This primer employs data from the European Social Survey Round 11. Please find the questionnaire here. For variable naming conventions refer to the code book pp. 61ff.
With question C18 of the survey all 40156 survey participants were asked if they would describe themselves as being a member of a group that is discrimated against in their counry of residence. Respondents who agreed were subsequently asked on what grounds their group is discriminated against (question C19). Possible answers were coded as separate binary variable [variable name in squared brackets] with the values “Marked” and “Not Marked”, respondents could select all that applied:
Now we have to compile these ten multiple response questions (C19a…C19j) into a single table that informs us about the grounds to feel discriminated against.
A trivial approach includes reporting the frequency tables of all single responses separately:
# collect original variable names
mrnames <- c(
"dscrrce","dscrntn","dscrrlg","dscrlng","dscretn",
"dscrage","dscrgnd","dscrsex","dscrdsb","dscroth"
)
# provide frequency table for each answer
sapply(df[,mrnames], table)## dscrrce dscrntn dscrrlg dscrlng dscretn dscrage dscrgnd dscrsex
## Not marked 39675 39575 39660 39964 39767 39772 39537 39767
## Marked 481 581 496 192 389 384 619 389
## dscrdsb dscroth
## Not marked 39909 39521
## Marked 247 635
In order to refer the “Marked” answers to the respective sub population we have to select respondents who marked at least one single item (i.e. who were asked question C19 and gave a valid response). Their total count is:
# create temporary data frame for multiple response items
mr_df = df[,mrnames]
# select all cases with at east one Marked answer
mr_df = mr_df[rowSums(mr_df == "Marked") > 0,]
# save and check row count
n = nrow(mr_df)
n## [1] 3053
Now we know the number of items that were not marked by all respondents who were asked for the grounds of discrimination:
# provide frequency table for each answer and
# save as multiple response (mr) table
mr_table = sapply(mrnames, function (x) table(mr_df[,x]))
# convert into data frame
mr_table = as.data.frame(mr_table)
# print simple table
mr_table## dscrrce dscrntn dscrrlg dscrlng dscretn dscrage dscrgnd dscrsex
## Not marked 2572 2472 2557 2861 2664 2669 2434 2664
## Marked 481 581 496 192 389 384 619 389
## dscrdsb dscroth
## Not marked 2806 2418
## Marked 247 635
We will build on this table to get a more dateild overview. First, add totals, namely the numer of respondents (n) and the number of valid answers (sum) provided by them.
# add totals (valid respondents (n), valid answers (sum))
mr_table$Total = c(n, rowSums(mr_table[2,]))
mr_table # print, outcomment if not needed## dscrrce dscrntn dscrrlg dscrlng dscretn dscrage dscrgnd dscrsex
## Not marked 2572 2472 2557 2861 2664 2669 2434 2664
## Marked 481 581 496 192 389 384 619 389
## dscrdsb dscroth Total
## Not marked 2806 2418 3053
## Marked 247 635 4413
Now we transpose the table, the transposition is again (re)converted into a data frame.
# transpose, save as data frame again
mr_table = as.data.frame(t(mr_table))
mr_table # print, outcomment if not needed## Not marked Marked
## dscrrce 2572 481
## dscrntn 2472 581
## dscrrlg 2557 496
## dscrlng 2861 192
## dscretn 2664 389
## dscrage 2669 384
## dscrgnd 2434 619
## dscrsex 2664 389
## dscrdsb 2806 247
## dscroth 2418 635
## Total 3053 4413
We have to calculate the share of “Marked” answers from all valid answers. The latter was calculated above:
n## [1] 3053which is equal to summarizing “Marked” and “Not Marked” row wise:
mr_table[,"Marked"] + mr_table[,"Not marked"]## [1] 3053 3053 3053 3053 3053 3053 3053 3053 3053 3053 7466
The percentages can now be computed as
# add percentages
mr_table[,"% of Respondents"] = round(mr_table$Marked*100 / n, 1)
mr_table # print, outcomment if not needed## Not marked Marked % of Respondents
## dscrrce 2572 481 15.8
## dscrntn 2472 581 19.0
## dscrrlg 2557 496 16.2
## dscrlng 2861 192 6.3
## dscretn 2664 389 12.7
## dscrage 2669 384 12.6
## dscrgnd 2434 619 20.3
## dscrsex 2664 389 12.7
## dscrdsb 2806 247 8.1
## dscroth 2418 635 20.8
## Total 3053 4413 144.5
The table’s row names show the variable short names which cam now be replaced by their respective lables:
# replace item names
rownames(mr_table) = c(
"colour or race",
"nationality",
"religion",
"language",
"ethnic group",
"age",
"gender",
"sexuality",
"disability",
"other grounds",
"Total"
)
mr_table # print, outcomment if not needed## Not marked Marked % of Respondents
## colour or race 2572 481 15.8
## nationality 2472 581 19.0
## religion 2557 496 16.2
## language 2861 192 6.3
## ethnic group 2664 389 12.7
## age 2669 384 12.6
## gender 2434 619 20.3
## sexuality 2664 389 12.7
## disability 2806 247 8.1
## other grounds 2418 635 20.8
## Total 3053 4413 144.5
Delete column “Not marked” as it is redundant:
# remove "Not marked"
mr_table = mr_table[,2:3]
mr_table # print, outcomment if not needed## Marked % of Respondents
## colour or race 481 15.8
## nationality 581 19.0
## religion 496 16.2
## language 192 6.3
## ethnic group 389 12.7
## age 384 12.6
## gender 619 20.3
## sexuality 389 12.7
## disability 247 8.1
## other grounds 635 20.8
## Total 4413 144.5
Finally, create a ranking to improve our message:
# get order of items, ignore Total
sort_order = order(mr_table[1:nrow(mr_table)-1, "% of Respondents"])
# reverse order, add Total
sort_order = c(rev(sort_order), nrow(mr_table))
# rank items
mr_table = mr_table[sort_order,]
mr_table # print, outcomment if not needed## Marked % of Respondents
## other grounds 635 20.8
## gender 619 20.3
## nationality 581 19.0
## religion 496 16.2
## colour or race 481 15.8
## sexuality 389 12.7
## ethnic group 389 12.7
## age 384 12.6
## disability 247 8.1
## language 192 6.3
## Total 4413 144.5Now we can easily identify the TOP 3 grounds to feel discriminated against, namely other grounds (20.8%), gender (20.3%), and nationality (19%), while disability (8.1%) and language (6.3%) are the least important factors. On average, each respondent has reported 1.45 grouns to feel discriminated against.
This result can now be labeled and styled to your personal preferences.
# show formatted table
footnote(
kable_styling(
kable(mr_table,
caption = paste0("Grounds to feel discriminated against (n=",n,")")
),
font_size = 13, full_width = F,
bootstrap_options = c("hover", "condensed"),
),
c("ESS round 11, all countries
Only respondents who describe themselves as being
a member of a group that is discriminated against"), general_title=""
)| Marked | % of Respondents | |
|---|---|---|
| other grounds | 635 | 20.8 |
| gender | 619 | 20.3 |
| nationality | 581 | 19.0 |
| religion | 496 | 16.2 |
| colour or race | 481 | 15.8 |
| sexuality | 389 | 12.7 |
| ethnic group | 389 | 12.7 |
| age | 384 | 12.6 |
| disability | 247 | 8.1 |
| language | 192 | 6.3 |
| Total | 4413 | 144.5 |
|
ESS round 11, all countries Only respondents who describe themselves as being a member of a group that is discriminated against |