SpecifyCoefficient {cmm}R Documentation



Gives the generalized exp-log specification for various coefficients


SpecifyCoefficient(name, arg = NULL, rep = 1, data = NULL)



character: name of desired coefficient


an argument specific to the coefficient, e.g., a vector of scores or number of rows and colums.


data set. Necessary for MEL estimation


number of repetitions of the coefficient


Currently the following coefficients are implemented:

 SpecifyCoefficient("Mean",arg = scores)
 SpecifyCoefficient("Variance", arg = scores)
 SpecifyCoefficient("StandardDeviation", arg = scores)
 SpecifyCoefficient("GiniMeanDifference", arg = scores)
 SpecifyCoefficient("Entropy", arg = k)
 SpecifyCoefficient("DiversityIndex", arg = k)
 SpecifyCoefficient("Covariance",arg = list(xscores,yscores))
 SpecifyCoefficient("Correlation",arg = list(xscores,yscores))
 SpecifyCoefficient("KendallTau",arg = list(r,c))
 SpecifyCoefficient("GoodmanKruskalGammma",arg = list(r,c))
 SpecifyCoefficient("CronbachAlpha",arg = list(items,item.scores), data = X) 
 SpecifyCoefficient("DifferenceInProportions",arg = m)
 SpecifyCoefficient("LoglinearParameters",arg = dim)
 SpecifyCoefficient("Probabilities",arg = dim)
 SpecifyCoefficient("LogProbabilities",arg = dim)
 SpecifyCoefficient("ConditionalProbabilities",arg = list(var,condvar,dim))
 SpecifyCoefficient("AllMokkenHj", arg = list(items,item.scores), data = X)
 SpecifyCoefficient("MokkenH", arg = list(items,item.scores), data = X)
 SpecifyCoefficient("OrdinalLocation-A", arg = arg)
 SpecifyCoefficient("OrdinalLocation-L", arg = arg)
 SpecifyCoefficient("OrdinalDispersion-A", arg = arg)
 SpecifyCoefficient("OrdinalDispersion-L", arg = arg)

Here, scores is a score vector, e.g., c(1,2,3,4,5); k is the number of cells in a table; r is the number of rows and columns of a square table; dim is the dimension of the table; items are the columns in the data matrix that should be used for compuing the statistic; item.scores is the number of item scores (e.g., 2 for dichotomous items, or 5 for Likert items); X is the NxJ data matrix. "LoglinearParameters" gives the highest order loglinear parameters (loglinear parameters can also be obtained as output of SampleStatistics, ModelStatistics or MarginalModelFit by setting ShowParameters=TRUE and the coefficients equal to log probabilities.


output is of the form list(matlist,funlist) where matlist is a list of matrices and funlist is a list of function names, which can currently be "log", "exp", "identity", "xlogx" (i.e., f(x)=x\log(x)), "xbarx" (i.e., f(x)=x(1-x)), "sqrt"


W. P. Bergsma w.p.bergsma@lse.ac.uk


Bergsma, W. P. (1997). Marginal models for categorical data. Tilburg, The Netherlands: Tilburg University Press. http://stats.lse.ac.uk/bergsma/pdf/bergsma_phdthesis.pdf

Bergsma, W. P., Croon, M. A., & Hagenaars, J. A. P. (2009). Marginal models for dependent, clustered, and longitudunal categorical data. Berlin: Springer.

See Also

ConstraintMatrix, DesignMatrix, MarginalMatrix


   SpecifyCoefficient("Mean",arg = c(1,2,3))
   SpecifyCoefficient("Variance",arg = c(1,2,3))
   SpecifyCoefficient("StandardDeviation",arg = c(1,2,3))
   SpecifyCoefficient("GiniMeanDifference",arg = c(1,2,3))
   SpecifyCoefficient("Entropy",arg = 5)
   SpecifyCoefficient("DiversityIndex",arg = 5)
   SpecifyCoefficient("Covariance",arg = list(c(1,2,3),c(1,2,3)))
   SpecifyCoefficient("Correlation",arg = list(c(1,2,3),c(1,2,3)))
   SpecifyCoefficient("KendallTau",arg = list(3,4))
   SpecifyCoefficient("GoodmanKruskalGammma",arg = list(3,4))
   SpecifyCoefficient("CohenKappa",arg = 3)
   SpecifyCoefficient("DifferenceInProportions",arg = 1)
   SpecifyCoefficient("LogOddsRatio",arg = 1)
   SpecifyCoefficient("LoglinearParameters",arg = c(3,3))
   SpecifyCoefficient("Probabilities",arg = 8)
   SpecifyCoefficient("LogProbabilities",arg = 8)
   # conditional probabilities for 3x3 table, conditioning on first variable
   SpecifyCoefficient("ConditionalProbabilities",arg = list(c(1,2),c(1),c(3,3)))

[Package cmm version 1.0 Index]