participation in study segments is not recorded by respective variables, e.g. a participant's refusal to attend a specific examination is not recorded.
participation in study segments is recorded by respective variables.
Use case (1) will be common in smaller studies. For the calculation of segment missingness it is assumed that study variables are nested in respective segments. This structure must be specified in the static metadata. The R-function identifies all variables within each segment and returns TRUE if all variables within a segment are missing, otherwise FALSE.
Use case (2) assumes a more complex structure of study data and metadata.
The study data comprise so-called intro-variables (either TRUE/FALSE or codes
for non-participation). The column PART_VAR in the metadata is
filled by variable-IDs indicating for each variable the respective
intro-variable. This structure has the benefit that subsequent calculation of
item missingness obtains correct denominators for the calculation of
missingness rates.
Descriptor
com_segment_missingness(
study_data,
meta_data,
group_vars = NULL,
meta_data_segment,
strata_vars = NULL,
label_col,
threshold_value,
direction,
color_gradient_direction,
expected_observations = c("HIERARCHY", "ALL", "SEGMENT"),
exclude_roles = c(VARIABLE_ROLES$PROCESS)
)a list with:
SummaryData: data frame about segment missingness
SummaryPlot: ggplot2 heatmap plot: a heatmap-like graphic that
highlights critical values depending on the respective
threshold_value and direction.
data.frame the data frame that contains the measurements
data.frame the data frame that contains metadata attributes of study data
variable the name of a variable used for grouping, defaults to NULL for not grouping output
data.frame Segment level metadata. Optional.
variable the name of a variable used for stratification, defaults to NULL for not grouping output
variable attribute the name of the column in the metadata with labels of variables
numeric from=0 to=100. a numerical value ranging from 0-100
enum low | high. "high" or "low", i.e. are deviations above/below the threshold critical. This argument is deprecated and replaced by color_gradient_direction.
enum above | below. "above" or "below", i.e. are deviations above or below the threshold critical? (default: above)
enum HIERARCHY | ALL | SEGMENT. If ALL, all
observations are expected to comprise
all study segments. If SEGMENT, the
PART_VAR is expected to point
to a variable with values of 0 and 1,
indicating whether the variable was
expected to be observed for each data
row. If HIERARCHY, this is also
checked recursively, so, if a variable
points to such a participation variable,
and that other variable does has also
a PART_VAR entry pointing
to a variable, the observation of the
initial variable is only
expected, if both segment variables are
1.
variable roles a character (vector) of variable roles not included
This implementation uses one threshold to discriminate critical from non-critical values. If direction is above than all values below the threshold_value are normal (displayed in dark blue in the plot and flagged with GRADING = 0 in the dataframe). All values above the threshold_value are considered critical. The more they deviate from the threshold the displayed color shifts to dark red. All critical values are highlighted with GRADING = 1 in the summary data frame. By default, highest values are always shown in dark red irrespective of the absolute deviation.
If direction is below than all values above the threshold_value are normal (displayed in dark blue, GRADING = 0).
This function does not support a resp_vars argument but exclude_roles to
specify variables not relevant for detecting a missing segment.
List function.