datadesk.gif (2520 bytes)

Data Desk is a fast, easy-to-use data analysis package that has been helping people understand their data since 1986. Data Desk provides interactive tools for data analysis and display based on the concepts and philosophy of Exploratory Data Analysis. Data Desk implements many traditional statistics techniques suitable for data from planned experiments and sample surveys. However, the program's true strength is its powerful tools for data exploration. These tools simplify intuitive examination of your data. No special training in statistics is needed for these insightful graphic displays. When you explore your data with Data Desk, you will find patterns and relationships. But you will also bring to light the elements that don't fit -- often the most important discovery you can make about your data.

Overview Summary ActivStats

 

OVERVIEW


Putting Exploratory Data Analysis to work for you means displaying your data in many related ways. It means computations fast enough to try out several alternative analyses in the time that you might have expected to spend on a single analysis. It means linking all these views of your data to get a deeper understanding of the patterns, relationships, and exceptions in your data.

Data Desk's interface is designed to be simple and efficient. Modify plots and tables directly with your mouse. Add or change variables with drag and drops. Plots and tables are stored in icons so that they can be opened and modified later.

Data Desk provides many sources of help and guidance. HyperView® menus, attached to every Data Desk plot and table, provide gentle guidance and suggestions. The on-line Help System gives quick access to information on the program's commands. The Quickstart Guide, Handbook and Statistics Guide offer descriptions of commands, background statistics information and many detailed examples. And free technical support is provided to all registered users.

Data Desk Plus provides an even greater level of assistance. This package combines Data Desk with ActivStats, our statistics refresh and reference multimedia CD. ActivStats uses video, narration, interactive tools, simulations, and web links to help you learn and understand statistics concepts. It also shows you how to apply those concepts to real-world data in Data Desk with special guided analysis files.

A significant amount of data is stored in spreadsheet programs like Microsoft Excel. Data Desk ships with an Excel add-in that moves data directly from Excel into Data Desk. The data are moved without the hassle of creating and opening an ASCII file.

Data Desk is also one of the fastest statistics packages available on a PC or a Macintosh. For example, a regression with three predictors and 50,000 datapoints run on a 133 MHz Pentium PC with 16 MB of Ram computes in 1 second. A scatterplot of two of those variables displays instantly and a 3-D rotating plot of three of those 50,000 datapoint variables rotates smoothly.

Data Desk works comfortably with large files - 10,000 datapoints, 100,000 datapoints, even a million datapoints. Data Desk's fast visual tools allow effective explorations through almost any size dataset.

Data Desk uses animation to help you see patterns you might miss in a static display. For example, you can easily link a sliding control to any part of an equation and see the effects of sliding the value as displays update. Data Desk automatically makes sliders to help you find optimal transformations of variables, to learn about the sensitivity of analyses to small shifts in variables, and to assess the sensitivity of nonlinear regressions. You can easily build your own animations.


The example above selects reexpressions of the variables Assets and Sales for data on large companies. Before transformation the variables are skewed. Moving these slider changes the reexpression of the variables, and the displays update to show the result.
 

Data Desk's direct manipulation interface provides "first person" interaction with your data rather than the "third person" interaction found in most programs. Instead of giving the computer a command (by typing it or filling in a dialog screen), you just pick up the appropriate tool or icon and do the job directly. So, if you want to reposition the points in a plot, don't tell Data Desk where to plot them verbally, pick them up with the hand tool and move them. If you want to identify a point, click on it. If you want to change the color or symbol of a point select it and click on the desired color or symbol in the palettes. If you want to rotate points in a rotating plot, push them. If you want a new variable on the x-axis of a plot, pick up its icon and place it on the axis. This direct interface lets you act more naturally, which simplifies and speeds analyses.


The example above shows the Query identifying a point, the Pointer tool selecting that point, a click in the symbols palette to change the display symbol, and the Hand tool repositioning the points to make it easier to see the point's y-axis value.

You can modify any Data Desk plot or table in place. Simply pick up a variable with your mouse, and drag it into the table. The table updates instantly. To change a variable in a plot, drag a new variable onto an axis.

Data Desk's dynamic plots and tables show patterns and structure among several variables. Rotating plots and sliders use animation to reveal aspects of your data that static plots cannot show. Changing the color, symbol and selection state of datapoints instantly changes all plots to show patterns, clusters, and outliers. Hot Set selections restrict displays to subsets of your data with a single click.

In the example below, the scatterplot of gas mileage by weight for a sample of automobiles is colored by the number of Cylinders, and a line fit to each colored subset. As different countries are selected by clicking on cells of the frequency table, the scatterplot shows only the selected cars. It is easy to see how the cars from different countries compare.


All Data Desk's plots and tables are linked together so that points selected in one display highlight in all other displays. Select bars in a bar chart or histogram, slices of a pie chart, rows, columns, or cells of a table, or any points in a scatterplot, rotating plot, dotplot, or boxplot, and the selected points highlight in all plots. Such consistency reveals relationships among several variables, identify clusters and subgroups and helps you to catch errors and outliers.

Colors and symbols are also linked across plots. Change the color or symbol of a group of points in one plot and the points change in all plots.

In the example above, points selected in a scatterplot of Miles per Gallon vs. Weight are highlighted in the bar chart of Country. Selecting the points in the high end of the weight scale shows which country makes the heaviest cars. Selecting the US bar in the bar chart shows how the cars built in the US compare to those built in other countries.

Any Data Desk analysis can become a template for similar analyses of other data. Data Desk's template tools make it easy to build a fully self-documented drag-and-drop interface for combinations of analyses and plots. Just do the analysis once to show Data Desk what you want. Data Desk's visual Action Control Language lets you add loops, conditional decisions, and other basic controls with no need ever to write a program or macro. Template files can be shared with others -- even with those using Data Desk on another operating system.


SUMMARY

Statistical Features:

  • Summary Statistics: for individual variables, broken down by categories and saved as variables: mean, median, midrange, midquartile range, biweight, standard deviation, variance, population standard deviation, standard error, interquartile range, range, sum of squares, sum, count, skewness, kurtosis, order statistics, rank, minimum and maximum.
  • Frequency Tables and Contingency Tables with chi-square and Fisher's Exact Test, expected values, standardized residuals, cumulative statistics and percents.
  • Standard Tests and Confidence Intervals: Including z-tests, t-tests, two-sample t-tests, paired t-tests, pooled t-tests, F-tests and a Chi-Square test of individual variances.
  • Nonparametric Tests: Sign, Paired Sign, Wilcoxon Signed Rank and Mann-Whitney U. Option to include or remove ties.
  • Correlation: Pearson Correlation, Spearman Rank Correlation, Kendall's Tau and Covariance.
  • Regression: Simple Regression, Multiple Regression, Weighted Regression, Polynomial Regression, Stepwise Regression. Full suite of diagnostic plots and statistics, all linked to update automatically if the data are changed. Diagnostics include: residuals - internally and externally studentized, leverage, predicted values, DFFits, Cook's distance, Hadi's Influence and Likelihood. Special diagnostic plots include partial regression plots, potential-residual plots, scatterplots of predicted values vs. residual and normal probability plots of all diagnostic statistics. Drag new response and predictor variables into the regression or remove them with mouse clicks; diagnostic statistics and plots update as well.
  • Nonlinear Regression: Fit virtually any function. Interactive visual tools help you find starting values and test the sensitivity of your results. Pre-built templates include: Exponential, Power, Growth, S, Log, lnverse, Quadratic, Cubic, or customize your own template.
  • ANOVA: Special one-step command to compute One-way and Multi-way, balanced or unbalanced, ANOVA and ANOVA with interactions.
  • General Linear Model: N-way ANOVA, ANCOVA, MANOVA, MANCOVA and Repeated Measures; fixed effect, random-effect and mixed models; unbalanced designs, nested designs, designs with missing cells and weighted analyses. Results include a complete ANOVA table, a cell means table and a coefficients table. Full set of diagnostic statistics and plots. Post-Hoc tests, including Scheffe, Bonferroni and Least Significant Difference. Modify the data, factor type or model and recompute the analysis in place. Unique OverView Table for multivariate models. Multivariate tests for MANOVA models include Pillai's trace, Wilk's Lambda, Hotteling-Lawley trace and Roy's Maximum root.
  • Logistic Regression, including support for discrete factors, interaction terms and unbalanced designs, likelihood, coefficients and Wald statistics.
  • Cluster Analysis with single or complete linkage clustering and interactive dendogram.
  • Principal Component Analysis: Saves principal axes as equations so "nearby" versions with simple coefficients are easy to find. The analysis computes a summary table and variables that hold the eigenvalues, eigenvectors and orthogonal basis vectors.
  • Transformations: more than 100 functions which include rounding, arithmetic, logical, summary, exponential, trigonometric, string, probabilities and many others. Select simple transformations from menus or type in arbitrary complex functions, including IF/THEN/FALSE. Transformed variables recompute automatically if the data are changed, so they are always up-to-date. Pre-fabricated, user-controlled, dynamic transformations include a Box-Cox transformation, Mix X and Y, Tukey's Lambda and Dynamic Lag.

Other Features

  • Unlimited number of cases and variables.
  • HyperView menus offer expert, context-sensitive suggestions.
  • Action language: automate calculation and implement new functions beyond the standard Data Desk menus. Easy-to-use, object-based programming interface. Many standard programming commands including: Let Y:=X, If, If else, While, Repeat, Loop, Done, Block, Abort, Call, Initialize, Update, Case Attributes, Random, Interface and more.
  • Templates: Create pre-built analyses for use with other data. lncorporate drag and drop interface, push button controls and notes for annotation and instruction.
  • Efficient and flexible data management with relational structure including random sampling, transposing, sorting, ranking, generating patterned data, generating random numbers, replicating, appending, splitting variables by group, variable tables and scalar tables.
  • Consistency and Updating: Windows offer to update if the underlying data are changed. Alternatively, view before vs. after side-by-side.
  • Data Importing: Copy' and Paste from spreadsheets, word processors, database programs and others. Read ASCII text files with arbitrary delimiters and Fixed Format files.
  • Combine Data Desk files; import the contents of a saved Data Desk file into another open file. Combine data, analyses, and derived variables.
  • Exporting: Export data (as ASCII text tables), plots (in color PICT on Mac or bitmap format on Windows) and computed results (as tab-delimited text or picture).
  • Compact datafiles dramatically reduces the size of datafiles to make them easier to store and share with others.
  • Graphical Control: Customized graphical controls 50 you can build your own dynamic graphics.
  • Full control over Font, Size and Scaling for all plots and tables (Mac only').
  • Preferences file to save defaults.
  • Selector button allows subset analysis.
  • Group button allows parallel category' analysis.
  • HotSet variables allow subsets and categories to be changed dynamically.
  • Colors can be added to any display by group if the variable is discrete or by rank if the variable is continuous.
  • Symbols: assign any plot symbol to the selected points in the plot. Data plotted with that symbol will show this symbol in all other plots.
  • Slide Show Presentations let you step through sets of windows, including text windows that provide commentary and advice.
  • Notebook Windows: Record history of analysis, group plots and text together for printing. Plot and tables placed into notebook are time-stamped and linked to original analyses.
  • Documentation:Tutorial Quickstart Guide, Handbook and Statistics Guide written by a statistics professor. All contain complete discussions of the statistical methods, examples (with data included), instructions and a full index.
  • Extremely fast computations and display.
  • The application takes up about 1 MB of disk space. Total installation is 4 MB. Requires any Mac running system 7.0 or better with 2 MB of free RAM. Any 486 or Pentium running Windows95 or Windows NT 4.0 or better.
  • Comprehensive Help system.
  • Free technical support for registered users.

Visit the Data Desk website to view descriptions and examples of some important Data Desk features, download a demo or contact Red Dog Scientific Services if you are based in Southern Africa for more information or a demo CD.

 

ActivStats - Learning Stats with Multimedia

ActivStats is a complete multimedia presentation of the material commonly found in an Introductory Statistics course. ActivStats uses a full range of multimedia to motivate, explain, visualize, and apply introductory statistics concepts. It integrates video, simulation, animation, narration, text, interactive experiments, web access, and Data Desk into a rich learning environment.

An ActivStats for Everyone

ActivStats Professional is designed to help you review the statistics you have forgotten from that course you once had (and hated) or to learn these valuable skills for the first time on your own. It includes a specially developed textbook written by David Moore, a world famous teacher and textbook author. Registered ActivStats Professional owners have the support of a web based "Teaching Assistant" free for six months to answer your questions about statistics.

ActivStats (student version) is designed for students to use along with an introductory statistics course or for review or study on your own. It is designed to work with any standard statistics text. It can be the text for a course, can supplement any standard text (providing a built-in statistics package at no extra cost), or can be used individually by students wishing for study aids. The student version of ActivStats is available for course adoption and to registered students.

Teacher Support Materials are available to support teachers wishing to teach with ActivStats. ActivStats can provide vivid in-class demonstrations. Any teacher can add his or her own materials to any page of the Lesson book.

To find out more about ActivStats and receive a demo CD please contact Red Dog Scientific Services.