TrialSize (version 1.4)

Propensity.Score.nostrata: Propensity Score ignoring strata

Description

Combining data across J strata. Still use weighted Mantel_Haenszel test.

Ho: \(p_{j1}=p_{j2}\),

Ha: \(p_{j2} q_{j1}/(p_{j1} q_{j2})\)=phi, which is not equal to 1

Usage

Propensity.Score.nostrata(alpha, beta, J, a, b, p1, phi)

Arguments

alpha

significance level

beta

power = 1-beta

J

There are totally J stratas.

a

a=c(a1,a2,...,aJ), aj=nj/n denote the allocation proportion for stratuum j (sum(aj)=1)

b

b=c(b11,b21,...,bJ1), bjk=njk/nj, k=1,2 denote the allocation proportion for group k within stratum j (bj1+bj2=1). Assume group 1 is the control.

p1

p1=c(p11,p21,....,pj1), pjk denote the response probability for group k in stratum j. qjk=1-pjk.

phi

\(p_{j2} q_{j1}/(p_{j1} q_{j2})\)=phi, so that \(p_{j2} = phi p_{j1} /( q_{j1}+ phi p_{j1})\)

References

Chow SC, Shao J, Wang H. Sample Size Calculation in Clinical Research. New York: Marcel Dekker, 2003

Examples

# NOT RUN {
a=c(0.15,0.15,0.2,0.25,0.25);
b=c(0.4,0.4,0.5,0.6,0.6);
p1=c(0.5,0.6,0.7,0.8,0.9);

Example.15.2.3.2<-Propensity.Score.nostrata(alpha=0.05,beta=0.2,J=5,a,b,p1,phi=2)
Example.15.2.3.2
# 1151

# }