Skip to content

Problem detect cheating with CopyDetect Package

2 messages · Ahmad Nursalim, Jeff Newmiller

Original
Ahmad Nursalim
# 
Dear All list members,
Please Help me
I have the problem detect cheating analysis with use of CopyDetect Package
I use my own data
namely data.abcd and slopintrc
and when I tried to count by R studio with a script

for (i in 1: replication) {

   x <- CopyDetect2 (data = data.abcd,
                    item.par = slopintrc,
                    pair = c (pairs [i, 1], pairs [i, 2]),
                    options = c ("A", "B", "C", "D", "E"))

   pairs [i,] $ W = x $ W.index $ p.value
   pairs [i,] $ GBT = x $ GBT.index $ p.value
   pairs [i,] $ K = x $ K.index $ k.index
   pairs [i,] $ K1 = x $ K.variants $ K1.index
   pairs [i,] $ K2 = x $ K.variants $ K2.index
   pairs [i,] $ S1 = x $ K.variants $ S1.index
   pairs [i,] $ S2 = x $ K.variants $ S2.index
}

an error

Error in solve.default(object$hessian) :
  system is computationally singular: reciprocal condition number =
2.52915e-29In addition: Warning message:In ltm.fit(X, betas,
constraint, formula, con) :
  Hessian matrix at convergence is not positive definite; unstable solution.

Please help me
--
1 day later
Jeff Newmiller
# 
If we R users unfamiliar with the details of the CopyDetect package are to be able to help,  we would need a reproducible example that includes data. See to http://www.catb.org/esr/faqs/smart-questions.html and http://adv-r.had.co.nz/Reproducibility.html for discussion on how to ask for help online.

I will make a wild guess that your data is nowhere near as broadly variable as this algorithm needs in order to work... but it might just be incorrect specification of parameters. Either way you have hidden the answers from us. If you think the package is broken or you need help with the theory you should identify the maintainer with the maintainer() function and correspond with them.