morpateFit#

Purpose#

Estimates average treatment effects (ATE) using a multivariate ordered response probit (MORP) model by systematically modifying selected covariates and computing the resulting predicted probabilities.

Format#

out = morpATEFit(fname, dvordname, davordname, ivord, changevar, changeval[, ctl])#
Parameters:

  • fname (string) – Name of the dataset file to load.

  • dvordname (Kx1 string vector) – Vector of dependent ordinal variable names.

  • davordname (Kx1 string vector) – Vector of alternative availability variable names.

  • ivord (KxM string matrix) – Matrix of independent variable names for the ordered response.

  • changevar (Px1 string vector) – Vector of variable names to modify for counterfactual evaluation.

  • changeval (Px1 vector) – Vector of values to assign to changevar for counterfactual evaluation.

  • ctl (struct) –

    Optional. Instance of a morpControl structure for advanced control of estimation options. If not provided, defaults are used.

    Member

    Type

    Default

    Description

    ctl.method

    string

    "OVUS"

    Analytic approximation method to use in estimation.

    ctl.spher

    scalar

    0

    If 1, uses spherical parameterization; if 0, uses radial parameterization.

    ctl.indep

    scalar

    0

    If 1, assumes independence across equations; if 0, allows correlation.

    ctl.indepfirst

    scalar

    0

    If 1, estimates the independence model first before correlated estimation.

    ctl.correst

    matrix

    {}

    Correlation restriction matrix for advanced restriction specifications.

Returns:

out (Cx(L+1) matrix, where C = total combinations evaluated, L = number of ordinal variables) – Matrix containing each evaluated combination of ordinal levels and its corresponding predicted mean probability.

Details#

  • Loads the dataset, modifies changevar columns to changeval for counterfactual estimation.

  • Evaluates all possible combinations of ordinal outcome levels systematically.

  • Computes predicted probabilities for each combination using the fitted MORP model (lpr1_morp).

  • Outputs a matrix combining the evaluated level combinations with their average predicted probabilities.

  • Reports average predicted probabilities for specific target levels (e.g., level 3) for key outcomes such as happiness, meaningfulness, stress, and tiredness.

Library#

bhatlib

Source#

morpfit.src

See also

morpfit()

On this page