Review of SigmaPlot
By Joseph M. HILBE

March, 2005

SigmaPlot 9.0 is a publication-quality graphics program that is currently sold through SYSTAT Software, Inc. It is selfcontained software that gives users the capability to analyze data with built-in basic descriptive statistical and graphical capability, as further examined in this review. SigmaPlot also provides the capability to perform various linear-based regression analyses with a comprehensive list of fit assessment tools, ANOVA, tests, power analysis, and a host of nonlinear curve fitting mechanisms.

As I shall later detail, SigmaPlot has incorporated a wide variety of graphical and charting forms into its toolbox of options. For many graphics, the user may produce a quality graph by simply selecting a pictured graph-type from the left side toolbar, indicating the variables desired from the variables that are in the active editor, and clicking on the execute button. Of course, the graph may be enhanced in the Graph Editor, allowing the user to annotate anywhere on the graph, and to select line types, thicknesses and colors, and to indicate a file format with which to save the graph. It is a very nicely produced product, as we shall learn with further examination.


1. NEW 9.0 FEATURES

A partial listing of features new to version 9.0:

• Notebook Manager, with password protection of all notebooks
• Categorical Data Support, including graph-by-groups capability
• More than 30 additional rotateable symbols and shapes (half filled, BMW propeller, and so forth)
• PDF and HTML export of both reports and graphs
• Automatic placing of equations used on the graph page
• New 3D engine for increased stability and printing speed
• Plot equation and solver improvements
• ODBC import; import from major databases
• Fully integrated with SigmaStat 3.1

• SPSS (*sav)
• SigmaStat (*spw)
• Systat (*.sys, *.syd)
• Excel (*.xls)
• Lotus (*.wk*)
• Quattro and Quattro Pro
• Dbase (*.dbf)
• Paradox (*.db)
• Symphony (*.wk1, *.wr*)
• Access
• Axon (text and binary)
• Mocha
• SigmaScan
• SigmaScan Pro
• SigmaScan Pro Image
• Table curve (2D and 3D)
• ODBC Database format
• text (*.txt, *.prn, *.dat, *.asc); space, comma delimited; fixed format

I have found that unless the dataset is rather large it is easier to simply copy the data from the Data Editor of another statistical package, spreadsheet, or database and paste it directly into the Data Editor of SigmaPlot. For importation of data from statistical packages that do not allow direct importation, this is usually the least time-consuming method. The only pitfall to using this method, however, is that the top line of the newly pasted data consists of the variable names, not the first rowof data. What I do in this situation is to copy the individual variable names (in text format) from the first row into the variable renaming box. I then delete the first row, cut the dataset using the mouse, and paste it into the Data Editor starting at cell A1. After naming and saving the resultant dataset, I am prepared to analyze or graph the data in the Editor. Itwould be helpful if the next version of SigmaPlot llows for the possibility of importing data without the variable names being assigned as the first row of data. I would prefer that the variable names, if transferred at all, would be assigned to their respective variable cells rather than to the data section.

I might add here that SigmaPlot 9.0 can be used from within Excel spreadsheets as well as from within SigmaPlot. The integration of packages is quite remarkable.

Although SigmaPlot can directly import all major spreadsheet and database data into its Data Editor, Iwould suggest to the vendors that they seek importation capability for files saved in SAS, S-Plus, Stata,BMDP, NCSS, Minitab, and other commonly used statistical packages.SYSTAThas this capability and it should not be difficult to provide this capability into SigmaPlot as well. Exporting SigmaPlot files to these other statistical packages should equally be possible.


3. GRAPH TYPES

SigmaPlot 9.0 likely has one of, if not the most, complete range of graphical types of any existing graphical software on the market.

Types include:

• Scatterplots
• Line plots
• Line and scatterplots
• Area plots
• Polar plots
• Ternary plots
• Vertical bar charts
• Horizontal bar charts
• Box plots
• Pie charts
• Contour plots
• 3D scatterplots
• 3D line plots
• 3D bar plots
• 3D mesh plots

Each type of plot or chart can be constructed in a variety of ways; for instance, Area Plots may take the form of Simple or Vertical, with formatting as Simple, Multiple, or Complex. Composite graphs are easily created whereby various types of plots can be merged for use with the same data. An example presented in the Reference Manual shows a single graph composed of two vertical area plots and simple area and scatterplots. The graph depicts relationships between the three states of matter to pressure and temperature, together with an indicator of their triple point intersection at a particular set of values. In addition to the above, SigmaPlot has an extensive Regression Equation Library. The Library consists of 110 separate regression types and 12 Ligand Binding formats. A sample of regression types include (with options in parentheses):

• Linear, quadratic, cubic
• 3 to 5 parameters, most with modified and with pseudo-
  • Voigt parameterization, of:
  • Gaussian (2D, 3D)
  • Lorentzian (2D, 3D)
  • Log-normal
  • Logistic (2 to 4D)
  • Weibull (+Sigmoidal)
  • Sigmoidal
  • Gompertz
  • Hill function
• 2 to 7 parameter exponential decay (single, double, and triple)
• 2 to 5 parameter single and double rise to Maximum
• 1 to 5 parameter exponential growth (single, double, and modified)
• 2 to 5 parameter rectangular hyperbola (single, double, decay, modified, waveform)
• 3 to 5 parameter sine (squared, damped, modified, power)
• 3 to 4 parameter Pareto (modified)
• 3 to 4 parameter symmetric (modified two)
• 1 to 11 parameter rational (types I, II, and III)
• Varieties of plane, paraboloid, saturation, dose–response, and competition

4. DESCRIPTIVE STATISTICS

A user may request an array of descriptive statistics for any column or row of numeric values in the Data Editor. Included in the output are values for:

• Mean (with 95% CI)
• Standard deviation
• Standard error
• Minimum
• Maximum
• Total
• Number of occupied cells
• Smallest positive value
• Number of missing values
• Number of text or empty cells.

These statistics provide the user with useful information that can be used to good purpose in the construction of informative graphs.


5. HELP FACILITIES

The software comes with a 852-page Reference Manual, which is also provided in PDF format on the Installation Disk. Full onscreen help is available using the help screens. Tutorial screens provide the user with comprehensive step-by-step help in using all of the major graph types. I strongly advise new users to traverse the various tutorials prior to engaging the software for complex work projects. Finally, I should mention that Technical Support is very helpful and responds to questions in a timely and informative manner.


6. CONCLUDING REMARKS

I have attempted to provide an overview of SigmaPlot 9.0 and indicate its various features. I also identified a difficulty in data importation using the Windows copy-and-paste facility. I also suggest that the vendors expand the range of import/export capability. Aside from those points, however, I found SigmaPlot 9.0 to be a relatively easy, yet comprehensive and helpful, graphics tool. The data management capability in SigmaPlot is also quite good and is invaluable when one must transform data prior to analysis and graphing.

I have reviewed many graphics packages in the past, including Harvard Graphics, Stanford Graphics, and others. If I were forced to select only one stand-alone graphics package for use with my work, I would not hesitate to select SigmaPlot 9.0.