By AOLYTIX Research Desk 15 min read · Data Analysis · Nigerian Postgraduate Research
Let me tell you something I've seen too many times.
A student spends four months collecting data — questionnaires, fieldwork, follow-up calls, the whole process. Then they sit down to analyse it and freeze completely. Not because they aren't smart. But because nobody actually taught them what to do after the data comes in.
That's what this guide is for.
SPSS is the most widely used data analysis tool in Nigerian universities, and it isn't going anywhere. UNILAG, ABU, UNN, LASU, OAU — departments across the country require SPSS output in submitted dissertations. But most students figure it out by trial, error, and panicked WhatsApp messages to classmates at 11pm.
You deserve better than that.
Excel is fine for budgets and basic calculations. It's not built for research. SPSS was purpose-built for social science data — it handles hundreds of variables, runs complex statistical tests in seconds, and produces output that examiners and journal reviewers recognise instantly.
More importantly: your supervisor will ask for it. Your defence committee will expect it. And if you ever want to publish your findings, SPSS output is universally understood.
That said — knowing the software isn't enough. You need to know what to run, when, and why.
This sounds obvious. It isn't.
I've reviewed project chapters where students ran five different statistical tests with no clear reason for any of them. The data was there, the tables were there — but the analysis told no story. When the examiner asked "why did you use regression here?", they had no answer.
Before you open SPSS, ask yourself three questions:
Your answers determine everything — which tests you run, how you structure Chapter 4, and how you interpret your results.
The most common mistake isn't in the analysis. It's in the data entry.
Garbage in, garbage out — and I mean that literally. We once had a client bring us a dataset where "Strongly Agree" had been coded as both 1 and 5 in different sections of the same questionnaire. Three hours of analysis, all worthless.
Here's how to do it right:
Assign numerical codes to every response option. For a 5-point Likert scale: - Strongly Agree = 5 - Agree = 4 - Neutral = 3 - Disagree = 2 - Strongly Disagree = 1
In SPSS, every variable is a column. Every respondent is a row. So if you had 150 respondents answer 30 questions, your data view should have 150 rows and 30+ columns.
In Variable View, set three things for every variable: - A short, clear name (Q1, Gender, Income — not "Question_One_About_The_Gender_Of_The_Respondent") - The correct measurement level: Nominal for categories like gender or religion, Ordinal for ranked scales, Scale for continuous data like age or income - Value labels — so when you see "2" in your output, SPSS can tell you it means "Female"
Get this right and everything downstream becomes easier. Get it wrong and you'll be redoing it under deadline pressure.
Don't skip this. Not even if you're running late.
Go to Analyze > Descriptive Statistics > Frequencies and run it on all your variables. You're looking for:
Examiners spot uncleaned data immediately. The dead giveaway is a table where N doesn't match across sections — it tells them you have missing data you haven't accounted for.
Before you test any hypothesis, describe your sample. This is what goes in Chapter 4, Section 1 — the demographic profile of your respondents.
Go to: - Analyze > Descriptive Statistics > Frequencies for categorical variables (gender, education level, department) - Analyze > Descriptive Statistics > Descriptives for continuous variables (age, years of experience, income)
Report means, standard deviations, and frequencies with percentages. A table showing that 63% of your respondents are female, the average age is 34, and 71% hold a first degree — that's not filler. It's essential context for interpreting everything that follows.
This is where most students freeze. There's a fear that choosing the "wrong" test will invalidate everything. But the logic is actually straightforward once you know the decision rule.
| What you're trying to find out | Test to use |
|---|---|
| Is there a relationship between two scale variables? | Pearson Correlation |
| Is there a relationship between two categorical variables? | Chi-Square Test |
| Is there a significant difference between two groups? | Independent Samples T-Test |
| Is there a difference among three or more groups? | One-Way ANOVA |
| Can one or more variables predict an outcome? | Linear or Multiple Regression |
| Is my Likert scale internally consistent? | Cronbach's Alpha (Reliability Analysis) |
A quick note on Cronbach's Alpha — a lot of students skip this because their supervisor didn't specifically ask for it. Don't. If you used a multi-item scale (say, 5 questions all measuring "job satisfaction"), you need to show the scale is reliable before you use it in any further analysis. Anything below 0.70 is a problem you need to explain.
If you're a 400-level student: you most likely only need descriptive statistics, frequencies, and one correlation or chi-square test. Skip the regression section below and focus on Steps 3–5. That covers 90% of final year project requirements.
SPSS gives you a lot of numbers. Most of them don't belong in your dissertation.
Here's what matters:
Pearson Correlation - r — the correlation coefficient. Ranges from -1 to +1. Values above 0.5 or below -0.5 are typically considered moderate to strong. - Sig. (2-tailed) — your p-value. Below 0.05 = statistically significant. That's your threshold.
Independent Samples T-Test - Sig. (2-tailed) — again, below 0.05 means a significant difference exists between your two groups - Also report the mean for each group so readers understand the direction of the difference, not just that one exists
Regression - R² — how much of the variation in your outcome variable is explained by your predictors. An R² of 0.45 means your model explains 45% of the variance. - Beta (β) coefficients — which predictors are significant and in what direction? A positive beta means more of that variable leads to more of the outcome. Negative means the opposite. - Sig. — for each predictor, below 0.05 = significant contribution to the model
One thing I see constantly in student work: they paste a table showing Sig. = .000 and write "the result is 100% significant." That's not what .000 means. It means the p-value is less than 0.001 — which is significant, yes, but it's not certainty. It's a probability statement. Write it correctly: "The result was statistically significant (r = .67, p < .001)."
SPSS output is not dissertation-ready. You need to reformat it.
Most Nigerian universities require APA 7th edition table formatting. That means: - No vertical lines in tables (APA tables use horizontal lines only) - Table number and title above the table, notes below - Every table referenced in your text before it appears
For charts, SPSS's built-in graphics are passable but not beautiful. If presentation matters, export your data to Excel for cleaner charts — or use R's ggplot2 if you're comfortable with it.
And the golden rule: don't drop tables into your dissertation and move on. Every table must be followed by a paragraph that interprets what it shows and links it back to your research question or hypothesis.
I've reviewed enough dissertations to have a list.
Running a mean on a nominal variable — Gender is not a scale. Religion is not a scale. Running descriptives on these gives you a "mean gender" of 1.6, which means absolutely nothing.
Ignoring test assumptions — Parametric tests like T-Test and ANOVA assume your data is normally distributed and variances are roughly equal. Violate these without acknowledgement and a sharp examiner will catch it.
Misreading .000 as "100% significant" — Already covered above, but it bears repeating because it's that common.
Pasting all your SPSS output into the appendix and calling it analysis — Output is not analysis. Interpretation is analysis.
No link between your tests and your research objectives — Every statistical test you run must connect explicitly to a stated objective or hypothesis. If you can't say "This test addresses Objective 3," don't run it.
There's no shame in getting help. Some of the strongest dissertations I've seen were produced by students who knew exactly what they needed and went and got it.
If your dataset is complex — multiple scales, several hypotheses, a large sample — professional analysis support can save you weeks of back-and-forth with your supervisor. If you're close to a submission deadline and stuck on a specific section, targeted help on that section alone is often enough.
At AOLYTIX Group, we work with master's and PhD students across Nigeria on SPSS data analysis — from data coding and cleaning all the way through to results write-up. We've handled everything from 80-respondent surveys for master's projects to multi-site datasets for doctoral dissertations.
Stuck on Cronbach's alpha or unsure which test fits your hypothesis? Send us your questionnaire and research questions. We'll give you a free recommendation — no strings attached — on the right tests to run.
Get in touch with AOLYTIX Group →
SPSS is not that hard once you understand the logic:
Your dissertation is four or five years of academic work culminating in one document. The analysis chapter shouldn't be the part that lets you down.
AOLYTIX Research Desk is the publishing arm of AOLYTIX Group — a Nigerian academic research and consulting firm supporting postgraduate students, researchers, and organisations across Africa with data analysis, dissertation support, and research consulting.