EgoCor (version 1.3.4)

vario.reg.prep: Adjustment for covariates before semi-variogram model fitting

Description

Adjustment for covariates provides the option to eliminate non-spatial effects on the variable of interest. Given a linear regression output of class 'lm' or 'lmerMod' with the attribute of interest as dependent variable, the function provides a dataset containing the coordinates of the original observations and the studentized residuals of the regression model.

Usage

vario.reg.prep(reg, data = NULL)

Value

A data frame with three columns:

x

x-coordinate in the first column.

y

y-coordinate in the second column.

adj

Studentized residuals to be used as new variable adjusted for covariates.

Arguments

reg

An object of class 'lm' obtained as result of a linear regression using the function lm from the package stats or an object of class 'lmerMod' obtained as result of a linear mixed model regression using the function lmer from the package lme4.

data

Only needed if the data argument within the regression function lm/lmer is not provided: A data frame containing the geo-coded dataset containing the Cartesian x- and y-coordinates in the first and second column, the outcome of interest in the third column and all covariates used for the regression in further columns.

Details

The adjusted outcome is defined as the student residuals of the linear or linear mixed regression model. They are calculated using the rstudent function from package stats. In case of a mixed model, the adjusted variable vector resembles the conditional studentized residuals.

The geo-coded dataset used for the regression is extracted from the current environment. In order to work, the dataset has to be loaded into the environment prior to the use of vario.reg.prep.

If the data argument was specified within the regression function lm/lmer, vario.reg.prep automatically extracts the name of the dataset used for regression and calls it from the current environment. Otherwise, the dataset has to be provided manually as input argument within vario.reg.prep.

See Also

lm in the stats package for information on the fitting of a linear regression model;

lmer in the lme4 package for information on the fitting of a linear mixed regression model;

rstudent in the stats package for information on how the attribute of interest is adjusted for covariates.

Examples

## Example 1
head(birth) #geo-coded dataset
hist(birth$birthweight) # attribute of interest

# Linear regression model
mod1 = lm(birthweight ~ primiparous + datediff + bmi
+ inc, data = birth)
summary(mod1)
data.adj1 = vario.reg.prep(mod1)

head(data.adj1)
hist(data.adj1$adj) # adjusted attribute of interest
# The data frame can be used as input for the vario.mod function.


## Example 2
# No data argument provided within lm (not recommended, but possible):
mod2 = lm(birth$birthweight ~ birth$primiparous + birth$datediff + birth$bmi
+ birth$inc)
summary(mod2)
# In this case, make sure to provide the data argument here:
data.adj2 = vario.reg.prep(reg = mod2, data = birth)


if (requireNamespace("lme4", quietly = TRUE)) {
## Example 3
# Linear mixed regression model
mod3 = lme4::lmer(birthweight ~ primiparous + datediff
                + bmi + (1|inc), data = birth)
summary(mod3)
data.adj3 = vario.reg.prep(mod3)


## Example 4
# Data argument within lmer not provided (not recommended, but possible):
mod4 = lme4::lmer(birth$birthweight ~ birth$primiparous + birth$datediff
            + birth$bmi + (1|birth$inc))
summary(mod4)
# In this case, make sure to provide the data argument here:
data.adj4 = vario.reg.prep(reg = mod4, data = birth)
}