Display exact p-values in `etable' #590
Description
Hello Laurent,
A journal asked me to report all my tables with exact p-values instead of the usual asterisk pattern for 0.01, 0.05, 0.1.
According to the documentation:
The coefstat parameter accepts only one of: "se" (default), "tstat", or "confint".
I need to display coefficients, standard errors, and exact p-values in my regression tables, so I'm thinking of something like:
coefstat = c("coef", "se", "p")
I couldn't find a way to do this in fixest. As always, if I missed something that's already available, I apologize.
If this isn't possible right now, I'll handle it using postprocess.tex, but this seems a bit hacky, so I'd prefer to avoid it if there's a cleaner solution.
library(fixest)
data("iris")
# Fit a linear regression model using fixest
model1 <- feols(Sepal.Length ~ Sepal.Width + Petal.Length + Petal.Width, data = iris)
model2 <- feols(Sepal.Length ~ Sepal.Width + Petal.Length + Petal.Width | Species, data = iris)
# Replace stars with p-values
sign_dict<-seq(0,1,by=0.001)
names_dict<-paste0(",p<",sign_dict)
names(sign_dict) <- names_dict
rm_signif_legend<- function(tab) {
# Remove the line that include the text " Signif. codes:"
tab <- tab[!grepl("Signif", tab)]
# Replace $^{,p<0.1}$ with (p<0.1)
tab <- gsub("\\$\\^\\{,p<([0-9.]+)\\}\\$", "$p<\\1$", tab)
return(tab)
}
etable(model1, model2,
title = "Linear Regression Models for Sepal Length",
digits = 3,
signif.code = sign_dict,
postprocess.tex = rm_signif_legend,
tex=TRUE
)
Which returns:
\begin{table}[htbp]
\caption{Linear Regression Models for Sepal Length}
\centering
\begin{tabular}{lcc}
\tabularnewline \midrule \midrule
Dependent Variable: & \multicolumn{2}{c}{Sepal.Length}\\
Model: & (1) & (2)\\
\midrule
\emph{Variables}\\
Constant & 1.86, p<0.001 & \\
& (0.251) & \\
Sepal.Width & 0.651, p<0.001 & 0.496, p<0.055\\
& (0.067) & (0.121)\\
Petal.Length & 0.709, p<0.001 & 0.829, p<0.014\\
& (0.057) & (0.097)\\
Petal.Width & -0.556, p<0.001 & -0.315, p<0.103\\
& (0.128) & (0.110)\\
\midrule
\emph{Fixed-effects}\\
Species & & Yes\\
\midrule
\emph{Fit statistics}\\
Observations & 150 & 150\\
R$^2$ & 0.85861 & 0.86731\\
Within R$^2$ & & 0.65201\\
\end{tabular}
\end{table}