Fuel cycle crosstabs help uncover significant differences in data. The process begins with a broad analysis to check for any relationships between row and column variables. If noteworthy findings are revealed, a deeper examination of specific differences follows.
Summary
- Start with a global chi-square test.
- Only if it's significant, move to a pairwise column proportion test.
- Skip any comparison with low counts to maintain result quality.
- Mark significant differences directly in the crosstab.
This approach strikes a balance between rigor and clarity, enabling you to focus on what matters most.
Chi-Square Test (Global)
Testing begins by examining the overall relationship between the row variable (for example, "How happy are you?") and the column variable (for example, gender).
- If the chi-square test shows no significant relationship (p ≥ 0.05), we stop there. No pairwise comparisons are shown.
- If the test is significant (p < 0.05), we move on to test individual column differences.
This gatekeeping step reduces false positives and keeps your output clean.
Pairwise Column Proportion Tests
If the global test finds something, it checks which columns differ from each other.
- For each row (for example, "Very happy," "Pretty happy"), we compare every pair of columns (for example, Male vs. Female, Female vs. Non-binary).
- We use a pairwise column proportion test to compare column percentages.
- If any cell in a pair has a count below your selected threshold (5, 10, 20, or 30), we skip that pair to avoid unreliable results.
Example:
If 47% of respondents in Column A chose "Very happy" (150 out of 320) and 35% in Column B did the same (105 out of 300), we test whether that 12-point gap is statistically meaningful.
Cell Annotations
When a difference is significant, the cell is highlighted with a color change to light blue. Hovering your mouse over the cell shows the exact details.