Inconsistent renaming of statistic and estimate columns

[mps9506]

Thanks for adding the ability to easily rename statistics and estimates in the development version of modelsummary(). This is very handy, but I came across a few inconsistencies in application when reshaping the table to have statistics in a separate column (I could very well be missing something obvious here).

• Example 1a renaming confidence intervals splits the column and replaces the estimate column name with NA:

library(modelsummary)
mod <- lm(mpg ~ factor(cyl), mtcars)

modelsummary(
  mod, 
  estimate = c("$\\hat{\\beta}$" = "estimate"),
  statistic = c("Confidence Interval" = "conf.int"),
  shape = term ~ model + statistic)

+--------------+---------+---------------------+--------+
|              | (1)                           |        |
+--------------+---------+---------------------+--------+
|              | NA      | Confidence Interval | (1)    |
+==============+=========+=====================+========+
| (Intercept)  | 26.664  | 24.676              | 28.651 |
+--------------+---------+---------------------+--------+
| factor(cyl)6 | -6.921  | -10.108             | -3.734 |
+--------------+---------+---------------------+--------+
| factor(cyl)8 | -11.564 | -14.220             | -8.908 |
+--------------+---------+---------------------+--------+
| Num.Obs.     | 32      |                     |        |
+--------------+---------+---------------------+--------+
| R2           | 0.732   |                     |        |
+--------------+---------+---------------------+--------+
| R2 Adj.      | 0.714   |                     |        |
+--------------+---------+---------------------+--------+
| AIC          | 170.6   |                     |        |
+--------------+---------+---------------------+--------+
| BIC          | 176.4   |                     |        |
+--------------+---------+---------------------+--------+
| Log.Lik.     | -81.282 |                     |        |
+--------------+---------+---------------------+--------+
| RMSE         | 3.07    |                     |        |
+--------------+---------+---------------------+--------+ 

Example 1b: using "[{conf.low}, {conf.high}]" fails to rename the confidence interval column:

modelsummary(
  mod, 
  estimate = c("$\\hat{\\beta}$" = "estimate"),
  statistic = c("Confidence Interval" = "[{conf.low}, {conf.high}]"),
  shape = term ~ model + statistic)

+--------------+---------------+---------------------------+
|              | (1)                                       |
+--------------+---------------+---------------------------+
|              | $\hat{\beta}$ | [{conf.low}, {conf.high}] |
+==============+===============+===========================+
| (Intercept)  | 26.664        | [24.676, 28.651]          |
+--------------+---------------+---------------------------+
| factor(cyl)6 | -6.921        | [-10.108, -3.734]         |
+--------------+---------------+---------------------------+
| factor(cyl)8 | -11.564       | [-14.220, -8.908]         |
+--------------+---------------+---------------------------+
| Num.Obs.     | 32            |                           |
+--------------+---------------+---------------------------+
| R2           | 0.732         |                           |
+--------------+---------------+---------------------------+
| R2 Adj.      | 0.714         |                           |
+--------------+---------------+---------------------------+
| AIC          | 170.6         |                           |
+--------------+---------------+---------------------------+
| BIC          | 176.4         |                           |
+--------------+---------------+---------------------------+
| Log.Lik.     | -81.282       |                           |
+--------------+---------------+---------------------------+
| RMSE         | 3.07          |                           |
+--------------+---------------+---------------------------+ 

and, example 2, using a named vector greater than length equal 1 for the statistics columns causes the estimate column to be named NA (maybe I'm using this incorrectly?):

modelsummary(
  mod, 
  estimate = c("$\\hat{\\beta}$" = "estimate"),
  statistic = c("t-stat" = "statistic",
                "p-value" = "p.value"),
  shape = term ~ model + statistic)

+--------------+---------+--------+---------+
|              | (1)                        |
+--------------+---------+--------+---------+
|              | NA      | t-stat | p-value |
+==============+=========+========+=========+
| (Intercept)  | 26.664  | 27.437 | <0.001  |
+--------------+---------+--------+---------+
| factor(cyl)6 | -6.921  | -4.441 | <0.001  |
+--------------+---------+--------+---------+
| factor(cyl)8 | -11.564 | -8.905 | <0.001  |
+--------------+---------+--------+---------+
| Num.Obs.     | 32      |        |         |
+--------------+---------+--------+---------+
| R2           | 0.732   |        |         |
+--------------+---------+--------+---------+
| R2 Adj.      | 0.714   |        |         |
+--------------+---------+--------+---------+
| AIC          | 170.6   |        |         |
+--------------+---------+--------+---------+
| BIC          | 176.4   |        |         |
+--------------+---------+--------+---------+
| Log.Lik.     | -81.282 |        |         |
+--------------+---------+--------+---------+
| RMSE         | 3.07    |        |         |
+--------------+---------+--------+---------+ 

---

output from sessionInfo()

R version 4.3.1 (2023-06-16 ucrt)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 11 x64 (build 22631)

Matrix products: default


locale:
[1] LC_COLLATE=English_United States.utf8  LC_CTYPE=English_United States.utf8    LC_MONETARY=English_United States.utf8 LC_NUMERIC=C                          
[5] LC_TIME=English_United States.utf8    

time zone: America/New_York
tzcode source: internal

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

other attached packages:
[1] survey_4.4-1            survival_3.5-7          Matrix_1.6-1.1          modelsummary_1.9.9.9999

loaded via a namespace (and not attached):
 [1] compiler_4.3.1       crayon_1.5.2         Rcpp_1.0.11          tinytable_0.1.0.9000 callr_3.7.3          splines_4.3.1        fastmap_1.1.1        lattice_0.21-8      
 [9] R6_2.5.1             tables_0.9.17        generics_0.1.3       curl_5.0.2           knitr_1.45           MASS_7.3-60          backports_1.4.1      checkmate_2.3.1     
[17] desc_1.4.2           insight_0.19.8.10    rprojroot_2.0.3      DBI_1.1.3            rlang_1.1.3          performance_0.10.9.6 xfun_0.42            parameters_0.21.5.7 
[25] datawizard_0.9.1.7   cli_3.6.2            ps_1.7.5             digest_0.6.33        processx_3.8.2       rstudioapi_0.15.0    remotes_2.4.2.1      prettyunits_1.2.0   
[33] glue_1.6.2           data.table_1.14.8    mitools_2.4          pkgbuild_1.4.2       fansi_1.0.4          bayestestR_0.13.2    tools_4.3.1          htmltools_0.5.6     

[vincentarelbundock]

thanks a lot for the report. I really appreciate that you took the time to craft minimal and useful example.

I believe that all the issues highlighted are now fixed on Github.