MBS Thesis Data Analysis with SPSS: From Questionnaire to Chapter Four

15 July 2026 · 7 min read

The analysis chapter is where most MBS theses at TU stall. The questionnaire is collected, the deadline is close, and SPSS is a wall of menus. But the typical MBS thesis — Likert-scale questionnaire, one dependent variable, a few independent variables — follows one standard analysis path. Here it is, end to end.

Code your questionnaire into SPSS

In Variable View, create one variable per question with a short name (Q1_ServiceQuality), the right type (ordinal for Likert items, nominal for gender or bank name), and value labels (1 = Strongly Disagree through 5 = Strongly Agree). Then compute each construct's score by averaging its items — Transform → Compute Variable. Analysis happens on these construct scores, not individual questions.

Reliability first

Before any real analysis, run Cronbach's alpha on each construct (Analyze → Scale → Reliability Analysis). Supervisors and external examiners look for alpha above 0.7. If a construct falls short, check whether one negatively worded item wasn't reverse-scored — that's the cause more often than genuinely bad data.

The standard MBS analysis sequence

Descriptive statistics — demographics table, then means and standard deviations of each construct.
Correlation analysis — Pearson correlations between independent variables and the dependent variable.
Group comparisons where relevant — t-test (two groups) or ANOVA (three plus) for demographic differences.
Multiple regression — testing which independent variables significantly predict the dependent variable, with R², beta coefficients and p-values.

Presenting chapter four

Structure the chapter in the same sequence: respondent profile, descriptives, correlation, regression, then hypothesis results. Number every table, title it above, and interpret it in a sentence or two below — never drop in a raw SPSS output screenshot. State each hypothesis result explicitly: "H1 is supported (β = 0.42, p < .001)."

Common mistakes external examiners catch

Running regression without reporting or checking correlation first.
Interpreting a non-significant result (p > .05) as if it proved the hypothesis.
Tables copied from SPSS with variable names like VAR00003 still visible.
A sample size far below what the proposal promised, with no explanation.

If your data is collected but the analysis has you stuck, Assignment Nepal's statisticians can check your test choices and table presentation, or walk you through the output — message us on WhatsApp at +977 9768768340.