FAQs

General

Statistics

SPSS

SPSS Book

R Book

SAS Book

Books Generally:

Responses

Will I analyze your data or enter into consultancy about data analysis? The short answer is no. I'm fairly busy with my own research, answering emails, generally trying to help everyone in the universe. I probably could charge people lots of money for analysing their data for them, but I'm happier just answering what questions I have time for for free or in return for charity donations. I don't have time to answer long questions, but I do answer some short ones when I have time. Don't be offended though if you don't get a reply. Twitter is also good for very short questions (@profandyfield) or my official facebook page (not my personal one). If you want a detailed discussion about statistics then try Adam Hafdahl or any other proper consultant.

Will I reply to questions by email? I prefer to answer questions on a public forum so people can search previous discussions. So try methodspace (to which I contribute), Facebook, or Twitter in preference to email.

Will I help with assignments or masters/PhD work? It is completely inappropriate to ask me to help with assignments for which you get marked. First, it's deceitful to take credit for work when you have been advised by a third party without the markers knowledge. Secondly, I don't want to be sued if my advice is perceived to contribute to a poor mark. For masters and PhD work your first port of call should be your supervisor: it's their job to help you, not mine. If they feel out of their depth then they are welcome to contact me. However, I will not get embroiled in 'My supervisor says x, but what do you think?' type of discussions because your supervisor knows the nuances of your research area whereas I do not, and I do not want to give bad advice.

Is there a nonparametric equivalent of x? Apart from the obvious nonparametric tests there are now a huge family of tests known as robust tests. These tests typically rely on bootstrapping, trimmed means or both to produce results that are not dependent on the shape of your data/population/sampling distribution. SPSS has a bootstrapping module (tutorial here), but it's pretty limited. If you want to do Robust methods, the best reference source is Rand Wilcox's book. As part of this book he provides files to run robust ANOVA, ANOVA, MANOVA and multiple regression as well as a whole host of other procedures. If there is a robust method, it is in Rand's book, and he has written a macro procedure to run it! You can also download these macros from his website. His Macros work in the freeware package R. We have practical examples of how to use Wilcox's tests in R in our book Discovering Statistics Using R.

What is the 'corrected model' in SPSS ANOVA output? The corrected model represents the overall effect of all main effects and interactions in the model.In other words it is the amount of variability in the outcome that can be explained simultaneously by all main effects and interactions. This is not the same as the corrected total, which also includes the error in the model. Usually, this effect is not particularly interesting for interpreting the results because hypotheses are typically tested by any interaction terms, or in the absence of interaction terms the main effects.

Why does my sphericity test look weird? Sphericity applies only when you have three or more repeated measures conditions (with only two conditions sphericity must hold true ... if you read the explanation of what sphercitiy is in one of my books you'll see why, because it relates to the variances of DIFFERENCES between conditions being equal, and with only 2 conditions there is only one set of differences). Anyway, if you have only two conditions SPSS will produce lots of blanks in the table for Mauchley's test (but note the epsilon values are 1, which tells you that the data have perfect sphericity), these blanks might seem confusing, but they're just there because it can't test for sphericity with only two conditions. It probably just shouldn't produce a table at all, but if it did that you'd wonder where the table had gone.

I can't find something in the menus. SPSS is a modular program so not all versions have all features. Up until IBM bought SPSS, it was modular, so universities/companies would buy access to certain modules, so if you can't find something it probably means your license doesn't include the module you need. A list of modules is here. Since IBM bought SPSS, they seem to have reduced things down to three versions: standard, professional and premium. You can see a comparison of versions here.

Is there an ebook version? The SPSS book has an eBook bundle available from the publishers (here). The SAS and R versions of the book do not (as far as I know). I have actually never seen the eBook, and so I have no idea what it is like. Personally i would like to do something proper and exciting with eBooks but the publishers are less keen until the market for them in Higher education improves. Watch this space patiently.

Why doesn't the student version of the book work with the software provided? This is an interesting question:) I first want to make clear that I wasn't consulted about this version of the book ... at all. The first I knew of its existence was when someone emailed me asking why the software with the book wouldn't run a logistic regression! The SPSS student edition (which looks like it might not exist for much longer because IBM don't seem keen to continue it, but I could be wrong) is limited to certain modules, and number of cases/variables. The SPSS book is based on having all of the modules you might want to have, and had I been consulted I would have mentioned that it was a bad idea. However I wasn't. All I can say is don't buy this version. If you are at an institution that has a site licence just get the normal version of the book, if you're spending your own money then probably use the freeware package R instead of SPSS, because it's free and incredibly powerful (although the learning curve is a little steeper). If you go down this route you might find our book Discovering Statistics Using R useful when it comes out.

What happened to the R-plugin podcast?At the time when I promised to do a podcasts, I'd never used the R plugin in SPSS, but I thought it seemed like a good thing to have a podcast about. Then I used it. It basically involves installing the plugin which may or may not work, and generally creates a lot of faff. Even if you get it working you simply type Input Program R ... End Program and stick a bunch of R code in between. Basically it means that you might as well just do it in R and spare yourself the misery of trying to get the R plugin to actually work. So, rather than do a podcast about using R in SPSS, which wouldn't actually be very useful, I wrote a book about R instead. This took me quite a bit longer than the podcast would have. THe next edition of the SPSS book (due 2013) will cover bootstrapping.

Will I write a book on X? I've just finished a book about R, and in 2012 I'll be updating Discovering Statistics Using SPSS. I'm also writing an Introductory Statistics Textbook. It's fair to say that's all I'll be doing for the next few years.

When am I updating x (insert name of one of my books)? My current scheule is:

  • Discovering Statistics Using SPSS (4th edition): writing in 2012 for 2013 release.
  • Discovering Statistics Using SAS (2nd edition): no update planned at present.
  • Discovering Statistics Using R (2nd edition): for goodness sake, let me get the first edition out before you ask me this:)
  • How to design and Report Experiments: supposed to be doing it now but it's not really happening:)