Eight Tiny Experiments for Honest Distributions & Clear Decisions

1) Report Two Centers When Skewed  
If the tail is long, report both mean and median. The gap is the story.

2) Write the Five Numbers  
Before plotting, write min, Q1, median, Q3, max on a sticky note. If the note is clear, the chart will be.

3) Outliers: Trim, Transform, or Tell  
Three honest moves: investigate and fix errors, use robust stats (median/IQR), or analyze twice (with/without) and tell.

4) Spread Drives Strategy  
Two classes with the same median can need different plans if one is wide and one is tight. Teach to the spread you have.

5) Percentiles Talk Human  
“Top 15%” lands better than “z = 1.04.” Translate for decisions, not for impressing statisticians.

6) Compare Boxes, Not Whiskers  
When comparing groups, focus on medians and boxes (IQR). Whiskers are scenery; the box is the neighborhood.

7) Plot Before You Correlate  
A quick scatterplot will save you from nonsense r’s. Shape first, statistic second.

8) One Conclusion, One Constraint  
End your write-up with one clear inference and one explicit limit: “We can compare these sections this term; we can’t generalize beyond it.”

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.