Use area weights when survey effort is deliberately unequal

Define the ecological target

Stratification divides the target population into defined groups before sampling. It can ensure representation of rare habitats and improve precision when groups differ. Allocating equal effort to two habitats does not imply they occupy equal area. Distinguish why samples were allocated from the population quantity being estimated.

Specify the observation model

For area-average density, multiply each stratum mean by its fraction of target area and add. Total abundance is each stratum’s density times its area, summed across strata. These calculations assume that the supplied stratum estimates represent their strata; an area weight cannot correct biased site selection inside a stratum.

Work through the sampling logic

The model contains 100 hectares with a high-density and low-density habitat. The low-density mean is fixed at 5 individuals per hectare, while high-density mean and area share vary. Equal sample allocation gives an unweighted mean of the two means, which generally differs from the target-area mean.

Challenge the inference

A survey may intentionally oversample a small habitat because it is rare, variable or especially important. That design can be scientifically useful if its inclusion probabilities and analysis weights are retained. Dropping the weights afterwards changes the estimand. Conversely, an area-weighted overall mean can conceal a serious decline within the small habitat, so report relevant stratum results as well.

Design a defensible survey

Uncertainty requires more than two mean values. Sample sizes, within-stratum variation, finite-population structure and design dependence inform a variance estimate. The displayed point calculation does not estimate those quantities. Document the habitat map date and test sensitivity if actual areas are uncertain or changing.

Worked original packet

At 20 hectares of high-density habitat with mean 20 and 80 hectares at mean 5, the target mean is 8 individuals/ha and total is 800. Equal sample allocation would give unweighted mean 12.5, overemphasizing the small dense habitat for this area-average target.

Investigation

Compare high-habitat shares 20 and 80 at density 20. Keep the within-habitat means fixed, reconstruct totals and explain what an oversampling strategy can and cannot improve.

Changed-case transfer

A target has 40 hectares at density 30 and 60 hectares at density 5. Find the area mean, total and equal-allocation mean.

Feedback

Area mean 0.4×30+0.6×5=15 individuals/ha; total 1500. Equal-allocation mean 17.5 differs because sample allocation and area composition are different.

Sources and evidence boundary

Original instructional prose, values and exercises. Context checked 30 October 2026: primary agency source. All observations are synthetic. No organism collection, trapping, tagging, release or unsafe-site access is requested; a real study requires qualified design, permission and supervision. Public feedback is practice, not independently verified mastery.