Free DOE Software

I spent quite a long time looking into statistical analysis software that’s free, and will remain free, yet allow you to design and analyze DOE style experiments, with emphasis on two level factorial experiments that pack the most punch.

I came up with a list of three options that you can download and jump right in. One of them requires Excel but the other two are stand alone. Let’s take a look.

DEVELVE

This software tool was developed by Frank Pauw, in the Netherlands. Frank is a seasoned professional of over twenty years experience at practical applications of DOE. He offers a free version for Develve that you can renew for free every year, and asks for a seventy five year license to use it for commercial applications. The software features a robust suite of tools for statistical analysis, as well as supporting the design and analysis of factorial designs up to twelve levels. I get the impression that the support documentation on his website is very helpful for the newcomer to DOE in general and to factorial experiments in particular.

Real – Statistics

After Develve, I recommend you take Real-Statistics for a test drive. It’s an extensive and comprehensive suite of statistical analysis tools in the form of add-ins for Excel. That means if you don’t already have access to Excel, then you’ll need to spend about seventy dollars a year for your own personal copy of the Microsoft 365 suite of office tools, that includes Word and Excel. This software is free to download and appears quite straightforward to use as an Excel Add-in. The main reason I list it here is that it’s not only completely free, apart from Excel, but it includes a most impressive collection of tools for statistical analysis, including a full suite of analysis and diagnostic tests that support its two level factorial experiments. 

I recommend you view the YouTube review of this software by Steven Bradburn, from 2022. He offers rave reviews and also covers a similar, free suite of Excel Add-ins for statistical analysis.

JASP

Third on my list is a user -friendly version of software that enables you to use the full power of the R programming language that’s been specifically engineered for statistical analysis. But when you use JASP as the user interface, you won’t need to write a single line of code. 

However, I list JASP at number three because it might not be quite ready for prime time (version 18 – 2024) when it comes to DOE and factorial experiments.

JASP is completely free and open source, as is R, and is maintained by the University of Amsterdam

JASP stands for Jeffrey’s Awesome Statistical Platform.

The software options are presented as a number of modules. You’ll need to open the “Quality Control” module to set up or design factorial experiments. 

Within this module, I was able to quickly set up a two level factorial experiment with four factors, but I ran into some glitches when it came time to analyze the results of the experiment.

What I did was use the example problem data from the Montgomery and DOE Simplified texts on DOE factorial experiments. 

What happened was that I was able to list all of the interaction terms for the ANOVA part of the analysis on my first attempt, but then I was only permitted to include the main factors and the two factor interactions on subsequent attempts, even after repeating the whole experiment. What was particularly vexing was that I was unable to reject any of the two factor interactions, even the ones I could safely attribute to pure noise.

When I wanted to go further and plot the results of the best fit model, neither the surface plot nor the contour plot tools worked at all.

In summary, JASP promises to offer the full power of R but hasn’t quite arrived. It’s got an impressive suite of statistical tools but the module for “Quality Control” is not quite fully operational. But when it is, it will rock your socks off, so keep an eye on this one.

Combination of AI with R and Python

In my search for powerful yet free ways to design and analyze factorial experiments, I came across a very interesting video by Marcel Butsche, in which he demonstrates how to use ChatGPT in to write snippets of code in the Python language, another programming language that excels at statistical analysis by virtue of is vast collection of subroutines already written. Marcel demonstrates how you can design and analyze a two level factorial experiment with four factors, again, using ChatGPT and Python.

I really experienced an “Ah Ha” moment when I watched his video. In other words, I had an epiphany. “Of course!” I said out loud. Many of the large language AI models out there are primarily intended to write code, so why not use them to write code in R or Python to both design and analyse a DOE style factorial experiment.

Moreover, the AI could provide assistance and guide you along, and help to lay out a workflow for conducting your experiments, from the inception of your experimental idea right through to presenting your results.

Then I tried it for myself, using Perplexity.AI with both R and Python. And presto! Perplexity.AI delivered. I realized then that the possibilities were limitless, or at least worthy of their own dedicated series of blog posts and videos.

Free Trials with Leading DOE Platforms

Now I realize that some of my audience may still shy away from AI and the powerful languages of R and Python. For those of you who want a complete package with no fuss and no muss, then I recommend you take the free version of the Design Expert software for a test drive. But must caution you to only activate the trial version once your schedule is relatively stable and open because you’ll need to focus on it quite intensively before the brief trial period runs out. Definitely get your hands on the Box and Hunter, Montgomery and DOE simplified texts in advance.

Beyond Design Expert, you may also find trial versions of JMP and Mintab, but neither of these offer the user experience dedicated to DOE that Design Expert offers.