Quarto tables, figures, and stats

Chunks can produce figures and tables

```{r}
#| label: tbl-one
#| tbl-cap: "This is a great table"
knitr::kable(mtcars)
```

Chunks can produce figures or tables

```{r}
#| label: fig-hist
#| fig-cap: "This is a histogram"
hist(rnorm(100))
```

Figure 1: This is a histogram

Cross-referencing

You can then refer to those with @tbl-one and @fig-hist and the Table and Figure ordering will be correct (and linked)

@fig-hist contains a histogram and @tbl-one a table.

gets printed as:

Figure 1 contains a histogram and Table 1 a table.

Inline R

Along with just regular text, you can also run R code within the text:

There were `r 3 + 4` participants

becomes:

There were 7 participants

Inline stats

I often create a list of stats that I want to report in a manuscript:

stats <- list(n = nrow(data),
              mean_age = mean(data$age))

I can then print these numbers in the text with:

There were `r stats$n` participants with a mean age of `r stats$mean_age`.

which turns into:

There were 1123 participants with a mean age of 43.5.

Inline stats from {gtsummary}

We saw very, very briefly yesterday:

inline_text(income_table, variable = "age_bir")

[1] "595 (95% CI 538, 652; p<0.001)"

We pulled a statistic from our univariate table

If we’re making a table, we probably want to report numbers from it

```{r}
#| label: tbl-descr
#| tbl-cap: "Descriptive statistics"
#| output-location: slide
table1 <- tbl_summary(
  nlsy,
  by = sex_cat,
  include = c(sex_cat, race_eth_cat, region_cat,
              eyesight_cat, glasses, age_bir)) |> 
  add_overall(last = TRUE)
table1
```

I want to report some stats!

How about the median (IQR) age of the male participants at the birth of their first child?

inline_text(table1, variable = "age_bir", column = "Male")

[1] "25 (21, 29)"

Or the frequency and percentage of women from the South?

inline_text(table1, variable = "region_cat", level = "South", column = "Female")

[1] "2,317 (38%)"

And the overall stats on people who wear glasses?

inline_text(table1, variable = "glasses", column = "stat_0",
            pattern = "{n}/{N} ({p}%)")

[1] "3,894/8,450 (46%)"

Better yet…

We can integrate these into the text of our manuscript:

A greater proportion of female (`r inline_text(table1, variable = "glasses", column = "Female")`) than male 

(`r inline_text(table1, variable = "glasses", column = "Male")`) participants wore glasses.

Which becomes:

A greater proportion of female (2,328 (54%)) than male (1,566 (38%)) participants wore glasses.

Readability

Because this can be hard to read, I’d suggest storing those stats in a chunk before the text:

```{r}
glasses_f <- inline_text(table1, variable = "glasses", 
                         column = "Female")
glasses_m <- inline_text(table1, variable = "glasses", 
                         column = "Male")
```
A greater proportion of female (`r glasses_f`) than male (`r glasses_m`) participants wore glasses.

Exercises

Return to the quarto document with the tables.

• Choose a table to label and caption, and then write a sentence that cross-references it (e.g., Table 1 shows the descriptive statistics)
• From that table, choose at least two statistics to pull out of the table and include in the text using inline_text().
• Add another statistic to the text that you calculate yourself using the nlsy data, e.g., the mean number of hours of sleep on weekends.