add_p() t.test method change rendering issue

[postgres-newbie]

When generating a tbl_summary, if an add_p(test = everything() ~ "t.test") was used and later the tbl_summary object (let's call it: sample_tbl_summary) was run through the statement:
add_p(sample_tbl_summary, test = everything() ~ aov),
the output in viewer still displays the p-values from the t.test instead of updating them to those obtained from an ANOVA.

[ddsjoberg]

Thank you for your post. Please update your post with a reproducible example, aka a reprex. A reprex includes both data and code I can run on my machine that replicates the issue you're reporting. Take just a few minutes to review reprex.tidyverse.org for detailed instructions on how to create a reprex.

I am going to go ahead and close this issue for now. But please do re-open the issue after you've added your reproducible example.

[postgres-newbie]

Here's some reproducible code:

library(gtsummary);
library(gt);
library(survival)
library(dplyr);
library(stringr);
library(purrr);
library(forcats);
library(tidyr)

save_tbl_summary<-trial %>%
select(response, age, grade) %>%
mutate(response = factor(response, labels = c("No Tumor Response", "Tumor Responded"))) %>%
tbl_summary(
by = response,
missing = "no",
label = list(age ~ "Patient Age", grade ~ "Tumor Grade")
) %>%
add_p(pvalue_fun = ~style_pvalue(.x, digits = 2)) %>%
add_q()

The p-value that gets rendered is >0.9 when it is 0.813

add_p(save_tbl_summary, test = everything() ~ "aov")

Here is the p-value when calculated using "aov"

aov(as.numeric(as.factor(grade)) ~ as.factor(response),
data = subset(trial, subset= response %in% c(1, 0))) %>% summary()

Df Sum Sq Mean Sq F value Pr(>F)

as.factor(response) 1 0.04 0.0383 0.056 0.813 # The p-value that gets rendered is >0.9 when it is 0.813

Residuals 191 129.88 0.6800

Here's the code to show that the chisel.test p-value gets reported which is 0.93

chisq.test(x = trial$response[which(trial$response %in% c(0, 1))],
y = trial$grade[which(trial$response%in% c(0, 1))])

Pearson's Chi-squared test

data: trial$response[which(trial$response %in% c(0, 1))] and trial$grade[which(trial$response %in% c(0, 1))]

X-squared = 0.15014, df = 2, p-value = 0.9277

[ddsjoberg]

Thanks for adding the example.

I wasn't expecting users to run add_p() multiple times on the same table. It looks like I need add a check that the p-value column is not already present, and if it is then throw an error.

[postgres-newbie]

Yes, an error message would be nice or what would be better is to allow for the p-values to be overwritten in the rendered version with the new method.

[ddsjoberg]

you can create another table and merge them together with tbl_merge(). We rely on the column names, so cannot allow more than one run without causing issues downstream

[ddsjoberg]

error messaging added to main branch