Exploring Related Columns
AddMaple automatically analyzes your data to find statistically significant relationships between columns. When you select a column, AddMaple tests it against all other columns in your dataset using appropriate statistical tests.
How It Works
AddMaple performs different statistical tests depending on the data types:
- Categorical vs Categorical: Chi-square test with Cramer's V effect size
- Numeric vs Categorical: ANOVA (for 3+ groups) or T-test (for 2 groups) with Cohen's d effect size
- Numeric vs Numeric: Pearson correlation (for normal data) or Spearman correlation (for non-normal data)
Results are ordered by statistical significance and effect size, showing you the strongest relationships first.
Viewing Related Columns
If AddMaple detects columns that are significantly related to your selected column, they will show up in the Stats tab:
The results are ordered by significance. Click on any column to automatically create a pivot chart showing the relationship.
Understanding Statistical Results
When you are viewing 2 columns pivoted together, AddMaple automatically shows a Stats Overview card that summarizes the relationship strength at a glance. This card uses color coding (green for strong relationships, orange for moderate, gray for none) and provides a plain-English interpretation of the statistical results.
You can also analyze relationships directly in pivot tables using significance testing, which highlights cells with z-score shading to show where segments perform higher or lower than expected.
In the "stats" tab, click the toggle "Statistical Test and Calculations" to see the detailed calculations behind each result. This shows you:
- The specific statistical test used (Chi-square, ANOVA, T-test, or correlation)
- P-value (statistical significance)
- Effect size measures (Cramer's V, Cohen's d, or correlation coefficient)
- Sample size and degrees of freedom
Click the toggle "Further Insights Between Column Categories" to discover which specific groups are most significantly different from the rest. This detailed breakdown only appears for chi-square and ANOVA comparisons.
Why You Might See Few or No Related Columns
If AddMaple finds few or no statistically significant relationships, this could be due to several factors:
Insufficient Data
- Small sample size: Statistical tests require adequate data to detect relationships
- Low counts: Category pairs with very few responses (expected counts less than 5) tend to reduce statistical significance
- Missing data: High amounts of missing or empty values reduce the effective sample size
Data Characteristics
- Uniform distributions: If most categories have similar response patterns, relationships may not be detectable
- Weak relationships: Some relationships exist but are too weak to reach statistical significance
- Data quality: Inconsistent or unclear category definitions can mask real relationships
Statistical Thresholds
AddMaple uses conservative statistical criteria:
- P-value must be less than 0.05 (95% confidence level)
- Effect sizes must meet minimum thresholds for practical significance
- At least 80% of contingency table cells must have expected counts ≥ 5