inst/doc/plot_marginal_effects.R

In sjPlot: Data Visualization for Statistics in Social Science

params <-
list(EVAL = FALSE)

## ----set-options, echo = FALSE------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
knitr::opts_chunk$set(
  collapse = TRUE, 
  comment = "#>", 
  message = FALSE,
  dev = "png", 
  fig.width = 7, 
  fig.height = 3.5, 
  warning = FALSE,
  eval = TRUE
  # eval = if (isTRUE(exists("params"))) params$EVAL else FALSE
)
options(width = 800, tibble.width = Inf)

if (!requireNamespace("sjmisc", quietly = TRUE) ||
    !requireNamespace("splines", quietly = TRUE) ||
    !requireNamespace("ggeffects", quietly = TRUE) ||
    !requireNamespace("effects", quietly = TRUE) ||
    !requireNamespace("ggplot2", quietly = TRUE) ||
    !requireNamespace("haven", quietly = TRUE) ||
    !requireNamespace("sjlabelled", quietly = TRUE)) {
  knitr::opts_chunk$set(eval = FALSE)
}


## -----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
library(sjPlot)
library(ggplot2)
data(efc)
theme_set(theme_sjplot())

fit <- lm(barthtot ~ c12hour + neg_c_7 + c161sex + c172code, data = efc)

plot_model(fit, type = "pred", terms = "c12hour")

## -----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
plot_model(fit, type = "pred", terms = c("c12hour", "c172code"))

## -----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
plot_model(fit, type = "pred", terms = c("c12hour", "c172code", "c161sex"))

## -----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
plot_model(fit, type = "pred", terms = c("c12hour [30, 50, 80]", "c172code [1,3]"))

## -----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
data(mtcars)
mpg_model <- lm(mpg ~ log(hp), data = mtcars)

# x-values and predictions based on the log(hp)-values
plot_model(mpg_model, type = "pred", terms = "hp")

# x-values and predictions based on hp-values from 50 to 150
plot_model(mpg_model, type = "pred", terms = "hp [50:150]")

## -----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
# x-values and predictions based on exponentiated hp-values
plot_model(mpg_model, type = "pred", terms = "hp [exp]")

## -----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
library(splines)
data(women)

fm1 <- lm(weight ~ bs(height, df = 5), data = women)
plot_model(fm1, type = "pred", terms = "height")

## -----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
data(efc)
efc$c172code <- sjlabelled::as_factor(efc$c172code)
fit <- lm(neg_c_7 ~ c12hour + c172code, data = efc)

# reference category is used for "c172code", i.e. c172code
# used the first level as value for predictions
plot_model(fit, type = "pred", terms = "c12hour")

## -----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
# proportion is used for "c172code", i.e. it is set to
# mean(sjlabelled::as_numeric(efc$c172code), na.rm = T),
# which is about 1.9715
plot_model(fit, type = "eff", terms = "c12hour")