JASP is a free alternative to SPSS that also includes Bayesian methods.

https://jasp-stats.org/

Dear Christian,

perhaps you could consider also  R, www.r-project.org...  It is platform independent.

Cheers, Vojtech

www.r-project.org

Dear Christian,

yes R is shoul be considered. It's free and has a great variety of packages. Also the results may be easily reproduced by any person.

JASP is a free alternative to SPSS that also includes Bayesian methods.

https://jasp-stats.org/

SPSS is statistical package for social sciences - practically no full equivalent. Nextel is JASP, but not completely ...... Load one more package that covers the social needs ..

I recommend GeNIe and SMILE, implementing Bayesian networks, data analysis, causal discovery from data.  They are commercial products these days (originally developed at the University of Pittsburgh) but available free of charge for academic research and teaching use.  Here is the link: http://www.bayesfusion.com/.  Follow the Academia link to academic downloads.

I recommend the use of R. It is the best and more complete opensource software for statistical analysis. To be used by studets, it is required a bit effort in the beginning. We use it on first year of BA in Economics and Management and, at the end of the course, almost all students are able to use it autonomously.

 KNIME (free and OS) has quite a few statistics functions (you'll need to get the extra statisics packages from the Labs site). It also has a workflow interface that would be familiar to the SPSS users.

http://www.knime.org

I would like to recommend the R, too. Of course, for your students, I would also like to recommend the process of checking that a software/program meets specifications (e.g. testing with known dataset or known software).

I would like to recommend the R software.

With R there is flexibility but sometimes you just want to input values and get a result and not go into the algorithm part

Since there are so many recommendations for R, I'll definitely recommend my students to check out R as well. Thanks for all the great answers and suggestions. Maybe a combined programming / statistics course utilizing R would be a good addition to our curriculum in the future.

If R is preferred, it is also useful the InfoStat, free version for student, that has the capability to connect to R.

http://www.infostat.com.ar/?lang=en

While on the subject of (free) statistical packages, has anyone tested the packages against each other using the same data set? One area of concern is how the individual packages round off calculations. Is their a default value or does one have to set it? Can anyone comment on this? This brings me to another issue. What is the best statistical method to use for a data set? Can the same data set be made more significant using a different technique, ie, by transforming the variables in a different way or using more or less robust techniques? Perhaps a standard data set (or sets) can be developed from real or made up figures that can be used to test the various methods and their transformations much like a reference material is used to calibrate analytical instrumentation. Just a thought.

I highly recommend JASP (at least for the most basic of SPSS functions). Very intuitive, and creates APA-formatted tables in real time (in the same window), that can then be pasted directly into Word. Reads csv files only

Free statistical page R is a Good candidate - very flexibal and simple interface. My student like working with R

Along with PSPP, JASP and SOFA, I would highly recommend Jamovi, another easy-to use software from the developers of JASP. They have launched Jamovi as a (intended) complete replacement for SPSS. Do have a look at the latest current version 0.9.5.6

https://www.jamovi.org/

Thank you very much for the suggestion - jamovi looks really great. I have an addition to this list myself: Epi Info, a public domain statistics suite released by the US Centers for Disease Control (CDC). Available for Windows and Android. Works great, is simple to use and generates very appealing graphics.

https://www.cdc.gov/epiinfo/index.html

Except for R, does any other package support Mixed model analysis? I tried Jamovi, and it doesn't have that function. Thank you

-- Just saw that you can add packages. Very cool product!

This will sound snooty, but I'd really like to know what people think?

Why use SPSS, SAS, or any of the clones when you can just use R?

Given that R is one of the two main languages being used by data scientists, it is a powerful, full-fledged programming language, comes with an fantastic IDE/workbench, can do anything from simple one-line analysis and one-line plots to whatever your heart desires, has a zillion packages which have been tested and published in peer-reviewed articles, it seems like there is really no need for paid software or software that accomplishes only a subset of what R does with very little to benefit in intuitiveness.

I want get window supported free statistical software, which can be learned and used by beginners, please find for me.

@Motilal Ghimire sir as recommended above JASP would be best choice for beginners. This works software works on basis of R without the need of coding syntax!

In response to Chandramani Aryal , there is no way for a scientist, let alone a student, in today's world to avoid "coding syntax." Avoiding it will just hold you back. Coding is not technical anymore. It's just writing recipes with a few keywords. Most high school children in the US learn at least a little bit of programming. Don't give into the fear. Just learn to write a few lines of code. In the end, it's much faster than remembering which menus to go to or which keystrokes to use.

Ariel Balter Thank you for your concern. I partially agree with you! I prefer R over click and run software and encourage students to do the same. But the case is quite different in Nepal. Majority of students hardly get opportunity to learn programming in high school or later. Thus, they prefer click and run type software such as excel, SPSS including other.

Chandramani Aryal Understood! What I said about high school is certainly not true everywhere here--we have a wide variety of socioeconomic states, cities, and neighborhoods. But for the most part, our students are lucky to live in a country that has so many resources.

Ariel Balter Agree! Saying this I firmly believe that choice of software is of secondary importance. Understanding statistical concept and using it in right instances is what students/researchers should focus to learn. Syntax, code or use of software can be learned at any point of time and switching between software are quite easy while you are conceptually enriched.

Chandramani Aryal I also agree. You make an excellent point that the statistics itself is of primary importance. Statistical software is of no use if you cannot apply it correctly and interpret the results correctly!

To answer Ariel Balter, one reason to use SPSS and SAS (I am an exclusive R user) is that when submitting analyses as a part of new drug or devices for approval by the FDA, the FDA requires "The computer software used for data management and statistical analysis should be reliable, and documentation of appropriate software testing procedures should be available." (https://www.fda.gov/media/109552/download). Because of the way R is maintained this may be difficult. Of course R does go through testing; however, I am not sure the testing and documentation that is done would meet the rigors required by the FDA (speculation). SAS does. I do not know about SPSS.

I still can't see any reason (other than Chandramani Aryal 's situation) to use anything but R. It is fast becoming the industry standard in biomedical research. Where I work, some of the older statisticians use SPSS or Stata, but everyone else is using R. Check pubmed for recent publications in your field and see what they are using.

Ariel Balter does not address the point some multibillion dollar industries (pharma and CROs come to mind) are not going to be comfortable with a product that does not have a solid industry behind it and been extensively vetted (e.g. SAS). I have had numerous discussions with several different CROs and all use SAS and will not consider using R.

The question posed by Ariel Balter was, "Why use SPSS, SAS, or any of the clones when you can just use R?" I was simply positing one possible reason, I am sure there are others.

Gregory E Gilbert you will have a VERY hard time convincing me that R lacks the reputation of SPSS and SAS. I would bet $ that "pharma" IS using R. Actually I'm very confident in that because I know people who have come to where I work FROM pharma and they use R. I also see job adverts from pharma asking for R.

As far as the FDA, the majority of the research that will be used to back up result will have been done using R. So I think it would be pretty hard to say that the FDA wouldn't like it.

Finally, I just don't get what you are saying in the first place about "comfortable" and "solid industry behind" (RStudio is a BIG company with commercial accounts), and "extensively vetted." We're not talking about designing rockets or quantum chemistry here.

80% of what people are using R for is doing things like t-tests, anova, linear regression, etc. And then plotting. A highschool senior taking calculus can derive the equations for these things and any computer science sophomore could write computer code to calculate them.

Sorry, but I just don't think your arguments have any weight at all. In 20 years when all the baby boomers using SPSS and SAS are retired, virtually no one will be using them.

Reading Material:

https://rviews.rstudio.com/2018/10/03/some-thoughts-on-r-pharma-2018/

https://www.r-bloggers.com/using-r-in-the-pharmaceutical-industry/

http://rinpharma.com/

https://www.rstudio.com/about/customer-spotlight/

Ariel Balter I greatly appreciate the references. However, I have spoken recently with several large CROs where I have mentioned my experience with R and they have told me, explicitly they are a SAS shop because the FDA "does not accept R". After further research, they are misinformed, what they meant to say is that "We (the CRO), does not want to put the effort into assembling the documentation required by the FDA to accept R."

What I think you still fail to see is that I am an R advocate and use R exclusively. I was pointing out that if I am an established CRO that has already invested in SAS, as a business decision, after investing in resources to assemble the documentation required by the FDA and retraining my programmers it would be years and even decades before I saw any cost savings.

All that being said my clients may still not be comfortable with me doing analyses in R as opposed to SAS.

Therefore, I come to the conclusion that *one* reason some may choose not to switch from SAS, SPSS, and STATA is purely monetary and a business decision. As a student, I think it is important to become proficient in all three major statistical packages, including R because some businesses are exclusively SAS, SPSS or STATA shops. (I recently had one interview that required me to write SAS code as a part of the interview.)

From an individual researcher standpoint, I think R makes the most sense; but then again that was not the question. The question asked was, Why use SPSS, SAS, or any of the clones when you can just use R? I have proposed some practical reasons , whether you agree with them or not, why "...SPSS, SAS, or any clones..." might be used.

Gregory E Gilbert Thank you for your clarifications. However, to make sure that any future readers are properly informed, I want to post some references that specifically address the FDA issue.

Since at least 2012, which in computer years is a lifetime ago, R has been approved for FDA drug trials.

https://blog.revolutionanalytics.com/2012/06/fda-r-ok.html

As for your colleagues worrying about preparing documentation, they are ill informed. The documentation has already been prepared.

https://www.r-project.org/doc/R-FDA.pdf

Here is a document directly from the FDA explaining how they can benefit from using R:

http://washstat.org/presentations/20181024/Schuette.pdf

In searching around for these documents I found many anecdotal statements from people in the pharma industry, and the FDA that they are predominantly using R in-house.

So, to finally return to Christian Reinboth 's original question, if I were training future researchers, I would want their time to be spent in the best way possible. I would teach them the software that is predominantly used in clinical research (not to mention public health and many other fields), whether that is commercial or academic, and is most likely to become the standard software in the coming decades.

I think the apparent dichotomy is mooted by the fact that now, in both SAS and SPSS, you can write R code to your heart's content in the shells (training studios, really) they provide you, and once the code is readable, SAS or SPSS gobbles and executes it, spitting out SAS and SPSS-formatted output. Perhaps this was a work around for the FDA and other government agencies.

(Disclaimer, I'm the author of Past)

Ariel, I often use R myself, like most people. It's fine, free, and insanely wide-ranging of course. But why argue for a monopoly? The day when R is alone, there will be no more development of new types of statistical software, and there will be no cross-checking of results in different softwares. It is, in fact, possible to do better than R in the future. For example, the programming language itself is absurdly awful compared with modern languages. And don't pooh-pooh the learning threshold, even for experienced programmers. There are, and will always be, many teaching situations where a simpler user interface saves a lot of time and makes it easier to focus on the concepts. And although I know R quite well and am a decent programmer, I often use Past when it saves me a few seconds of precious research time. In short, I advocate diversity in the ecology of statistical software.

Oyvind Hammer: I don't want to miss this opportunity to personally thank you for PAST. I work a lot with business students who study part-time and who are always very interested in applying what they have learned directly in their working environment. The fact that I don't have to tell these students what a subscription to SPSS costs (privately or for their companies), but that I can provide them with easy-to-use, free statistics software, is of great benefit for my lectures.

@Oyvi

Oyvind Hammer I COMPLETELY agree with you. In fact, for students heading towards data science in general, Python is an excellent choice. For business oriented students, it is true that a great many tests can be done easily in Excel which is a standard tool that all business students will know.

But given the question of needing to choose a single statistical software to use for an introductory class it would be R all the say. As far as the learning curve, most things that can be done in one line in any other tool can be done so in R as well. These days, many young people have already done a tiny bit of programming. And if they haven't, it's about time they do anyway. It will come up again in their professional lives at some point for certain.

Ariel, I certainly agree with your basic points, and I definitely agree that students must learn to code. But first, I shudder to think that millions of students now learn R as their first language, and get the impression that this is what programming is about. Secondly, why would you want to program even a single line of code to accomplish a simple statistical task? I don't want to have to type in a cryptic command to change font or open a file in Word. That's how it was done in 1973 :-)

I don't think `lm(Case~Age+Gender, mydata)` or `t.test(dat1, dat2)` is really much harder than pointing and clicking. And probably faster after you have done it a few times. Seeing as how older stats software is heading out, I would teach with excel for point and click or R for one-liners. Anyway, apparently there are many ways to skin a cat.

Dear

Thank you Michael D. Nunez for sharing this link https://jasp-stats.org/

Cordialy

PAST is definitely a good choice.

Cheers,

Annette

I’d also like to point out my favourite (at the moment at least) jamovi - available at www.jamovi.org. It is similar to JASP but more oriented towards frequentist statistics and has nice extensions and a really nifty data editing capabilities (computed variables, transformations and filtering).

I started out with PAST, but since RM-ANOVA is a bit complicated, I switched to jamovi and JASP, and started learning R and pROC, because I needed more advanced ROC analysis which are not achievable with free point and click solutions. Right now I can do most analyses i need in R (thanks to many great tutorials on the internet), but i prefer point and click solutions which save me time.

While I was taking stats classes in graduate school, we used MedCalc, but although it has nice features tailored for clinical research, i think that for the asking price it should have a nicer (i.e. not looking like Windows 95) UI and graphs.

Andrej Sribar good for you for taking the dive into R. Eventually, you will find it as fast as point and click, and also have routines you have used in the past saved that you can use again in the future.

For a free Excel add-in for linear and logistic regression that is very good for teaching, try RegressIt (https://regressit.com). It is vastly superior to Excel's built-in regression tool. It has an efficient menu interface and produces very high quality output, and it includes many innovative features to guide students in learning and applying good practices of analysis and to make life easy for instructors as well. It includes built-in teaching notes, and it is very good for interactive classroom exercises. It also includes an interface with R that allows users to perform very sophisticated linear and logistic analysis in R (including a number of training and testing options) without writing any code. The analysis is controlled from a menu in Excel with well-designed output produced in both Excel and RStudio. This feature does not require knowledge of R programming, and it gives the user a gentle introduction to the RStudio environment.

Dear Christian Reinboth

You can also add Mini tab software which can be a good alternative for SPSS.

Greetings!

JASP, Jamovi, MATLAB, R. These are all good alternatives.

Origin pro is also a good alternative!

For a GUI, I second the recommendation for jamovi (JASP is also great, but I prefer jamovi). I didn't hear about FreeMat, but would rather recommend Octave https://www.gnu.org/software/octave/ for a free and open source MATLAB alterntive, FreeMat seems to be kinda passive since 2013 while Octave is actively developed and maintained. I'd rather avoid MATLAB, it's great but it's closed source and commercial (hope it gets extinct in 10-20 years in science applications...); R is of course a very good choice if it's not problematic to write a script. I wonder if there are R Markdown templates for simple analyses that are usually present in statistical GUIs, that would be really great for teaching.

R is a good alternative to all of these statistical softwares.