2x2 Contingency Table Calculator

Prediction-powered odds ratio and relative risk sample size planning

This calculator uses prediction-powered inference (PPI/PPI++) formulas from this package for planning studies that compare two groups via a 2x2 contingency table. It supports both odds ratio (OR) and relative risk (RR) effect measures.

Interactive PPI Tool

2×2 contingency table calculator

Compare two groups using odds ratio or relative risk with a binary PPI++ surrogate.

Setup: Two groups (exposed/control) with a binary outcome Y and a binary surrogate prediction f with known sensitivity and specificity.

Odds ratio: Tested via logistic regression with PPI++ sandwich variance. Equivalent to a Wald test for log(OR).

Relative risk: Tested via delta method on per-group PPI++ mean estimates. Wald test for log(RR).

Methods: details and derivations are on Variance & Methods.

Odds Ratio (OR) Relative Risk (RR)

Inputs

Group probabilities

Control P(Y=1) (%)

Exposed P(Y=1) (%)

Prevalence of exposure (%)

Unlabeled N (total)

Alpha (two-sided)

Target power (%)

Expected non-response (%)

Surrogate quality

Sensitivity

Specificity

Results

PPI++ n (total labeled) -

PPI++ recruitment n (adj.) -

Classical n (no surrogate) -

Labeled saving vs classical -

True OR -

Oracle lambda -

Classic

-

PPI++

-

PPI adj.

-

Formula note will appear here.

Inputs

Group probabilities

Control P(Y=1) (%)

Exposed P(Y=1) (%)

Prevalence of exposure (%)

Unlabeled N (total)

Alpha (two-sided)

Target power (%)

Expected non-response (%)

Surrogate quality

Sensitivity

Specificity

Results

PPI++ n (total labeled) -

PPI++ recruitment n (adj.) -

Classical n (no surrogate) -

Labeled saving vs classical -

True RR -

log(RR) -

Classic

-

PPI++

-

PPI adj.

-

Formula note will appear here.

Uses the same formulas as power_ppi_2x2() in the pppower R package. See Variance & Methods for derivations.