ezz: Score test for survival data.

Description Usage Arguments Details Value Author(s) References See Also

Description

This function takes survival data from two groups and a list of quantiles and returns a table indicating whether one group or the other is significantly overrepresented in each quantile.

Usage

1
ezz(d, d2 = NULL, g, quant = 0.9, step = 0.01, thresh = 0.05, gnames = NULL, noob = F, thetas = c(0.05, 0.95), debug = F)

Arguments

d

An integer vector of survival times (usually days).

d2

An optional second integer vector of survival times (usually days). If this argument is used, 'd' is assumed to be the control group and 'd2' is assumed to be the experimental group. If this argument is left at its default of NULL, both groups are assumed to be in 'd' and the 'g' argument is used to separate them out.

g

A charcter vector containing two distinct group names, and equal in length to the combined length of 'd' and 'd2'. However, if both 'd' and 'd2' are used, this argument may also just have a length of 2.

quant

The quantiles to compare. The default is 0.9, but this argument can take a vector of numbers between 0 and 1.

step

Internal parameter used by score algorithm.

thresh

Significance threshold.

gnames

A character vector of length two indicating the group names as they should appear in the output. If not used, the group names are obtained from the 'g' argument instead.

noob

If this argument is set to TRUE then a simplified version of the output is printed.

thetas

Internal parameter used by score algorithm.

debug

Debug flag. Not necessary for normal use.

Details

This test divides the observed difference in survival fractions at each quantile (of the combined dataset) by the estimated standard error under the null hyopthesis. The resulting table indicates whether one group or the other is overrepresented in each quantile beyond a level that would be expected by random chance.

Value

Returns a data.frame with quantiles as the row names and the following columns:

age

Age at the given quantile.

srv.A

Number of surviving animals in group 'A' at the given age (A is substituted with the control group's name).

srv.A.frc

Fraction of surviving animals in group 'A' at the given age (A is substituted with the control group's name).

srv.B

Number of surviving animals in group 'B' at the given age (B is substituted with the experimental group's name).

srv.B.frc

Fraction of surviving animals in group 'B' at the given age (B is substituted with the experimental group's name).

survtotal.frc

Fraction of total surviving animals.

zp

Score test statistic.

p

Significance level.

sig

Indicator of significance.

Author(s)

Alex F. Bokov

References

Wang,C., Li,Q., Redden,D.T., Weindruch,R., Allison,D.B. (2004). Statistical methods for testing effects on "maximal lifespan". Mechanisms of Ageing and Development 125, 629-632

Mehotra,D.V., Chan,I.S.F., Berger,R.L. (2003). A cautionary note on exact unconditional inference for a difference between two independent binomial proportions. Biometrics 59, 441-450

See Also

survgui


Survomatic documentation built on May 2, 2019, 4:09 p.m.

Related to ezz in Survomatic...