Software for data analysis and graphing
The choice of analysis and graphing software depends on the dataset, the required analysis, the figure’s output format and who will edit it later. Dedicated scientific packages, spreadsheet add-ins and standard office tools support different combinations of these needs.
Selection requirements
A medical writer may need several packages because their functions, output formats and interfaces differ. One project might require advanced regression, while another needs an interactive three-dimensional graph exported as a movie.
Three questions can narrow the choice of software:
- How should the data be processed and handled?
- What output format is required?
- Which interface is comfortable to use, and on which operating system?
Specific requirements include a command-line or graphical interface, programmable functions, advanced regression, raster and vector export, real-time updates, video output, and statistical analysis integrated with graphing. Usability, output formats, programmability and advanced features should be considered together. After narrowing the candidates, hands-on testing with the project’s own data is the best way to judge suitability, using trial versions where available.
Dedicated packages
Scientific graphing packages differ in the analyses they integrate and the kinds of figure they produce. Their capabilities include:
| Package | Analysis and graphing capabilities |
|---|---|
| DataGraph | Two-dimensional scatter and line plots, histograms, bar charts and function fitting, with data organised in rows and columns and graph settings controlled through panels. |
| DataTank | Two- and three-dimensional graphing, computational operations and three-dimensional video production. |
| DeltaGraph | Two- and basic three-dimensional graphs, mathematical and statistical functions, Pantone colour matching, and raster and vector export. |
| Grapher | Two- and three-dimensional function, polar, ternary, vector and contour plots, fitted curves, raster and vector export, and programmable automation. |
| GraphPad Prism | Nonlinear regression, curve fitting, t-tests, non-parametric comparisons, analysis of variance, contingency tables and survival analysis, with statistical results integrated into graphs. |
| IGOR Pro | Curve fitting, peak analysis, signal processing and statistics, with volumetric imaging displays, three-dimensional animation and data acquisition through plugins. |
| KaleidaGraph | Two-dimensional scientific plots, extensive curve fitting, formulae in datasheets, reusable templates and control over presentation. |
| MagicPlot | Two-dimensional X–Y and waterfall plots with nonlinear curve fitting. |
| Maple | Statistical analysis, differential equations, two- and three-dimensional plotting, control over surface appearance, and generation of code in other languages. |
| Mathematica | Statistical and computational analysis, volumetric and three-dimensional imaging, and bioinformatics functions including sequence alignment and protein-structure rendering. |
| MATLAB | Numerical computation, programming, two- and three-dimensional models, signal and image processing, and real-time data acquisition through add-ons. |
| OriginPro | Curve fitting, peak analysis, descriptive and inferential statistics, multivariate and survival analysis, power and sample-size calculations, receiver operating characteristic curves and mathematical-function analysis, with three-dimensional graphing and animation. |
| SciDAVis | Two- and three-dimensional plots, curve fitting, filtering, interpolation, Fourier analysis, correlation and deconvolution, with raster and vector export and Python-supported calculations. |
| SigmaPlot | Two- and three-dimensional graphing, programmable tasks, descriptive statistics, Deming regression, three-way analysis of variance, data transformation and an enzyme-kinetics module. |
MATLAB users should allow time to learn its programming language.
Spreadsheet analysis and add-ins
Excel supports simple datasets, routine calculations and graphing, and can organise data before export to a scientific graphing package. Copying data between applications can fragment the analysis unless each copy is documented and labelled.
Daniel’s XL Toolbox is a free, open-source Excel add-in designed to keep data storage, analysis and visualisation together. Its source code can be inspected online. It requires Excel on Windows and does not run in Mac Excel, OpenOffice or LibreOffice. The capabilities described in 2014 included analysis of variance, correlation and regression, chart styling and export, time-stamped backups, data transposition and worksheet management.
Planning and arranging data
The Toolbox offered two group-allocation algorithms: one minimised differences using Kullback–Leibler divergences, while the other maximised the analysis-of-variance P value. Both allowed some randomisation to make allocation unpredictable and were intended for small or medium experiments or trials stratifying on a few characteristics, with less sophistication than dedicated allocation software.
Its Transpose Wizard automated reorganisation of laboratory-machine output, summarised transposed data and calculated standard curves. A Formula Builder supported unordered observations from several groups, so data could be recorded in arrival order rather than rearranged for analysis.
Statistical analysis
The 2014 Toolbox’s one-way analysis of variance accepted consistently labelled data in different layouts and offered Bonferroni–Holm, Holm–Šidák and Tukey post hoc comparisons. It checked homogeneity of variance using a modified Levene’s test, but offered no non-parametric alternative. Its two-way analysis accepted combined group labels and repeated intra-individual measurements, but did not offer post hoc testing.
Charts and export
The Toolbox could apply reusable chart designs and add custom error bars through automatic detection or an interactive mode, directing line-graph error bars to reduce overlap where requested. Export options included TIFF, PNG and EMF, black and white, greyscale, RGB and CMYK, colour management for chart export, and any resolution within reasonable limits.
A spreadsheet could also serve as a layout grid for a multi-panel figure. Holding Alt while dragging a graph snapped it to the grid, and inserting a text box after selecting the graph attached a panel letter that moved with it. The figures could then be grouped, or all graphic objects on the worksheet could be exported together.
Office graphics and editable handoff
Apple’s Numbers offers presentation-oriented three-dimensional graphics, with basic curve fitting and error bars on some two-dimensional charts. It can synchronise work between desktop and mobile devices through iCloud.
PowerPoint’s embedded Excel functionality can create editable visual elements without the cost and learning demands of specialised graphic-design software. Data-driven charts keep their data in an embedded spreadsheet, so changes to the data update the slide’s graph. Available forms include bar, scatter, line and pie charts, with forest plots and box-and-whisker plots also possible through creation or modification.
Word and PowerPoint tables can be transferred between the programs by copying and pasting as tables, with reformatting where needed. They do not need to be recreated for each format.
In 2023, one visual-communication practitioner described starting charts in Excel and finishing them in PowerPoint or Adobe tools, choosing office tools when sharing with clients. Another preferred office formats because clients requested files they could edit, whereas Adobe Creative Cloud files were often inaccessible to them. An earlier practitioner recommendation favoured Excel’s chart wizard over Word’s chart function because the resulting figure was easier to edit and manipulate.