Scorio - v0.2.3
    Preparing search index...

    Interface MirtOptions

    Options for compensatory multidimensional IRT.

    interface MirtOptions {
        emIter?: number;
        fixGuessing?: number | null;
        guessingUpper?: number;
        maxIter?: number;
        method?: RankMethod;
        model?: "2pl" | "3pl";
        nFactors?: number;
        nQuadrature?: number;
        regDiscrimination?: number;
        regGuessing?: number;
        returnItemParams?: boolean;
        tol?: number;
    }

    Hierarchy (View Summary)

    Index
    emIter? fixGuessing? guessingUpper? maxIter? method? model? nFactors? nQuadrature? regDiscrimination? regGuessing? returnItemParams? tol?
    emIter?: number

    Max EM iterations.

    fixGuessing?: number | null

    Fixed guessing in [0, guessingUpper] (3PL only); otherwise estimated.

    guessingUpper?: number

    Upper bound for item guessing in (0, 1).

    maxIter?: number

    Max L-BFGS iterations per EM M-step.

    method?: RankMethod

    Tie-handling convention. Default "competition".

    model?: "2pl" | "3pl"

    "2pl" (no guessing) or "3pl" (item pseudo-guessing).

    nFactors?: number

    Number of latent ability dimensions D (default 2).

    nQuadrature?: number

    Gauss-Hermite nodes per dimension (grid is nQuadrature ** nFactors).

    regDiscrimination?: number

    L2 (ridge) penalty on slope vectors.

    regGuessing?: number

    L2 penalty on guessing logits (3PL only).

    returnItemParams?: boolean

    Include item parameters and multidimensional posterior summaries.

    tol?: number

    Convergence tolerance on the max item-parameter change between EM steps.

    Settings
    Member Visibility

    On This Page
    Properties