New from SPSS: SPSS 8.0; text analysis program
SPSS Inc., Chicago, has released SPSS 8.0 for Windows, which features an overhauled graphics system based on SPSS’ pivoting technology. Users can now visually explore their data by creating and editing graphs. In addition, SPSS 8.0 for Windows features more statistics, better data access and management, and improved help.
SPSS’ pivoting technology enables users to interact with "live" multidimensional tables and charts. Results can be seen from new angles by dragging and dropping variables. Users can actively explore their data as they drag and drop new variables onto their graph and watch their chart update, or change from one chart type to another with the click of a mouse. 8.0 includes new chart types, including 3-D bar charts, plottedpie charts and ribbon charts.
SPSS 8.0 for Windows users can determine if differences between multiple groups are statistically significant in experiments with an improved ANOVA. Users can perform all post-hoc analyses (in ONEWAY) and analyze their custom-made model with four types of sums of squares. SPSS 8.0 for Windows also features enhancements in other statistical procedures including: EXAMINE, CROSSTABS, MEANS, N-PAR TESTS and RELIABILITY.
Up to 256-character labels can be displayed in the dialog boxes, so whole survey questions or formula names can be more easily selected. Users can save and create prompted ODBC queries so they only need to answer a question such as "Which quarter?" to get automatic retrieval of data from the ODBC database.
The new Results Coach explains each element of the table, and highlights its analytical role on assessing report findings. With the "Ask Me" feature, by typing simple English statements such as "How do I make a bar chart?" users can find the help topics they need to proceed with their analysis.
SPSS has also released TextSmart, a software product which enables researchers to quantify and analyze a customer’s responses to questions such as Why do you shop here? and What can we do to improve our service? TextSmart uses statistical algorithms to cluster terms in responses that tend to occur together. The auto-categorize feature uses a matrix of similarities to create categories from these clusters. Users can set the auto-categorize feature to create categories from clusters only, a combination of clusters and individual terms, or individual terms only.
TextSmart creates the Automatic Categories Graph every time the auto-categorize feature is run. This graph uses color and distance to give an overall picture of how terms and categories cluster within the set of responses. To focus on the detail, the Responses table provides views of a single response, all responses or only responses that have certain characteristics.
Automated filtering tools prepare text for the creation of categories. TextSmart’s stemming tool finds all forms of words and combines them with the root word (try = try, tries, tried). Users can combine words into aliases with a few mouse clicks.
TextSmart does not require the user to have a background in text coding, text analysis or statistics. SPSS has designed a system with an Import Wizard that walks users through the process of importing their files for analysis. Online Help answers questions as they move through the process. Users can run the procedures they want with the click of a toolbar button or by using TextSmart’s complete menu system.
Skila debuts health care intelligence tool
Skila Inc., Mahwah, N.J., has introduced Skila, a turnkey competitive inteliigence tool for the health care industry. Using computer agent technology, it searches on-line information sources and pushes the information to the user’s desktop. The product helps marketers compile data on medical diagnoses and procedures, medical product and equipment suppliers, and health care providers, from periodicals, Internet sources, and public and private databases. It is designed to provide big-picture overviews of the health care market and more focused insights into specific enrifles or aspects. A demo version of the product is available in the "Our Products" section of the company’s Web site, http://www.skila.com.
Lite version of Web-Publisher
U.K.-based ISPC has released a lite version of its ite Web-Publisher program. The program will accept both the crosstabs and any other electronic documents and convert them to HTML format. These reports are readable by any Interact browser on any platform.
New product from Sawtooth Software
Sawtooth Software, Sequim, Wash., has released a new software product called ICE for estimating individual-level conjoint utilities from discrete choice (CBC) data. ICE stands for "individual choice estimation." Choice analysts have generally used agregate analysis, which is limiting from an analytical and theoretical approach. Individual-level modeling with ICE captures the heterogeneity in the data and can result in more accurate market share simulations. ILA (independence from irrelevant alternatives, aka red-bus/blue-bus) problems inherent to aggregate logit analysis are also less problematic with individual-level models.
Neural network application from StatSoft
StatSoft, Inc., Tulsa, Okla., has released a new data analysis application, STATISTICA: Neural Networks, a comprehensive application capable of desiging a range of neural network.architectures, employing both widelyused and highly-specialized training algorithms.
STATISTICA: Neural Networks offers features such as training algorithms, an Automatic Network Designer, a Neuro-Genefic Input Selection facility, complete API (Application Programming Interface) support, and the ability to interface with STATISTICA data files and graphs. STATISTICA is a comprehensive statistics package from StatSoft, Inc.
Typically, choosing the fight architecture of a neural network is a trial-and-error process, but STATISTICA: Neural Networks does this for the user via an Automatic Network Designer that utilizes heuristics and sophisticated optimization strategies to determine the best network architecture.
Neuro-Genetic Input Selection procedures aid in determining the input variables that should be used in training the network. It uses an optimization strategy to compare the possible combinations of input variables to determine which set is most effective.
Software codes open-ended responses
StatPac Inc., Minneapolis, has released Verbatim Blaster, a program to assist researchers in coding open-ended responses from surveys. Verbatim Blaster will automatically evaluate respondents’ answers and code them into consistent response categories with minimal help from a researcher. It works by examining respondents’ comments to extract key words and phrases, and then uses them to create the response categories. At the heart of the program is an artificial intelligence engine that checks grammar, spelling, and context.
Microtab updates crosstab software
Microtab, Inc., Roswell, Ga., has released a new version of its crosstabulation software. This release includes Z-test, t-test and chi-square statistical testing capabilities. One-tailed or two-tailed Z-tests for independent proportions can be performed on either adjacent or non-adjacent banner points. One-tailed or two-tailed Z-tests for independent proportions can be performed on either adjacent or non-adjacent banner points. One-tailed or two-tailed t-tests for independent means can likewise be provided on either adjacent or non-adjacent banner points. Additionally, chi-square testing for independent cells can be performed to test for significant differences. The crosstabulated tables that are produced will display the results of the requested statistical tests on the tables December 1997 www.quirks.com as they are printed - and there is no slow down in the compilation and printing time.
Survey of buying power now on-line
Market Statistics, New York, in partnership with Sales & Marketing Management magazine, has unveiled an Intemet site, Survey Of Buying Power On-Line. For an annual subscription fee users are granted unlimited access to its demographic reporting and mapping capabilities.
SBP On-Line allows users to analyze over 300 of Market Statistics’ demographic and economic data variables (including population, income, retail sales, and consumer purchasing data) for any U.S. market, such as states, MSAs, DMAs, counties, ZIP codes and block groups.
A planning tool available to marketing and research professionals for the past 68 years, the Survey of Buying Power (produced by Market Statistics and published by Sales & Marketing Management), has provided subscribers with retail sales forecasts, media planning, advertising sales, and site selection information.
Survey of Buying Power On-Line offers three levels of functionality to users, enabling them to tailor data to their specific needs. They can! generate consumer and economic reports for any market area; rank markets by any one of Market Statistics’ 300+ data variables; create presentation-quality market and trade area maps.