latentTransform function

Transform Mann-Whiteny parameter to latent Mann-Whitney parameter

Take data and Mann-Whitney parameter values and transform them to latent continuous Mann-Whitney parameters.

latentTransform(x, y, phiValues, output = c("mw", "po"), epsilon = 10^(-6))

Arguments

• x: numeric vector of responses from group 1
• y: numeric vector of responses from group 2
• phiValues: vector of Mann-Whitney parameters to be transformed
• output: either 'mw' (to outpout latnet continuous Mann-Whitney parameter) or 'po' (to output proportional odds parameter)
• epsilon: small value for limits of proportional odds parameter, smallest is epsilon and largest is 1/epsilon

Details

Calculates the empirical distribution from both groups and uses that and the sample sizes to transform the phiValues to their associated latent continuous Mann-Whitney parameter. Extreme values may not be transformed by this method and are replaced by 0 or 1, where those extreme values are data dependent (and depend on epsilon). See the code and its comments for details.

Returns

a vector of latent continuous Mann-Whitney parameters (when output='mw') or proportional odds parameters (when output='po')