SPSS for UAE dissertations — the analyses your supervisor expects

SPSS is the default statistical software at UAE master’s and undergraduate programs running quantitative research methods. R and Python are gaining ground at the more research-intensive institutions, and Stata appears in economics tracks, but SPSS remains the workhorse — the software your supervisor expects you to use and the one most UAE dissertation methodology chapters reference.

This is the working summary of the analyses UAE master’s dissertation supervisors expect, the assumptions you need to check, and the reporting conventions that signal competence.

The standard sequence

A typical UAE quantitative master’s dissertation analyses data in this sequence:

1. Descriptive statistics for the sample.
2. Reliability testing if you used a multi-item scale.
3. Assumption checking for the inferential tests you plan.
4. Bivariate analysis — correlations or simple group comparisons.
5. Multivariate analysis — regression or other model-based analysis.
6. Robustness checks — sensitivity or alternative specifications.

Each step has SPSS commands and reporting conventions.

Step 1: Descriptive statistics

For each variable, report mean, standard deviation, range, and sample size. For categorical variables, frequency counts and percentages.

In SPSS: Analyze → Descriptive Statistics → Descriptives for continuous variables; Analyze → Descriptive Statistics → Frequencies for categorical.

Reporting convention: present descriptive statistics in a table (Table 4.1 or similar) before any inferential analysis. The table should include all variables that appear in the inferential analysis.

Step 2: Reliability (Cronbach’s alpha)

If your survey used multi-item scales (e.g., a 5-item scale measuring organisational commitment), report Cronbach’s alpha for each scale.

In SPSS: Analyze → Scale → Reliability Analysis. Add the scale items, run, report alpha.

Convention thresholds:

• α ≥ 0.9: Excellent
• α ≥ 0.8: Good
• α ≥ 0.7: Acceptable
• α < 0.7: Questionable; consider scale modification

UAE master’s dissertations report alpha for every multi-item scale. Below 0.7, the methodology chapter should discuss what you did about it.

Step 3: Assumption checking

Before running parametric tests (t-tests, ANOVA, regression), check the assumptions. Most UAE undergraduate dissertations skip this; UAE master’s dissertations should not.

The four standard checks:

1. Normality. Analyze → Descriptive Statistics → Explore. Look at Shapiro-Wilk (for n < 50) or Kolmogorov-Smirnov (for n ≥ 50). Also check skewness and kurtosis values. For n > 100, the central limit theorem usually saves parametric tests even with mild non-normality.

2. Homogeneity of variance. Levene’s test, automatically included in SPSS t-test and ANOVA output. p > 0.05 means equal variances assumed; p < 0.05 means use the unequal-variances correction.

3. Linearity. For regression, examine scatter plots between predictors and outcome. Non-linear relationships need transformation or non-linear models.

4. Multicollinearity. For multiple regression, check VIF (Variance Inflation Factor). VIF > 5 (some say > 10) suggests problematic multicollinearity.

Report the assumption checks explicitly in your methodology or findings chapter. Don’t hide them.

Step 4: Bivariate analysis

For two-variable relationships:

• Continuous × continuous: Pearson correlation (if both normal) or Spearman (if not).
• Continuous × binary categorical: independent-samples t-test.
• Continuous × multi-level categorical: one-way ANOVA.
• Categorical × categorical: chi-square test of independence.

SPSS paths:

• Correlation: Analyze → Correlate → Bivariate.
• t-test: Analyze → Compare Means → Independent-Samples T Test.
• ANOVA: Analyze → Compare Means → One-Way ANOVA.
• Chi-square: Analyze → Descriptive Statistics → Crosstabs with chi-square option.

Reporting convention: state the test, report the test statistic, df, p-value, and effect size. Effect sizes that UAE supervisors expect: r for correlation, Cohen’s d for t-tests, η² for ANOVA, Cramer’s V for chi-square.

Step 5: Multivariate analysis

Most UAE master’s dissertations build to multiple linear regression as the headline analysis. SPSS path: Analyze → Regression → Linear.

Reporting standard:

Predictor	β	t	p	VIF
Age	0.12	1.87	.063	1.34
Experience	0.34	4.92	< .001	2.18
Education	0.18	2.71	.007	1.45

Plus model-level statistics: R², adjusted R², F statistic, df, p-value for the overall model.

For logistic regression (binary outcome variable), use Analyze → Regression → Binary Logistic. Report odds ratios, 95% confidence intervals, and the Hosmer-Lemeshow goodness-of-fit test.

Step 6: Robustness checks

UAE master’s dissertations increasingly include robustness sections. Three common approaches:

1. Alternative specifications. Run the regression with different control variables or different operationalisations of the dependent variable.

2. Subsample analysis. Test whether the main finding holds in important subgroups (e.g., male vs female, public vs private sector).

3. Sensitivity to outliers. Re-run the analysis with outliers excluded.

A robustness section that confirms the main finding strengthens the work. A robustness section that contradicts the main finding is a finding in itself — discuss it honestly.

Where SPSS analyses lose marks

Five recurring patterns in underwritten quantitative analyses:

1. Missing assumption checks. I ran a t-test without checking normality or homogeneity of variance.

2. No effect sizes. p < 0.05 without Cohen’s d, η², or r. p-values without effect sizes are half the story.

3. Single-method reporting. Just regression coefficients, no descriptive statistics first.

4. Tables formatted as SPSS output. SPSS output tables look ugly in a dissertation. Reformat in Word with proper headers and footnotes.

5. Missing sample sizes. Every analysis should note the n that was actually included (different from the total sample if there’s missing data).

When The Essay Atelier writes SPSS analyses

Our quantitative writers handle SPSS analyses across the full range of techniques described above, plus more specialist methods (SEM in AMOS, multilevel modelling, factor analysis). We deliver the analysis with reformatted output tables, methodology-chapter prose describing what was done and why, and findings-chapter prose interpreting the results.

If you have data you’ve collected but don’t know how to analyse, message the editors with a description of the data and the research question. We’ll tell you what analyses make sense before drafting.