I know SAS, Statistica and R (there are many others out there). Any of them looks nicer than SPSS to me. I suggest trying R, a fast-growing highly-popular multi-platform open-source free statistical software. It is very flexible, you can get support from paid versions and it has a wide user base willing to help.

I know SAS, Statistica and R (there are many others out there). Any of them looks nicer than SPSS to me. I suggest trying R, a fast-growing highly-popular multi-platform open-source free statistical software. It is very flexible, you can get support from paid versions and it has a wide user base willing to help.

Great suggestions Felipe. Much appreciated!

Let's see: the SPSS environment is designed to look much like Excel and to require the least amount of knowledge of for the tests run or models made (so much so that programming or customizing the statistical models/methods one wishes to use for one's data is at least as difficult as doing so in C or Java and harder than Python). So, of the software packages I've used and/or taught others to use that are superior to SPSS include (but are not limited to) the following:

R, SAS, MATLAB, Mathematica, Maple, RExcel, Stata, Statistica, Minitab, S-Plus, Octave, SageMath, IPython, an abacus, a slide rule, and a magic 8-ball. Ok, the last 3 are sarcastic. However, SPSS is popular mainly because it is easy to teach students who don't like math how to use statistical methods to analyze data in numerous fields. I started tutoring undergrads in statistics using SPSS in my 2nd semester as an undergrad before I knew what else existed. That was many years ago, and yet I continue to find students plugging data in with the justification that they were taught a particular research question corresponded to a specific kind of SPSS procedure. More than any other popular statistical/computational software package, SPSS enables not only courses but the most popular undergraduate and graduate statistics textbooks to orient their content towards using SPSS rather than statistics. After all, all you need to do is enter in number, select an option from a drop-down menu, select a few options, and analyze. And bingo! Garbage in, garbage out.

well, andrew, much of your critique of using spss is not spss' fault but the users fault :-)

just because it's easy to use spss nobody says that you need no knowledge of statistics. if you do - for example - an anova or a correlation or whatever in spss you need exactly the same knowledge of statistics than you will need in R.

i mean, i could teach a 5 year old kid with absolutely no knowledge of statistics to download a R script, change variable names and run it...

so what i'm saying is: we have to distinguish between a software and a user (same with ms word and latex, or with powerpoint and it's alternatives, etc.)

Dear Manfred Hammeri:

It is little known that Heroin isn't just a street drug but a brand name: Bayer's Heroin. It was an over-the-counter cough suppressant Bayer developed in the early 20th century. The problem was it turned out to be highly addictive. Now, one could say that it isn't Bayer's fault that those who went to their local store and bought Bayer's Heroin for reasons that had nothing to do with cough suppressants, but the fact is that the product created a situation in which the consumers were manipulated and the market altered.

SPSS doesn't just make it easy to perform advanced statistics without any clue what one is doing. It makes it difficult to customize one's models or methods and thereby makes it harder for those who know statistics to make use of their knowledge. If I wish to go beyond plugging data into an Excel like spreadsheet and selecting from options over which I have no control, I have to do a great deal of work. Rarely do courses teach that this is even an option. Meanwhile, in e.g., R., I must customize my approach to my data. It absolutely IS NOT TRUE that the user is to blame when the software makes it difficult for an advanced user to perform the kind of data analysis desired or create the models wished while ensuring that it is so much easier to plug'n'chug.

In order to analyze data in R, it is often necessary to write code that specifies details of tests/models that, if they are even available in SPSS, can be selected by the click of a mouse by pure guess. Many of the procedures in SPSS aren't known by those who use them, because they merely require applying a sequence of selection steps and hitting "enter" rather than writing the code specifying the specific procedures, functions, etc., associated with even commonly used simple statistical techniques/models.

It is deliberately designed to be easy to use without needing anything other than plugging in data and making selections and hard to customize for the desired approach to one's data. What, exactly, makes SPSS require "exactly the same knowledge of statistics that you will need in R"?

My primary issue with SPSS, before even getting to the data analysis, is plain data management. They SAVE icon command was not working!!! IT support advised that users should used File>Save As instead of clicking on a SAVE icon...Really? Ridiculous isn't it. IBM ended up creating a 'fix-pack for this issue" so other users don't have the same problem in the future.  And then, it get even better, perfectly well installed SPSS suddenly decided that it is not so perfectly installed after a year or so and errors out all SPSS files. In order to open the files IBM wants me to renew SPSS for another $1500 annual fee to gain access to my files. Likely I was able to install their trial version and convers all spss files back to Excel.

By no means am I a biostat, but I do have programming knowledge and SPSS definitely has programming bugs that user catch and fix packs come out (but not communicated  to the community). So, at the end of the day, I am looking for a system that is user friendly and does not require PhD in Biostatistics to operate. Thank you.

What about Inform, Medidata or Oracle? Do you have experience with these applications?

"It makes it difficult to customize one's models or methods and thereby makes it harder for those who know statistics to make use of their knowledge." - that's a different thing. this doesn't mean that one who just needs a method to answer his/her research question which is already implemented in spss doesn't know statistics.

"if they are even available in SPSS, can be selected by the click of a mouse by pure guess." - as i said before, that refers to the users not to the software.

"What, exactly, makes SPSS require "exactly the same knowledge of statistics that you will need in R"?" - both doesn't require any knowledge of statistics in the first place. everybody can click around in those programms or type in some code found in the internet. but if you want to analyze data and interpret the outcomes scientifically correct, you need to know the same things. you need to know the underlying processes.

i'm sure you wouldn't say that a great poet shouldn't use microsoft word just because it's easy to use, right? the great poet knows how to write. the same with spss and other software.

of course i know that R (and other programms) offer things which are not available in spss, but that's not topic of this discussion. in general i always find such discussions on diverse software products somewhat strange. e.g. the big discussion for almost 20 years now that powerpoint is very bad and not a good presentation tool - most of it's critics do not really know powerpoint in detail but just saw some weak standard stylesheet presentation of some unexperienced user (again: the user! not the programm).

"that's a different thing. this doesn't mean that one who just needs a method to answer his/her research question which is already implemented in spss doesn't know statistics."

As Ola Sinelnikova correctly pointed out, the problems with SPSS begins before one can apply any statistical/data analysis. However, even if one has SPSS, your statement doesn't remotely address what I stated. Your reply is simply "granted that one understands the statistical methods one uses and granted that one is capable of implementing these in SPSS, then one is capable of implementing these in SPSS." The conclusion that SPSS is as useful for "one who just needs a method to answer...[a] question which is already implemented" simply asserts that if you know how the entirety of methods proffered by SPSS can be implemented, then one can implement them. Basically, you assert that if you can use SPSS, then you can use SPSS. I can name many, many, many computational and statistical models that one can't implement with SPSS (and have on this site), which is nothing compared to the difficulty involved in determining how specific models/methods are implemented. SPSS is fundamentally an environment to plug data in, select methods that can be chosen via drop-down menus and other such options, and run these regardless of any understanding of the basics of the methods used. One cannot do this when one's software package requires that one specify that which SPSS not only determines by default, but doesn't even indicate that it does this.

I have found that defenders of SPSS tend fall into two categories: those who have to teach undergrads who don't like math and have trouble with statistics, and those who had trouble with statistics, didn't like math, and learned to use SPSS so as not to learn statistics. There are plenty of exceptions, of course. It's just that I've what I described to be the norm.And SPSS has advertised (as have leading statistical textbooks) the same.

"Basically, you assert that if you can use SPSS, then you can use SPSS."

did you really read my comments? but maybe there is just some misconception because i'm not a native speaker.

what i say is the following: the user is responsible for his/her knowledge of statistics, and he/she needs to know what to do with a certain software and which software to use for certain analyses.

what you say is: spss is bad.

the reasons you gave for this are not convincing. especially in the context of the original question of Ola.

that's my understanding of the discussion so far. but as i said, i'm not a native speaker, maybe i missed something.

Alex have made an excellent point that I omitted in my too-short answer. R can be difficult to learn if you are not used to "programming" (SAS users might be more used to it). Anyway, you can install Rcommander, which provides nice menus.

I see a slight contradiction in Alex's statement, calling the SPSS approach 'click and pray' and deterring users from learning R due to its 'steep learning curve'. Learning SPSS to a level beyond 'click and pray' is a similar effort, only you may shortcut it by clicking and prefer to pray. To me, you apply double standards. Ever tried to "program" a loop or use vectors in SPSS Syntax? Or read a complex data structure with SPSS (disentangling UTF8 with/out BOMs)? Let me exaggerate a bit: Allowing users ignorant of statistical details to use SPSS is like allowing a ten year old child to drive a 10-ton-truck: The power-steering power-breaks etc. make it pretty easy to actuate the vehicle, but not to pilot it... ;-)

Alex, thank you for your reply! I absolutely agree on your statement regarding good and bad teachers (and your example of pairwise t-tests makes me shiver as well. ouch!) Further, it was not my intention to indicate that you tell your students such a thing, just to be clear on that point too. I also agree that many researchers will be able to perform solid analyses with a certain set of tools, easily accessible via the SPSS interface. My response was driven by the impression of a dichotomy, R = difficult, SPSS = easy, end of line. This I do not believe.

Let me give an example from my own experience: I teach statistics for more than 12 years now, and I adopted R a few years ago. My students have to learn both programs (SPSS, because psychologists are still expected to know SPSS, and R, for its variety and flexibility). Usually, I started with SPSS and switched after a few weeks to R. Students heavily complained, of course. Eventually, I started with R (very basic and in thin slices, like a well hung salami) and switched later to SPSS. I was suprised: some students complained that they are distracted by the many windows popping up! They were used to a concise script file where they could overlook all analysis steps performed so far (I admit, I did not start with the syntax window, you got me - but, hey, we are talking about SPSS, after all...).

Hence my point is: To me, it appears crucial how beginners are introduced to either program. As long as they are able to build a reliable mind map of what is going on, they may feel pretty comfortable with the software they use.

Maybe, there is some kind of Freudian projection going on: We (ourselves) started many years ago with SPSS and had to learn R after having gained lots of experience with the logic of SPSS. And I confess, my first R-experiences were traumatic, although I already knew "real" programming languages (e.g. Fortran). But our freshmen face an entirely different situation, they are unencumbered. And I think, this could be a chance. So, let us maybe question our own premises - and the dichotomy in our head? ;-)

best,

Rainer

I have been using R statistical software. It has several packages which will cover almost everything in statistics. The coding is easier. Also produced beautiful plots.

Recently, I came across STATA, which found appealing because of its excellent user friendly interface.

I am surprised, that no one mentioned JMP of SAS Company. http://www.jmp.com/en_us/software.html

This package is rather handy and bases on SAS Routines.

Compared to SPSS it has only one table to care of both, variables and data.

It also has database connections. Furthermore a simple SAS Scripting is possible for calculation routines and also graphics. Last mentioned can be transferred as  vector-graphics for further processing and adaptation.

Finally, the support is brilliant.

Dear Sinelnikova

I agree with other answers from others. I will write a short answer for your question. In my opinion, R is choice for most of researchers because it is free and innovative regarding compared to other programs. Of course it has also some disadvantages but its user increasing day by day because of its advantages more than its disadvantages. I want to prefer  some packages as follows;

-Rcmdr package

-Rattle packgae

-Deducer package

These three packages solve your data mining problems. Also you can download individual packages from CRAN as your special purpose. I should said that the most favorite part of the R is the absence of packages according to needs for a wide audience.

Fatih KAHRIMAN