Skip to content

Configure Post-Stratified Variance Estimation

For each stratum hh, the estimator computes the sample mean

yˉh=1nh∑iyhi\bar{y}_h = \frac{1}{n_h}\sum_i y_{hi}

and sample variance

sh2=∑iyhi2−nhyˉh2nh(nh−1).s_h^2 = \frac{\sum_i y_{hi}^2 - n_h\bar{y}_h^2}{n_h(n_h - 1)}.

These are combined within estimation unit gg using the FIA post-stratification variance formula

vg=1n∑h[whnh+(1−wh)nhn]sh2,v_g = \frac{1}{n}\sum_h\left[w_h n_h + (1-w_h)\frac{n_h}{n}\right]s_h^2,

and then across independent estimation units as

v=∑gwg2vg.v = \sum_g w_g^2 v_g.

For total estimates, the final wgw_g coefficients are replaced by estimation-unit areas. The standard error is v\sqrt{v}.

The calculation is sparse: it aggregates only observed plot rows and uses ∑iyhi\sum_i y_{hi} and ∑iyhi2\sum_i y_{hi}^2 as sufficient statistics. Implicit zero-valued rows do not need to be materialized because they contribute zero to both sums; nhn_h remains the full stratum sample size.

ve_post_strat()

A PostStratifiedVarianceEstimator object.