I'm going to disagree with some of the answers' comments in praise of being "user friendly". There's a danger with menu-driven statistical software, that the user isn't, uh, --- reminded --- to understand and check the assumptions of an analysis. There's no advantage to users being able to run analyses that are not valid or that they don't understand.
If you start with a good, complete example, using a language-based software (R and language-based interface for SAS and SPSS) is relatively quick and easy. It is also easy to save the steps of the analysis so that it can be reproduced, which is another drawback for menu-based software.
R is worth the time. There is so much more power to perform a lot of analytical techniques, but it will frustrate you!
No software is the best for statistical purpose and modelling. Every software has its own area of emphasis depending on the vision of its authors. Eviews is good for time series analysis, minitab is good for multivariate analysis and experimental design, matlab is good for simulations, R is good for graphics, spss is good for modelling. The most desirable characteristic is user-friendliness. No matter how versatile a software might be, it is not useful if it is not user-friendly.
I no longer use SPSS, in part because it isn't remotely user-friendly. It has the illusion of user-friendliness by making it easy to run some analyses, but the interface in insanely inconsistent throughout making it awful for teaching using any approach that tries to focus in underlying concepts and principles. It often has terrible defaults and many procedures are poorly documented.
R is far more powerful, flexible and more consistent (at least for base packages - user contributed packages do vary; but that is true for macros etc. in any package). R is also supported by a very knowledgable user base that is growing rapidly, runs on all three main desktop platforms and is free. It is also supported by many excellent free packages to help workflow, disseminate code and create applications.
It can be hard to learn - especially if switching from something like SPSS - but using tools such as R studio makes it much easier. JASP is promising (but runs using R underneath).
I'm going to disagree with some of the answers' comments in praise of being "user friendly". There's a danger with menu-driven statistical software, that the user isn't, uh, --- reminded --- to understand and check the assumptions of an analysis. There's no advantage to users being able to run analyses that are not valid or that they don't understand.
If you start with a good, complete example, using a language-based software (R and language-based interface for SAS and SPSS) is relatively quick and easy. It is also easy to save the steps of the analysis so that it can be reproduced, which is another drawback for menu-based software.
I hadn't seen JASP before. It seems to have great promise. I like the format of the output. And that it reports effect sizes. And has a fair number of options for each analysis.
But, of course, there are always limitations in a menu-based system. My pet peeve so far is that it uses Levene's test to test for homogeneity in ANOVA. I'm theoretically opposed to using tests in this manner to check for normality or homogeneity. But this is an example of the limitations of menu-based software: I can't ask it to plot the residuals for me to check for homogeneity.
But I think I like JASP better than other menu-driven interfaces for R. These would be R Commander. And Blue Sky Statistics. I suppose each has its advantages. R Commander is nice in that it shows you the code it is using, so that you can learn. Blue Sky has some nice features, but is a commercial enterprise.
R is free, open source and multi-platform. SPSS is probably easier to use due to its "click to get results" approach. This being said, RCommander enables the "click" approach in R. And it is possible to do some coding in SPSS. I think R is the most used statistical software among statisticians nowadays. But you will need to spend some time to learn it before getting good results.
My experience with R as opposed to SPSS is very positive. R is free open and provides far more power than SPSS. It taks some time learning, but in my opinion this small 'investment' pays off
- Badges
- Science topic
- Similar topics
- Engineering
- Modeling