PSPP vs SPSS: What the Free GNU Clone Does (and Doesn't) in 2026
PSPP is a free, open-source program that reads SPSS syntax and .sav files and covers the procedures most coursework needs: frequencies, crosstabs, t tests, ANOVA, correlation, linear and logistic regression, factor analysis and reliability. What it lacks compared with SPSS is breadth (no mixed models, hierarchical clustering, discriminant analysis or ordinal regression, among others), charting and polish, so it suits standard analyses and syntax-driven work better than advanced modelling.
This comparison is based on PSPP's official manual for version 2.1.2 (released 19 September 2026) and IBM's own pages, checked in October 2026. It's published by the team behind TRTT, which uses PSPP as one of the references its results are tested against.
PSPP in one paragraph
PSPP is part of the GNU project and describes itself as a free replacement for SPSS. It's licensed under GPLv3, has no licence fees and no expiry date, and its site says it handles over a billion cases and over a billion variables. You can use it through a graphical interface (PSPPIRE, with a data editor, dialogs, a syntax editor and an output window) or from the command line. The menus follow the classic SPSS structure, so a crosstab is under Analyze ▸ Descriptive Statistics ▸ Crosstabs, as in SPSS.
Procedures side by side
SPSS's coverage depends on what you pay for. IBM's subscription has a Base tier and add-on bundles; regression, advanced statistics and custom tables are one such add-on. In the table, "SPSS" means SPSS with the relevant module.
| Analysis | PSPP 2.1.2 | SPSS |
|---|---|---|
| Frequencies, Descriptives, Explore, Means | Yes | Yes |
| Crosstabs with chi-square | Yes | Yes |
| t tests (one-sample, independent, paired) | Yes | Yes |
| One-way ANOVA with post hoc tests | Yes (Bonferroni, Games-Howell, LSD, Scheffé, Šidák, Tukey) | Yes, more options |
| Univariate GLM (factorial ANOVA, ANCOVA) | Yes (fixed effects) | Yes |
| Repeated-measures and multivariate GLM | No | Yes |
| Mixed models | No | Yes |
| Bivariate correlations | Yes | Yes |
| Partial and nonparametric correlations | No | Yes |
| Linear regression | Yes | Yes |
| Binary logistic regression | Yes | Yes |
| Multinomial and ordinal regression | No | Yes |
| Factor analysis | Yes (principal components, principal axis) | Yes, more extraction methods |
| Reliability (Cronbach's alpha, split-half) | Yes | Yes |
| K-means clustering | Yes | Yes |
| Hierarchical and TwoStep clustering | No | Yes |
| Discriminant analysis | No | Yes |
| Nonparametric tests | Yes | Yes |
| ROC curves | Yes | Yes |
| Custom tables (CTABLES) | Yes | Yes |
| Multiple response tables | No (sets can be defined) | Yes |
| Survival analysis, time series | No | Yes |
| Bayesian procedures | No | Yes (since version 25) |
The "No" entries come from PSPP's own "Not Implemented" chapter, which is refreshingly frank. It lists about a hundred SPSS commands PSPP doesn't have yet, from CLUSTER and DISCRIMINANT to MIXED, PLUM and GGRAPH.
Syntax compatibility
This is PSPP's strongest suit. It runs SPSS command syntax, including the parts other free tools skip: DO IF, LOOP, DO REPEAT, DEFINE macros, AGGREGATE, ADD FILES, MATCH FILES, UPDATE and the MATRIX language. A typical .sps file that cleans data, recodes variables and runs standard procedures has a good chance of working unchanged.
Where scripts break:
- Unimplemented procedures: any command in the "Not Implemented" list stops with an error.
- Subcommands: PSPP doesn't always support every subcommand of a command it does implement, so check the manual for the specific one.
- Programmability: SPSS can run Python and R code inside BEGIN PROGRAM blocks; the PSPP manual doesn't mention BEGIN PROGRAM, and output management (OMS) is on the not-implemented list.
Files: .sav, .zsav, .por
PSPP handles the main SPSS data formats well:
- .sav: reads and writes, with variable labels, value labels, missing values and formats.
- .zsav: writes it with
SAVE /ZCOMPRESSEDand reads it with GET. - .por: imports and exports SPSS portable files.
- Spreadsheets and databases:
GET DATAreads Gnumeric and OpenDocument spreadsheets, text files and PostgreSQL queries. The manual doesn't list .xlsx, so for Excel files you may need to save as .ods or .csv first. - Export:
SAVE TRANSLATEwrites CSV and tab-separated files.
SPSS itself reads and writes a wider range, including Excel, Stata and SAS formats.
Output and charts
PSPP produces SPSS-style tables and writes output as text, PostScript, PDF, HTML and OpenDocument. From the command line it can also write .spv files, and the pspp-output utility can read and convert .spv files from SPSS 16 and later.
Charts are the clearest gap. PSPP's GRAPH command draws scatterplots, histograms and bar charts, and some procedures add their own (histograms and pie charts in FREQUENCIES, plots in EXAMINE, scree plots in FACTOR). There's no Chart Builder and no GGRAPH, so SPSS's Chart Builder and chart editor are far more capable.
Installing on Windows, macOS, Linux
- Linux: the easiest platform.
apt-get install psppon Debian and Ubuntu,dnf install psppon Fedora, or Flathub. - Windows: installers are linked from the PSPP download page; the project recommends the smaller non-debug installer.
- macOS: three routes: an application bundle, MacPorts or a Homebrew tap. The ready-made bundle has sometimes lagged behind the newest release, so if you need the latest version, Homebrew is the usual answer.
SPSS Statistics 32, by comparison, installs on 64-bit Windows and macOS. IBM doesn't offer a desktop client for Linux (only SPSS Statistics Server), and it costs money: IBM lists the base subscription "starting at $99 USD per authorized user", with monthly, quarterly and annual plans (checked October 2026). Many universities provide it to students free of charge.
Who should use PSPP
PSPP is a good fit if:
- you work on Linux, or want open-source software;
- you have SPSS syntax that uses DO IF, AGGREGATE, MATCH FILES or macros;
- your analyses are standard: descriptives, crosstabs, t tests, ANOVA, regression (including binary logistic), factor analysis, reliability;
- you need SPSS-compatible .sav, .zsav or .por files on the way out.
Stay with SPSS (or look elsewhere) if you need mixed models, repeated-measures ANOVA, hierarchical clustering, ordinal or multinomial regression, Bayesian tests or publication-quality charts. For repeated measures and Bayesian tests specifically, jamovi and JASP are free options; see free SPSS alternatives.
Where TRTT fits
TRTT is another free tool that follows SPSS, with a different trade-off. It runs in a desktop browser with nothing to install (handy if installing PSPP on a Mac or a locked-down laptop is the problem) and as a native app on iPhone and iPad (free for data sets up to 100 cases). Its Analyze menu mirrors SPSS 32, and it covers several procedures PSPP doesn't: TwoStep and hierarchical clustering, discriminant analysis, nearest neighbour, ordinal regression, Ratio statistics, multiple response and the Chart Builder.
PSPP is ahead on syntax. TRTT runs procedures, transformations and labels from .sps files, but not DO IF, IF, LOOP, variable ranges with TO, AGGREGATE, MATCH FILES, GET FILE, SAVE OUTFILE, logistic regression or CTABLES (full list). The web version also has no PDF export and isn't designed for phone screens, and TRTT isn't open source. The detailed comparison is TRTT vs PSPP. On a Mac or on Linux, see SPSS alternatives for Mac and SPSS alternatives for Linux.
Open TRTT in your browser – free
FAQ
What is the difference between SPSS and PSPP?
SPSS is IBM's commercial statistics package with a very wide range of procedures, advanced charts and paid support. PSPP is a free GNU program that reads SPSS syntax and data files and covers the most common procedures, but lacks many advanced ones such as mixed models, hierarchical clustering and ordinal regression.
Is PSPP as good as SPSS?
For standard coursework analyses (descriptives, crosstabs, t tests, ANOVA, regression, reliability), PSPP gives you what you need for free. For advanced models, extensive charting or Python/R programmability inside syntax, SPSS is clearly ahead.
Can PSPP open SPSS .sav files?
Yes. PSPP reads and writes .sav files with labels and missing values, and also handles compressed .zsav and portable .por files.
Does SPSS syntax run in PSPP?
Much of it does, including DO IF, LOOP, AGGREGATE, MATCH FILES and macros. Commands for procedures PSPP hasn't implemented, and Python or R program blocks, will fail.
Is PSPP safe and legal to use?
Yes. PSPP is free software published by the GNU project under GPLv3. Download it from the official GNU PSPP page or your Linux distribution's repositories.
TRTT is an independent product. It is not affiliated with or endorsed by IBM. IBM and SPSS are trademarks of International Business Machines Corporation.