The following is an admittedly biased list of what we feel are some benefits and issues seen with our more popular packages. It is based within the context of academic research and the typical needs of those who come to our office whose primary use of statistical software is scientific research. It is not exhaustive, but may give some additional insight into what's available if you are wondering what to use. There is a wealth of information on the web for all of the below packages for you to further use to make your decision.
Note that the best choice however, is to be flexible and hop around as your specific needs suit you. At UNT, you have free access to all of these in one way or another.
Package | Strengths | Weaknesses |
R | Free (Open-Source) so not restricted to site licenses Has kept up with the times with regard to methods appropriate to the academic researcher across a wide range of disciplines Near-daily updates Can teach/demonstrate statistical concepts with it (even has an add-on package devoted specifically to that) Graphics capabilities practically unmatched Extensive documentation, even free complete books on line Easy to use menu system (as an addon) that has even advanced methods available |
More difficult to learn than other stat programs, unless you already do real programming. No real viable 'spreadsheet style' maneuverability within the data (though for analyzing parts of a dataset it is actually much easier than many gui-oriented products). |
Summary: There simply isn't much you can't do with R as an academic researcher, and while it can be difficult to learn, that will eventually turn into time saved had you tried to do similar things with other packages. It is Rich's package of choice. | ||
SPSS | Easy to use menu system for basic analyses (going beyond the basics for most research needs will typically require syntax or another package) Until v. 16, very easy to move around within the data Can get output easily and in easy to read format |
Not geared specifically to academic research needs Prohibitively expensive for students Consistent license issues New versions are not fully backwards compatible by design For some disciplines' typical analyses, no significant updates in years Not a true programming language Lacks basic things like testing of assumptions, computing of some standard effect sizes etc. Only a handful of 'official' statistic specifc macros available and most regarding analyses much better implemented in other packages (e.g. subset regression) Typically released notably buggy Base install is decades behind the times in most analyses often used such as regression, ANOVA, factor analysis Few base offerings for dealing with problematic data Terrible graphics Cost |
Summary: We don't feel SPSS has academic research needs first and foremost anymore, and in general it makes doing good, modern data analysis (in that academic arena) difficult to go along with being cost prohibitive to students. The R&SS group further feels that release 16 was possibly their poorest effort ever. Its only advantage over other packages for years has been ease of use, and that has been dwindling rapidly as the other packages catch up. | ||
SAS | Powerful out of the box Includes many modern techniques |
Not geared specifically to academic research needs Painful installation Even more expensive than SPSS student version Not a true programming language Not user friendly compared to other packages Cost |
Summary: Of the standard popular packages one comes across, it has more to offer than most, though not very user-friendly. It appears to offer more modern academic computing than e.g. SPSS and is widely used across many disciplines. | ||
STATA | Essentially a stat package geared specifically to the social scientist Very strong on Generalized Linear Models and Time Series analysis Straightforward programming language User add-ons easy to implement Has its own journal |
Not so easy on the eyes for most people (unless you miss DOS) Cost |
Summary: A nice package, particularly for the political science, economics crowd, where its offerings are quite advanced. Patrick's chosen package. | ||
S-Plus |
Gui implementation of the S-language while retaining all the power of that language | Less flexible than its R counterpart Cost |
Summary: For the academic crowd, it doesn't really offer anything R doesn't except an easy to use gui out of the box. If you're willing to pay for that and not have access to the multitude of packages in R, feel free. For academics, the S-language offers more than the others available, in our opinion. |