How to analyze questionnaire data for a research project

Analyzing questionnaire data follows a fixed order: clean the data, define your variables, check reliability, describe the sample, run the test that matches each research question, and report the results. Most students who feel stuck have skipped a step or chosen the test before they looked at their question.

You decide the plan before you collect any data. The analysis plan belongs in Chapter 3, and your committee will check that every research question has a matching test.

1. Clean the data

Open the spreadsheet and look at it before you run anything.

• Check for values outside the possible range, such as a 7 on a 1 to 5 scale.
• Decide what to do with missing answers, and write the rule down. Common choices are to exclude the case from that analysis or to use a stated method for replacing the value.
• Reverse-score items that are worded in the opposite direction, so a high score means the same thing across the scale.
• Remove duplicate or incomplete responses, and record how many you removed and why.

2. Define the variables

List each variable, how it was measured, and its type. A single 1 to 5 agreement item produces ordinal data. A scale score made by averaging several items is often treated as continuous, and your methodology chapter should state and justify that choice. Age in years is continuous. Gender, department, and yes-or-no answers are categorical.

The type decides which tests you can use.

3. Check the reliability of your scales

If you used a multi-item scale, report its reliability for your own sample. Cronbach's alpha is the usual coefficient. Report the value and the benchmark your field accepts, and name the instrument and its published reliability next to yours.

4. Describe your sample

Before any test, report who answered. Give frequencies and percentages for categorical variables, and means and standard deviations for continuous ones. A demographic table opening Chapter 4 is standard.

5. Match the test to the question

The research question and the variable types decide the test.

Your question	Common test
Are two continuous variables related?	Pearson correlation (Spearman if the data are ordinal or not normal)
Do two groups differ on a score?	Independent-samples t-test (Mann-Whitney U if assumptions fail)
Do three or more groups differ?	One-way ANOVA (Kruskal-Wallis if assumptions fail)
Do several variables predict an outcome?	Multiple regression
Are two categorical variables associated?	Chi-square test

Do not pick the test because a classmate used it. Pick it because it answers your question with your kind of data.

6. Check the assumptions

Each test assumes something about the data, such as normal distribution, similar spread across groups, or independent responses. Check the assumptions that apply to your test, report what you found, and switch to the alternative test when they fail. Skipping this step is a common reason committees question a results chapter.

7. Report the results

APA 7 asks for the test statistic, the degrees of freedom, the exact p value, and an effect size, for example t(58) = 2.31, p = .024, d = 0.60. Add a confidence interval where your field expects one. Put full results in a table, and use the text to say what the table shows. Our guide to reporting quantitative results in Chapter 4 covers the layout.

Software

SPSS is common in the social sciences and education, and your university may license it. R is free and powerful. Jamovi is free and has a point-and-click interface that suits students new to statistics. Excel handles descriptive statistics and simple charts, and it is limited for the rest.

When to ask for help

If your questions involve regression with many predictors, mediation, or a scale you built yourself, ask your supervisor or your university's statistics support early. Wrong tests found at the defense cost far more than a conversation in week three.

Not sure your analysis plan matches your questions?

The Two Dissertation Moms offers a Research Design and Methodology Review that checks whether each research question has a matching method, sample, and test before your committee does.