Extract coefficients and other estimates from msel object.
# S3 method for msel
coef(
object,
...,
eq = NULL,
eq2 = NULL,
eq3 = NULL,
regime = NULL,
type = "coef"
)See 'Details' section.
an object of class "msel".
further arguments (currently ignored).
an integer representing the index of the ordered equation.
an integer representing the index of the continuous equation.
an integer representing the index of the alternative of the multinomial equation.
an integer representing a regime of the continuous equation.
a character representing a type of the output. Possible options
are "coef", "coef2", coef_lambda,
"coef_var", "coef3", "cuts", "cov",
"cov1", "var", "cov2", "cov3",
and marginal.
See 'Details' for additional information.
Consider the notations from the 'Details' section of
msel.
Mean coefficients of the ordinal equations
Suppose that type = "coef". Then estimates of the \(\gamma_{j}\)
coefficients are returned for each \(j\in\{1,...,J\}\).
If eq = j then only estimates of the \(\gamma_{j}\) coefficients
are returned.
Variance coefficients of the ordinal equations
Suppose that type = "coef_var". Then estimates of the
\(\gamma_{j}^{*}\) coefficients are returned for each
\(j\in\{1,...,J\}\). If eq = j then only estimates of
\(\gamma_{j}^{*}\) coefficients are returned.
Coefficients of the continuous equations
Suppose that type = "coef2". Then estimates of the \(\beta_{r}\)
coefficients are returned for each \(r\in\{0,...,R - 1\}\).
If eq2 = k then only estimates for the \(k\)-th continuous equation
are returned. If regime = r then estimates of the \(\beta_{r}\)
coefficients are returned for the eq2-th continuous equation.
Herewith if regime is not NULL and eq2 is NULL
it is assumed that eq2 = 1.
Selectivity terms
Suppose that type = "coef_lambda". Then estimates of the coefficients
associated with the selectivity terms are returned for each
\(r\in\{0,...,R - 1\}\). If eq2 = k then only estimates for the
\(k\)-th continuous equation are returned. If regime = r then
estimates of the coefficients of the selectivity terms are returned for the
eq2-th continuous equation.
Thresholds of the ordinal equations
Suppose that type = "cuts" or type = "thresholds". Then
estimates of the \(c_{j}\) cuts (thresholds) are returned for each
\(j\in\{1,...,J\}\). If eq = j then only estimates of the
\(c_{j}\) cuts are returned.
Covariances between the random errors of the ordinal equations
Suppose that type = "cov1". Then estimate of the covariance matrix of
\(u_{i}\) is returned. If eq = c(a, b) then the function returns
\((a, b)\)-th element of this matrix i.e. an element from the
a-th row and the b-th column which represents an estimate of
\(Cov(u_{ai}, u_{bi})\).
Covariances between the random errors of the ordinal and continuous equations
Suppose that type = "cov12". Then estimates of the covariances between
random errors of the ordinal \(u_{i}\) and cotninuous \(\varepsilon_{i}\)
equations are returned. If eq2 = k then covariances with random errors
of the k-th continuous equation are returned. If in addition
eq = j and regime = r then the function returns an estimate of
\(Cov(u_{ji}, \varepsilon_{ri})\) for the k-th continuous equation.
If eq2 = NULL it is assumed that eq2 = 1.
Variances of the random errors of the continuous equations
Suppose that type = "var". Then estimates of the variances of
\(\varepsilon_{i}\) are returned. If eq2 = k then estimates only for
the \(k\)-th continuous equation are returned. If in addition
regime = r then estimate of the \(Var(\varepsilon_{ri})\) is
returned. Herewith if regime is not NULL and eq2 is
NULL it is assumed that eq2 = 1.
Covariances between the random errors of the continuous equations
Suppose that type = "cov2". Then estimates of the covariances between
random errors of different continuous equations in different regimes are
returned. If eq2 = c(a, b) and regime = c(c, d) then function
returns an estimate of the covariance of random errors of the a-th
and b-th continuous equations in the regimes c and d
correspondingly. If this covariance is not identifiable then NA value
is returned.
Coefficients of the multinomial equation
Suppose that type = "coef3". Then estimates of the
\(\tilde{\gamma}_{j}\) coefficients are returned for each
\(j\in\{0,...,\tilde{J} - 1\}\). If eq3 = j then only estimates of
the \(\tilde{\gamma}_{j}\) coefficients are returned.
Covariances between the random errors of the multinomial equations
Suppose that type = "cov3". Then estimate of the covariance matrix of
\(\tilde{u}_{i}\) is returned. If eq3 = c(a, b) then the function
returns \((a, b)\)-th element of this matrix i.e. an element from the
a-th row and the b-th column which represents an estimate of
\(Cov(\tilde{u}_{(a+1)i}, \tilde{u}_{(b+1)i})\).
Parameters of the marginal distributions
Suppose that type = "marginal". Then a list is returned which
j-th element is a numeric vector of estimates of the parameters
of the marginal distribution of \(u_{ji}\).
Asymptotic covariance matrix
Suppose that type = "cov". Then estimate of the asymptotic covariance
matrix of the estimator is returned. Note that this estimate depends
on the cov_type argument of msel.