test_highest {lmhelprs}

Test the Highest Order Term by ANOVA

Description

Identify the highest order terms in a model fitted by 'lm()', and compare this model to a model with this term removed using ANOVA.

Usage

test_highest(lm_out)

highest_order(lm_out)

Arguments

lm_out

The output of stats::lm().

Details

The function test_highest() first check if a model fitted by stats::lm() has a unique highest order term (e.g., the term x1:x2, in the model y ~ x1 + x2 + x1:x2). If yes, it will fit a model with this term removed, and then call hierarchical_lm() to compare the original model with this reduced model.

If the model does not have a unique highest order term, an error will be raised.

Value

A hierarchical_lm-class object, which is the output of hierarchical_lm(). Two models are compared, the original model and the model with the unique highest order term removed.

Functions

• test_highest(): Test the highest order term.

• highest_order(): Find the highest order term.

Limitation

It relies on terms created by stats::lm() to determine the order of each term. If a higher order term is created manually (e.g., I(x1 * x2)), then it cannot know that this term is a second order term.

Author(s)

Shu Fai Cheung https://orcid.org/0000-0002-9871-9448

See Also

hierarchical_lm()

Examples

dat <- data_test1

lm1 <- lm(y ~ x1 + x2 + cat1*x3, dat)
lm2 <- lm(y ~ x1 + x2*x3 + x4, dat)

test_highest(lm1)
test_highest(lm2)

highest_order(lm1)
highest_order(lm2)

# The followings will yield an error

lm3 <- lm(y ~ x1 + x2 + x3, dat)
summary(lm3)
tryCatch(test_highest(lm3), error = function(e) e)
tryCatch(highest_order(lm3), error = function(e) e)

lm4 <- lm(y ~ x1 + x2*x3 + x4*x5, dat)
summary(lm4)
tryCatch(test_highest(lm4), error = function(e) e)
tryCatch(highest_order(lm4), error = function(e) e)