1

I have an object one that contains the results of a kruskal-wallis test. The p-value is really tiny i.e. 6.86e-09. Is there a way in inline code in R markdown to say p < 0.0001 instead of writing the exact p-value?

one <- kruskal.test(Petal.Width ~ Species, data = iris)

I have tried:

There was a statistically significant difference in the Petal Width between species
H(`r one[[2]]`) = `r round(one[[1]], 3)`, *p* = `r round(one[[3]], 11)`.

And:

`r if(one[[3]] < 0.0001){ print("< 0.0001") } else { round(one[[3]], 4) }`

Desired output:

There was a statistically significant difference in the Petal Width between species H(2) = 131.19, p < 0.0001.

There are lot's out there on how to recode into significance codes etc, but not really how to do this for inline code. I am new to this so any help is appreciated!

Darren Tsai
  • 32,117
  • 5
  • 21
  • 51
MM1
  • 478
  • 15

3 Answers3

2

You can use a vectorised ifelse for this. If the value is smaller than 0.0001, print a "<" symbol together with 0.0001, else print a "=" symbol with the actual rounded value.

An addition one[[2]] is included to demonstrate the case where the value is greater than 0.0001.

You can add an extra ifelse statement at the beginning of your sentence (ifelse(one[[3]] < 0.0001, "was", "wasn't")) to determine whether it is statistically significant or not.

---
title: "Untitled"
output: pdf_document
date: "2023-01-10"
---

```{r}
one <- kruskal.test(Petal.Width ~ Species, data = iris)
```

There `r ifelse(one[[3]] < 0.0001, "was", "wasn't")` a statistically significant difference in the Petal Width between species H(`r one[[2]]`) = `r round(one[[1]], 3)`, p `r ifelse(one[[3]] < 0.0001, paste("<", 0.0001), paste("=", round(one[[3]], 4)))`.

There `r ifelse(one[[2]] < 0.0001, "was", "wasn't")` a statistically significant difference in the Petal Width between species H(`r one[[2]]`) = `r round(one[[1]], 3)`, p `r ifelse(one[[2]] < 0.0001, paste("<", 0.0001), paste("=", round(one[[2]], 4)))`.

Rmarkdown PDF output example:

rmd_out

benson23
  • 16,369
  • 9
  • 19
  • 38
2

You can use format.pval() to format p-values.

format.pval(one$p.value, eps = 0.0001, scientific = FALSE)
# [1] "< 0.0001"

where eps is a threshold; those less than eps are formatted as "< [eps]" (where [eps] stands for format(eps, digits)).

Darren Tsai
  • 32,117
  • 5
  • 21
  • 51
1

For one, the print statement is probably unnecessary. On my end, it helped splitting the ifelse out of the inline, writing its result to a variable, and then just calling on that variable in the inline

one <- kruskal.test(Petal.Width ~ Species, data = iris)
pval <- if(one[[3]] < 0.0001){"< 0.0001"}else{round(one[[3]], 4)}

There was a statistically significant difference in the Petal Width between species H(`r one[[2]]`) = `r round(one[[1]], 3)`, p = `r pval`.

If you only have two conditions you can also use ifelse -

pval <- ifelse(one[[3]] < 0.0001, "< 0.0001", round(one[[3]],4))
mweylandt
  • 108
  • 5