Is it possible for esttex output to respect common (i.e. unique) dictionary aliases?
I often want to output similar variables on the same row in a regression table. For instance, to make them immediately comparable even if they are named differently in my actual R model calls. The way this works for other regression outputting packages in R (e.g. huxtable or stargazer) is to give them the same alias. However, that strategy does not appear to work here.
Consider the following example, where I want to compare full vs partial marginal effects, using different model syntax. The coefficient on Wind in mod_2 is actually the same as Month5:Wind in mod_1. Yet, even if I explicitly give it the same alias, the two coefficients are still printed on separate rows.
library(fixest)
aq = airquality
aq$Month = factor(aq$Month)
## Full marginal effects
mod_1 = feols(Ozone ~ Month / Wind, data = aq)
#> NOTE: 37 observations removed because of NA values (Breakup: LHS: 37, RHS: 0).
## Partial marginal effects
mod_2 = feols(Ozone ~ Month * Wind, data = aq)
#> NOTE: 37 observations removed because of NA values (Breakup: LHS: 37, RHS: 0).
## "Wind" in mod_2 is actual "Month5:Wind"...
myDict = c("Wind" = "Month5:Wind")
## Doesn't work as expected: Get separate "Month5:Wind" rows instead of one.
esttex(
mod_1, mod_2, titles = c("Full MEs", "Partial MEs"),
dict = myDict,
drop = c("Int", paste0("Month", 6:9)) ## Optional: Just to highlight rows of interest
)
#> \begin{table}[htbp]\centering
#> \caption{no title}
#> \begin{tabular}{lcc}
#> & & \tabularnewline
#> \hline
#> \hline
#> Dependent Variable:&\multicolumn{2}{c}{Ozone}\\
#> Model:&(1)&(2)\\
#> \hline
#> \emph{Variables}\tabularnewline
#> Month5:Wind&-2.3681$^{*}$& \\
#> &(1.3162)& \\
#> Month5:Wind& &-2.3681$^{*}$\\
#> & &(1.3162)\\
#> \hline
#> \emph{Fit statistics}& & \\
#> Observations& 116&116\\
#> R$^2$ & 0.54732&0.54732\\
#> Adjusted R$^2$ & 0.50889&0.50889\\
#> \hline
#> \hline
#> \multicolumn{3}{l}{\emph{Normal standard-errors in parentheses. Signif Codes: ***: 0.01, **: 0.05, *: 0.1}}\\
#> \end{tabular}
#> \end{table}
As promised, a second
esttex
issue...Is it possible for
esttex
output to respect common (i.e. unique) dictionary aliases?I often want to output similar variables on the same row in a regression table. For instance, to make them immediately comparable even if they are named differently in my actual R model calls. The way this works for other regression outputting packages in R (e.g. huxtable or stargazer) is to give them the same alias. However, that strategy does not appear to work here.
Consider the following example, where I want to compare full vs partial marginal effects, using different model syntax. The coefficient on Wind in
mod_2
is actually the same as Month5:Wind inmod_1
. Yet, even if I explicitly give it the same alias, the two coefficients are still printed on separate rows.Created on 2019-12-29 by the reprex package (v0.3.0)