vlt.math.sample_difference_error

  SAMPLE_DIFFERENCE_ERROR - calculate the uncertainty in the difference between 2 samples

  [SD, SDE] = vlt.math.sample_difference_error(X1, X2, ...)

  Computes the mean sample difference SD and the "error" or "uncertainty" in that
  difference SDE in the mean of X2 minus the mean of X1.

  This function takes name/value pairs that modify the behavior of the function.
  Parameter (default)            | Description
  --------------------------------------------------------------------------------
  algorithm ('assume_normality') | If 'assume_normality', then
                                 |  uses standard error based on 'normal' assumption
                                 |  (that is, SDE = sqrt(SME1^2 + SME2^2), where
                                 |   SME is standard error of the mean.
                                 | If 'bootstrap', then 
                                 |  simulations are performed that draw from samples X1
                                 |  and X2 with replacement (sample sizes remain fixed)
                                 |  and the difference in means computed.
  bootstrap_samples (10000)      | Number of simulations to perform when bootstrap is used
  bootstrap_confidence ...       | The bootstrap confidence interval to report for SDE
      ([cdf('norm',-1,0,1) ...   |   (the default is the percentiles that correspond to
        cdf('norm', 1,0,1)])     |    +/-1 standard deviation around the normal
                                 |    distribution). SDE is half of this
                                 |    confidence interval.
  meanfunction ('nanmean')       | The mean function to use (could use 'nanmedian')


  Example:
     x1 = randn(50,1) + 5;
     x2 = randn(50,1) + 0;
     [sd,sde]=vlt.math.sample_difference_error(x1,x2),
     % show the bootstrap version is similar to standard when data is normally distributed
     [sdb,sdeb]=vlt.math.sample_difference_error(x1,x2,'algorithm','bootstrap'),