Why Supervisors Reject SPSS Results: 15 Common Mistakes Researchers Make

“I did my analysis carefully, so why were my marks cut?” If you are asking yourself this, you are not alone. Most SPSS problems in dissertations come from a small set of repeated mistakes, and almost all of them can be avoided.

15 common SPSS mistakes students make

1. Selecting the wrong statistical test

Using a t-test where ANOVA or regression is needed can invalidate your results. Match the test to your research question, variable types and design. If you are unsure, compare the Mann-Whitney U test and the independent t-test.

2. Misinterpreting the results

A significant p-value does not “prove” a hypothesis. Wrong interpretation is one of the quickest ways to lose credibility with a marker.

3. Skipping assumption testing

Did you check normality, homogeneity of variance or multicollinearity? Supervisors expect it and examiners look for it.

4. Weak reporting of results

Copying SPSS tables is not enough. Explain in clear academic language what the numbers mean for your research question.

5. Confusing variable names with variable labels

SPSS variable names cannot contain spaces or special characters, or start with a number. Use short, standard codes in the Name column and put the full description in the Label column.

6. Forgetting to define missing values

Codes such as 99 or 999 must be declared as missing in Variable View. Otherwise SPSS treats them as real data and your means and correlations will be wrong.

7. Treating Likert categories as truly continuous

A single Likert item is ordinal. The distance between “Agree” and “Strongly agree” is not necessarily the same as between “Disagree” and “Neutral”. Choose tests with this in mind.

8. Running tests before checking statistical assumptions

Check the distribution of your data before running parametric tests, not after.

9. Not checking the sample size in the output

If your dataset has 500 rows but the output shows N = 50, a filter, split file or missing data problem is hiding cases. Always check N.

10. Entering data with inconsistent coding

Using Male = 1 in one section and Male = 0 in another creates misleading results. Set a clear coding scheme before entering data.

11. Ignoring scale reliability

Check Cronbach’s alpha or another reliability measure before testing hypotheses with questionnaire scales. Before testing, also complete basic SPSS data cleaning.

12. Failing to reverse-code negatively worded items

If negatively worded questions are not reverse-coded, reliability drops and relationships between variables are distorted.

13. Using parametric tests on non-normal data

The independent t-test, ANOVA and Pearson correlation assume approximately normal data. If that fails, consider Mann-Whitney U, Kruskal-Wallis or Spearman correlation.

14. Reporting only p-values

Supervisors increasingly expect effect sizes such as Cohen’s d, eta squared or R² to show how large and practically important a result is.

15. Not saving the SPSS syntax file

A syntax file is a complete record of every step. It makes corrections, revisions and replication much easier.

My advice before you submit

Before submitting, ask yourself: Am I using the right test? Have I checked every assumption? Have I interpreted the findings correctly? A little extra work now saves weeks of revisions later.

Frequently asked questions

Why do supervisors reject SPSS results?

Usually because the wrong test was used, assumptions were not checked, results were misinterpreted, or the write-up only copied SPSS tables without explaining them.

What is the most common SPSS mistake?

Choosing a statistical test that does not match the research question and variable types is one of the most common and most serious mistakes.

Should I report effect sizes in SPSS results?

Yes. Report effect sizes such as Cohen’s d, eta squared or R² alongside p-values to show the size and practical importance of the result.