Note: While the following review is specifically about SPSS 14.0 (Student Version), it applies to all SPSS 14.0 versions. The basic difference between the student version and the full version is the number of variables allowed (50 with student version), “cases” (e.g., participants) (1,500 with student version), and statistics it performs (some more “complex” stats aren’t available). The full version does not have these variable and case limits. Students probably will not notice the difference, however.

SPSS 14.0 rarely goes by its full name Statistical (I think) Package for the Social Sciences, possibly because name is misleading. SPSS is useful in all types of fields (e.g., hard sciences, engineering, etc.) and in applied settings (e.g., government, business). SPSS performs a variety of descriptive and inferential statistical analyses, and includes many other features (e.g., creating simple graphs). (Note: I prefer Prism for graphs for a variety of reasons, but for simple bar graphs or scatterplots, SPSS works fine). When not limited by the student version constraints (and students can always upgrade to the full version if necessary), SPSS can handle a huge amount of data. I had a data set with over 100,000 “cases,” and SPSS performed statistical analyses just fine, albeit slowly.

One reason to appreciate SPSS (generally) and SPSS 14.0 is the user-friendly, point-and-click interface. In other words, SPSS 14.0 allows a user to conduct relatively simple or complex statistics with a few clicks of the mouse, just like one can change formatting on a word processor document through the “tools” or “options” tabs. While syntax (i.e., programming) is available, and sometimes necessary for complex statistics, the main focus of SPSS 14.0 for Students is that the students can learn statistics and not have to spend all of their time trying to learn software. The SPSS analyses options are grouped and well-organized, so a student never really struggles to figure out where an option is located.

SPSS 14.0 includes, but is not limited to, the following analyses: descriptive summaries (e.g., mean, median, mode, standard deviation, etc.); t-tests; multivariate analyses (e.g., ANOVA, MANOVA, ANCOVA, MANCOVA); correlations and regression analyses; and multidimensional scaling. The student using SPSS 14.0 can spend more time learning about statistics and less time doing pointless hand calculations.

Output files (i.e., the actual output from statistical analyses); data files (i.e., where the data is saved in spreadsheet fashion); and syntax (i.e., the written programming if applicable) can all be stored as separate files. Additionally, within a data set a student can compute a new variable (e.g., if the student has a score on some test at 3 different time-points entered into SPSS, then the student can make a new variable which is the average of those scores—without hand calculations). A variety of other file-types are compatible with SPSS (e.g., Excel files; tab delimited text; etc.).

Why might students want/need SPSS 14.0?

Admittedly, most students using SPSS 14.0 at home will do so because SPSS 14.0 is required in a specific class (or sequence of classes). The student will appreciate having SPSS at home instead of having to find a computer lab on campus with the program installed (as is the case on many campuses). If students only need SPSS 14.0 for a short time, trial versions are available on the SPSS website (http://www.spss.com) that last for 30 days and perform most stats a student would use in a lower-level statistics course (i.e., no multivariate analyses; some within-group versions of select analyses also not available). (Note: I think the only demo version available online now is SPSS 15.0, which is very similar).

Even if a student is not in a class requiring SPSS 14.0, the program might be useful for writing reports for class (many classes require APA style research reports with analyses). Students working on Honor’s Theses or some equivalent also would appreciate having SPSS on a personal computer.

Improvements from Older Versions

I upgraded to SPSS 14.0 from SPSS 12.0 this past summer. SPSS 14.0 is somewhat more user-friendly than the previous version, mostly because of a few “speed” improvements. However, I notice very few differences in the analysis aspects of SPSS 14.0. SPSS 12.0 worked just fine for the statistics I typically use; version 14.0 adds nothing new that I personally use, although I assume that they added features and fixed any bugs from earlier versions.

With each subsequent version, SPSS seems to give better effect size estimates. For most students, this may sound like gibberish, and the issue is notably minor. I still never use SPSS to calculate effect size, instead relying on one of the many effect size calculators available free online.

I do like the output a bit more on SPSS 14.0 (compared to previous versions). Editing, copying, pasting, and navigating through output is somewhat easier with this version. The output in SPSS 14.0 also seems better labeled and organized. One of the output features I really appreciate is the “bug” descriptions. If there is an error in the analyses (i.e., user-error of some sort), SPSS 14.0 describes the error in the output (e.g., “Not enough variables for reliable analysis”); older versions would simply provide an error number (e.g., “Error number 17.”).

Bugs and Limitations

SPSS has a few limitations, in general, though most students will not have a major issue with these. A statistics instructor may not understand or be aware of these limitations (thus the student will never notice); an instructor may explain the limitations and offer alternatives or warnings; or students will not need any analyses that are limited. These limitations I mention are not exhaustive, and are potentially controversial (more of a statistical “philosophy” issue than a “right or wrong” issue). I will be brief and list the two limitations that I commonly face:

1. Post-hoc follow-up tests: The Bonferroni follow-up is not as common in much of the literature, but it seems to be the post-hoc test of choice for SPSS. While SPSS performs LSD (least significant difference) follow-ups, the value tends to be based on more “overall” trends. Without going into detail (partially because I do not feel skilled enough to explain), many find SPSS’s post-hoc tests unreliable and use other methods (e.g., calculators available online…just search for “post-hoc follow-up calculators” on Google).

2. Effect size estimates (by clicking the “provide estimates of effect size” box when conducting analyses) just is not helpful to most people who need estimates of effect size. Just don’t use them.

Bugs

Since upgrading to SPSS 14.0, colleagues and I have noticed only three “major” bugs. Though I have never had the first problem, several colleagues complain that they cannot open output files (and sometimes data sets) saved from earlier versions of SPSS and saved on SPSS 15.0. I know at least two people who upgraded to SPSS 14.0 then almost immediately to SPSS 15.0 because of this bug. The second “bug” might be user-error, but in earlier versions the option was more user-friendly so if it isn’t a bug, it is a deficiency in usability. I am not able to “pin” columns. Pinning is helpful because I can scroll horizontally across the data set (much like in Excel) and keep one column “pinned” so that it remains on the screen. This is helpful for matching certain variables. It used to be a one-click option. Now that option appears grey for me and just fails to work.

The other “bug(??)” I’ve encountered could be a really major one, but I have trouble replicating it consistently across data files. I initially saved a small data set on SPSS 12.0. When I opened the data set on SPSS 14.0, it appears just fine. I had two columns that I needed to average together (to form a third column). I used the compute statement on SPSS 14.0 and made that third column. Because the data set was small enough and someone luckily noticed a discrepancy in data, I hand calculated the mean for each of the data points. The SPSS 14.0 calculated individual means (i.e., for each data point) worked just fine, but the overall mean of that data was off by quite a bit. This means that any statistical tests I performed on the data also might be fraudulent. I am unable to replicate the bug when I hand enter data, so it might only occur when I upgrade from old files. It is just something to be wary of; relying on computers does have its limitations.

Should I purchase SPSS 14.0 (Student Version) or Upgrade?

I use SPSS 14.0 now with few major problems that I know of, though I am not completely comfortable considering the aforementioned “bug.” I am upgrading…again…in the fall. Overall, I suggest NOT upgrading to or purchasing SPSS 14.0 unless:

1. SPSS 14.0 is used in a class you are taking. Even if there is a bug, etc., an instructor might not catch it. You want your data to match the instructor’s answer key, even if that key is “wrong.” Sad, but true. Additionally, though the changes are probably very minor, there are advantages to working on the same version of SPSS used in training until mastering a version. A possible “best of both worlds” is to use SPSS 14.0 in a lab and to try the free trial of SPSS 15.0 at home. If using the version at home seems sufficient, then one might consider shelling out the money to purchase it.

2. IF someone is having problems with opening older and new files, it would be worthwhile to use a different version of SPSS (i.e., SPSS 15.0). However, if SPSS is cheaper or more readily available, it would do just fine.

Overall, SPSS is a lifesaving program for many people, but SPS 14.0 fails to deliver on upgrade expectations. My rating is based strictly on SPSS 14.0 compared to other versions. Additionally, SAS and mPlus are commonly used alternatives to SPSS. The former does basically the same things SPSS does and is more common among engineers; the latter is the premier tool for modeling and surpasses SPSS on those elements. While SPSS has several “add-on” programs, I use SPSS for the “base” statistics and other packages for specialized “stuff” (e.g., Prism, mPlus, etc.).


Recommended: No