Appendix D — Summary tables

While we need to report the parameters and statistics of our models in text with any reports, it can also be extremely useful to provide a model summary table, especially when reporting complex analyses. Below I present examples from two of my favourite packages for model reporting.

Both packages come with lots of options for customisation to produce presentation ready tables

D.1 sjPlot

sjPlot A really nice package that helps produce model summaries for you automatically, currently it will only output in HTML format

library(sjPlot)
tab_model(William_ls2)

	William
Predictors	Estimates	CI	p
(Intercept)	8.53	8.38 – 8.68	<0.001
O3	-7.14	-9.15 – -5.13	<0.001
Stress [Well-watered]	0.18	0.07 – 0.29	0.003
Observations	30
R² / R² adjusted	0.703 / 0.681

tab_model(
  William_ls2, 
  pred.labels = c("Intercept", "O³", "Stress[Well-watered]"),
  dv.labels = "Biomass",
  string.pred = "Coefficient",
  string.ci = "Conf. Int (95%)",
  string.p = "P-Value"
)

	Biomass
Coefficient	Estimates	Conf. Int (95%)	P-Value
Intercept	8.53	8.38 – 8.68	<0.001
O³	-7.14	-9.15 – -5.13	<0.001
Stress[Well-watered]	0.18	0.07 – 0.29	0.003
Observations	30
R² / R² adjusted	0.703 / 0.681

gtsummary Is an extension of the gt package and provides even greater levels of flexibility and customisation. It is also compatible with HTML, word or pdf. Perhaps the only downside is the level of options and customisability can be slightly overwhelming. In addition the gt package has been undergoing a lot of active development, so sometimes older code examples are now defunct.

library(gtsummary)
tbl_regression(William_ls2)

Characteristic	Beta	95% CI¹	p-value
O3	-7.1	-9.2, -5.1	<0.001
Stress
Stressed	—	—
Well-watered	0.18	0.07, 0.29	0.003
¹ CI = Confidence Interval

tbl_regression(William_ls2,
               intercept = TRUE,
               label = list(O3 ~ "O³", Stress ~ "Stress"),
               pvalue_fun = ~ style_pvalue(.x, digits = 2)) %>% 
  modify_header(label = "Coefficient",
                estimate = "Estimate") %>% 
  bold_labels() %>% 
  bold_p(t = 0.05) %>% 
  as_gt () %>% 
   gt::tab_source_note(gt::md("*This is an ordinary least squares model*"))

Coefficient	Estimate	95% CI¹	p-value
(Intercept)	8.5	8.4, 8.7	<0.001
O³	-7.1	-9.2, -5.1	<0.001
Stress
Stressed	—	—
Well-watered	0.18	0.07, 0.29	0.003
This is an ordinary least squares model
¹ CI = Confidence Interval