andersenplot

Plot hazards for two strata for each time. At times with an event in one but not the other group, the fitted hazard remains constant, and so the plot is a step function. If hazards are proportional between strata, then the plot should be close to a straight line.

Proportional hazards estimation in the presence of a partially monotone likelihood has difficulties, in that finite estimators do not exist.  These difficulties are related to those arising from logistic and multinomial regression.  References for methods are given in the separate function documents.  Supported by grant NSF DMS 1712839.

John E Kolassa

PHInfiniteEstimates

Tools for Inference in the Presence of a Monotone Likelihood

andersenplot function

<dl><dt>fit</dt>
<dd>A coxph fit with a stratification term.</dd></dl>

Arguments

Plot hazards for two strata for each time.  At times with an event in one but not the other group, the fitted hazard remains constant, and so the plot is a step function.  If hazards are proportional between strata, then the plot should be close to a straight line. — andersenplot

<dl>

<dt>fit</dt>
<dd>A coxph fit with a stratification term.</dd>

</dl>

Plot hazards for two strata for each time.  At times with an event in one but not the other group, the fitted hazard remains constant, and so the plot is a step function.  If hazards are proportional between strata, then the plot should be close to a straight line.

andersenplot: Plot hazards for two strata for each time. At times with an event in one but not the other group, the fitted hazard remains constant, and so the plot is a step function. If hazards are proportional between strata, then the plot should be close to a straight line.

Description

Usage

Arguments