Data analysis turns your survey or interview answers into results. Good analysis does not look for complicated tests, but for the right ones for your objectives and data. This guide summarises the usual steps in a TFG.
Preparing the data
Export the responses to a spreadsheet and check their quality. Remove duplicates, check incomplete answers and code the variables: for example, 1 to 5 for a Likert scale.
Keep an untouched copy of the original data. If the board asks, you can explain every change.
Descriptive statistics
Always start by describing. For categorical variables, such as sex or sector, use frequencies and percentages. For numerical variables or scales, use the mean, median and standard deviation.
In small samples, give the number of cases with every percentage. “50%” says little if it means 4 people out of 8.
Choosing the test
The test depends on the type of variables and your objective:
- Relationship between two categorical variables: chi-square.
- Difference in means between two groups: Student’s t-test, or Mann-Whitney U if the data are not normally distributed.
- Difference between more than two groups: ANOVA, or Kruskal-Wallis as a non-parametric alternative.
- Relationship between two numerical variables: Pearson or Spearman correlation.
If you are unsure, ask your supervisor before calculating. A wrongly chosen test invalidates the result.
Interpreting without overstating
A significant result means the relationship is probably not due to chance. It does not prove causation. A correlation between remote work and satisfaction does not show that one causes the other.
Always report the test value and its significance, and explain the result in plain words.
Qualitative data
If you conducted interviews, thematic analysis is the most common approach. You read the transcripts, mark passages related to your objectives and group them into categories. Then you compare categories across participants.
Software
Excel is enough for descriptive statistics and some simple tests. Jamovi is free and easy to use. SPSS is common at universities, and R offers more options. For qualitative analysis there are programs such as ATLAS.ti or NVivo, though a well-organised spreadsheet also works.
Presenting the results
Present the results in the order of your objectives. Each table or chart has a number and a title, and is discussed in the text. Deeper interpretation belongs in the discussion and conclusions.
Common mistakes
- Applying tests without checking they fit the data.
- Confusing correlation with causation.
- Tables with no commentary in the text.
- Percentages without case counts in small samples.
Further reading
See our guides to the TFG survey and the methodology. If you need help with the calculations, our statistical analysis service works with your data and explains each result.
The data and the interpretation in the text remain yours.