Which SPSS modules do I actually need?
One of the most common questions we get asked is some version of “which SPSS modules do I need?” It usually comes up at renewal time, or when someone has inherited a licence from a colleague and has no idea what’s on it, or when a procedure they were expecting to find in the menus turns out to be greyed out.
It’s a good question, and the answer is that it depends entirely on the kind of analysis you do. But there’s a sensible way to work it out, and that’s what this post is about.
If you first want to check what you’ve already got, our short FAQ shows you how. It takes about thirty seconds.
First, how SPSS Statistics is put together
SPSS Statistics is a modular product. Everyone starts with SPSS Statistics Base, which is the core software and covers the descriptive statistics, crosstabs, t tests, one way ANOVA, correlation, linear regression, factor analysis, cluster analysis, nonparametric tests and charting that most people use day to day. Data preparation and validation tools, which used to be a separate module, are now included in Base as well.
Everything beyond that is an add on module. There are 15 of them and they aren’t standalone products, so you can’t buy Forecasting on its own and expect it to run without Base underneath it.
You can buy modules individually, or you can buy one of three bundles.
- The Standard bundle adds Advanced Statistics, Regression and Custom Tables to Base.
- Professional adds Data Preparation, Missing Values, Forecasting, Categories and Decision Trees on top of that.
- Premium includes everything, right up to Neural Networks, Complex Samples and Amos.
Bundles are usually better value if you need three or more modules, but they only make sense if you’ll use most of what’s in them.
Start with the work, not the module list
The mistake people make is to go through the list of 15 modules asking “might I use this?” The answer to that is almost always yes, and if you go down that road then you’ll end up with Premium and a large invoice.
The better approach is to start from the other end. What analysis do you do regularly? What’s in the reports you produce every month or every quarter? What have you been doing in Excel because SPSS didn’t seem to have it, or doing in Base in a roundabout way because the proper tool wasn’t there?
Once you’ve got that list, matching it to modules is fairly quick. Here’s a guide to the ones that come up most often. Each module name links to its page on our site, and every one of those pages has a quick overview video if you’d rather watch than read.
If you produce lots of tables and reports
Custom Tables is the module we recommend most, and it’s the one people are most surprised they’ve been managing without. Base will give you crosstabs and means tables, but if you want them formatted for a report you’ll typically end up copying them into Excel and tidying them up by hand.
Custom Tables lets you build the table you actually want, with nested variables, subtotals, significance tests and your own layout, and drop it straight into a report. If you spend more than an hour a week reformatting SPSS output, this pays for itself very quickly.
If your statistics have moved beyond the basics
Advanced Statistics is where the general linear model and generalised linear model procedures live, which means MANOVA, ANCOVA, repeated measures, mixed models, survival analysis and a lot else. It’s the standard next step for researchers in medical, pharmaceutical, manufacturing and market research settings once the tests in Base stop being enough.
Regression adds the nonlinear regression family that Base doesn’t include. Binary and multinomial logistic regression are the big ones, and if you’re predicting a yes/no or categorical outcome you’ll need this module. It also covers probit, nonlinear regression, weighted and two stage least squares.
These two are the reason the Standard bundle exists. If you need one, there’s a fair chance you’ll need the other.
If you work with survey data
Complex Samples matters if your surveys use stratified, clustered or multistage designs rather than simple random samples. Classical tests assume a simple random sample, so if that’s not what you’ve got your standard errors and significance levels will be wrong. Complex Samples corrects for the design. If you’re a survey or public opinion researcher, or you work with large government or NHS datasets, this one is close to essential.
Categories is for optimal scaling and correspondence analysis. In plain terms, it’s how you get perceptual maps, and how you run regression style analysis on nominal and ordinal data. Market researchers working on brand perception and product positioning use it a lot.
Conjoint is specifically for conjoint analysis, working out how people trade off the features of a product or service. If you do pricing or product development research you’ll know whether you need it.
Exact Tests is worth a look if you routinely work with small samples or sparse tables, where the usual asymptotic p values become unreliable. It gives you exact significance levels for more than 30 nonparametric and categorical tests, and there’s nothing new to learn because you interpret the output exactly as you would in Base.
If your data is incomplete
Missing Values does three things. It shows you the pattern of missing data so you can see whether it’s random or systematic, it estimates statistics that account for the gaps, and it imputes plausible values using regression or EM methods. If you’ve been relying on listwise deletion and watching your sample size shrink with every variable you add, this is the fix.
If you’re forecasting
Forecasting is the time series module. It will automatically pick the best fitting ARIMA or exponential smoothing model for your data, run hundreds of series at once and save the models so you can update forecasts as new data arrives. Call centre staffing, A&E attendances, demand planning and sales forecasting are the typical uses. If you’re currently forecasting in a spreadsheet, this is a significant step up.
If you’re moving into predictive modelling
Decision Trees builds classification and regression trees, and its main advantage is that the output is visual and easy to explain to a non technical audience. It’s widely used for segmentation, profiling, credit risk scoring and campaign targeting, and it’s a good alternative to logistic regression when you need to show people how the model reached its answer.
Neural Networks goes further, with multilayer perceptron and radial basis function models for complex nonlinear relationships. The trade off is interpretability. Neural networks are powerful but they’re hard to explain, so we’d normally suggest Decision Trees first unless you have a specific reason to go straight to neural networks.
Bootstrapping is a different kind of module. Rather than adding new procedures it makes the ones you already have more robust, by resampling your data thousands of times to produce more reliable standard errors and confidence intervals. It’s particularly useful when the assumptions behind a classical test are shaky.
Two specialist ones
Direct Marketing wraps up RFM analysis, cluster analysis, prospect profiling, postcode analysis, propensity scoring and control package testing in a single simplified interface for marketing analysts. Most of the underlying techniques exist elsewhere in SPSS, but if you’re a marketer rather than a statistician this module makes them a lot more approachable.
Amos is structural equation modelling, with a drag and drop interface for building path models and testing causal hypotheses. It’s mainly used in academic research, particularly psychology, education and social science, and it’s the one module that behaves like a separate application rather than an extra menu.
A few things people often get wrong
Buying Premium “to be safe” is the most common one. Unless you’re a large research team with a wide mix of work, you’ll be paying for modules that are never opened.
The second is the opposite: buying Base on its own for a team that spends half its time reformatting output in Excel, when Custom Tables would have saved them that time for a modest additional cost.
The third is not checking what’s already licensed before renewing. We regularly find organisations paying for modules that were added for one project years ago and haven’t been used since. That’s money that could go on training, or on a module you’d actually use.
If you’d rather just ask
We’ve been helping people choose SPSS modules for a long time, and it’s a conversation we’re happy to have without any obligation. Tell us what you’re trying to do and we’ll tell you what you need, and just as importantly what you don’t. Get in touch or call us on 020 7786 3568.
You can also read the full description of every module, and watch the two minute overview videos, on our IBM SPSS Statistics modules page, and see how the three bundles compare on the bundles page.
