SPSS vs Excel for Thesis Statistics: Which Should You Use? (2026)
| Tool | Cost to you | Covers | Best for |
|---|---|---|---|
| Excel + Analysis ToolPak | Already on most machines; the ToolPak is a free built-in add-in you must switch on | Descriptive statistics, t-tests, ANOVA, correlation, regression | Straightforward descriptive studies with parametric tests |
| SPSS | Paid — but frequently licensed by the school | The full standard range including non-parametric tests and reliability analysis | Studies needing tests beyond the ToolPak, especially with a school license |
| JASP | Free, open source | Wide standard range with a point-and-click interface | Students with no school license |
| jamovi | Free, open source — desktop or cloud | Wide standard range, spreadsheet-style interface | Students who want something that feels like a spreadsheet |
Ask the right question first
Students ask “which software should I use?” when the answer depends entirely on a question they have not yet answered: which tests does your Chapter 3 say you will run?
Your statistical treatment section names specific tests tied to specific research questions. That list is your requirement. Choose the cheapest tool that runs every test on it — and if you do not have that list yet, you are not ready to choose software. Go back to your research design and settle it.
What Excel actually does — and the add-in nobody switches on
Most students underestimate Excel because they have only ever used it for averages. Excel ships with an add-in called the Analysis ToolPak, and Microsoft’s own documentation describes what it provides.
It is not enabled by default. In Excel for Windows you load it through File, then Options, then Add-Ins, then select Excel Add-ins in the Manage box and choose Go, then tick Analysis ToolPak. If it is not listed you can browse to locate it, and if it is not installed Excel offers to install it.
Once loaded, the ToolPak’s documented tools cover the ground most descriptive theses need: descriptive statistics, correlation, regression, t-tests and analysis of variance. Microsoft documents several ANOVA variants — single factor, two-factor with replication, and two-factor without replication — and notes that for a two-sample comparison you can use the worksheet function T.TEST, while for more than two samples the single factor ANOVA model applies instead.
For a study whose statistical treatment is weighted means, frequencies, a t-test and a one-way ANOVA, that is the whole requirement met with software you already have.
Where Excel stops
The ToolPak’s documented tools are parametric. Two things commonly required in Philippine theses are not among them.
Non-parametric tests. If your data are ordinal — and Likert-scale responses often are treated as such — or if your groups are small and not normally distributed, your adviser may require Mann-Whitney U, Kruskal-Wallis, Wilcoxon signed-rank or Spearman’s rho. These are not part of the ToolPak’s documented tool set.
Reliability analysis. If you wrote your own questionnaire, you will almost certainly be asked for Cronbach’s alpha. That is not a ToolPak tool either.
You can compute some of these in Excel by building the formulas yourself. Do not. A hand-built statistical formula in a spreadsheet is the single most error-prone artefact in an undergraduate thesis, it is nearly impossible for a panel to verify, and one mis-anchored cell reference silently changes your findings.
The ranked shortlist
1. Whatever your school licenses — check this before anything else
Why it ranks first. Many Philippine institutions license SPSS or an equivalent for student use through the computer laboratory or an IT services page. If yours does, the decision is made: you get the full standard range at no cost to you, your adviser almost certainly knows that package, and the output format is one your panel recognises.
How to find out. Ask your adviser, your program’s laboratory staff, or the library. Do not conclude from a friend’s guess that no license exists — this is the single cheapest question in this entire article and students routinely skip it, then pay for software the school already provides.
2. JASP or jamovi — the free options that are genuinely good
Who they suit. Any student without a school license, which in practice is a lot of people.
JASP describes itself as a fresh way to do statistics and is a free, open-source project supported by a university. jamovi describes itself as a free and open statistical spreadsheet, designed from the ground up to be easy to use and positioned explicitly as an alternative to costly proprietary statistics products. jamovi offers both a desktop download and a free cloud version, which matters if your laptop is old or shared.
Where they fall short. Your adviser may not have used them, so screenshots in your appendix will look unfamiliar and you may have to explain the output. That is a small cost and an honest one. Neither tool changes what a t-test is.
3. Excel with the ToolPak — right for a real subset of theses
Who it suits. Descriptive studies whose treatment is confined to frequencies, means, standard deviations, t-tests, ANOVA, correlation and regression.
Where it falls short. The moment your adviser asks for a non-parametric test or Cronbach’s alpha, you are switching tools mid-project — with your data already encoded to a layout the new tool may not want.
4. Paying for statistical software yourself — almost never necessary
Between a possible school license and two capable free packages, an undergraduate paying out of pocket for statistics software is usually a student who did not check the first two options. Check them.
The recommendation
Ask whether your school licenses a statistics package. If it does, use that. If it does not, use JASP or jamovi — both are free and open source. Use Excel only when your entire statistical treatment sits inside the Analysis ToolPak’s documented tools.
And whichever you choose, choose it before you encode your data. Every package expects a tidy layout — one row per respondent, one column per variable, codes rather than words — and re-encoding two hundred questionnaires because you switched tools in February is a week nobody has.
If you have already encoded everything the wrong way
This happens often enough to be worth its own answer, and it is usually recoverable without re-typing anything.
The two most common encoding mistakes are storing answers as words rather than codes — “Strongly Agree” instead of 5 — and spreading one variable across several columns when it should be one. Both can normally be fixed with a lookup or a find-and-replace on a copy of the sheet, not by re-entering questionnaires.
The mistake that genuinely costs you is one nobody notices until analysis: several rows per respondent, because a group member encoded their batch with a different layout. Check this the moment the batches are merged. Count the rows and compare against the number of returned questionnaires — if the two disagree, find out why before you run anything, because every statistic you produce afterwards will be built on a file whose row count you cannot explain to a panel.
Five habits that prevent a Chapter 4 disaster
- Encode in the standard layout from day one. One row per respondent, one column per variable, numeric codes for categories, and a separate sheet or document listing what each code means.
- Never overwrite your raw data. Keep the original encoding untouched and do every transformation in a copy. If an analysis goes wrong you want to be able to go back.
- Run one test as soon as you have thirty responses. Do not wait until collection closes to discover the tool cannot do what you need.
- Keep the output. Many manuals require the statistical output in an appendix, and panels do ask to see it.
- Write down which test answered which research question as you go. That mapping is what your Chapter 1 questions promised, and reconstructing it later from memory is how mismatches appear.
Frequently asked questions
Can I do my whole thesis in Excel?
Yes, if your statistical treatment is limited to descriptive statistics, t-tests, ANOVA, correlation and regression — the ground the Analysis ToolPak documents.
Is the Analysis ToolPak free?
It is a built-in Excel add-in that you enable through Options and Add-Ins. It is not switched on by default.
Can Excel compute Cronbach’s alpha?
It is not among the ToolPak’s documented tools. Use a statistics package rather than building the formula by hand.
Can Excel run non-parametric tests?
Mann-Whitney, Kruskal-Wallis and Wilcoxon are not among the ToolPak’s documented tools. A dedicated package handles them properly.
Is SPSS free for students?
It is commercial software, but many institutions license it for student use. Ask your school before assuming you must buy it.
Are JASP and jamovi good enough for a thesis?
Both are free, open-source statistics packages covering the standard range. The statistics they produce are the same statistics.
Will my panel accept output from free software?
Panels assess whether the correct test was run and correctly interpreted. If you expect unfamiliarity, label your output clearly.
Which should I learn if I have never used any?
Whichever your adviser can help you with. Support matters more than features when you are stuck at 11pm.
Do qualitative studies need any of this?
No. Qualitative analysis is thematic, not statistical — see our guide to choosing between the two designs.
Where does the tool get named in my thesis?
In the statistical treatment section of Chapter 3, alongside each test and the research question it answers.
The software is not the hard part
Running the test takes minutes. Knowing which test your question requires, and writing up what the result means rather than just reporting it, is the work your panel is actually assessing.
Tesify keeps your research questions, statistical treatment and results aligned across all five chapters, so nothing appears in Chapter 4 that Chapter 1 did not promise — 100% written by you, alongside 9,000+ students and 15,000+ chapters.