Tables.Rmd

Communicating Results

We don’t just do data analysis to do it. Usually, we need to share it with someone else. Whether that is a member of a lab, our boss, the general public, we need to be able to communicate results effectively. Most people probably do not want to look at data the way we’ve been looking for it.

The purpose of this notebook is to look at making tables prettier. We want to enable whoever we’re talking to to turn the data we have and the analysis we’ve done into insights.

Let us consider this case. You are designing a new trapping experiment out at the Portal, Arizona field site. You want to know what types of animal in- and exclosures are catching in the most animals

surveys  %>% 
  na.omit() %>% 
  group_by(plot_type) %>% 
  filter(species_id == "NL") %>% 
  count()

## # A tibble: 5 × 2
## # Groups:   plot_type [5]
##   plot_type                     n
##   <chr>                     <int>
## 1 Control                     506
## 2 Long-term Krat Exclosure    141
## 3 Rodent Exclosure            121
## 4 Short-term Krat Exclosure   209
## 5 Spectab exclosure            68

We’ve seen this before. It looks fine. It gets the message across. We can, however, make it look a little nicer:

table <- surveys  %>% 
  na.omit() %>% 
  group_by(plot_type) %>% 
  filter(species_id == "NL") %>% 
  count()

knitr::kable(table )

plot_type	n
Control	506
Long-term Krat Exclosure	141
Rodent Exclosure	121
Short-term Krat Exclosure	209
Spectab exclosure	68

We can make a few edits to display more intuitive labeling:

table <- table %>% 
  rename_at('plot_type', ~'Experimental Treatment')

table <- table %>% 
  rename_at('n', ~'Number Sampled')

knitr::kable(table)

Experimental Treatment	Number Sampled
Control	506
Long-term Krat Exclosure	141
Rodent Exclosure	121
Short-term Krat Exclosure	209
Spectab exclosure	68

These are some viable options for small tables. Let’s think about bigger tables. Big things are hard to look at. Let’s get a good, but not overwhelming, amount of data.

data_for_table <- surveys %>% 
  na.omit() %>% 
  select(species_id, weight, year) %>% 
  group_by(year) %>% 
  summarize(mean_weight = mean(weight)) %>% 
  arrange()
data_for_table

## # A tibble: 26 × 2
##     year mean_weight
##    <dbl>       <dbl>
##  1  1977        46.6
##  2  1978        66.2
##  3  1979        63.2
##  4  1980        64.6
##  5  1981        65.2
##  6  1982        52.7
##  7  1983        54.7
##  8  1984        50.7
##  9  1985        46.4
## 10  1986        54.9
## # ℹ 16 more rows

We can try a visual with kable.

data_for_table <- surveys %>% 
  na.omit() %>% 
  select(species_id, weight, year) %>% 
  group_by(year) %>% 
  summarize(mean_weight = mean(weight)) %>% 
  arrange()
knitr::kable(data_for_table)

year	mean_weight
1977	46.55512
1978	66.21739
1979	63.19056
1980	64.57789
1981	65.18014
1982	52.67731
1983	54.66106
1984	50.74800
1985	46.41154
1986	54.88319
1987	49.32500
1988	44.58118
1989	35.35225
1990	34.89904
1991	32.03571
1992	33.27897
1993	34.18859
1994	34.36813
1995	29.66978
1996	28.13416
1997	31.67637
1998	34.83116
1999	36.22316
2000	32.07805
2001	36.21897
2002	35.52647

This isn’t bad - it at least allows us to scroll and look at things. The DT package, however, allows us to look very comfortably at these data. We can even search!

data_for_table <- surveys %>% 
  na.omit() %>% 
  select(species_id, weight, year) %>% 
  group_by(year) %>% 
  summarize(mean_weight = mean(weight)) %>% 
  arrange()
DT::datatable(data_for_table)

Even looking at surveys isn’t so painful with datatable. It’s a real memory hog, though, so don’t everyone do it at once.

April Wright

2023-08-31

Communicating Results