plot_interaction: Plot interaction effects for regression models

In ha-pu/supportR: Helpful supporters

Description

The function plot_interaction provides interaction plots for linear regression models or linear fixed effects models. For linear models, the user can provide a cluster variable to cluster the standard errors and can select to use bootstraping for clustering.

Usage

  plot_interaction(input, iv, moderator, ci = 0.9, type = 0, var_cluster = NULL)

Arguments

input	Object of class "mod_vcov", linear model of class "lm" or "glm", or linear fixed effects model of class "lmerMod" or "glmerMod"
iv	character giving the name of the independt variable for which the interaction effect is plotted - for input is class "lmerMod" or "glmerMod" a logical iv must be indicated as "variableTRUE""
moderator	character giving the name of the variable that moderates the interaction effect - the moderator variable mustnot be of class "logical"
ci	Confidence interval shown in interaction plot - must be numeric between 0 and 1
type	Indicator which type of standard errors is to be computed: 0 - no clustering, 1 - clustering using var_cluster, 2 - clustering using var_cluster with bootstraping
var_cluster	Vector, matrix, or data.frame containing the variables that are used for clustering

Details

For input of class "lm" or "glm" the function uses cplot from the margins package to prepare the interaction plot. For input of class "lmerMod" and "glmerMod" the funciton uses interplot from the interplot package to prepare the interaction plot. For input of class "lm" or "glm" the options type == 0, type == 1, and type == 2 control how the variance-covariance matrix for standard error estimation is computed. The options type == 1 and type == 2 require a cluster variable defined in var_cluster.

Value

Interaction plot as ggplot2 object

See Also

cplot, interplot, ggplot

Examples

  data <- supportR::create_data()

  mod1 <- lm(firm_value ~ profit + cogs + rnd + competition * board_size, data = data)
  mod1_vcov <- model_vcov(model = mod1, type = 1, var_cluster = data$country)
  plot_interaction(input = mod1_vcov, iv = "competition", moderator = "board_size", ci = 0.9)
  plot_interaction(input = mod1, iv = "competition", moderator = "board_size", ci = 0.9, type = 0, var_cluster = NULL)


  mod2 <- glm(female_ceo ~ profit + cogs + rnd + ceo_age * board_size, data = data, family = "binomial")
  plot_interaction(input = mod2, iv = "ceo_age", moderator = "board_size", ci = 0.9, type = 2, var_cluster = data$industry)

  mod3 <- lme4::lmer(firm_value ~ profit + cogs + rnd + competition * board_size + (1 | country), data = data)
  plot_interaction(input = mod3, iv = "competition", moderator = "board_size", ci = 0.9)

  mod4 <- lme4::glmer(female_ceo ~ profit + cogs + rnd + ceo_age * board_size + (1 | industry), data = data, family = "binomial")
  plot_interaction(input = mod4, iv = "ceo_age", moderator = "board_size", ci = 0.9)

ha-pu/supportR documentation built on Sept. 13, 2020, 5:52 p.m.