Overview
IBM SPSS Statistics remains the default package in many psychology, nursing, education and business programs because its dialog boxes lower the barrier to entry. The trade-off is that the menus hide important choices: which variance assumption a t test row reflects, whether a scale was scored with reverse-keyed items flipped, or whether a regression silently dropped cases listwise. Most of the SPSS problems we see come from these hidden decisions rather than from the statistics themselves.
We work inside SPSS the way a careful analyst would. Every dialog we use is pasted to a syntax file (.sps), so the full analysis can be rerun from the raw data file (.sav) in one step, checked by your supervisor and adjusted later without repeating dozens of clicks. Where a question needs something SPSS does not do well, we tell you and suggest an extension, AMOS, R or another route.
You can choose a delivered analysis with an output walkthrough, a review of analyses you already ran, or live sessions where you drive SPSS and we guide each step. In every format the aim is the same: you can explain what was done and why.
Who it’s for and the problems it solves
This is for you if you are…
- Undergraduate and master's students whose program requires SPSS for a dissertation or capstone project
- Doctoral researchers in health, education or psychology with questionnaire data and a supervisor who reads SPSS output
- Researchers who ran analyses by clicking menus and now need a clean, repeatable syntax file
- Staff in institutions or NGOs analyzing survey data in a site-licensed copy of SPSS
Common problems we help with
- The Variable View is messy: string variables that should be numeric, missing value codes such as 99 counted as real answers, or no value labels
- You are not sure which row of the Independent-Samples Test table to report, or what Levene's test is telling you
- Reliability analysis gives a low Cronbach's alpha and a warning about negative average covariance
- Regression output shows fewer cases than your sample and you do not know why
- You ran everything through menus and cannot reproduce the numbers in your draft chapter
- Your version or license lacks a module (for example Advanced Statistics or Custom Tables) that a tutorial assumes
What’s included — and what isn’t
Included
- Audit of your .sav file: variable types, measurement levels, value labels, user-missing codes and duplicate cases
- Recoding and computing scale scores with RECODE, COMPUTE and MEAN.n functions, including reverse-keyed items
- Descriptive statistics, frequencies, crosstabs and assumption checks (normality plots, Levene's test, residual diagnostics)
- Inferential procedures: t tests, one-way and factorial ANOVA, ANCOVA, repeated measures via GLM, chi-square and non-parametric tests
- Correlation, linear and logistic regression, reliability analysis and exploratory factor analysis
- Mediation and moderation using the PROCESS macro, or the built-in mediation procedure in recent releases, with bootstrapped intervals
- Split-file and select-cases setups for subgroup analyses, documented so nothing is left filtered by accident
- A commented syntax file that recreates every step from the raw data
- Output pruned to the tables you actually need, with notes explaining each one
Not included
- Choosing tests to produce a significant result, or rerunning analyses on subsets until one works
- Writing the results or discussion chapter you will submit for assessment — we explain the output so you can write it yourself
- Supplying, sharing or activating SPSS licenses; you need access through your institution or IBM
- Structural equation models in AMOS beyond what your project scope includes (see our SEM service for that work)
What you receive
- Cleaned data file
- A .sav file with consistent variable names, labels, measurement levels and missing-value definitions, saved separately from your original.
- Syntax file (.sps)
- Pasted and commented syntax for every recode, compute and analysis, ordered so it runs top to bottom on the raw data.
- Annotated output
- An exported output file (.spv, PDF or Word) with the relevant tables kept and short notes on what each value means.
- Results tables
- Key tables re-laid out in APA 7 style or your department's format, ready for you to place and describe in your chapter.
- Walkthrough session
- A live or recorded explanation of the output so you can answer your supervisor's or examiner's questions about it.
How it works
Share your brief and data
Send your research questions or hypotheses, the questionnaire or codebook, the .sav or spreadsheet file and your SPSS version.
File audit and analysis plan
We check the data setup, list any issues that need your decision (for example, how to treat 'don't know' answers) and confirm which procedure answers each question.
Preparation in syntax
Recodes, scale scores and exclusions are written as syntax and checked with frequencies before any hypothesis testing.
Analysis and output review
We run the agreed procedures, check assumptions, and trim and annotate the output.
Handover and questions
You receive files and tables, then a walkthrough and a round of follow-up questions.
SPSS procedures and tools we use
- Data preparation commands
- RECODE INTO DIFFERENT, COMPUTE with MEAN.n to allow a set number of missing items, MISSING VALUES, and AUTORECODE for text categories.
- Compare Means and GLM
- Independent and paired t tests, one-way ANOVA with post hoc tests (Tukey, Games-Howell when variances differ), factorial and repeated-measures designs.
- Regression
- Linear regression in blocks for hierarchical models, collinearity diagnostics (VIF, tolerance), saved residuals, and binary logistic regression with classification tables.
- Scale and dimension reduction
- RELIABILITY with item-total statistics, and FACTOR for exploratory factor analysis with parallel-analysis support where needed.
- Non-parametric tests
- Mann–Whitney U, Wilcoxon signed-rank, Kruskal–Wallis and Friedman tests when assumptions for parametric tests are not met.
- Extensions and macros
- Andrew Hayes's PROCESS macro for moderation and mediation, and Python- or R-based extensions installed through the Extension Hub.
- Missing value analysis
- Pattern summaries and, where the module is licensed and justified, multiple imputation with pooled estimates.
Why we always paste to syntax
Every SPSS dialog has a Paste button that writes the equivalent command into a syntax window. Using it turns a series of clicks into a script you can read, comment and rerun. If your supervisor asks you to drop three participants or add a covariate, you edit one line and run the file again instead of repeating the whole analysis from memory.
Syntax is also the clearest record of what was done. A reviewer can see that a scale was computed as MEAN.4 (at least four of five items answered) rather than a plain sum, or that a SELECT IF permanently deleted cases rather than filtering them. We comment each block so you can follow it even if you have never written syntax.
- Keep the original data file read-only and save cleaned versions under new names
- Use FILTER rather than SELECT IF unless deleting cases is intended
- Turn off split file and filters at the end of each block to avoid silently subsetting later analyses
Reading SPSS output without guessing
SPSS prints more than you need, and the important number is not always where you expect. In the Independent-Samples Test table, Levene's test decides whether you read the 'Equal variances assumed' or 'not assumed' row. In regression, the Model Summary gives R² for the whole block while the Coefficients table gives the unique contribution of each predictor. In logistic regression, the Exp(B) column is the odds ratio you usually interpret, not B.
Recent SPSS releases also print effect sizes for t tests and one-way ANOVA, but not for every procedure, so partial eta squared, Cohen's d or odds-ratio confidence intervals sometimes need to be requested or calculated. We point out which values belong in your write-up and which can be left in the appendix.
Common SPSS errors and what they usually mean
Many error messages and warnings have routine causes once you know where to look. We diagnose them from your syntax and data rather than by trial and error.
- "The determinant of the covariance matrix is zero" or a non-positive definite matrix in factor analysis — usually a duplicated or perfectly correlated item
- Negative average covariance warning in RELIABILITY — reverse-keyed items were not recoded before computing alpha
- A variable missing from a dialog list — it is stored as a string or has the wrong measurement level
- Fewer cases in the output than in the file — listwise deletion across predictors, or a filter still active
- Post hoc tests not performed because a factor has fewer than three groups — the omnibus test is the result
Versions, licensing and modules
IBM releases SPSS Statistics roughly once a year; version 32 appeared in April 2026 with a built-in mediation procedure and vector autoregression, following version 31. IBM has also renamed its subscription offering to IBM SPSS Statistics Digital. Universities often license an older release or a particular edition, and some procedures (for example Complex Samples, Custom Tables or multiple imputation) depend on what your license includes. Check Help > About and your license details before starting, and tell us what you have.
We write syntax that runs on the version you use and avoid newer features where your institution is behind. If a procedure is unavailable, we suggest an alternative such as a free extension, the PROCESS macro, or running one part of the analysis in R or JASP.
What the work can look like
Illustrative example — fictional data
Sample walkthrough note: independent-samples t test
A master's student compared exam anxiety scores between students who did and did not attend a mindfulness workshop (n = 58 and n = 61). The numbers below are invented to show how we annotate SPSS output.
Which row to read
Levene's test was significant (F = 5.12, p = .026), so the variances differ between groups. Read the 'Equal variances not assumed' row (Welch's t test), where df is adjusted to 108.4 rather than 117.
What the result says
Workshop attendees reported lower anxiety (M = 2.84, SD = 0.61) than non-attendees (M = 3.19, SD = 0.88), t(108.4) = −2.53, p = .013, Cohen's d = −0.46, 95% CI [−0.83, −0.10]. The confidence interval for d excludes zero but is wide, which is worth mentioning.
Syntax used
T-TEST GROUPS=workshop(0 1) /VARIABLES=anxiety_mean /ES DISPLAY(TRUE) /CRITERIA=CI(.95). The anxiety_mean score was computed earlier as MEAN.5 of six items after recoding items 2 and 5.
Illustrative example with invented data. Real walkthroughs follow your variables, SPSS version and reporting requirements.
What you’ll need to provide
- The data file (.sav, .xlsx or .csv), anonymized before sharing
- Your questionnaire or codebook, including which items are reverse-worded
- Research questions or hypotheses and any analysis plan from your proposal
- Your SPSS version and which modules your license includes, if known
- Department or journal formatting requirements, and your deadline
You don’t need everything ready to ask for a quote — tell us what you have and we’ll say what’s missing.
What affects price and turnaround
Price depends on
- Number of research questions and procedures to run
- State of the data file: text responses to code, many reverse-worded items or several files to merge
- Whether you need a delivered analysis, a review of your own output, or guided live sessions
- Advanced procedures such as repeated-measures designs, multiple imputation or PROCESS models
- Depth of table formatting and explanation required
Turnaround depends on
- How complete and consistent your data file and codebook are when you send them
- Time needed for decisions only you can make, such as exclusion criteria
- Number of procedures and whether live sessions must be scheduled
Every project is quoted individually. How we price our work.
Questions people ask before starting
Which SPSS version do you use?
We match yours. Tell us the version shown under Help > About and whether you have a site license or a student subscription. Syntax is written to run on your release, and we flag any procedure that needs a module you may not have.
Can you just check the analysis I already ran?
Yes. Send your .sav, syntax (or a description of the menus you used) and output. We check the data preparation, the choice of procedure and the reading of each table, and list anything that should be corrected.
I only have Excel data. Can we still use SPSS?
Yes. We import the spreadsheet, set variable types and labels, and save a proper .sav file with a syntax record of the import, so the conversion is transparent.
My supervisor says my Cronbach's alpha is too low. What can be done?
First we check that reverse-keyed items were recoded and that the scale was computed on the right items. Then we look at item-total statistics. Removing items is only reasonable when you can justify it on content grounds, and we explain how to report the decision rather than deleting items silently.
Will you write up the results for me?
We provide formatted tables and explain what each result means. Writing the chapter is your work; we can then review your draft for accurate reporting of the statistics.
Is my data kept confidential?
We ask you to remove names and direct identifiers before sending. Files are used only for your project and deleted when it closes unless you ask us to keep them for revisions.