Standard Deviation (SD) -Game

Standard Deviation (SD) — Interactive Lesson

Standard Deviation (SD) — Interactive Lesson

See how SD reacts to outliers, skew, and sample size. Compare SD vs Median/IQR and SD vs SE with live numbers.

Examples mode: Tip: switch modes to show everyday vs clinical versions in questions and hints.

How to use

  • Pick a data shape (Normal-ish or Skewed).
  • Set sample size, mean, and spread; add/remove outliers.
  • Watch Mean, SD, Median, IQR, SE and a mini-histogram update.
Score: 0 Solved: 0/6

Quick refresher (plain English)

  • SD = how spread out values are around the mean (same units).
  • Use mean±SD for bell-ish data; use median (IQR) for skewed data.
  • SD ≠ SE: SD = spread of people; SE = precision of the mean (SE = SD/√n).
  • Rule: if SD ≈ mean (or bigger), suspect skew/outliers → prefer median/IQR.

Everyday vs Clinical examples

  • Cookie weights: Mean 20 g, SD 1 g → most cookies are near 19–21 g. A batch with SD 4 g is inconsistent.
  • Commute time: Mean 30 min, SD 3 → predictable commute; SD 15 → unreliable.
  • Class quiz: Same mean, SD 5 vs SD 15 → same average, different consistency.
  • Pain scores: Mean 4.5, SD 1.0 vs SD 3.0 → same average relief, different consistency across patients.
  • Length of stay: Mean 6 d, SD 7.5 (skewed) → report median 3 (IQR 2–5) for planning.
  • ALT: 42 ± 6 vs 42 ± 18 U/L → same mean, different variability → different follow-up needs.

Dataset Builder

n = 60
spread = 6
outliers = 1
Mean:
SD:
Median:
IQR (Q1–Q3):
SE (SD/√n):
95% CI for mean:
What am I seeing?

The chart bins your data into a simple histogram. With Skewed data or when you add an outlier, notice how mean and SD move more than median and IQR. That’s why we use median (IQR) for skewed data.

Q1 — Same mean, different SD

Two cookie brands have the same average weight (20 g). Brand A SD = 0.8 g; Brand B SD = 2.5 g. What does that mean?

Two pain-score groups have the same mean (4.5/10). Group A SD = 0.8; Group B SD = 2.5. What’s the clean takeaway?

Q2 — Skew alert

Your team tracks daily website visits. Most days are 200–400, but a few viral days hit 5,000. Which summary helps day-to-day planning?

Your LOS data look like the Skewed setting with a long tail. Which summary should you use for day-to-day planning?

Q3 — SD vs SE

A coffee shop’s average order value is 7.80 ± 0.10 (SE), n = 400, and someone calls it “customer variability.” What’s the right response?

A paper reports 130 ± 2 (SE), n = 400 and calls it “data variability.” What’s the right response?

Q4 — Shrink the SE (n effect)

You’re averaging cookie weights. The SD of individual cookies is fixed at 10 g. Slide n and watch how the SE of the mean changes as the sample grows.

Assume population SD = 10 units (fixed). Move n and watch SE = SD/√n change. What changes as n grows?

n = 25 SD (fixed): 10.0 SE: 2.00 95% CI half-width: 3.92

Q5 — Which bars are SD vs SE?

Two bar charts of class test scores: one shows tall error bars (scores spread out), another shows short error bars (mean is precise). Which label belongs to which?

You see a figure with two kinds of error bars: one tall (spread of people), one short (precision of mean). Which label belongs to which?

Q6 — Reporting coach

You’re summarizing daily sales, which look roughly bell-shaped. You want to describe how much sales vary day-to-day and also how precisely you know the mean. What should you show?

A trial’s primary endpoint is continuous and roughly symmetric. In Results, you want to describe patient variability and show the mean’s precision. What’s a good combo?

Diagnostics & Self‑tests

Click Run self‑tests to validate key behaviors (mode buttons present, SE math, basic stats render). These do not change your score.

Results:

  

Debrief — Keep SD honest

  • Use mean±SD when data are roughly symmetric.
  • Use median (IQR) when data are skewed or have big outliers.
  • SD ≠ SE — don’t mix them. If you see mean±SE, it’s about the mean’s precision.
  • Always show n; tiny n makes SD unstable.