escape=FALSE not functioning for panel titles with fixest_multi objects when exporting table in latex_tabular

[chickymonkeys]

Hi!

I would like to share a potential issue with you as I have tried to circumvent it in several ways without any success. In a nutshell, the escape=FALSE option for modelsummary seems to ignore the titles of panels when using rbind or rcollapse, that keep to be escaped in case of latex content in there.

Here is a minimal reproducible example with mtcars:

library(modelsummary)
library(fixest)

# regression with fixest package
reg <- fixest::feols(
    c(mpg, hp) ~ sw(log(qsec), log(drat)) | gear + vs, data = mtcars
)

# grep panels 
panels_reg <- list(
    "Panel A. $\\log(qsec)_{gt}" = reg[grepl("qsec", names(reg))],
    "Panel B. $\\Delta(drat)_{gt}" = reg[grepl("qsec", names(reg))]
)

# some latex math mode for the coefficients 
coef_labels_reg <- rep(
        "$\\log({SO_{X}})_{g} \\times Post_{t}$",
    length(panels_reg[[1]])
)
names(coef_labels_reg) <- c("log(qsec)", "(logdrat)")

# adding some rows at the end (these are not escaped)
rows_reg <- tibble(
    a = c("\\hspace{1em}Gear FE", rep("$\\checkmark$", 2)),
    b = c("\\hspace{1em}VS FE", rep("$\\checkmark$", 2))
)
rows_reg <- data.frame(t(rows_reg))

# run modelsummary with `escape=FALSE` option on
modelsummary::modelsummary(
    panels_reg,
    output = "latex_tabular",
    coef_map = coef_labels_reg,
    shape = "rbind",
    stars = c("*" = 0.1, "**" = 0.05, "***" = 0.01),
    col.names = NULL,
    add_rows = rows_reg,
    escape = FALSE
)

Here is my sessionInfo() output:

R version 4.2.3 (2023-03-15)
Platform: aarch64-apple-darwin22.3.0 (64-bit)
Running under: macOS Ventura 13.2.1

Matrix products: default
LAPACK: /opt/homebrew/Cellar/r/4.2.3/lib/R/lib/libRlapack.dylib

locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/POSIX/en_US.UTF-8/en_US.UTF-8

attached base packages:
[1] stats     graphics  grDevices utils     datasets 
[6] methods   base     

other attached packages:
 [1] fstcore_0.9.14          tmaptools_3.1-1        
 [3] tmap_3.3-3              terra_1.7-21           
 [5] sf_1.0-11               rio_0.5.29             
 [7] fst_0.9.8               broom_1.0.4            
 [9] did_2.1.2               fixest_0.11.1          
[11] data.table_1.14.8       modelsummary_1.3.0.9013
[13] ggplot2_3.4.1           collapse_1.9.3         
[15] lubridate_1.9.2         stringr_1.5.0          
[17] tidyr_1.3.0             dplyr_1.1.0            
[19] here_1.0.1              pacman_0.5.1           

loaded via a namespace (and not attached):
  [1] readxl_1.4.2         backports_1.4.1     
  [3] systemfonts_1.0.4    lwgeom_0.2-11       
  [5] sp_1.6-0             crosstalk_1.2.0     
  [7] leaflet_2.1.2        httpgd_1.3.1        
  [9] digest_0.6.31        htmltools_0.5.4     
 [11] fansi_1.0.4          magrittr_2.0.3      
 [13] checkmate_2.1.0      tzdb_0.3.0          
 [15] openxlsx_4.2.5.2     BMisc_1.4.5         
 [17] readr_2.1.4          vroom_1.6.1         
 [19] sandwich_3.0-2       svglite_2.1.1       
 [21] timechange_0.2.0     colorspace_2.1-0    
 [23] rvest_1.0.3          haven_2.5.2         
 [25] xfun_0.37            leafem_0.2.0        
 [27] crayon_1.5.2         jsonlite_1.8.4      
 [29] DRDID_1.0.4          zoo_1.8-11          
 [31] glue_1.6.2           kableExtra_1.3.4    
 [33] stars_0.6-0          gtable_0.3.2        
 [35] webshot_0.5.4        car_3.1-1           
 [37] abind_1.4-5          scales_1.2.1        
 [39] DBI_1.1.3            rstatix_0.7.2       
 [41] Rcpp_1.0.10          viridisLite_0.4.1   
 [43] performance_0.10.2.6 units_0.8-1         
 [45] foreign_0.8-84       bit_4.0.5           
 [47] proxy_0.4-27         Formula_1.2-5       
 [49] datawizard_0.6.5     htmlwidgets_1.6.2   
 [51] httr_1.4.5           RColorBrewer_1.1-3  
 [53] wk_0.7.2             ellipsis_0.3.2      
 [55] pkgconfig_2.0.3      XML_3.99-0.14       
 [57] farver_2.1.1         utf8_1.2.3          
 [59] tidyselect_1.2.0     labeling_0.4.2      
 [61] rlang_1.1.0          later_1.3.0         
 [63] munsell_0.5.0        cellranger_1.1.0    
 [65] tools_4.2.3          cli_3.6.0           
 [67] generics_0.1.3       evaluate_0.20       
 [69] fastmap_1.1.1        tables_0.9.10       
 [71] leafsync_0.1.0       knitr_1.42          
 [73] bit64_4.0.5          zip_2.2.2           
 [75] purrr_1.0.1          s2_1.1.2            
 [77] nlme_3.1-162         xml2_1.3.3          
 [79] compiler_4.2.3       rstudioapi_0.14     
 [81] curl_5.0.0           png_0.1-8           
 [83] e1071_1.7-13         ggsignif_0.6.4      
 [85] tibble_3.2.0         stringi_1.7.12      
 [87] parameters_0.20.2.10 forcats_1.0.0       
 [89] lattice_0.20-45      Matrix_1.5-3        
 [91] classInt_0.4-9       vctrs_0.6.0         
 [93] pillar_1.8.1         lifecycle_1.0.3     
 [95] trust_0.1-8          insight_0.19.1      
 [97] raster_3.6-20        R6_2.5.1            
 [99] KernSmooth_2.23-20   codetools_0.2-19    
[101] dichromat_2.0-0.1    rprojroot_2.0.3     
[103] withr_2.5.0          bayestestR_0.13.0   
[105] parallel_4.2.3       hms_1.1.2           
[107] dreamerr_1.2.3       grid_4.2.3          
[109] class_7.3-21         rmarkdown_2.20      
[111] carData_3.0-5        ggpubr_0.6.0        
[113] numDeriv_2016.8-1.1  base64enc_0.1-3  

I hope it is relevant! Or if it is not, please let me know whether I am doing something wrong :)

[vincentarelbundock]

Thanks a lot for your detailed and useful report! I am sorry for the delay; my “real” job has been keeping me very busy.

Version 1.3.0.9017 on Github should fix the issue. Please let me know if it works for you.

library(modelsummary)
tmp <- transform(mtcars, a_b = qsec, b_c = hp)
panels <- list(
"Panel B: $\\log(3)^2$" = list(
    "A_B" = lm(mpg ~ a_b, data = tmp),
    "B_C" = lm(mpg ~ a_b + b_c, data = tmp)),
"Panel B: $\\log(3)_2$" = list(
    "A_B" = lm(disp ~ a_b, data = tmp),
    "B_C" = lm(disp ~ a_b + b_c, data = tmp))
)
modelsummary(panels, output = "latex_tabular", shape = "rbind", title = "$\\beta_1$", escape = FALSE)
# \begin{table}
# 
# \caption{\label{tab:unnamed-chunk-2}$\beta_1$}
# \centering
# \begin{tabular}[t]{lcc}
# \toprule
#   & A_B & B_C\\
# \midrule
# \addlinespace[0.5em]
# \multicolumn{3}{l}{\textit{Panel B: $\log(3)^2$}}\\
# \midrule \hspace{1em}(Intercept) & -5.114 & 48.324\\
# \hspace{1em} & (10.030) & (11.103)\\
# \hspace{1em}a_b & 1.412 & -0.887\\
# \hspace{1em} & (0.559) & (0.535)\\
# \hspace{1em}b_c &  & -0.085\\
# \hspace{1em} &  & (0.014)\\
# \hspace{1em}Num.Obs. & 32 & \vphantom{1} 32\\
# \hspace{1em}R2 & 0.175 & 0.637\\
# \hspace{1em}R2 Adj. & 0.148 & 0.612\\
# \hspace{1em}AIC & 204.6 & 180.3\\
# \hspace{1em}BIC & 209.0 & 186.2\\
# \hspace{1em}Log.Lik. & -99.294 & -86.170\\
# \hspace{1em}F & 6.377 & 25.431\\
# \hspace{1em}RMSE & 5.39 & 3.57\\
# \addlinespace[0.5em]
# \multicolumn{3}{l}{\textit{Panel B: $\log(3)_2$}}\\
# \midrule \hspace{1em}(Intercept) & 767.619 & -340.780\\
# \hspace{1em} & (204.642) & (221.689)\\
# \hspace{1em}a_b & -30.080 & 17.599\\
# \hspace{1em} & (11.410) & (10.674)\\
# \hspace{1em}b_c &  & 1.755\\
# \hspace{1em} &  & (0.278)\\
# \hspace{1em}Num.Obs. & 32 & 32\\
# \hspace{1em}R2 & 0.188 & 0.658\\
# \hspace{1em}R2 Adj. & 0.161 & 0.634\\
# \hspace{1em}AIC & 397.6 & 372.0\\
# \hspace{1em}BIC & 402.0 & 377.8\\
# \hspace{1em}Log.Lik. & -195.797 & -181.979\\
# \hspace{1em}F & 6.950 & 27.859\\
# \hspace{1em}RMSE & 109.92 & 71.37\\
# \bottomrule
# \end{tabular}
# \end{table}