d2l_RSC: Hessian of log-likelihood of one-parameter RSC model.

In peterhalpin/cirt: Psychometric models of small group collaborations

Description

Computes the second derivatives of the log-likelihood, in c(u, theta1, theta2). Called by RSC. Calls the function bdiag_m written by Martin Maechler, ETH Zurich.

Usage

d2l_RSC(resp, u, parms, theta1, theta2, obs = T)

Arguments

resp	a matrix or data.frame containing the (conjunctively-scored, binary)item responses.
u	the logit of the weight parameter of the RSC model.
parms	a data.frame with columns parms$alpha and parms$beta corresponding to the discrimination and difficulty parameters of the 2PL model, respectively.
theta1	the latent trait of member 1.
theta2	the latent trait of member 2.
obs	logical: should the observed value be returned? If not, the expected value is returned.

Value

3 * K by 3 * K block-diagonal, symmmetrical matrix of second derivatives, with K = length(w) and rows/cols ordered as rep(c(w_k, theta1_k, theta2_k), times = K).

peterhalpin/cirt documentation built on May 25, 2019, 1:50 a.m.