Robustness: Alternative model specifications

Goal

Estimate CLPM1 and CLPM2 models to assess sensitivity to model specifications

Set up and data import

#robustness analysis.
#library
library(tidyverse)
library(lavaan)
library(data.table)
library(viridis)
library(doParallel)
library(parallel)
library(patchwork)

#data
load("results/predicted_means/240816_pred-means-cleaned-df.Rdata")

#extract datafile from pref_results
MyData <- pred_results$df_combined

Lavaan specifications

CLPM1

Without controls

#CLPM1#
#without controls
EU_clpm1_constrained_lag <- '

  # Estimate the lagged effects (constrained)
  eu_2   ~ a*eu_1 + b*Feduc_a_1
  eu_3   ~ a*eu_2 + b*Feduc_a_2
  eu_4   ~ a*eu_3 + b*Feduc_a_3
  eu_5   ~ a*eu_4 + b*Feduc_a_4
  eu_6   ~ a*eu_5 + b*Feduc_a_5
  eu_7   ~ a*eu_6 + b*Feduc_a_6
  eu_8   ~ a*eu_7 + b*Feduc_a_7
  eu_9   ~ a*eu_8 + b*Feduc_a_8
  eu_10  ~ a*eu_9 + b*Feduc_a_9
  eu_11  ~ a*eu_10 + b*Feduc_a_10
  
  
  Feduc_a_2  ~ c*eu_1 + d*Feduc_a_1
  Feduc_a_3  ~ c*eu_2 + d*Feduc_a_2
  Feduc_a_4  ~ c*eu_3 + d*Feduc_a_3
  Feduc_a_5  ~ c*eu_4 + d*Feduc_a_4
  Feduc_a_6  ~ c*eu_5 + d*Feduc_a_5
  Feduc_a_7  ~ c*eu_6 + d*Feduc_a_6
  Feduc_a_8  ~ c*eu_7 + d*Feduc_a_7
  Feduc_a_9  ~ c*eu_8 + d*Feduc_a_8
  Feduc_a_10  ~ c*eu_9 + d*Feduc_a_9
  Feduc_a_11  ~ c*eu_10 + d*Feduc_a_10
  
  # Estimate the covariance at the first wave and residual covariances
  eu_1 ~~ Feduc_a_1 # Covariance
  eu_2 ~~ Feduc_a_2 # Covariance
  eu_3 ~~ Feduc_a_3 # Covariance
  eu_4 ~~ Feduc_a_4 # Covariance
  eu_5 ~~ Feduc_a_5 # Covariance
  eu_6 ~~ Feduc_a_6 # Covariance
  eu_7 ~~ Feduc_a_7 # Covariance
  eu_8 ~~ Feduc_a_8 # Covariance
  eu_9 ~~ Feduc_a_9 # Covariance
  eu_10 ~~ Feduc_a_10 # Covariance
  eu_11 ~~ Feduc_a_11 # Covariance
  
  
  # Estimate the variance 
  eu_1 ~~ eu_1 
  Feduc_a_1 ~~ Feduc_a_1
'

cult_clpm1_constrained_lag <- '

  # Estimate the lagged effects (constrained)
  cult_2   ~ a*cult_1 + b*Feduc_a_1
  cult_3   ~ a*cult_2 + b*Feduc_a_2
  cult_4   ~ a*cult_3 + b*Feduc_a_3
  cult_5   ~ a*cult_4 + b*Feduc_a_4
  cult_6   ~ a*cult_5 + b*Feduc_a_5
  cult_7   ~ a*cult_6 + b*Feduc_a_6
  cult_8   ~ a*cult_7 + b*Feduc_a_7
  cult_9   ~ a*cult_8 + b*Feduc_a_8
  cult_10  ~ a*cult_9 + b*Feduc_a_9
  cult_11  ~ a*cult_10 + b*Feduc_a_10
  
  
  Feduc_a_2  ~ c*cult_1 + d*Feduc_a_1
  Feduc_a_3  ~ c*cult_2 + d*Feduc_a_2
  Feduc_a_4  ~ c*cult_3 + d*Feduc_a_3
  Feduc_a_5  ~ c*cult_4 + d*Feduc_a_4
  Feduc_a_6  ~ c*cult_5 + d*Feduc_a_5
  Feduc_a_7  ~ c*cult_6 + d*Feduc_a_6
  Feduc_a_8  ~ c*cult_7 + d*Feduc_a_7
  Feduc_a_9  ~ c*cult_8 + d*Feduc_a_8
  Feduc_a_10  ~ c*cult_9 + d*Feduc_a_9
  Feduc_a_11  ~ c*cult_10 + d*Feduc_a_10
  
  # Estimate the covariance at the first wave. 
  cult_1 ~~ Feduc_a_1 # Covariance
  cult_2 ~~ Feduc_a_2 # Covariance
  cult_3 ~~ Feduc_a_3 # Covariance
  cult_4 ~~ Feduc_a_4 # Covariance
  cult_5 ~~ Feduc_a_5 # Covariance
  cult_6 ~~ Feduc_a_6 # Covariance
  cult_7 ~~ Feduc_a_7 # Covariance
  cult_8 ~~ Feduc_a_8 # Covariance
  cult_9 ~~ Feduc_a_9 # Covariance
  cult_10 ~~ Feduc_a_10 # Covariance
  cult_11 ~~ Feduc_a_11 # Covariance
  
  # Estimate the variance 
  cult_1 ~~ cult_1 
  Feduc_a_1 ~~ Feduc_a_1
'

inc_diff_clpm1_constrained_lag <- '

  # Estimate the lagged effects (constrained)
  inc_diff_2   ~ a*inc_diff_1 + b*Feduc_a_1
  inc_diff_3   ~ a*inc_diff_2 + b*Feduc_a_2
  inc_diff_4   ~ a*inc_diff_3 + b*Feduc_a_3
  inc_diff_5   ~ a*inc_diff_4 + b*Feduc_a_4
  inc_diff_6   ~ a*inc_diff_5 + b*Feduc_a_5
  inc_diff_7   ~ a*inc_diff_6 + b*Feduc_a_6
  inc_diff_8   ~ a*inc_diff_7 + b*Feduc_a_7
  inc_diff_9   ~ a*inc_diff_8 + b*Feduc_a_8
  inc_diff_10  ~ a*inc_diff_9 + b*Feduc_a_9
  inc_diff_11  ~ a*inc_diff_10 + b*Feduc_a_10
  
  
  Feduc_a_2  ~ c*inc_diff_1 + d*Feduc_a_1
  Feduc_a_3  ~ c*inc_diff_2 + d*Feduc_a_2
  Feduc_a_4  ~ c*inc_diff_3 + d*Feduc_a_3
  Feduc_a_5  ~ c*inc_diff_4 + d*Feduc_a_4
  Feduc_a_6  ~ c*inc_diff_5 + d*Feduc_a_5
  Feduc_a_7  ~ c*inc_diff_6 + d*Feduc_a_6
  Feduc_a_8  ~ c*inc_diff_7 + d*Feduc_a_7
  Feduc_a_9  ~ c*inc_diff_8 + d*Feduc_a_8
  Feduc_a_10  ~ c*inc_diff_9 + d*Feduc_a_9
  Feduc_a_11  ~ c*inc_diff_10 + d*Feduc_a_10
  
  # Estimate the covariance at the first wave. 
  inc_diff_1 ~~ Feduc_a_1 # Covariance
  inc_diff_2 ~~ Feduc_a_2 # Covariance
  inc_diff_3 ~~ Feduc_a_3 # Covariance
  inc_diff_4 ~~ Feduc_a_4 # Covariance
  inc_diff_5 ~~ Feduc_a_5 # Covariance
  inc_diff_6 ~~ Feduc_a_6 # Covariance
  inc_diff_7 ~~ Feduc_a_7 # Covariance
  inc_diff_8 ~~ Feduc_a_8 # Covariance
  inc_diff_9 ~~ Feduc_a_9 # Covariance
  inc_diff_10 ~~ Feduc_a_10 # Covariance
  inc_diff_11 ~~ Feduc_a_11 # Covariance
  
  # Estimate the variance 
  inc_diff_1 ~~ inc_diff_1 
  Feduc_a_1 ~~ Feduc_a_1
'

#store formulas in list
formulas_clpm1_nocontrols <- list(EU_clpm1_constrained_lag,
     cult_clpm1_constrained_lag,
     inc_diff_clpm1_constrained_lag)

With control variables

# With control variables
EU_clpm1_control_constrained_lag <- '

  # Estimate the lagged effects (constrained)
  eu_2   ~ a*eu_1 + b*Feduc_a_1 + age_1 + educ_1 + work_1 + married_1
  eu_3   ~ a*eu_2 + b*Feduc_a_2 + age_2 + educ_2 + work_2 + married_2
  eu_4   ~ a*eu_3 + b*Feduc_a_3 + age_3 + educ_3 + work_3 + married_3
  eu_5   ~ a*eu_4 + b*Feduc_a_4 + age_4 + educ_4 + work_4 + married_4
  eu_6   ~ a*eu_5 + b*Feduc_a_5 + age_5 + educ_5 + work_5 + married_5
  eu_7   ~ a*eu_6 + b*Feduc_a_6 + age_6 + educ_6 + work_6 + married_6
  eu_8   ~ a*eu_7 + b*Feduc_a_7 + age_7 + educ_7 + work_7 + married_7
  eu_9   ~ a*eu_8 + b*Feduc_a_8 + age_8 + educ_8 + work_8 + married_8
  eu_10  ~ a*eu_9 + b*Feduc_a_9 + age_9 + educ_9 + work_9 + married_9
  eu_11  ~ a*eu_10 + b*Feduc_a_10 + age_10 + educ_10 + work_10 + married_10
  
  
  Feduc_a_2  ~ c*eu_1 + d*Feduc_a_1 + age_1 + educ_1 + work_1 + married_1
  Feduc_a_3  ~ c*eu_2 + d*Feduc_a_2 + age_2 + educ_2 + work_2 + married_2
  Feduc_a_4  ~ c*eu_3 + d*Feduc_a_3 + age_3 + educ_3 + work_3 + married_3
  Feduc_a_5  ~ c*eu_4 + d*Feduc_a_4 + age_4 + educ_4 + work_4 + married_4
  Feduc_a_6  ~ c*eu_5 + d*Feduc_a_5 + age_5 + educ_5 + work_5 + married_5
  Feduc_a_7  ~ c*eu_6 + d*Feduc_a_6 + age_6 + educ_6 + work_6 + married_6
  Feduc_a_8  ~ c*eu_7 + d*Feduc_a_7 + age_7 + educ_7 + work_7 + married_7
  Feduc_a_9  ~ c*eu_8 + d*Feduc_a_8 + age_8 + educ_8 + work_8 + married_8
  Feduc_a_10  ~ c*eu_9 + d*Feduc_a_9 + age_9 + educ_9 + work_9 + married_9
  Feduc_a_11  ~ c*eu_10 + d*Feduc_a_10 + age_10 + educ_10 + work_10 + married_10
  
  # Estimate the covariance at the first wave and residual covariances
  eu_1 ~~ Feduc_a_1 # Covariance
  eu_2 ~~ Feduc_a_2 # Covariance
  eu_3 ~~ Feduc_a_3 # Covariance
  eu_4 ~~ Feduc_a_4 # Covariance
  eu_5 ~~ Feduc_a_5 # Covariance
  eu_6 ~~ Feduc_a_6 # Covariance
  eu_7 ~~ Feduc_a_7 # Covariance
  eu_8 ~~ Feduc_a_8 # Covariance
  eu_9 ~~ Feduc_a_9 # Covariance
  eu_10 ~~ Feduc_a_10 # Covariance
  eu_11 ~~ Feduc_a_11 # Covariance
  
  # Estimate the variance 
  eu_1 ~~ eu_1 
  Feduc_a_1 ~~ Feduc_a_1
'

cult_clpm1_control_constrained_lag <- '
   # Estimate the lagged effects (constrained)
  cult_2   ~ a*cult_1 + b*Feduc_a_1 + age_1 + educ_1 + work_1 + married_1
  cult_3   ~ a*cult_2 + b*Feduc_a_2 + age_2 + educ_2 + work_2 + married_2
  cult_4   ~ a*cult_3 + b*Feduc_a_3 + age_3 + educ_3 + work_3 + married_3
  cult_5   ~ a*cult_4 + b*Feduc_a_4 + age_4 + educ_4 + work_4 + married_4
  cult_6   ~ a*cult_5 + b*Feduc_a_5 + age_5 + educ_5 + work_5 + married_5
  cult_7   ~ a*cult_6 + b*Feduc_a_6 + age_6 + educ_6 + work_6 + married_6
  cult_8   ~ a*cult_7 + b*Feduc_a_7 + age_7 + educ_7 + work_7 + married_7
  cult_9   ~ a*cult_8 + b*Feduc_a_8 + age_8 + educ_8 + work_8 + married_8
  cult_10  ~ a*cult_9 + b*Feduc_a_9 + age_9 + educ_9 + work_9 + married_9
  cult_11  ~ a*cult_10 + b*Feduc_a_10 + age_10 + educ_10 + work_10 + married_10
  
  
  Feduc_a_2  ~ c*cult_1 + d*Feduc_a_1 + age_1 + educ_1 + work_1 + married_1
  Feduc_a_3  ~ c*cult_2 + d*Feduc_a_2 + age_2 + educ_2 + work_2 + married_2
  Feduc_a_4  ~ c*cult_3 + d*Feduc_a_3 + age_3 + educ_3 + work_3 + married_3
  Feduc_a_5  ~ c*cult_4 + d*Feduc_a_4 + age_4 + educ_4 + work_4 + married_4
  Feduc_a_6  ~ c*cult_5 + d*Feduc_a_5 + age_5 + educ_5 + work_5 + married_5
  Feduc_a_7  ~ c*cult_6 + d*Feduc_a_6 + age_6 + educ_6 + work_6 + married_6
  Feduc_a_8  ~ c*cult_7 + d*Feduc_a_7 + age_7 + educ_7 + work_7 + married_7
  Feduc_a_9  ~ c*cult_8 + d*Feduc_a_8 + age_8 + educ_8 + work_8 + married_8
  Feduc_a_10  ~ c*cult_9 + d*Feduc_a_9 + age_9 + educ_9 + work_9 + married_9
  Feduc_a_11  ~ c*cult_10 + d*Feduc_a_10 + age_10 + educ_10 + work_10 + married_10
  
  # Estimate the covariance at the first wave. 
  cult_1 ~~ Feduc_a_1 # Covariance
  cult_2 ~~ Feduc_a_2 # Covariance
  cult_3 ~~ Feduc_a_3 # Covariance
  cult_4 ~~ Feduc_a_4 # Covariance
  cult_5 ~~ Feduc_a_5 # Covariance
  cult_6 ~~ Feduc_a_6 # Covariance
  cult_7 ~~ Feduc_a_7 # Covariance
  cult_8 ~~ Feduc_a_8 # Covariance
  cult_9 ~~ Feduc_a_9 # Covariance
  cult_10 ~~ Feduc_a_10 # Covariance
  cult_11 ~~ Feduc_a_11 # Covariance
  
  # Estimate the variance 
  cult_1 ~~ cult_1 
  Feduc_a_1 ~~ Feduc_a_1
'

inc_diff_clpm1_control_constrained_lag <- '
  # Estimate the lagged effects (constrained)
  inc_diff_2   ~ a*inc_diff_1 + b*Feduc_a_1 + age_1 + educ_1 + work_1 + married_1
  inc_diff_3   ~ a*inc_diff_2 + b*Feduc_a_2 + age_2 + educ_2 + work_2 + married_2
  inc_diff_4   ~ a*inc_diff_3 + b*Feduc_a_3 + age_3 + educ_3 + work_3 + married_3
  inc_diff_5   ~ a*inc_diff_4 + b*Feduc_a_4 + age_4 + educ_4 + work_4 + married_4
  inc_diff_6   ~ a*inc_diff_5 + b*Feduc_a_5 + age_5 + educ_5 + work_5 + married_5
  inc_diff_7   ~ a*inc_diff_6 + b*Feduc_a_6 + age_6 + educ_6 + work_6 + married_6
  inc_diff_8   ~ a*inc_diff_7 + b*Feduc_a_7 + age_7 + educ_7 + work_7 + married_7
  inc_diff_9   ~ a*inc_diff_8 + b*Feduc_a_8 + age_8 + educ_8 + work_8 + married_8
  inc_diff_10  ~ a*inc_diff_9 + b*Feduc_a_9 + age_9 + educ_9 + work_9 + married_9
  inc_diff_11  ~ a*inc_diff_10 + b*Feduc_a_10 + age_10 + educ_10 + work_10 + married_10
  
  
  Feduc_a_2  ~ c*inc_diff_1 + d*Feduc_a_1 + age_1 + educ_1 + work_1 + married_1
  Feduc_a_3  ~ c*inc_diff_2 + d*Feduc_a_2 + age_2 + educ_2 + work_2 + married_2
  Feduc_a_4  ~ c*inc_diff_3 + d*Feduc_a_3 + age_3 + educ_3 + work_3 + married_3
  Feduc_a_5  ~ c*inc_diff_4 + d*Feduc_a_4 + age_4 + educ_4 + work_4 + married_4
  Feduc_a_6  ~ c*inc_diff_5 + d*Feduc_a_5 + age_5 + educ_5 + work_5 + married_5
  Feduc_a_7  ~ c*inc_diff_6 + d*Feduc_a_6 + age_6 + educ_6 + work_6 + married_6
  Feduc_a_8  ~ c*inc_diff_7 + d*Feduc_a_7 + age_7 + educ_7 + work_7 + married_7
  Feduc_a_9  ~ c*inc_diff_8 + d*Feduc_a_8 + age_8 + educ_8 + work_8 + married_8
  Feduc_a_10  ~ c*inc_diff_9 + d*Feduc_a_9 + age_9 + educ_9 + work_9 + married_9
  Feduc_a_11  ~ c*inc_diff_10 + d*Feduc_a_10 + age_10 + educ_10 + work_10 + married_10
  
  # Estimate the covariance at the first wave and residual covariances 
  inc_diff_1 ~~ Feduc_a_1 # Covariance
  inc_diff_2 ~~ Feduc_a_2 # Covariance
  inc_diff_3 ~~ Feduc_a_3 # Covariance
  inc_diff_4 ~~ Feduc_a_4 # Covariance
  inc_diff_5 ~~ Feduc_a_5 # Covariance
  inc_diff_6 ~~ Feduc_a_6 # Covariance
  inc_diff_7 ~~ Feduc_a_7 # Covariance
  inc_diff_8 ~~ Feduc_a_8 # Covariance
  inc_diff_9 ~~ Feduc_a_9 # Covariance
  inc_diff_10 ~~ Feduc_a_10 # Covariance
  inc_diff_11 ~~ Feduc_a_11 # Covariance
  
  # Estimate the variance 
  inc_diff_1 ~~ inc_diff_1 
  Feduc_a_1 ~~ Feduc_a_1
'

#store formulas in list
formulas_clpm1_controls <- list(EU_clpm1_control_constrained_lag,
                                  cult_clpm1_control_constrained_lag,
                                  inc_diff_clpm1_control_constrained_lag)

CLPM2

Without controls

#CLPM2#
#create formulas with no controls
EU_clpm2_constrained_lag <- '

  # Estimate the lagged effects (constrained)
  eu_2   ~ a*eu_1 + b*Feduc_a_1
  eu_3   ~ a*eu_2 + b*Feduc_a_2 + c*eu_1 + d*Feduc_a_1
  eu_4   ~ a*eu_3 + b*Feduc_a_3 + c*eu_2 + d*Feduc_a_2
  eu_5   ~ a*eu_4 + b*Feduc_a_4 + c*eu_3 + d*Feduc_a_3
  eu_6   ~ a*eu_5 + b*Feduc_a_5 + c*eu_4 + d*Feduc_a_4
  eu_7   ~ a*eu_6 + b*Feduc_a_6 + c*eu_5 + d*Feduc_a_5
  eu_8   ~ a*eu_7 + b*Feduc_a_7 + c*eu_6 + d*Feduc_a_6
  eu_9   ~ a*eu_8 + b*Feduc_a_8 + c*eu_7 + d*Feduc_a_7
  eu_10  ~ a*eu_9 + b*Feduc_a_9 + c*eu_8 + d*Feduc_a_8
  eu_11  ~ a*eu_10 + b*Feduc_a_10 + c*eu_9 + d*Feduc_a_9
  
  
  Feduc_a_2  ~ e*eu_1 + f*Feduc_a_1
  Feduc_a_3  ~ e*eu_2 + f*Feduc_a_2 + g*eu_1 + h*Feduc_a_1
  Feduc_a_4  ~ e*eu_3 + f*Feduc_a_3 + g*eu_2 + h*Feduc_a_2
  Feduc_a_5  ~ e*eu_4 + f*Feduc_a_4 + g*eu_3 + h*Feduc_a_3
  Feduc_a_6  ~ e*eu_5 + f*Feduc_a_5 + g*eu_4 + h*Feduc_a_4
  Feduc_a_7  ~ e*eu_6 + f*Feduc_a_6 + g*eu_5 + h*Feduc_a_5
  Feduc_a_8  ~ e*eu_7 + f*Feduc_a_7 + g*eu_6 + h*Feduc_a_6
  Feduc_a_9  ~ e*eu_8 + f*Feduc_a_8 + g*eu_7 + h*Feduc_a_7
  Feduc_a_10  ~ e*eu_9 + f*Feduc_a_9 + g*eu_8 + h*Feduc_a_8
  Feduc_a_11  ~ e*eu_10 + f*Feduc_a_10 + g*eu_9 + h*Feduc_a_9
  
  # Estimate the covariance at the first wave and residual covariances
  eu_1 ~~ Feduc_a_1 # Covariance
  eu_2 ~~ Feduc_a_2 # Covariance
  eu_3 ~~ Feduc_a_3 # Covariance
  eu_4 ~~ Feduc_a_4 # Covariance
  eu_5 ~~ Feduc_a_5 # Covariance
  eu_6 ~~ Feduc_a_6 # Covariance
  eu_7 ~~ Feduc_a_7 # Covariance
  eu_8 ~~ Feduc_a_8 # Covariance
  eu_9 ~~ Feduc_a_9 # Covariance
  eu_10 ~~ Feduc_a_10 # Covariance
  eu_11 ~~ Feduc_a_11 # Covariance
  
  # Estimate the variance 
  eu_1 ~~ eu_1 
  Feduc_a_1 ~~ Feduc_a_1
'

cult_clpm2_constrained_lag <- '

   # Estimate the lagged effects (constrained)
  eu_2   ~ a*eu_1 + b*Feduc_a_1
  eu_3   ~ a*eu_2 + b*Feduc_a_2 + c*eu_1 + d*Feduc_a_1
  eu_4   ~ a*eu_3 + b*Feduc_a_3 + c*eu_2 + d*Feduc_a_2
  eu_5   ~ a*eu_4 + b*Feduc_a_4 + c*eu_3 + d*Feduc_a_3
  eu_6   ~ a*eu_5 + b*Feduc_a_5 + c*eu_4 + d*Feduc_a_4
  eu_7   ~ a*eu_6 + b*Feduc_a_6 + c*eu_5 + d*Feduc_a_5
  eu_8   ~ a*eu_7 + b*Feduc_a_7 + c*eu_6 + d*Feduc_a_6
  eu_9   ~ a*eu_8 + b*Feduc_a_8 + c*eu_7 + d*Feduc_a_7
  eu_10  ~ a*eu_9 + b*Feduc_a_9 + c*eu_8 + d*Feduc_a_8
  eu_11  ~ a*eu_10 + b*Feduc_a_10 + c*eu_9 + d*Feduc_a_9
  
  
  Feduc_a_2  ~ e*eu_1 + f*Feduc_a_1
  Feduc_a_3  ~ e*eu_2 + f*Feduc_a_2 + g*eu_1 + h*Feduc_a_1
  Feduc_a_4  ~ e*eu_3 + f*Feduc_a_3 + g*eu_2 + h*Feduc_a_2
  Feduc_a_5  ~ e*eu_4 + f*Feduc_a_4 + g*eu_3 + h*Feduc_a_3
  Feduc_a_6  ~ e*eu_5 + f*Feduc_a_5 + g*eu_4 + h*Feduc_a_4
  Feduc_a_7  ~ e*eu_6 + f*Feduc_a_6 + g*eu_5 + h*Feduc_a_5
  Feduc_a_8  ~ e*eu_7 + f*Feduc_a_7 + g*eu_6 + h*Feduc_a_6
  Feduc_a_9  ~ e*eu_8 + f*Feduc_a_8 + g*eu_7 + h*Feduc_a_7
  Feduc_a_10  ~ e*eu_9 + f*Feduc_a_9 + g*eu_8 + h*Feduc_a_8
  Feduc_a_11  ~ e*eu_10 + f*Feduc_a_10 + g*eu_9 + h*Feduc_a_9
  
  # Estimate the covariance at the first wave. 
  cult_1 ~~ Feduc_a_1 # Covariance
  cult_2 ~~ Feduc_a_2 # Covariance
  cult_3 ~~ Feduc_a_3 # Covariance
  cult_4 ~~ Feduc_a_4 # Covariance
  cult_5 ~~ Feduc_a_5 # Covariance
  cult_6 ~~ Feduc_a_6 # Covariance
  cult_7 ~~ Feduc_a_7 # Covariance
  cult_8 ~~ Feduc_a_8 # Covariance
  cult_9 ~~ Feduc_a_9 # Covariance
  cult_10 ~~ Feduc_a_10 # Covariance
  cult_11 ~~ Feduc_a_11 # Covariance
  
  # Estimate the variance 
  eu_1 ~~ eu_1 
  Feduc_a_1 ~~ Feduc_a_1
'

inc_diff_clpm2_constrained_lag <- '

  # Estimate the lagged effects (constrained)
  inc_diff_2   ~ a*inc_diff_1 + b*Feduc_a_1
  inc_diff_3   ~ a*inc_diff_2 + b*Feduc_a_2 + c*inc_diff_1 + d*Feduc_a_1
  inc_diff_4   ~ a*inc_diff_3 + b*Feduc_a_3 + c*inc_diff_2 + d*Feduc_a_2
  inc_diff_5   ~ a*inc_diff_4 + b*Feduc_a_4 + c*inc_diff_3 + d*Feduc_a_3
  inc_diff_6   ~ a*inc_diff_5 + b*Feduc_a_5 + c*inc_diff_4 + d*Feduc_a_4
  inc_diff_7   ~ a*inc_diff_6 + b*Feduc_a_6 + c*inc_diff_5 + d*Feduc_a_5
  inc_diff_8   ~ a*inc_diff_7 + b*Feduc_a_7 + c*inc_diff_6 + d*Feduc_a_6
  inc_diff_9   ~ a*inc_diff_8 + b*Feduc_a_8 + c*inc_diff_7 + d*Feduc_a_7
  inc_diff_10  ~ a*inc_diff_9 + b*Feduc_a_9 + c*inc_diff_8 + d*Feduc_a_8
  inc_diff_11  ~ a*inc_diff_10 + b*Feduc_a_10 + c*inc_diff_9 + d*Feduc_a_9
  
  
  Feduc_a_2  ~ e*inc_diff_1 + f*Feduc_a_1
  Feduc_a_3  ~ e*inc_diff_2 + f*Feduc_a_2 + g*inc_diff_1 + h*Feduc_a_1
  Feduc_a_4  ~ e*inc_diff_3 + f*Feduc_a_3 + g*inc_diff_2 + h*Feduc_a_2
  Feduc_a_5  ~ e*inc_diff_4 + f*Feduc_a_4 + g*inc_diff_3 + h*Feduc_a_3
  Feduc_a_6  ~ e*inc_diff_5 + f*Feduc_a_5 + g*inc_diff_4 + h*Feduc_a_4
  Feduc_a_7  ~ e*inc_diff_6 + f*Feduc_a_6 + g*inc_diff_5 + h*Feduc_a_5
  Feduc_a_8  ~ e*inc_diff_7 + f*Feduc_a_7 + g*inc_diff_6 + h*Feduc_a_6
  Feduc_a_9  ~ e*inc_diff_8 + f*Feduc_a_8 + g*inc_diff_7 + h*Feduc_a_7
  Feduc_a_10  ~ e*inc_diff_9 + f*Feduc_a_9 + g*inc_diff_8 + h*Feduc_a_8
  Feduc_a_11  ~ e*inc_diff_10 + f*Feduc_a_10 + g*inc_diff_9 + h*Feduc_a_9
  
  # Estimate the covariance at the first wave and residual covariances 
  inc_diff_1 ~~ Feduc_a_1 # Covariance
  inc_diff_2 ~~ Feduc_a_2 # Covariance
  inc_diff_3 ~~ Feduc_a_3 # Covariance
  inc_diff_4 ~~ Feduc_a_4 # Covariance
  inc_diff_5 ~~ Feduc_a_5 # Covariance
  inc_diff_6 ~~ Feduc_a_6 # Covariance
  inc_diff_7 ~~ Feduc_a_7 # Covariance
  inc_diff_8 ~~ Feduc_a_8 # Covariance
  inc_diff_9 ~~ Feduc_a_9 # Covariance
  inc_diff_10 ~~ Feduc_a_10 # Covariance
  inc_diff_11 ~~ Feduc_a_11 # Covariance
  
  # Estimate the variance 
  inc_diff_1 ~~ inc_diff_1 
  Feduc_a_1 ~~ Feduc_a_1
'

#store formulas in list
formulas_clpm2_nocontrols <- list(EU_clpm2_constrained_lag,
                                  cult_clpm2_constrained_lag,
                                  inc_diff_clpm2_constrained_lag)

With controls

#create formulas with controls
EU_clpm2_control_constrained_lag <- '

  # Estimate the lagged effects (constrained)
  eu_2   ~ a*eu_1 + b*Feduc_a_1 + age_1 + educ_1 + work_1 + married_1
  eu_3   ~ a*eu_2 + b*Feduc_a_2 + c*eu_1 + d*Feduc_a_1 + age_2 + educ_2 + work_2 + married_2
  eu_4   ~ a*eu_3 + b*Feduc_a_3 + c*eu_2 + d*Feduc_a_2 + age_3 + educ_3 + work_3 + married_3 
  eu_5   ~ a*eu_4 + b*Feduc_a_4 + c*eu_3 + d*Feduc_a_3 + age_4 + educ_4 + work_4 + married_4
  eu_6   ~ a*eu_5 + b*Feduc_a_5 + c*eu_4 + d*Feduc_a_4 + age_5 + educ_5 + work_5 + married_5
  eu_7   ~ a*eu_6 + b*Feduc_a_6 + c*eu_5 + d*Feduc_a_5 + age_6 + educ_6 + work_6 + married_6
  eu_8   ~ a*eu_7 + b*Feduc_a_7 + c*eu_6 + d*Feduc_a_6 + age_7 + educ_7 + work_7 + married_7 
  eu_9   ~ a*eu_8 + b*Feduc_a_8 + c*eu_7 + d*Feduc_a_7 + age_8 + educ_8 + work_8 + married_8
  eu_10  ~ a*eu_9 + b*Feduc_a_9 + c*eu_8 + d*Feduc_a_8 + age_9 + educ_9 + work_9 + married_9
  eu_11  ~ a*eu_10 + b*Feduc_a_10 + c*eu_9 + d*Feduc_a_9 + age_10 + educ_10 + work_10 + married_10
  
  
  Feduc_a_2  ~ e*eu_1 + f*Feduc_a_1 + age_1 + educ_1 + work_1 + married_1
  Feduc_a_3  ~ e*eu_2 + f*Feduc_a_2 + g*eu_1 + h*Feduc_a_1 + age_2 + educ_2 + work_2 + married_2
  Feduc_a_4  ~ e*eu_3 + f*Feduc_a_3 + g*eu_2 + h*Feduc_a_2 + age_3 + educ_3 + work_3 + married_3
  Feduc_a_5  ~ e*eu_4 + f*Feduc_a_4 + g*eu_3 + h*Feduc_a_3 + age_4 + educ_4 + work_4 + married_4 
  Feduc_a_6  ~ e*eu_5 + f*Feduc_a_5 + g*eu_4 + h*Feduc_a_4 + age_5 + educ_5 + work_5 + married_5
  Feduc_a_7  ~ e*eu_6 + f*Feduc_a_6 + g*eu_5 + h*Feduc_a_5 + age_6 + educ_6 + work_6 + married_6
  Feduc_a_8  ~ e*eu_7 + f*Feduc_a_7 + g*eu_6 + h*Feduc_a_6 + age_7 + educ_7 + work_7 + married_7
  Feduc_a_9  ~ e*eu_8 + f*Feduc_a_8 + g*eu_7 + h*Feduc_a_7 + age_8 + educ_8 + work_8 + married_8
  Feduc_a_10  ~ e*eu_9 + f*Feduc_a_9 + g*eu_8 + h*Feduc_a_8 + age_9 + educ_9 + work_9 + married_9
  Feduc_a_11  ~ e*eu_10 + f*Feduc_a_10 + g*eu_9 + h*Feduc_a_9 + age_10 + educ_10 + work_10 + married_10
  
  # Estimate the covariance at the first wave and residual covariances
  eu_1 ~~ Feduc_a_1 # Covariance
  eu_2 ~~ Feduc_a_2 # Covariance
  eu_3 ~~ Feduc_a_3 # Covariance
  eu_4 ~~ Feduc_a_4 # Covariance
  eu_5 ~~ Feduc_a_5 # Covariance
  eu_6 ~~ Feduc_a_6 # Covariance
  eu_7 ~~ Feduc_a_7 # Covariance
  eu_8 ~~ Feduc_a_8 # Covariance
  eu_9 ~~ Feduc_a_9 # Covariance
  eu_10 ~~ Feduc_a_10 # Covariance
  eu_11 ~~ Feduc_a_11 # Covariance
  
  # Estimate the variance 
  eu_1 ~~ eu_1 
  Feduc_a_1 ~~ Feduc_a_1
'

cult_clpm2_control_constrained_lag <- '

  # Estimate the lagged effects (constrained)
  cult_2   ~ a*cult_1 + b*Feduc_a_1 + age_1 + educ_1 + work_1 + married_1
  cult_3   ~ a*cult_2 + b*Feduc_a_2 + c*cult_1 + d*Feduc_a_1 + age_2 + educ_2 + work_2 + married_2
  cult_4   ~ a*cult_3 + b*Feduc_a_3 + c*cult_2 + d*Feduc_a_2 + age_3 + educ_3 + work_3 + married_3 
  cult_5   ~ a*cult_4 + b*Feduc_a_4 + c*cult_3 + d*Feduc_a_3 + age_4 + educ_4 + work_4 + married_4
  cult_6   ~ a*cult_5 + b*Feduc_a_5 + c*cult_4 + d*Feduc_a_4 + age_5 + educ_5 + work_5 + married_5
  cult_7   ~ a*cult_6 + b*Feduc_a_6 + c*cult_5 + d*Feduc_a_5 + age_6 + educ_6 + work_6 + married_6
  cult_8   ~ a*cult_7 + b*Feduc_a_7 + c*cult_6 + d*Feduc_a_6 + age_7 + educ_7 + work_7 + married_7 
  cult_9   ~ a*cult_8 + b*Feduc_a_8 + c*cult_7 + d*Feduc_a_7 + age_8 + educ_8 + work_8 + married_8
  cult_10  ~ a*cult_9 + b*Feduc_a_9 + c*cult_8 + d*Feduc_a_8 + age_9 + educ_9 + work_9 + married_9
  cult_11  ~ a*cult_10 + b*Feduc_a_10 + c*cult_9 + d*Feduc_a_9 + age_10 + educ_10 + work_10 + married_10
  
  
  Feduc_a_2  ~ e*cult_1 + f*Feduc_a_1 + age_1 + educ_1 + work_1 + married_1
  Feduc_a_3  ~ e*cult_2 + f*Feduc_a_2 + g*cult_1 + h*Feduc_a_1 + age_2 + educ_2 + work_2 + married_2
  Feduc_a_4  ~ e*cult_3 + f*Feduc_a_3 + g*cult_2 + h*Feduc_a_2 + age_3 + educ_3 + work_3 + married_3
  Feduc_a_5  ~ e*cult_4 + f*Feduc_a_4 + g*cult_3 + h*Feduc_a_3 + age_4 + educ_4 + work_4 + married_4 
  Feduc_a_6  ~ e*cult_5 + f*Feduc_a_5 + g*cult_4 + h*Feduc_a_4 + age_5 + educ_5 + work_5 + married_5
  Feduc_a_7  ~ e*cult_6 + f*Feduc_a_6 + g*cult_5 + h*Feduc_a_5 + age_6 + educ_6 + work_6 + married_6
  Feduc_a_8  ~ e*cult_7 + f*Feduc_a_7 + g*cult_6 + h*Feduc_a_6 + age_7 + educ_7 + work_7 + married_7
  Feduc_a_9  ~ e*cult_8 + f*Feduc_a_8 + g*cult_7 + h*Feduc_a_7 + age_8 + educ_8 + work_8 + married_8
  Feduc_a_10  ~ e*cult_9 + f*Feduc_a_9 + g*cult_8 + h*Feduc_a_8 + age_9 + educ_9 + work_9 + married_9
  Feduc_a_11  ~ e*cult_10 + f*Feduc_a_10 + g*cult_9 + h*Feduc_a_9 + age_10 + educ_10 + work_10 + married_10
  
  # Estimate the covariance at the first wave. 
  cult_1 ~~ Feduc_a_1 # Covariance
  cult_2 ~~ Feduc_a_2 # Covariance
  cult_3 ~~ Feduc_a_3 # Covariance
  cult_4 ~~ Feduc_a_4 # Covariance
  cult_5 ~~ Feduc_a_5 # Covariance
  cult_6 ~~ Feduc_a_6 # Covariance
  cult_7 ~~ Feduc_a_7 # Covariance
  cult_8 ~~ Feduc_a_8 # Covariance
  cult_9 ~~ Feduc_a_9 # Covariance
  cult_10 ~~ Feduc_a_10 # Covariance
  cult_11 ~~ Feduc_a_11 # Covariance
  
  # Estimate the variance 
  eu_1 ~~ eu_1 
  Feduc_a_1 ~~ Feduc_a_1
'

inc_diff_clpm2_control_constrained_lag <- '

  # Estimate the lagged effects (constrained)
  inc_diff_2   ~ a*inc_diff_1 + b*Feduc_a_1 + age_1 + educ_1 + work_1 + married_1
  inc_diff_3   ~ a*inc_diff_2 + b*Feduc_a_2 + c*inc_diff_1 + d*Feduc_a_1 + age_2 + educ_2 + work_2 + married_2
  inc_diff_4   ~ a*inc_diff_3 + b*Feduc_a_3 + c*inc_diff_2 + d*Feduc_a_2 + age_3 + educ_3 + work_3 + married_3 
  inc_diff_5   ~ a*inc_diff_4 + b*Feduc_a_4 + c*inc_diff_3 + d*Feduc_a_3 + age_4 + educ_4 + work_4 + married_4
  inc_diff_6   ~ a*inc_diff_5 + b*Feduc_a_5 + c*inc_diff_4 + d*Feduc_a_4 + age_5 + educ_5 + work_5 + married_5
  inc_diff_7   ~ a*inc_diff_6 + b*Feduc_a_6 + c*inc_diff_5 + d*Feduc_a_5 + age_6 + educ_6 + work_6 + married_6
  inc_diff_8   ~ a*inc_diff_7 + b*Feduc_a_7 + c*inc_diff_6 + d*Feduc_a_6 + age_7 + educ_7 + work_7 + married_7 
  inc_diff_9   ~ a*inc_diff_8 + b*Feduc_a_8 + c*inc_diff_7 + d*Feduc_a_7 + age_8 + educ_8 + work_8 + married_8
  inc_diff_10  ~ a*inc_diff_9 + b*Feduc_a_9 + c*inc_diff_8 + d*Feduc_a_8 + age_9 + educ_9 + work_9 + married_9
  inc_diff_11  ~ a*inc_diff_10 + b*Feduc_a_10 + c*inc_diff_9 + d*Feduc_a_9 + age_10 + educ_10 + work_10 + married_10
  
  
  Feduc_a_2  ~ e*inc_diff_1 + f*Feduc_a_1 + age_1 + educ_1 + work_1 + married_1
  Feduc_a_3  ~ e*inc_diff_2 + f*Feduc_a_2 + g*inc_diff_1 + h*Feduc_a_1 + age_2 + educ_2 + work_2 + married_2
  Feduc_a_4  ~ e*inc_diff_3 + f*Feduc_a_3 + g*inc_diff_2 + h*Feduc_a_2 + age_3 + educ_3 + work_3 + married_3
  Feduc_a_5  ~ e*inc_diff_4 + f*Feduc_a_4 + g*inc_diff_3 + h*Feduc_a_3 + age_4 + educ_4 + work_4 + married_4 
  Feduc_a_6  ~ e*inc_diff_5 + f*Feduc_a_5 + g*inc_diff_4 + h*Feduc_a_4 + age_5 + educ_5 + work_5 + married_5
  Feduc_a_7  ~ e*inc_diff_6 + f*Feduc_a_6 + g*inc_diff_5 + h*Feduc_a_5 + age_6 + educ_6 + work_6 + married_6
  Feduc_a_8  ~ e*inc_diff_7 + f*Feduc_a_7 + g*inc_diff_6 + h*Feduc_a_6 + age_7 + educ_7 + work_7 + married_7
  Feduc_a_9  ~ e*inc_diff_8 + f*Feduc_a_8 + g*inc_diff_7 + h*Feduc_a_7 + age_8 + educ_8 + work_8 + married_8
  Feduc_a_10  ~ e*inc_diff_9 + f*Feduc_a_9 + g*inc_diff_8 + h*Feduc_a_8 + age_9 + educ_9 + work_9 + married_9
  Feduc_a_11  ~ e*inc_diff_10 + f*Feduc_a_10 + g*inc_diff_9 + h*Feduc_a_9 + age_10 + educ_10 + work_10 + married_10
  
  # Estimate the covariance at the first wave and residual covariances 
  inc_diff_1 ~~ Feduc_a_1 # Covariance
  inc_diff_2 ~~ Feduc_a_2 # Covariance
  inc_diff_3 ~~ Feduc_a_3 # Covariance
  inc_diff_4 ~~ Feduc_a_4 # Covariance
  inc_diff_5 ~~ Feduc_a_5 # Covariance
  inc_diff_6 ~~ Feduc_a_6 # Covariance
  inc_diff_7 ~~ Feduc_a_7 # Covariance
  inc_diff_8 ~~ Feduc_a_8 # Covariance
  inc_diff_9 ~~ Feduc_a_9 # Covariance
  inc_diff_10 ~~ Feduc_a_10 # Covariance
  inc_diff_11 ~~ Feduc_a_11 # Covariance
  
  # Estimate the variance 
  inc_diff_1 ~~ inc_diff_1 
  Feduc_a_1 ~~ Feduc_a_1
'

#store formulas in list
formulas_clpm2_controls <- list(EU_clpm2_control_constrained_lag,
                                cult_clpm2_control_constrained_lag,
                                inc_diff_clpm2_control_constrained_lag)

Model estimation

filename <- file.path("results",
                      "riclpm",
                      "240821_lavaan-clpm-results.Rdata")

#full model list
clpm_model_formulas <- list(
  EU_clpm1_constrained_lag,
  cult_clpm1_constrained_lag,
  inc_diff_clpm1_constrained_lag,
  EU_clpm1_control_constrained_lag,
  cult_clpm1_control_constrained_lag,
  inc_diff_clpm1_control_constrained_lag,
  EU_clpm2_constrained_lag,
  cult_clpm2_constrained_lag,
  inc_diff_clpm2_constrained_lag,
  EU_clpm2_control_constrained_lag,
  cult_clpm2_control_constrained_lag,
  inc_diff_clpm2_control_constrained_lag
)

#estimation
if (!file.exists(filename)) {
  # paralellize the estimation
  numCores <- detectCores()
  registerDoParallel(core = 6)
  
  #estimate models and store in list
  clpm_results_list <- foreach(a = 1:length(clpm_model_formulas),
                               .packages = "lavaan") %dopar% {
                                 sem(
                                   clpm_model_formulas[[a]],
                                   data = MyData,
                                   estimator = 'MLR',
                                   missing = "ML",
                                   meanstructure = T,
                                   int.ov.free = T
                                 )
                               }
  
  #stop parralellization
  stopImplicitCluster()
  
  save(clpm_results_list,
       file = filename)
} else
  (
    load(filename)
  )

[1] “clpm_results_list”

Results

#create sublist to loop over
clpm1_nocontrol <- clpm_results_list[1:3] 
clpm1_control <- clpm_results_list[4:6]
clpm2_nocontrol <- clpm_results_list[7:9] 
clpm2_control <- clpm_results_list[10:12]


#extract infro from lists
clpm1_nocontrol_df <- foreach(a = 1:3,
                              .combine = rbind) %do% {
                                standardizedSolution(clpm1_nocontrol[[a]]) %>%
                                  mutate(depvar = a,
                                         model = "clpm1",
                                         control = "none") %>% 
                                  filter(row_number() %in% c(2,21)) %>% 
                                  mutate(effect = case_when(
                                    label == "b" ~ 1,
                                    label == "c" ~ 2
                                  ))
                              }

clpm1_control_df <- foreach(a = 1:3,
                              .combine = rbind) %do% {
                                standardizedSolution(clpm1_control[[a]]) %>%
                                  mutate(depvar = a,
                                         model = "clpm1",
                                         control = "control") %>% 
                                  filter(row_number() %in% c(2,61)) %>% 
                                  mutate(effect = case_when(
                                    label == "b" ~ 1,
                                    label == "c" ~ 2
                                  ))
                              }

clpm2_nocontrol_df <- foreach(a = 1:3,
                            .combine = rbind) %do% {
                              standardizedSolution(clpm2_nocontrol[[a]]) %>%
                                mutate(depvar = a,
                                       model = "clpm2",
                                       control = "none") %>% 
                                filter(row_number() %in% c(4,6,41,43)) %>% 
                                mutate(effect = case_when(
                                  label == "b" ~ 1,
                                  label == "e" ~ 3,
                                  label == "d" ~ 2,
                                  label == "g" ~ 4,
                                ))
                            }

clpm2_control_df <- foreach(a = 1:3,
                            .combine = rbind) %do% {
                              standardizedSolution(clpm2_control[[a]]) %>%
                                mutate(depvar = a,
                                       model = "clpm2",
                                       control = "control") %>% 
                                filter(row_number() %in% c(8,10,85,87)) %>% 
                                mutate(effect = case_when(
                                  label == "b" ~ 1,
                                  label == "e" ~ 3,
                                  label == "d" ~ 2,
                                  label == "g" ~ 4,
                                ))
                            }
  
#create one df
results_df <- list(clpm1_control_df,
     clpm1_nocontrol_df,
     clpm2_control_df,
     clpm2_nocontrol_df) %>% 
  bind_rows()

#plot_clpm1
plot_clpm1 <- results_df %>% 
  filter(model == "clpm1") %>%
  filter(control == "control") %>% 
  rename(indep_var = rhs) %>% 
  mutate(effect = factor(effect, 
                         levels = 1:2,
                         labels = c(
                           "Educational \n influence",
                           "Selection")),
         depvar = case_when(
           depvar == 1 ~ 2,
           depvar == 2 ~ 1,
           depvar == 3 ~ 3
         ),
         depvar = factor(depvar,
                         levels = 1:3,
                         labels = c(
                           "Cultural Inclusion",
                           "European Integration",
                           "Income Equality"))) %>% 
  ggplot(aes(x = est.std, 
             y = effect,
             shape = effect)) +
  geom_vline(xintercept = 0) +
  geom_linerange(aes(xmin = est.std - (se*1.96), 
                     xmax = est.std + (se*1.96))) +
  geom_point(aes(colour = effect), 
             size = 2) +
  facet_wrap(vars(depvar),
             ncol = 1) +
  scale_x_continuous(limits = c(-0.10, 0.10)) +
  scale_colour_viridis(discrete = T,
                       option = "D")  +
  theme(panel.background = element_rect(fill = "#FFFFFF"),
        plot.background = element_rect(fill = "#FFFFFF"),
        panel.grid = element_line(colour = "grey"),
        panel.grid.major.y = element_blank(),
        text = element_text(family = "sans", size = 12),
        strip.background = element_rect(fill = "#A9A9A9"),
        panel.grid.minor = element_blank(),
        legend.position = "none",
        legend.title = element_blank(),
        legend.background = element_rect(fill = "#FFFFFF"),
        legend.key = element_rect(fill = "#FFFFFF")) +
  labs(y = "Path",
       x = "Standardized Estimate")

ggsave(plot = plot_clpm1, 
       file = "plots/results/riclpm_clpm1_240821.pdf",
       dpi = 600, width = 5, height = 4)

plot_clpm1

#plot_clpm2
plot_clpm2 <- results_df %>% 
  filter(model == "clpm2") %>%
  filter(control == "control") %>% 
  rename(indep_var = rhs) %>% 
  mutate(effect = factor(effect, 
                         levels = 1:4,
                         labels = c(
                           "Educational influence (lag 1)",
                           "Educational influence (lag 2)",
                           "Selection (lag 1)",
                           "Selection (lag 2)")),
         depvar = case_when(
           depvar == 1 ~ 2,
           depvar == 2 ~ 1,
           depvar == 3 ~ 3
         ),
         depvar = factor(depvar,
                         levels = 1:3,
                         labels = c(
                           "Cultural Inclusion",
                           "European Integration",
                           "Income Equality"))) %>% 
  ggplot(aes(x = est.std, 
             y = effect,
             shape = effect)) +
  geom_vline(xintercept = 0) +
  geom_linerange(aes(xmin = est.std - (se*1.96), 
                     xmax = est.std + (se*1.96))) +
  geom_point(aes(colour = effect,
                 fill = effect), 
             size = 2) +
  facet_wrap(vars(depvar),
             ncol = 1) +
  scale_shape_manual(values = c(21,22,23,24)) +
  scale_x_continuous(limits = c(-0.1, 0.1)) +
  scale_colour_viridis(discrete = T,
                       option = "D")  +
  scale_fill_viridis(discrete = T,
                       option = "D")  +
  theme(panel.background = element_rect(fill = "#FFFFFF"),
        plot.background = element_rect(fill = "#FFFFFF"),
        panel.grid = element_line(colour = "grey"),
        panel.grid.major.y = element_blank(),
        text = element_text(family = "sans", size = 12),
        strip.background = element_rect(fill = "#A9A9A9"),
        panel.grid.minor = element_blank(),
        legend.position = "none",
        legend.title = element_blank(),
        legend.background = element_rect(fill = "#FFFFFF"),
        legend.key = element_rect(fill = "#FFFFFF")) +
  labs(y = "Path",
       x = "Standardized Estimate")

ggsave(plot = plot_clpm2, 
       file = "plots/results/riclpm_clpm2_240821.pdf",
       dpi = 600, width = 5, height = 4)

plot_clpm2

---

Copyright © 2024 Jeroense Thijmen