BOOTSTRAP_MEDIAN

Name

BOOTSTRAP_MEDIAN

Purpose

Calculates the median and a confidence limit on the median based on
bootstrap resampling.

Calling Sequence

Result = BOOTSTRAP_MEDIAN(Values)

Inputs

Values: A vector of values whose median and error is to be calculated.

Keyword Parameters

NBOOT: Number of bootstrap resamplings. Default: 1000.
CONFLIMIT: Confidence limit. Default: 0.68 (equivalent to 1sigma
for a normal distribution).
UNIQLIST: If independent points are associated with more than one
element of Values, then they should all be included or
excluded together in the bootstrap resampling. In this
case, set UNIQLIST to the result of running UNIQ on
a list with the same length as Values containing the
unique identifier associated with each. Note that for
this to work, Values must be sorted in order of the
identifier.

Outputs

Returns a 3-element vector containing the lower limit, median, and
upper limit.

Example

Calculates the error in the median of 5000 values distributed normally:
IDL> vals = 2.5*RANDOMN(seed,5000)
IDL> PRINT, BOOTSTRAP_MEDIAN(vals)
-0.25968859 -0.15505694 0.095240064

Modification History

Written by: Jeremy Bailin
12 June 2008 Public release in JBIU