Statistics for Data Analysis V.32
What's New and Why It's Worth It
Version 32: even more powerful analysis, even more targeted tools
Statistics for Data Analysis, powered by IBM SPSS, just keeps getting better — more powerful, more complete.
Check out everything new in this release! 🔽
📊 Even more advanced features:
💡 Distance Correlation
It's a versatile metric that picks up on any kind of statistical dependency between variables — linear or non-linear. It spots non-linear relationships that often hide in real-world data, giving you a more joined-up workflow, and gets round the limits of standard analysis when real data has those messy, non-linear dependencies. A genuinely useful addition.
Where it comes in handy:
-
Medicine: non-linear biomarkers and clinical variables
-
Psychology: complex behavioural relationships
-
Sociology: social patterns that traditional Pearson correlation struggles to catch
-
Marketing research: user behaviour and customer analytics
💡 Proximity Mapping
Proximity mapping is a visualisation technique used to cut down the dimensionality of multivariate data and show how objects (cases, items, or other entities) relate to each other in a spatial layout.
It supports various sources of proximity, can bring in extra variables (attributes and properties), and gives you loads of options for transformation and restriction.
Great for exploratory clustering, perceptual analysis, segmentation, social research, and psychometrics.
Really handy for universities and academic research, customer satisfaction work, combined qualitative/quantitative analysis, market research, sociology, and psychology.
💡 Mixed Data Clustering
Picture wanting to split your data into groups (clusters), but you've got a "mixed" dataset of numbers and categories.
Mixed Data Clustering lets you run clustering analysis using numeric, categorical, and ordinal variables all at once — mixed data straight from surveys or real-world databases.
Unlike traditional clustering, Mixed Data Clustering is built to work with the actual structure of datasets used in social research, medicine, marketing, psychology, universities, and customer analytics.
This new version also brings stability improvements and better handling of complex datasets.
💡 Mediation Analysis
A statistical method for understanding the mechanism through which an independent variable affects a dependent variable via one or more mediating variables.
Why does it matter?
It helps you test hypotheses about indirect effects and get your head round the mechanisms of influence in your data.
Specifically, it's useful for:
-
Testing theoretical models of causal mechanisms
-
Pinpointing intervention targets by understanding mediating mechanisms
-
Weighing up the relative importance of different causal pathways
-
Checking whether effects vary across subgroups
-
Controlling for confounding variables in observational studies.
Widely used across health sciences, psychology, social sciences, business, and training.
⭐ New Add-Ons: more tools for better analysis!
It gives you reliable results even when outliers are lurking about.
Why does it matter?
Classic regression models (like OLS) are sensitive to outliers.
Even a handful of extreme values can throw off:
-
estimated coefficients
-
predictions
-
how you interpret the model.
Robust regression helps cut down the influence of outliers without chucking them out altogether, so you keep the useful info they contain.
Robust
Regression
A TreeMap is a chart that shows hierarchical data using nested rectangles. The size of each rectangle is proportional to a numeric variable.
Why does it matter?
-
Highlights how a dataset is made up (categories and subcategories)
-
Makes it easier to compare relative weights
-
Helps you spot patterns, imbalances, and anomalies
Tree Mapping lets you display large amounts of structured data in a really compact way.
Tree Mapping
⚡V32: real-world upgrades
✅ Vector Autoregressive Models
VAR models are especially handy anywhere you've got multiple correlated time series and want to look at how they evolve together and influence each other over time. Big in economics, finance, and marketing.
✅ T-Test Interface Improvement
In the independent-samples T-test dialogue box, you can now include Levene's statistic results in the output.
✅ Genomic Analysis
A new feature letting you import FASTQ/FQ sequencing files directly and turn them into an SPSS dataset with metrics ready to go for your usual statistical procedures (descriptives, frequencies, correlations, regression).
Genomic analysis is relevant (though not limited) to things like: spotting genetic mutations linked to diseases; finding potential drug targets and understanding disease mechanisms; developing treatments tailored to an individual's genetic profile.
Coefficient of Variation
A new option for calculating the coefficient of variation, now shown in the relevant output tables. It's a relative dispersion statistic that measures variability against the mean, letting you compare datasets with different units of measurement (e.g. weight in kg vs height in cm).
✅ Quick DOE Search
From the top search bar in Statistics for Data Analysis, you can now find all the Design of Experiments (DOE) commands grouped together, rather than digging through the Analyse menus by hand.