I hate to "contribute" with what is really the opposite of a contribution, but I blame experiences teaching others to use SPSS to run statistical tests that I knew to be inadequate given their data without even needing to know their data collection methods. In fact, too often many a researcher takes a multivariate statistics course which, of an unfortunate necessity, goes little beyond realizing how to recognize that some general type of question corresponds to some somewhat understood statistical method. So I would like to say now that "non-statisticians" can be thoroughly in command of the statistical measures/metrics/methods/tests that they use.

Modern statistical software packages are capable of incredibly powerful methods of data/statistical analyses. Unfortunately, even basic statistical measures like Pearson's r or the student's t/T (the uppercase being the distribution) can go very wrong very easily. If someone provided a program that could diagnose patients based on how you entered in their symptoms, would you trust the resulting diagnoses and medical treatments?

Even those in the natural scientists are required to take 3 semesters of a would-be calculus when every 2nd semester should use some combination of linear algebra and calculus as in e.g., Shifrin's Multivariate Mathematics or Hubbard or Hubbard's Vector Calculus, Linear Algebra, and Differential Forms. They don't. They continue to teach a calculus that isn't calculus so that it has to be learned by what is essentially pattern recognition and much of which must be replaced anyway.

Again, I'm sorry for the rant and the intrusion, especially as the original question did not say "How does someone use statistics without knowing what they are doing?" but a very, very different question (I'm not a statistician and I use statistics), I have simply experienced too many problems in too brief an interval with how mathematics is used, taught, and provides a foundation for research in the sciences.

Try JMP from SAS. This statistical software is very intuitive and user-friendly.

SPSS could be one , since as in its name "statistics package for social sicence,"

but I would say that SAS, MİNİTAB, SPSS or R are most used, I think sometimes it would be change what you would like to perform in terms of analysis.

I think the most simple software you can use need to have a very simple graphical interface and need to allow various statistical analyses.

A good list of simple software used are:

Statistics (Descriptive Statistics, Statistical Modeling, Data Analysis, and Advanced Statistical Methods):

- SPSS (and PSPP which is a free alternative)

- MINITAB

Other alternatives are in order of difficulty and time to be mastered:

Stata, SAS, R,

Time Series Analysis

- E-Views (and GRETL a free alternative)

Other alternatives are in this case:

Stata, SAS, R, Rats

Kind Regards

check out Statistica from Statsoft

Kind regards

Krzysztof

From my point of view, undoubtedly SPSS, because it has an interface so friendly and intuitive. But SPSS is not free.

On the other hand, R, is a free statistical program, less intuitive and friendly than SPSS, but R is free, there are a lot of people work with R and there are a lot of manual of R.

I think that if you have possibility with SPSS, use it. But if not, R is free and in R is easy to understand those commands that you need in your research.

And, there is a friendly R version that is RStudio

I hate to "contribute" with what is really the opposite of a contribution, but I blame experiences teaching others to use SPSS to run statistical tests that I knew to be inadequate given their data without even needing to know their data collection methods. In fact, too often many a researcher takes a multivariate statistics course which, of an unfortunate necessity, goes little beyond realizing how to recognize that some general type of question corresponds to some somewhat understood statistical method. So I would like to say now that "non-statisticians" can be thoroughly in command of the statistical measures/metrics/methods/tests that they use.

Modern statistical software packages are capable of incredibly powerful methods of data/statistical analyses. Unfortunately, even basic statistical measures like Pearson's r or the student's t/T (the uppercase being the distribution) can go very wrong very easily. If someone provided a program that could diagnose patients based on how you entered in their symptoms, would you trust the resulting diagnoses and medical treatments?

Even those in the natural scientists are required to take 3 semesters of a would-be calculus when every 2nd semester should use some combination of linear algebra and calculus as in e.g., Shifrin's Multivariate Mathematics or Hubbard or Hubbard's Vector Calculus, Linear Algebra, and Differential Forms. They don't. They continue to teach a calculus that isn't calculus so that it has to be learned by what is essentially pattern recognition and much of which must be replaced anyway.

Again, I'm sorry for the rant and the intrusion, especially as the original question did not say "How does someone use statistics without knowing what they are doing?" but a very, very different question (I'm not a statistician and I use statistics), I have simply experienced too many problems in too brief an interval with how mathematics is used, taught, and provides a foundation for research in the sciences.

Sorry, but non-statisticians should not use any statistics package at all. You can not use it as a "black box" until you know exactly what ity is doing inside, and that it is what you intend.

My choice would be "R" http://cran.r-project.org/

Good luck!

Peter.

For my opinion, the great problem of the easy statistical tools is that you pay their apparent semplicity with a great misunderstanding about results, analyses and models performed. I suggest you to learn R (maybe it is time consuming) because your statistical skills will increase as well as your R knowledge. Again, R is open source and involves all the statistical tolls for performing a good and correct statistical analysis. I hope to be useful.

i think you can try Statgraphics for exploring data and Statistica (statsoft) for statistical analyses. Of course you can do it with R, but you have to learn programming before, and it could be very time consuming for a new user.

Best regrads anad good luck

If you are looking for a very, very easy to use and understand curve-fitting software, I always recommend Distribution Analyzer at variation.com.

I think you should start with Excel and progress to Statistica and eventually move to R or other like Matlab or Mathematica (most mathematical modelling software have inbuilt statistical functions). Main difference is that using Excel and Statistica is like pressing buttons, but to use R you need to learn syntax, and I agree with Andrew Messing, some basics are required first before you start using software.

Try SOFA

http://www.sofastatistics.com/home.php

Free, open-source, easy-to-use.

(I am not associated with this product and I have used it only once, preferring SAS, SPSS and R)

And it is also very much subject to all of the problems of other easy-to-use statistical software that our colleagues have pointed out - you really do need to understand what is going on inside to be sure you are using the most appropriate procedure, with the appropriate options, and to be sure that your interpretation is sound.

It depends what you mean by non-statistician. If you mean someone with statistical training (or receiving statistical training) who is from another discipline then there are some good answers here. If you mean someone with no statistical training - then probably none (in the same way I wouldn't recommend a car to someone who can't drive).

The choice of software will depend on the discipline and type of stats they need.

Dear Thom,

By "non-statistician", I mean someone from a discipline other than Mathematics and Statistics, who has received statistical training.

You can try SPSS, it's easy to non-statistician. But if you want to opensource, you should choose R, it's free but little hard for non statistician.

I have four which I have found useful

1) Paper and pencil – it really is helpful when learning, to do things by hand, sometimes.

2) NSDSTAT to quote

“NSDstat is a fast, simple and fairly powerful data analysis software tool. It is an easy to use and suitable even for people with a random need for statistical computing. It includes a powerful data entry system, and handles both discrete and regional data.

Features of NSDstat include univariate, bivariate and multivariate analysis, graphical representation of data, regression, weighting and recoding. The original NSDstat program was made mainly for educational purposes, and this approach is one of the reasons why it stands out as an extremely user friendly program, not the least for users with a low or intermediate level of statistical knowledge. The package is now offered in two versions; "NSDstat Pro" for use at university level and "NSDstat Skole" for use at high school level”

It was given out to every high school in Norway and was developed by the Norwegian data archive so it allows for very good documentation of data; see http://www.nsd.uib.no/nsd/english/nsdstat.html

3) Minitab – this was specifically developed for teaching statistics and has excellent facilities for simulation and drawing samples. It also has an ‘assistant’ that help you choose the appropriate tool for the job – see http://www.minitab.com/en-us/products/minitab/assistant/

4) I have used MedCalc recently as it had an easy to use interface, and procedures that were not generally in other packages such as facilities for meta analysis ; method comparison & evaluation (eg Bland-Altman plot with multiple measurements per subject ) and methods for agreement (such as Intraclass correlation coefficient) http://www.medcalc.org/statistics.php

In terms of excel – you should read the NEGATIVE comments posted on

http://www.practicalstats.com/xlsstats/excelstats.html

To give one referenced comment: "it is not safe to assume that Microsoft Excel’s statistical procedures give the correct answer. Persons who wish to conduct statistical analyses should use some other package.” - McCullough and Heiser (2008) On the accuracy of statistical procedures in Microsoft Excel 2007 Computational Statistics & Data Analysis 52, pp. 4570-4578

You have been WARNED.

I have used R off and on for many years – and if it works, it is fine – but if something goes wrong , it can be very hard to debug.

I am not a great fan of SPSS but when you have a very large survey it can be a powerful tool.

Just some personal reflections to add to the mix.

Dear Isaac, although it is perfectly possible for any researcher to have practice in simpler statistical analysis, any statistical package will be complex for a beginner. So I suggest you try using excel that although not a 'statistical package', it allows you to perform 95% of the analysis in a simple search. That's what I do. When I need a more complex analysis I use STATA, but it is complex and requires a learning curve.

R (which has pretty much everything) and Python (with modules like scipy, PyMC and statsmodels). However, maybe they're not the easiest ones to use since they require some programming and manual reading skills.

Isaac:

Try SigmaStat!

I use JMP for all of my publications. It is not very high level, but it is a clear step above Excel and produces all of the statistics that I use regularly.

With all of that said, the key thing to remember when you are novice at stats is: garbage in, garbage out! Most of the elementary mistakes are made because of problems with the sample and a mismatch with your sample and statistical choices. A big one to watch out for is whether your sample is normally distributed.

If you don't understand this, then you are in jeopardy of finding statistical significance where none may exist.

In other words, no matter what program you use you might think you proved something that simply ain't there!

Dear Isaac,

InStat produced by GraphPad, Inc is what you are looking for. This package teaches while it works your data, thus helping you to become more proficient in using statistics. Here is the web address: http://www.graphpad.com/scientific-software/instat/

Best regards, Hans

I have had some experience with SPSS (Version 20) and STATA. My personal preference is SPSS which is very user friendly.

User friendly software is a direct way to the hell, because it offers you quick pre-boiled solutions that you can use without thinking and thoroughly analyzing your problem. SI would recommend either:

- read some good books about probability, statistics and experiment design, or

- hire a statistician and try to explain him your problem / motivation. Or both.

Always remember: G.I.G.O., as already said.

Truly,

Peter.

I completely share your opinion, Peter. All my research experience confirms what you wrote.

The first time when I was in need of the development of research results, I had no clue about statistics. I had no help from nowhere but I wanted to do my task as fair as possible. I reached for one of the "user friendly" programs. The plenty of ready-made procedures served by som software as "black boxes", for solving "the same" issue but leading to quite different conclusions, convinced me that this is definitely not the way to go.

So I diligently studied two introductory manuals on the basis of statistics, some knowledge I acquired intensively from the Internet, and I started the calculations with the use the language R.

That was a hit!

Completely free computing environment with tons of different well-documented procedures written in plain text including reference to the literature, many examples and manuals available on the Internet, the opportunity to experiment and create my own procedures allowing evaluations in a very individual way of treating very specific problems, meant that now the statistics became my passion.

The learning edge of R language (and understanding of statistics of course) was rather sharp, but THE BENEFITS that I got from it, constantly pays off in the form of consciously made ​​statistical analysis whose actions I am always able to rationally justify. After several years of experience in performing statistical calculations I am deeply convinced that the best and, maybe contrary to some appearances, THE EASIEST WAY to make a decent statistical evaluations in a very individual way of treating very specific problems, is simply learning statistics and acquisition of experience with the help of a good computing environment that encourages to further enhancing personal skills.

I know It hurts, but It's worth!

MS office Excel only Enough for non-statisticians

Jaroslaw has summed up my feelings on software. The problem with learning using something very easy is that you can run analyses using default settings, but these defaults may be entirely wrong for your research question or produce output you can not interpret.

I have a particular problem with SPSS because the menu interface is poorly designed - lacking internal consistency, having poor defaults for some analyses and giving vast quantities of output that is poorly documented. For example, if you run a one-way ANOVA you automatically get MANOVA output, a non-standard arrangement of ANOVA tables and a table of contrasts, but not (by default) cells means. If you do use SPSS is it important to use the syntax ... for instance getting students to paste the syntax and run it (as a precursor to editing syntax directly). There are (or were) some commands only accessible by syntax and many that are easier to run that way.

Yes, Excel is for secretaries, not for scientists...

I find Excel really useful for pre-processing data and sharing data. I even do simple stats using it, but it isn't a stats package and you need expertise to use it properly even for simple stats. For anything important (e.g., publication) I would run the stats in R (or similar) or at least check the Excel output with other software.

I use a spreadsheet (Calc, not Excel, but it doesn't matter) in a very simillar way. I use a worksheet to prepare the data for their use with R and I export it in CSV format. After statistical calculations in R I sometimes save the resulting data also in CSV format in order to be able to conveniently import them into office programs.

@ Thom Baguley: CSV is definitely better for sharing data, than xls(x)... it is human-readable, and platform/software independent, IMHO. And, when compressed, it is much, much smaller.

@ Peter: I agree, I generally use Excel to prepare my csv files (which also means I don't over-write the original file).

I use JMP Pro for my graduate level classes because its ease of use; if this is in a teaching context you can control and address the valid concerns voiced here. For my sophomore undergraduates, we use online tools to run a linear regression or ANOVA. I have no idea how good they are, but they are sufficient for teaching the concepts of R, F & p values and their implications for a null hypothesis. The URLs for these tools are:

http://www.danielsoper.com/statcalc3/calc.aspx?id=43 (for ANOVA) and

http://www.fon.hum.uva.nl/Service/Statistics/Correlation_coefficient.html

I am not a statistian but I use Statsoft's Statistica. During the first applications I studied all that I needed to use and understand what I was doing. Do in this way or hire someone knowing about this topic. Regards.

I would prefer to use http://www.danielsoper.com

Have you tried Origin software? if not, give it a try, it's wonderful and user friendly and the help is always available. I AM A CIVIL ENGINEER who is currently using Originpro. It's also cheap if you are a student.

you may try brightstat.com. If you have only one datafile it's for free. With brightstat.com you can do your statistics on your desktop computer or mobile device and you can also create and customize sophisticated interactive charts.

https://secure.brightstat.com

According to me the easiest is GMDH Shell provided the work is based on Neural Networks. Otherwise MATLAB has possibly been the most versatile along with SAAS. ‘R’ is not very user friendly but has vast features. SPSS is quite basic but probably the most used. EViews and Stata are excellent for Econometrics. Statistica is quite underrated but extremely efficient and user friendly in nature. Plenty of excel-add in‘s are available as well including NumXL and XLSTAT.

We are using IBM SPSS at school http://www.statistical-software.net/

Btw, there is one more interesting website about spss, video tutorials, user forum, spss news, everything you need. Take a look http://www.ibm-spss.com/

For basic comfort you can start with excel data package but if you want more deep statistical analysis then for better representation Minitab is best but you can't ignore the power of SPSS too. Both are great to use for day to day statistical analysis. But I will recommend you to learn R because in software like SPSS or Minitab it is quite difficult to manipulate data as per your requirement. As it is bound to give result as per the loaded code structure and it is quite difficult to manipulate code structure until you are a good programmer.

So it is better to learn basics of R and its statistical use. It takes not more than 1 month to understand statistics in R package. In long run once you learn how to use R for statistical analysis; I'm sure you will definitely enjoy it.

For R learning use datacamp.com and register the "Analytics Edge" tutorials in edx.org for statistical analysis and for best gain in R programming.

I would prefer to use: https://www.rstudio.com/

I learned R during the Biostatistics course that I took during my PhD journey however used SPSS and SAS when I was analysing my data. I think the Biostatistics course was vital to my ability to use other software.

Try JAMOVI and JASP. They are even free. Very intuitive and publication ready output

https://www.jamovi.org/

https://jasp-stats.org/