Why You Should Trust My Software Reviews

I’m reticent to “toot my own horn,” but a few people have asked if they can trust my reviews now that I’ve written the BlueSky Statistics User Guide. I think you can, and here is why. The short answer is that they are objective and easily verifiable. Each piece of software offers its unique style and features the others lack. So, no one piece of software will be the best for everyone. The longer answer is…

For 35 years, I managed the Statistical Consulting Center (which changed names several times) at The University of Tennessee, Knoxville. During that time, I helped thousands of people analyze their data. They came from nearly every department at UT. Some were expert programmers, but many lacked either the time or inclination to learn to code well. They used point-and-click style software such as Statgraphics, SPSS, and JMP. I also managed research software licenses for the statewide UT system. That alerted me to the ever-increasing cost of research software and made me advocate for free and open-source software.

During much of my career, I offered substantial assistance to improve JMP, Statgraphics, SAS, and SPSS. I served on advisory boards for the latter three. For my assistance with the SPSS User Interface, Dirk Willuhn, User Experience Researcher at IBM Research and Development, said:

Over the past two years, I’ve had the pleasure of meeting with Bob regularly in the context of our sponsor user program. Sponsor users help us shape the IBM SPSS Statistics user experience. Bob is a passionate advocate for the needs of statistics practitioners. He is always ready to share his valuable feedback based on his vast experience as a consultant and his remarkable knowledge of statistics. We have discussed ways to improve the ease of use of the software and efficiency for users, especially those less experienced. Our team highly appreciates Bob’s friendly and courteous disposition and his didactic style. Thank you so much on behalf of the SPSS Statistics Design team!


My first encounter with open source was the R language in 2005. I had learned many previous data science languages by then, most of which followed very similar rules. R, however, was different. My online notes about those differences ended up as the Springer books, R for SAS and SPSS Users, and later, R for Stata Users.

Back then, R Commander was the only graphical user interface (GUI) for R, and I learned much from studying its code. As more R GUIs appeared, I started reviewing them using a detailed template emphasizing the features I had learned to appreciate throughout my career. Before I published each review, I sent it to its development team to see if I had made any errors. They were always helpful, and in some cases, they quickly fixed problems I had found, allowing for more positive reviews. I received many kind words from the teams. One of the nicest came from Jonathon Love of the jamovi team:

jamovi has really benefitted from a lot of great suggestions from a lot of people, but sometimes someone comes along who goes above and beyond, providing compelling suggestions and nuanced feedback. Bob Muenchen is this man. It’s been great to be able to run design ideas by him, and draw upon his experience. More than a few of jamovi’s slick and sexy features have been inspired by Bob. Thanks Bob!


The BlueSky Statistics team asked me to help them between 2012 and 2018. I declined, telling them by that point I only supported free and open-source software. In 2018, they relented and made that version open-source. I contributed lots of data management and graphics code to the project and wrote the BlueSky Statistics User Guide, which you can read for free here. However, I continue to review each new version of the various R GUIs, and each review clearly points out how the software could be improved. I define each feature I look for and whether I consider it present or absent. When a feature, such as the number of analytic methods, can be counted, I do so. That too, is easily verified in each complete review. My biggest complaint about this methodology is that uncountable features are not weighted. For example, the ability to “split-file” or “by-group” every type of analysis is an important feature. However, only BlueSky currently offers it, and I don’t want to be the one who decides how many “points” such a feature is worth. Likewise, being able to search or filter the variable names list in dialog boxes is also very helpful. Currently, only jamovi and Deducer can do it. It’s worth more than one point to me, but how many more? You get to decide! Download the spreadsheet, apply your own weights, and you will find the GUI that is best for your purposes.

As you can probably tell, I’m rather obsessed with this topic, so please comment below or write me at muenchen.bob@gmail.com. Happy computing!

%d bloggers like this: