Plan precision using independent units rather than repeated readings

Define the decision and evidence boundary

Monitoring design should connect a decision-relevant precision target to independent sampling units. Many readings from one sensor or one site can improve its measurement but do not automatically create new independent sites. Temporal autocorrelation and clustering can make the nominal record count overstate information.

Reconstruct the transparent method

The original planning model assumes independent normal observations and a known population standard deviation σ. The standard error of a mean isσ/√n; a nominal 95% normal half-width is 1.96σ/√n. This is a prospective design calculation, not an estimated confidence interval from the four-pair packet.

Interpret the comparison

To target half-width at most 1 mg/L, solve n≥(1.96σ)² and round upward to an integer. Doubling n reduces the width by a factor√2, while quadrupling n halves it. Increasing variability increases required sample size quadratically. The control’s planning scenarios do not manufacture additional observed pairs.

Predeclare limitations and revision

Unknown variance, small nonnormal samples, unequal weights, attrition and correlated observations require a different uncertainty calculation or design adjustment. A confidence interval also concerns an estimand under assumptions, not a guarantee that a particular intervention is safe or effective. Specify which independent units could realistically be obtained.

Worked original packet

At planned n 8 and known σ2 mg/L, standard error 2/√8≈0.7071 and half-width≈1.3859. A1 mg/L target needs ceiling(3.92²)=16 independent units. Eight technical repeats at one site do not meet that design.

Investigation

Compare n 4,8,16 at the same σ, then double σ. State what is held fixed and identify a clustering mechanism that would invalidate the independence assumption.

Changed-case transfer

With known σ3 mg/L, how many independent units meet half-width 1? What half-width does n 16 give?

Feedback

Required ceiling(5.88²)=35 units. At n 16, half-width 1.96×3/4=1.47 mg/L, so the largest displayed n does not meet the target. The model reports the shortfall rather than pretending sixteen suffice.

Sources and evidence boundary

Original prose, dataset and numerical exercises. Primary method/mechanism source, checked 30 October 2026. The versioned packet is synthetic, not observed fieldwork. No treatment implementation, site access, pollutant handling or real management authorization is requested. Preserve first reasoning and critique in the course workspace.