trqwe: Performance oriented statistical metrics and utility functions.

Description Usage Arguments Value See Also Examples

Calculates the sample size required for a TOST test, given a sample size through exact statistics or simulation

1
2
3

tost_power(mu_A, mu_B, sd_A, sd_B, delta = 0.3, kappa = 1,
  alpha = 0.05, power = 0.8, method = "equivalence",
  n_iterations = 1000, paired = F, sample_var = 0)

`mu_A`	mean of group A
`mu_B`	mean of group B
`sd_A`	SD of group A
`sd_B`	SD of group B
`delta`	maximum tolerated difference (i.e., abs(mu_A - mu_B))
`kappa`	sample size of group A / sample size of group B (nA / nB)
`alpha`	significance threshold (0.05 default)
`power`	Desired power level (0.8 default)
`method`	Whether to use the exact statistics or simulation. "Exact" - exact statistics, "equivalence" - simulate using equivalence package tost function, "TOSTER" - simulate using TOSTER functions
`n_iterations`	Number of simulation iterations
`paired`	If the TOST is paired (kappa should be 1)
`paired_var`	Sample variance for paired TOST

For exact statistics, returns a list of two values: minimum sample size to achieve desired power and the power at that sample size. For simulation, returns a series of sample sizes and achieved power up to the desired power.

http://powerandsamplesize.com/Calculators/Compare-2-Means/2-Sample-Equality Chow S, Shao J, Wang H. 2008. Sample Size Calculations in Clinical Research. 2nd Ed. Chapman & Hall/CRC Biostatistics Series. page 58.

res_equivalence <- tost_power(mu_A=1, mu_B=1, sd_A=0.2, sd_B=0.2, delta=0.2, method="equivalence")
res_toster <- tost_power(mu_A=1, mu_B=1, sd_A=0.2, sd_B=0.2, delta=0.2, method="TOSTER")
res_exact <- tost_power(mu_A=1, mu_B=1, sd_A=0.2, sd_B=0.2, delta=0.2, method="exact")

plot(res_equivalence$nB, res_equivalence$Power, xlim=c(1,20), ylim=c(0,1), type="s", main="TOST power analysis")
lines(res_toster$nB, res_toster$Power, col="green", type="s")
abline(h=0.8, col="blue", lty=2)
points(res_exact$nB, res_exact$Power, col="red", pch=18, cex=2)
legend("topleft", legend=c("equivalence", "TOSTER", "exact"), col=c("black", "green", "red"), pch=1)

res_equivalence <- tost_power(mu_A=1, mu_B=1, sd_A=0.2, sd_B=0.2, delta=0.1, method="equivalence")
res_exact <- tost_power(mu_A=1, mu_B=1, sd_A=0.2, sd_B=0.2, delta=0.1, method="exact")
plot(res_equivalence$nB, res_equivalence$Power, xlim=c(1,80), ylim=c(0,1), type="s", main="TOST power analysis")
abline(h=0.8, col="blue", lty=2)
points(res_exact$nB, res_exact$Power, col="red", pch=18, cex=2)
legend("topleft", legend=c("equivalence", "exact"), col=c("black", "red"), pch=1)