reround()
takes one or more intermediate reconstructed values
and rounds them in some specific way – namely, the way they are supposed
to have been rounded originally, in the process that generated the reported
values.
This function provides an interface to all of scrutiny's rounding functions
as well as base::round()
. It is used as a helper within grim()
,
grimmer()
, and debit()
; and it might find use in other places for
consistency testing or reconstruction of statistical analyses.
Arguments
- x
Numeric. Vector of possibly original values.
- digits
Integer. Number of decimal places in the reported key values (i.e., mean or percentage within
grim()
, or standard deviation withingrimmer()
).- rounding
String. The rounding method that is supposed to have been used originally. See
vignette("rounding-options")
. Default is"up_or_down"
, which returns two values:x
rounded up and down.- threshold
Integer. If
rounding
is set to"up_from"
,"down_from"
, or"up_from_or_down_from"
,threshold
must be set to the number from which the reconstructed values should then be rounded up or down. Otherwise irrelevant. Default is5
.- symmetric
Logical. Set
symmetric
toTRUE
if the rounding of negative numbers with"up_or_down"
,"up"
,"down"
,"up_from_or_down_from"
,"up_from"
, or"down_from"
should mirror that of positive numbers so that their absolute values are always equal. Otherwise irrelevant. Default isFALSE
.
Value
Numeric vector of length 1 or 2. (It has length 1 unless rounding
is "up_or_down"
, "up_from_or_down_from"
, or"ceiling_or_floor"
, in
which case it has length 2.)
Details
reround()
internally calls the appropriate rounding function(s)
determined by the rounding
argument. See vignette("rounding-options")
for a complete list of values that rounding
can take.
For the specific rounding functions themselves, see documentation at
round_up()
, round_ceiling()
, and base::round()
.