benefit from an automatic imputation model that chooses the most suitable method based on characteristics of your data, or customize your imputation model.

• Model the individual data sets that are created, using techniques such as linear regression or expectation maximization algorithms, to produce parameter estimates for each.

• Obtain final parameter estimates by pooling estimates and computing inferential statistics that take into account variation within and between imputations.

Display and analyze patterns

• Display missing data for all cases and all variables using the data patterns table.

• Determine differences between missing and non-missing groups for a related variable with the separate t-test table.

• Assess how much the missing data for one variable relates to the missing data of another variable using the percent mismatch of patterns table.

Improve the accuracy of predictions with advanced regression procedures

IBM SPSS Regression software enables you to predict categorical outcomes and apply a range of nonlinear regression procedures. You can apply the procedures to business and analysis projects where ordinary regression techniques are limiting or inappropriate—such as studying consumer buying habits, responses to treatments or analyzing credit risk.

With SPSS Regression software, you can expand the capabilities of IBM SPSS Statistics Base for the data analysis stage in the analytical process.

• Predict categorical outcomes with more than two categories using multinomial logistic regression (MLR).

• Easily classify your data into groups using binary logistic regression.

• Estimate parameters of nonlinear models using nonlinear regression (NLR) and constrained nonlinear regression (CNLR).

• Meet statistical assumptions using weighted least squares and two-stage least squares.

• Evaluate the value of stimuli using probit analysis.

Predict categorical outcomes

• Using MLR, regress a categorical dependent variable with more than two categories on a set of independent variables. This helps you accurately predict group membership within key groups.

• Use stepwise functionality, including forward entry, backward elimination, forward stepwise or backward stepwise, to find the best predictor.

• For a large number of predictors, use Score and Wald methods to help you quickly reach results.

• Assess your model fit using Akaike information criterion (AIC) and Bayesian information criterion (BIC).

Easily classify your data

• Using binary logistic regression, build models in which the dependent variable is dichotomous; for example, buy versus not buy, pay versus default, graduate versus not graduate.

• Predict the probability of events such as solicitation responses or program participation.

• Select variables using six types of stepwise methods. This includes forward (select the strongest variables until there are no more significant predictors in the data set) and backward (at each step, remove the least significant predictor in the data set).

• Set inclusion or exclusion criteria.

Estimate parameters of nonlinear models

• Estimate nonlinear equations using NLR for unconstrained problems and CNLR for both constrained and unconstrained problems.

• Using NLR, estimate models with arbitrary relationships between independent and dependent variables using iterative estimation algorithms.

• With CNLR, use linear and nonlinear constraints on any combination of parameters.

• Estimate parameters by minimizing any smooth loss function (objective function), and compute bootstrap estimates of parameter standard errors and correlations.

Meet statistical assumptions

• If the spread of residuals is not constant, use weighted least squares to estimate the model. For example, when predicting stock values, stocks with higher share values fluctuate more than low-value shares.

• Use two-stage least squares to estimate the dependent variable when the independent variables are correlated with regression error terms. This allows you to control for correlations between predictor variables and error terms.

Evaluate the value of stimuli

• Use probit analysis to estimate the effects of one or more independent variables on a categorical dependent variable.

• Evaluate the value of stimuli using a logit or probit transformation of the proportion responding.

IBM SPSS SamplePower software enables you to quickly find the right sample size for your research, and test the possible results before you begin your study. The software provides advanced statistical techniques such as means and differences in means, correlation, one-way and factorial analysis of variance (ANOVA), regression and logistical regression, survival analysis and equivalence tests.

SPSS SamplePower helps you compare the effects of different study parameters using analytical tools.

• Get precise answers quickly with flexible, efficient tools.

• Compare results using different combinations of research parameters.

• Determine sample sizes and their impact using Cohen’s effect size conventions.

Get precise answers quickly

• Provides features to make finding accurate sample sizes easy.

• Offers a “classic” interface for those familiar with power analysis and an easy-to-use interface that walks you through the common procedures.

• Explains terms and takes you through the steps to determine an effective sample size, allowing you to move forward with your research.

Compare results

• An interactive summary panel gives you concise summaries of power and precision at any point, so you can see how each decision affects your results.

• A scenarios tool lets you control the flow of your research. You can vary alpha level, power, effect size or sample size, and store results as you continue.

• Tables and graphs enable you to assess how different combinations of research parameters (such as alpha levels and duration) will affect your statistical power.

Determine sample sizes and their impact

• Provides Cohen's effect size conventions, which allow you to determine effect sizes for particular tests.

• Provides a rule-of-thumb for determining ambiguous small, medium and large effect sizes. You can vary these effect sizes to see how they affect power or precision.

• Lets you easily find the sample size for the default power setting.

• Gives you the flexibility to choose different power size settings to compare results.

Create compelling visualizations to communicate analytic results

IBM SPSS Visualization Designer enables you to develop and share customized data visualizations, from simple charts to advanced graphics. The solution makes the creation of data visualizations accessible to non-technical users. With a drag-and-drop design interface, it eliminates the need for advanced programming skills.

SPSS Visualization Designer allows you to visually portray and communicate analytics to others.

• Provides templates and an intuitive visual design interface to help you create visualizations quickly and easily.

• Supports different data sources including databases and IBM SPSS Statistics files.

Provides templates and an intuitive visual design interface

• Get started quickly with dozens of built-in visualization templates.

• Use an advanced visual designer for drag-and-drop graph creation.

• Extend the capabilities of built-in templates or create your own.

• Share style sheets and graph templates tailored to your organization.

• Deploy graphs in operational systems using IBM SPSS Collaboration and Deployment Services, SPSS Statistics and IBM SPSS Modeler.

Supports different data sources

• Provides support for delimiter-separated SPSS Statistics data files.

• Supports common database sources such as IBM DB2®, Microsoft SQL Server, Oracle and Sybase systems.