How do you get significant differences to show with custom table, when using text in Expss?

Question

How do you get significant differences to show with custom table, when using text instead of labelled numerics as input? (should be + or – in each cell).

#Generate data
aw = function(n) { 
    sample(x = c("Aware","Consider","Buy"), n, replace = T, prob = c(0.1, 0.4, 0.5)) 
}
awa<-aw(3000)

gen = function(n) { 
    sample(x = c("Male","Female"), n, replace = T, prob = c(0.8, 0.2)) 
}
gende<-gen(3000)

ag = function(n) { 
    sample(x = c("<20","20 to 50","50+"), n, replace = T, prob = c(0.3, 0.2, 0.5)) 
}

age<-ag(3000)

data<-data.frame(awa,gende,age)
#Banner create
banner = calc(data, list(total(), gende,age))
#Custom table with significant differences calculated relative to total
data %>%
tab_significance_options(compare_type = "adjusted_first_column",subtable_marks = "both",sig_labels_first_column = c("-", "+"),mode = c("append")) %>%
tab_cells(awa) %>%
tab_cols(banner) %>%
tab_stat_cpct() %>%
tab_pivot()

score 0 · Accepted Answer · answered Mar 15 '19 at 09:18

There is a function tab_last_sig_cpct which calculate significance for the last statistics. Statistics should be column percent.

library(expss)
#Generate data
aw = function(n) { 
    sample(x = c("Aware","Consider","Buy"), n, replace = T, prob = c(0.1, 0.4, 0.5)) 
}
awa<-aw(3000)

gen = function(n) { 
    sample(x = c("Male","Female"), n, replace = T, prob = c(0.8, 0.2)) 
}
gende<-gen(3000)

ag = function(n) { 
    sample(x = c("<20","20 to 50","50+"), n, replace = T, prob = c(0.3, 0.2, 0.5)) 
}

age<-ag(3000)

data<-data.frame(awa,gende,age)
#Banner create
banner = calc(data, list(total(), gende,age))
#Custom table with significant differences calculated relative to total
data %>%
    tab_significance_options(compare_type = "adjusted_first_column",subtable_marks = "both",sig_labels_first_column = c("-", "+"),mode = c("append")) %>%
    tab_cells(awa) %>%
    tab_cols(banner) %>%
    tab_stat_cpct() %>%
    tab_last_sig_cpct(mode = "replace") %>% 
    tab_pivot()

 # |     |              | #Total | Female |   Male |    <20 | 20 to 50 |  50+ |
 # | --- | ------------ | ------ | ------ | ------ | ------ | -------- | ---- |
 # | awa |        Aware |   10.0 |  7.0 - | 10.8 + |  9.5   |     10.3 | 10.2 |
 # |     |          Buy |   51.1 | 53.8   | 50.4   | 48.6   |     51.4 | 52.5 |
 # |     |     Consider |   38.8 | 39.1   | 38.8   | 41.9 + |     38.3 | 37.3 |
 # |     | #Total cases |   3000 |  611   | 2389   |  884   |      601 | 1515 |

This works perfectly thanks. I misunderstood 'tab_last_sig_cpct' in the reference work to be only for cases where you wished to compare to previous adjacent column (for example when comparing waves). — CK7, Mar 15 '19 at 21:43

How do you get significant differences to show with custom table, when using text in Expss?

1 Answers1