1) Two Levers, One Bill
You can squeeze the margin by lowering confidence or raising n. One costs honesty. The other costs effort. Pick your bill and pay it on purpose.
2) Use t. Full Stop.
We don’t know σ. We almost never do. That’s why t exists. Feed it df = n − 1, take the critical value, and stop pretending.
3) n − 1: The Last Seat Is Taken
With a fixed mean, the final value has no freedom. That’s degrees of freedom: n − 1. Just like the last plate of broccoli at the cookout.
4) Guardrails, Not Red Tape
If your sampling is sloppy or your population shape is unknown or sample is tiny, don’t compute. Guardrails keep your conclusions roadworthy.
5) Speak Both Dialects
Leaders want bounds. Analysts want estimate ± margin. Say both in one breath. Everyone hears what they need.
6) Intervals That Move the World
If the interval doesn’t change staffing, budgets, timing, or thresholds, it’s not insight… it’s ornament. Measure what earns action.
7) Automate the Grind, Own the Meaning
Let Excel crunch. Your value is the name, the level, and the next step. Explain the “why,” not the formula.
8) Fix How Before How Much
More data can make you confidently wrong. First clean the method (sampling and measurement), then scale.