blm (version 2012.2.4)

excess.risk: Compute binned excess risk for BLM of LEXPIT fit

Description

Calculates the weighted average of the residual (excess) risk for a BLM or LEXPIT model by a specified grouping variable.

Usage

excess.risk(object, group, linear.only=FALSE)

Arguments

object

fitted blm of lexpit model

group

vector, grouping variable to bin excess risk on. Should be the same length as the design matrix for the model fit (after applying na.action).

linear.only

logical, indicator for LEXPIT objects whether excess risk should only adjust for linear component of model.

Value

• Vector of weighted average of residual risks whose names are the factor groups.

Details

The excess risk can be used to assess the functional relationship of exposures not included in the fitted model.

Examples

data(ccdata)

# INVESTIGATE THE FUNCTIONAL RELATIONSHIP BETWEEN RISK AND PACKYEARS
# AFTER ACCOUNTING FOR GENDER

fit <- blm(y~female, weights = ccdata$w,
       			     	strata=ccdata$strata,
				data=ccdata)

r <- excess.risk(fit, group = ccdata$packyear)
r

plot(y = r, x = as.numeric(names(r)),
       ylab = "Binned excess risk",
       xlab = "packyears", las = 1,
       type = "b")