get_all_corr makes it easy to calculate correlations across
every variable in a data frame or a select set of variables. It also
works with grouped data frames so you can check correlations among
the levels of several grouping variables.
Arguments
- data
A data frame or tibble object
- cols
<
tidy-select> The variables you want to get the correlations for.- group
<
tidy-select> A selection of columns to group the data by in addition totreats. This operates very similarly to.byfrom dplyr (for more info on that see ?dplyr_by). See examples to see how it operates.- wt
A variable to use as the weights for weighted correlations
- remove_redundant
Should rows where the two variables are the same be kept or removed? If
TRUE, the default, they are removed.- decimals
Number of decimals each number should be rounded to. Default is 3.
Examples
# load dplyr and adlgraphs
library(dplyr)
library(adlgraphs)
# To get correlations with three variables you can do it three ways
# 1. Create a new data frame with only the columns you want
new_data <- test_data %>% dplyr::select(top:dominate)
get_all_corr(new_data)
#> # A tibble: 6 × 8
#> x y correlation n conf.low conf.high p_value stars
#> <chr+lbl> <chr+lbl> <dbl> <int> <dbl> <dbl> <dbl> <chr>
#> 1 inferior [Some … top [An i… 0.498 250 0.389 0.606 0 ***
#> 2 dominate [No on… top [An i… -0.147 250 -0.271 -0.023 0.02 *
#> 3 top [An ideal s… inf… [Som… 0.498 250 0.389 0.606 0 ***
#> 4 dominate [No on… inf… [Som… -0.138 250 -0.262 -0.015 0.029 *
#> 5 top [An ideal s… dom… [No … -0.147 250 -0.271 -0.023 0.02 *
#> 6 inferior [Some … dom… [No … -0.138 250 -0.262 -0.015 0.029 *
# 2. Using dplyr::select() and pipes
test_data %>%
dplyr::select(c(top:dominate)) %>%
get_all_corr()
#> # A tibble: 6 × 8
#> x y correlation n conf.low conf.high p_value stars
#> <chr+lbl> <chr+lbl> <dbl> <int> <dbl> <dbl> <dbl> <chr>
#> 1 inferior [Some … top [An i… 0.498 250 0.389 0.606 0 ***
#> 2 dominate [No on… top [An i… -0.147 250 -0.271 -0.023 0.02 *
#> 3 top [An ideal s… inf… [Som… 0.498 250 0.389 0.606 0 ***
#> 4 dominate [No on… inf… [Som… -0.138 250 -0.262 -0.015 0.029 *
#> 5 top [An ideal s… dom… [No … -0.147 250 -0.271 -0.023 0.02 *
#> 6 inferior [Some … dom… [No … -0.138 250 -0.262 -0.015 0.029 *
# 3. Use the `cols` argument
get_all_corr(test_data, cols = c(top:dominate))
#> # A tibble: 6 × 8
#> x y correlation n conf.low conf.high p_value stars
#> <chr+lbl> <chr+lbl> <dbl> <int> <dbl> <dbl> <dbl> <chr>
#> 1 inferior [Some … top [An i… 0.498 250 0.389 0.606 0 ***
#> 2 dominate [No on… top [An i… -0.147 250 -0.271 -0.023 0.02 *
#> 3 top [An ideal s… inf… [Som… 0.498 250 0.389 0.606 0 ***
#> 4 dominate [No on… inf… [Som… -0.138 250 -0.262 -0.015 0.029 *
#> 5 top [An ideal s… dom… [No … -0.147 250 -0.271 -0.023 0.02 *
#> 6 inferior [Some … dom… [No … -0.138 250 -0.262 -0.015 0.029 *
# or
test_data %>% get_all_corr(c(top:dominate))
#> # A tibble: 6 × 8
#> x y correlation n conf.low conf.high p_value stars
#> <chr+lbl> <chr+lbl> <dbl> <int> <dbl> <dbl> <dbl> <chr>
#> 1 inferior [Some … top [An i… 0.498 250 0.389 0.606 0 ***
#> 2 dominate [No on… top [An i… -0.147 250 -0.271 -0.023 0.02 *
#> 3 top [An ideal s… inf… [Som… 0.498 250 0.389 0.606 0 ***
#> 4 dominate [No on… inf… [Som… -0.138 250 -0.262 -0.015 0.029 *
#> 5 top [An ideal s… dom… [No … -0.147 250 -0.271 -0.023 0.02 *
#> 6 inferior [Some … dom… [No … -0.138 250 -0.262 -0.015 0.029 *
# To get weighted correlations just specify the `wt` argument
test_data %>% get_all_corr(c(top:dominate), wt = wts)
#> # A tibble: 6 × 8
#> x y correlation n conf.low conf.high p_value stars
#> <chr+lbl> <chr+lbl> <dbl> <int> <dbl> <dbl> <dbl> <chr>
#> 1 inferior [Some … top [An i… 0.509 250 0.401 0.617 0 ***
#> 2 dominate [No on… top [An i… -0.112 250 -0.237 0.012 0.076 .
#> 3 top [An ideal s… inf… [Som… 0.509 250 0.401 0.617 0 ***
#> 4 dominate [No on… inf… [Som… -0.143 250 -0.267 -0.019 0.024 *
#> 5 top [An ideal s… dom… [No … -0.112 250 -0.237 0.012 0.076 .
#> 6 inferior [Some … dom… [No … -0.143 250 -0.267 -0.019 0.024 *
# You can also calculate grouped correlations. For example, if
# you were interested in comparing the weighted correlations
# among people with a college degree vs those without one, you
# would do it like this:
test_data %>%
dplyr::group_by(edu_f2) %>%
get_all_corr(c(top:dominate), wt = wts)
#> # A tibble: 12 × 9
#> edu_f2 x y correlation n conf.low conf.high p_value
#> <fct> <chr+lbl> <chr+lbl> <dbl> <int> <dbl> <dbl> <dbl>
#> 1 No Colleg… inf… [Som… top [An i… 0.526 142 0.384 0.668 0
#> 2 At Least … inf… [Som… top [An i… 0.479 108 0.31 0.648 0
#> 3 No Colleg… dom… [No … top [An i… -0.058 142 -0.225 0.108 0.49
#> 4 At Least … dom… [No … top [An i… -0.217 108 -0.405 -0.029 0.024
#> 5 No Colleg… top [An i… inf… [Som… 0.526 142 0.384 0.668 0
#> 6 At Least … top [An i… inf… [Som… 0.479 108 0.31 0.648 0
#> 7 No Colleg… dom… [No … inf… [Som… -0.121 142 -0.286 0.045 0.153
#> 8 At Least … dom… [No … inf… [Som… -0.18 108 -0.37 0.009 0.062
#> 9 No Colleg… top [An i… dom… [No … -0.058 142 -0.225 0.108 0.49
#> 10 At Least … top [An i… dom… [No … -0.217 108 -0.405 -0.029 0.024
#> 11 No Colleg… inf… [Som… dom… [No … -0.121 142 -0.286 0.045 0.153
#> 12 At Least … inf… [Som… dom… [No … -0.18 108 -0.37 0.009 0.062
#> # ℹ 1 more variable: stars <chr>
# Another way to calculate grouped correlations is to
# specify the group argument inside the function call:
get_all_corr(test_data, c(top:dominate), edu_f2, wts)
#> # A tibble: 12 × 9
#> edu_f2 x y correlation n conf.low conf.high p_value
#> <fct> <chr+lbl> <chr+lbl> <dbl> <int> <dbl> <dbl> <dbl>
#> 1 No Colleg… inf… [Som… top [An i… 0.526 142 0.384 0.668 0
#> 2 At Least … inf… [Som… top [An i… 0.479 108 0.31 0.648 0
#> 3 No Colleg… dom… [No … top [An i… -0.058 142 -0.225 0.108 0.49
#> 4 At Least … dom… [No … top [An i… -0.217 108 -0.405 -0.029 0.024
#> 5 No Colleg… top [An i… inf… [Som… 0.526 142 0.384 0.668 0
#> 6 At Least … top [An i… inf… [Som… 0.479 108 0.31 0.648 0
#> 7 No Colleg… dom… [No … inf… [Som… -0.121 142 -0.286 0.045 0.153
#> 8 At Least … dom… [No … inf… [Som… -0.18 108 -0.37 0.009 0.062
#> 9 No Colleg… top [An i… dom… [No … -0.058 142 -0.225 0.108 0.49
#> 10 At Least … top [An i… dom… [No … -0.217 108 -0.405 -0.029 0.024
#> 11 No Colleg… inf… [Som… dom… [No … -0.121 142 -0.286 0.045 0.153
#> 12 At Least … inf… [Som… dom… [No … -0.18 108 -0.37 0.009 0.062
#> # ℹ 1 more variable: stars <chr>
# You can also use multiple grouping variables
get_all_corr(test_data, c(top:dominate), c(edu_f2, pid_f3), wts)
#> # A tibble: 36 × 10
#> edu_f2 pid_f3 x y correlation n conf.low conf.high
#> <fct> <fct> <chr+lbl> <chr+lbl> <dbl> <int> <dbl> <dbl>
#> 1 No College … Democ… inf… [Som… top [An … 0.514 52 0.27 0.757
#> 2 No College … Indep… inf… [Som… top [An … 0.446 40 0.152 0.74
#> 3 No College … Repub… inf… [Som… top [An … 0.521 50 0.274 0.769
#> 4 At Least a … Democ… inf… [Som… top [An … 0.333 46 0.046 0.619
#> 5 At Least a … Indep… inf… [Som… top [An … 0.729 17 0.352 1.11
#> 6 At Least a … Repub… inf… [Som… top [An … 0.571 45 0.318 0.823
#> 7 No College … Democ… dom… [No … top [An … -0.05 52 -0.334 0.233
#> 8 No College … Indep… dom… [No … top [An … -0.301 40 -0.614 0.013
#> 9 No College … Repub… dom… [No … top [An … 0.044 50 -0.246 0.334
#> 10 At Least a … Democ… dom… [No … top [An … -0.227 46 -0.523 0.069
#> # ℹ 26 more rows
#> # ℹ 2 more variables: p_value <dbl>, stars <chr>