Infer a normal-model mean or standard deviation

What should be done first when an empirical-rule percentage is linked to a boundary?

Under the empirical-rule approximation, the boundary 42 seconds has z=−1. Keep that condition and vary the positive standard deviation s. Watch the compatible mean 42+s and predicted z=2 boundary 42+3s. Find the setting where that second boundary is 60 seconds

Does the single lower-boundary condition determine both parameters?

Inferring normal-model parameters

Empirical-rule percentages identify signed standard-deviation boundaries. Translate each boundary into x = mean + z × standard deviation, then solve for the missing parameter or parameters

1. 1Map percentages to signed z-values Under the stated empirical-rule normal approximation, a tail of about 2.5% below a boundary puts that boundary two standard deviations below the mean; a tail of about 2.5% above a boundary puts it two standard deviations above the mean \(\displaystyle P(X<x)\approx2.5\%\Rightarrow z_x\approx-2,\quad P(X>x)\approx2.5\%\Rightarrow z_x\approx2\)
2. 2Write the boundary equation Use the actual boundary, mean and standard deviation in x = mean + z × standard deviation \(\displaystyle x=\mu+zs\)
3. 3Use enough equations One boundary equation can find one missing parameter, while two independent boundary equations are needed when both mean and standard deviation are unknown \(\displaystyle x_1=\mu+z_1s,\quad x_2=\mu+z_2s\)
4. 4Check the result Substitute the parameters back into the boundary statements and check that the standard deviation is positive \(\displaystyle s>0\quad\text{and}\quad x=\mu+zs\)

Given values

\(\displaystyle x=38,\quad\mu=50\)

The mean is 50 marks and a boundary is 38 marks. Find the signed change from the mean to the boundary

Given values

\(\displaystyle x-\mu=-12,\quad z=-2\)

The boundary is 2 standard deviations below the mean. Use the signed change −12 marks to find the standard deviation

For a normal-model empirical-rule approximation, about 2.5% of observations lie below a boundary. Which signed boundary is intended?

Translate each tail into one equation

Use x=mean+z×s after mapping the stated tail to its signed boundary. With one parameter known, one informative boundary can determine the other. With both unknown, use two distinct signed z-boundaries: s=(x₂−x₁)/(z₂−z₁), then mean=x₁−z₁s

1. 1Check information and consistency Two copies of the same equation are not independent. A boundary at z=0 identifies the mean but gives no spread. Check s>0, preserve units, and substitute into every original condition. A negative inferred s signals inconsistent conditions or a mapping error, not a valid normal model
2. 2State the approximation The empirical rule uses rounded percentages. Treat its integer z-values as the stated approximation for this exercise; do not claim an exact population fit or silently replace them with inverse-normal calculator values

Given values

\(\displaystyle x=42,\quad z=-1,\quad s=6\)

Using the empirical-rule approximation, 16% lies below 42 seconds. If s=6 seconds, estimate the mean

Given values

\(\displaystyle 42=\mu-s,\quad60=\mu+2s\)

Two empirical-rule boundaries give 42=mean−s and 60=mean+2s. Find s.

Only 42=mean−s is known, and both parameters are unknown. Does it determine a unique model?

Infer parameters and test the conditions

For a fictional normal model of task durations, about 16% lie below 42 seconds and about 2.5% lie above 60 seconds. Use the empirical-rule approximation to map both tails, form two equations and estimate the mean and standard deviation. Sketch both boundaries relative to the mean and check both original conditions

1. 1Verify and investigate Find the implied central 95% interval. Explain why the first condition alone is insufficient by giving two positive-spread parameter pairs. Separately test the proposed conditions x=60 at z=−1 and x=42 at z=2; explain why they cannot define a positive-spread normal model
2. 2Choose and verify technology Choose simultaneous-equation or spreadsheet functionality, record the tail-to-z mapping and equations, and solve. Independently verify the subtraction and substitute into both conditions. State why these are estimates based on rounded empirical-rule percentages

Model response: two independent conditions, two checks

The lower 16% boundary maps to z=−1 and the upper 2.5% boundary to z=2. Thus 42=mean−s and 60=mean+2s. Subtraction gives 18=3s, so s=6 seconds and mean=48 seconds

1. 1Verify the model The two scores are (42−48)/6=−1 and (60−48)/6=2, matching the original tails. The central 95% interval is 48±2×6, or approximately 36–60 seconds. Label 36,42,48,60 on an actual axis with z=−2,−1,0,2 beneath them; shade below 42 and above 60 separately
2. 2Information and inconsistency Both (mean,s)=(48,6) and (52,10) satisfy the first condition alone. Repeating that condition adds no new information. For the separate proposal with a boundary of 60 at a standard score of −1 and a boundary of 42 at a standard score of 2, the calculation gives s=(42−60)/(2−(−1))=−6, impossible for a standard deviation. With s>0 the boundary x must increase with z, so the reversed ordering already reveals the inconsistency
3. 3One reproducible workflow Enter x₁=42,z₁=−1,x₂=60,z₂=2. Calculate s=(x₂−x₁)/(z₂−z₁) and mean=x₁−z₁s. Verify the denominator is nonzero and s positive, then check mean+z₁s=42 and mean+z₂s=60 independently. Record the empirical-rule convention: using numerical inverse-normal functions for exact 16% and 2.5% would be a different approximation

A fitted result satisfies one boundary but gives a negative standard deviation. Should it be accepted?