prob_rank_givenEffect_emp: Emperical ranks probbaility of the test given the effect size

In mshasan/empOPW: Empirical Method of Optimal Pvalue Weighting

Description

Emperical comnputation of the ranks probability of a test being higher than any other test given the effect size from the external information.

Usage

prob_rank_givenEffect_emp(pvalue, filter, group = 5L, h_breaks = 100L,
  bin_idx = 1L, smooth = TRUE, effectType = c("continuous", "binary"))

Arguments

pvalue	vector of test pvalues
filter	vector of filter statistics
group	number of groups, should be at least four
h_breaks	number of breaks for the histogram
bin_idx	Integer, bin number of the histogram. Almost alwyas it is one, becasue we are interested to obtain the ranks probability of the alternative tests
smooth	Character of ("TRUE" or "FALSE") to apply spline smoothing, default is TRUE
effectType	type of effect size c("binary","continuous")

Details

If one wants to test

H_0: \epsilion_i=0 vs. H_a: \epsilion_i > 0,

then et and ey should effect type continuous effect size and if one wants to test

H_0: \epsilion_i=0 vs. H_a: \epsilion_i = \epsilion,

one should the binary effect size

Value

ranksProb emperical probability of the rank of the test

Author(s)

Mohamad S. Hasan, [e-mail]

Examples

# generating data (known in practice)
set.seed(123)
X = runif(10000, min = 0, max = 2.5)         # covariate
H = rbinom(length(X), size = 1, prob = 0.1)   # hypothesis true or false
Z = rnorm(length(X), mean = H * X)            # Z-score
p = 1 - pnorm(Z)

# apply the function to compute the rank proabbility
grp = 10
ranksProb = prob_rank_givenEffect_emp(pvalue = p, filter = X, group = grp,
                               h_breaks = 71, effectType = "continuous")

# plot the probability
plot(1:grp, ranksProb, type="l", xlab = "ranks", ylab = "P(rank | effect)")

mshasan/empOPW documentation built on March 1, 2021, 4:19 a.m.