Try
A stipulated normal score model has mean μ=72 points and standard deviation σ=8 points. Find a score exceeded by only 10% of this distribution. Before using inverse probability, decide whether the answer belongs above or below 72. Sketch the upper 10% region and label the unshaded lower 90%.
Normality is an assumption of this example, not a conclusion from the words “score” or “average.” This calculator workflow evaluates a specified distribution. Suitable z-score tables and interpolation or other numerical methods can also solve the problem. Regardless of method, label the tail and the units before reporting a cutoff.
Predict
Compare three requests: an upper tail of 0.10, a central interval of 0.80, and two equal extreme tails totaling 0.20. Predict which boundary values they share. A central 80% interval leaves 20% outside, split as 10% on each side. Its upper endpoint therefore equals the one-sided upper-10% cutoff, while it also has a matching lower endpoint.
Do not confuse a central 90% interval with a one-sided 10% cutoff. The central 90% interval leaves only 5% in each tail, so its upper boundary is farther from the mean. Drawing the requested region makes the probability conversion visible.
Experiment
Set mean 72 and standard deviation 8. In Calculation direction, choose Find cutoffs from probability. Request an upper-tail cutoff with probability 0.10, then calculate. An independently evaluated normal-distribution reference is approximately 82.2524125244 points, or 82.2524 to four decimal places. The teaching calculator uses numerical approximations; its last displayed digits may differ from this reference and are not guaranteed accurate digits.
Copy the cutoff before changing controls. Choose Find probability from boundaries, keep the upper-tail region, and enter that cutoff in Lower boundary: the requested region lies above this lower edge. Calculate again; the upper probability should return approximately 0.10. A forward check can expose an inverted tail even when the inverse calculation itself succeeded. Editing a control clears the old result, so retain your values in your written working.
Return to Find cutoffs from probability. Change to central coverage 0.80, then to equal outside tails totaling 0.20. Record both boundaries and compare them. These differently worded region requests have the same boundary pair; independently evaluated reference values are about 61.7475874756 and 82.2524125244. Compare meaningful rounded values rather than demanding that every displayed digit match.
| Requested region | Lower CDF targets for boundaries | Worked boundaries |
|---|
| Upper 10% | 0.90 for the single cutoff | 82.2524 |
| Lower 10% | 0.10 for the single cutoff | 61.7476 |
| Central 80% | 0.10 and 0.90 | 61.7476 to 82.2524 |
| Equal extreme tails totaling 20% | 0.10 and 0.90 | Below 61.7476 or above 82.2524 |
Make the split-tail check explicitly using Find probability from boundaries. For the lower tail, put 61.7475874756 in Upper boundary; for the upper tail, put 82.2524125244 in Lower boundary. Add the two returned probabilities. The events are disjoint, giving approximately 0.10+0.10=0.20. Then choose the between region and enter both boundaries to check coverage 0.80, or the outside region to check total tail probability 0.20 directly. The same endpoints do not make the inside and outside events interchangeable. In forward mode the two boundaries need not be symmetric; inverse central and outside requests construct a symmetric pair deliberately.
Understand
Standardization measures distance from the mean in standard-deviation units: z=(x−μ)/σ. The standard normal lower CDF Φ(z) gives the area to the left. If the desired lower probability is q, an inverse query gives z=Φ⁻¹(q), and the original-unit cutoff is x=μ+σz. For an upper probability u, set q=1−u. For central coverage c, the lower and upper targets are (1−c)/2 and (1+c)/2.
The worked inverse value is z≈1.2815515655. Multiplying by 8 and adding 72 gives 82.2524125244. The lower matching standardized cutoff is its negative. Both cutoffs are equally far from 72. This symmetry check is independent of a forward CDF call and can reveal mistaken location or scale inputs.
The tool’s scale input is standard deviation. Here σ=8 points while variance is 64 squared points. Entering 64 stretches the distribution incorrectly and mixes measurement units. Changing the mean translates every cutoff without changing standardized distances; multiplying the standard deviation by a positive factor multiplies those distances in original units. Predict those changes, then test them with the controls.
For any two cutoffs a and b with a≤b, an interior probability is Φ((b−μ)/σ)−Φ((a−μ)/σ). The two extreme tails have probability Φ((a−μ)/σ)+1−Φ((b−μ)/σ). Those tails need not be equal for arbitrary endpoints; the inverse outside mode deliberately constructs an equal-tail pair. A normal model is continuous, so changing a strict inequality to an inclusive one at a boundary does not change its probability. This differs from the preceding binomial count lesson.
This teaching interface bounds the mean magnitude by 1000, standard deviation from 0.1 to 1000, single-tail probabilities from 0.001 to 0.999 and central or equal-outside probabilities from 0.002 to 0.998. These are computational teaching bounds, not statistical rules about real populations. Display precision is not a promise of model accuracy. Use an unrounded cutoff for a forward check where possible; rounding the cutoff first can change the returned probability slightly.
Apply
Try a fresh model with mean 20 seconds and standard deviation 3 seconds. Choose Find probability from boundaries. Find the probability below 17 seconds (upper boundary 17), the probability above 23 seconds (lower boundary 23), the probability between those endpoints and the probability outside them (both boundaries 17 and 23). Calculate rather than using a supplied CDF value. Predict how symmetry should connect your four answers.
Check the forward regions after calculating
The standardized endpoints are −1 and 1. Each extreme tail is approximately 0.1586552539. Their sum is approximately 0.3173105079, and the central probability is approximately 0.6826894921. The two tails agree by symmetry, and inside plus outside equals one. The result depends on the normal model; merely knowing a population’s mean and standard deviation would not determine these probabilities for an arbitrary distribution.
Continue
In the practice, record whether you need a forward probability or an inverse cutoff, then name the region. Keep μ, σ and their units alongside the result. Verify inverse answers with a forward calculation and a location or symmetry check. The calculator query itself does not save a demonstrated-understanding judgment; the explanation and unfamiliar transfer task provide the evidence to review.
