Biomedical Engineering Reference
In-Depth Information
After installing glrt, the package can be loaded as follows:
>library(Icens)
>library(gLRT)
The main function of glrt is
gLRT
. By selecting one of its four
method
options (glrt1, glrt2, glrt3, or score), the function will call function
gLRT1
,
gLRT2
,
gLRT3
, or
ScoreTest
, each of which can also be called directly by the
user. The test results of any of the tests returns the following:
•
Method used
•
Proposed test statistic U
•
Estimate of the covariance matrix of U, V
2
-statistic used to conduct the test
•
•
p-value of the test
Note that gLRT2 does not allow data to have exactly observed observations
(L
i
= R
i
), and if there is no exactly observed observation, gLRT3 is identical
to gLRT2.
The arguments of
gLRT
that are not shared by all four types of tests are
M
,
rho
, and
gamma
.
M
is the number of imputations used in gLRT1 for estimating
the covariance, while
rho
and
gamma
are the parameters of the link function
(x) = (x log x) x
(1 x)
used in gLRT2 and gLRT3. The default choice
for (
rho
,
gamma
) is (0, 0) and the choice should depend on application. See
Fleming and Harrington (1991) for explanations of choices of
rho
and
gamma
.
The rest of the arguments of
gLRT
:
A
,
k
,
EMstep
,
ICMstep
,
tol
,
maxiter
, and
inf
are shared by all four methods.
A
is the data matrix required to have
dimension n 3 with the rst two columns for the censored intervals in the
format of (Li,
i
;R
i
]. If the data use another number to represent infinity, specify
the number by using the option
inf=
. The third column of
A
is for treatment
indicators taking values between 0 and (k1). If k = 3, you would use 0, 1, or 2
Search WWH ::
Custom Search