This calculator helps you compare candidate extraction agents based on the inferential precision they provide and the cost of deploying them across your dataset.

Current mode

Planning Mode

Use assumptions or pilot estimates to explore a proposed study before running it. Results are projections under the assumptions you specify.

Observed data

Validation Mode

Use observed gold labels and candidate predictions to estimate each agent’s realized inferential precision.

Error-structure diagnostic

Downstream estimates by error structure

Ready to run

Simulate the distribution of downstream estimates for two equally accurate agents with different error structures.

How to read this panel

Holding accuracy fixed, correlated errors can displace the downstream estimate even after an oracle attenuation correction.

Specify the study

Set the corpus, gold sample, estimand, and data-generating assumptions.

Describe agents

Vary expected accuracy, latent error dependence, and deployment price.

Simulate precision

Estimate ordinary PPI variance and jackknife-corrected ESS across repeated samples.

Compare efficiently

Identify agents that deliver the greatest inferential precision at each cost.