Probability Using the Normal Distribution

Learning Objective

Use the normal distribution to compute probability

Why This Matters

When your professor curves an exam, they're using this exact math -- deciding what percentage of scores fall below 60 or above 90. When a fitness tracker flags your resting heart rate as "abnormal," it computed a probability from a normal distribution. The shaded area under this curve is behind every standardized test percentile, every medical lab reference range, and every quality control decision on every product you own.

How to Use This Simulation

Select a preset scenario or set μ and σ with the sliders. The curve updates immediately.
Choose a shading mode -- left tail, right tail, between, or outside -- to ask a probability question.
Drag the bound slider or type a value, and watch the shaded area and probability update in real time.
Read the formula display below the chart -- it shows every calculation step with substituted values.

Scenario:

Preset loaded: μ = , σ = . Sliders disabled -- values are outside the generic slider range.

μ (mean):

0.0

σ (std. dev.):

1.0

Probability Mode

Value (a):

Choose a shading mode and enter bound value(s) to see the probability calculation with substituted values.

Try switching modes. Same curve, same bounds, four different probabilities. Every normal distribution question reduces to one of these four area types.

Distribution

X ~ N(0, 1)

Probability

Z-Score(s)

What's Happening

Quick Check

For any normal distribution X ~ N(μ, σ), what is P(X < μ)?

Try This

Load the Final Exam Scores preset. The professor says scores below 60 earn an F. Before looking at the probability, predict: will more or fewer than 10% of the class fail?

Now read the probability. Were you right? Switch to Right Tail mode and enter a = 90. What fraction earns an A? Write one sentence comparing the two probabilities.

Spotify reports that the average song is 3.5 minutes with σ = 0.8 minutes. Load the Spotify Song Length preset and select Between mode with bounds a = 2.5 and b = 4.5. Standardize both bounds by hand: z₁ = (2.5 − 3.5)/0.8 and z₂ = (4.5 − 3.5)/0.8. Compute P(z₁ < Z < z₂) = Φ(z₂) − Φ(z₁) and verify against the simulation.

Now switch to the Standard Normal preset and enter those z-values as bounds. Explain why you get the same probability.

You're the quality engineer. Your bolt process has μ = 10.00mm, σ = 0.05mm. Current spec: reject anything outside 9.90–10.10mm. Management proposes widening tolerance to 9.85–10.15mm to cut waste. Load the Bolt Diameter preset and use Outside mode to compute the reject rate under both specs.

If you ship 50,000 bolts per day, how many additional defective bolts per day does the wider spec let through? Write a one-sentence recommendation that includes both reject rates.

Instructor Notes

Teaching Notes

The strongest moment in this simulation is when a student toggles through all four modes on the same distribution and sees four different probability values emerge from the same curve. That toggling action -- left tail, right tail, between, outside -- is the moment "every probability question is just an area question" clicks. Let them toggle freely before explaining the formulas.

The "between" mode with bounds set at μ ± σ produces 0.6827 -- the 68 in the 68-95-99.7 rule. The simulation now shows a badge when students discover this value, reinforcing the connection. Students who discover this by entering the bounds themselves, rather than being told, remember it better. Consider assigning the Starter tier challenge as a warmup activity.

The bound sliders are intentionally synced to the typed inputs. Some students explore better by dragging (continuous feedback), while others want to type exact values (precision). Let both approaches coexist -- the conversation about "why did you prefer dragging vs typing?" can surface different learning styles.

Common Student Errors

Treating P(X = a) as nonzero for a normal distribution. Students from discrete probability expect P(X = 72) to produce a real number. Reinforce: for continuous distributions, probability comes only from intervals (area), never from points.
Computing the right tail as "1 minus the left tail" but using the wrong CDF argument. P(X > a) = 1 − P(X ≤ a), not 1 − P(X ≤ −a).
In "between" mode, subtracting the CDFs in the wrong order: they compute CDF(a) − CDF(b) instead of CDF(b) − CDF(a), producing a negative probability. The formula display in this sim always shows CDF(upper) − CDF(lower).
Believing that changing σ changes P(X < μ). It doesn't -- that probability is always 0.5 by symmetry.

Discussion Questions

A medical lab test is "normal" if the result falls within μ ± 2σ. What percentage of perfectly healthy people will get an "abnormal" flag? What are the consequences of that false-positive rate in a population of 10 million?
Why do quality engineers care about the "outside" probability (both tails) instead of just one tail? What does it mean practically if a bolt is 0.2mm too wide versus 0.2mm too narrow?
If you knew only the probability P(X < 120) = 0.9332 and that the distribution is normal with σ = 15, could you determine μ? How?

Exam Connection

Most introductory statistics exams include 2-3 items requiring students to compute P(X < a), P(X > a), or P(a < X < b) for a given normal distribution. Common formats: "find the probability," "find the percentage," "what fraction of the population exceeds..." This simulation directly rehearses all three computation types. The Challenge tier (comparing reject rates under two specs) previews the logic of confidence interval construction.