Colours

The relevant data quality information should be available not only based on colours but also based on additional elements, e.g. the amount of an effect size, or a line indicating a range or variance.

Colours in plots

Using ggplot2 allows to manipulate the colours later. Nevertheless, we recommend the following conventions for colours.

Colours for discrete scales

For discrete scales, such as interviewers or centres, we recommend to generate colour-blind friendly figures (as suggested here; we have augmented the list by two colours: grey and brown).

QS_Name hex_code red green blue
qs_black #000000 0 0 0
qs_gray #B0B0B0 176 176 176
qs_orange #E69F00 230 159 0
qs_skyblue #56B4E9 86 180 233
qs_green #009E73 0 158 115
qs_yellow #F0E442 240 228 66
qs_blue #0072B2 0 114 178
qs_red #D55E00 213 94 0
qs_purple #CC79A7 204 121 167
qs_brown #8C510A 140 81 10

These colours are not recommended for the representation of data quality issues.

Colours for continuous scales

For continuous scales, such as magnitude of effects, we recommend to generate colour-blind friendly figures, as recommended here. Alternatively, we recommend the use of Viridis.

These colours are not recommended for the representation of data quality classifications.

Colours indicating the magnitude of data quality issues

To graph data quality issues, we recommend to generate colour-blind friendly figures, as recommended here.

The red pole should always be used to identify problems.

Colours in tables

We assume, a function produces the following data frame:

Using the R-package formattable this data frame can be formatted as follows (crude example):

Please see some annotations on the use of this R-package; other important options are datatable (DT) and knitr::kableExtra

Colours for tables should be based on the colours mentioned above for graphical output.