internal_validation_tests: Internal Validation Tests for DCPO Models

In fsolt/DCPOtools: Tools for Dynamic Comparative Public Opinion

Internal Validation Tests for DCPO Models

Description

Internal Validation Tests for DCPO Models

Usage

internal_validation_tests(
  dcpo_input,
  dcpo_output,
  model = c("dcpo", "claassen", "dgirt")
)

Arguments

model	the model employed to estimate the results, either "dcpo" or "claassen". The default is "dcpo".
input_data	the data object created by dcpo_setup used as input for DCPO::dcpo
output	either (a) the output of DCPO::dcpo or the extracted predicted number of responses from this output (via rstan::extract(dcpo_output, pars = "y_r_pred")) or (b) the output of rstan::stan using a Claassen (2019) Stan file or the extracted predicted number of responses from this output (via rstan::extract(dcpo_output, pars = "x_pred"))

Details

Details

Value

A data frame of validation test statistics, including the model's mean absolute error, the percentage improvement of this mean absolute error over that of the baseline provided by the country means in the available survey data, and the leave-one-out information criterion proposed by Ventari, Gelman, Gabry (2017). The country-means mean absolute error is also reported for reference.

fsolt/DCPOtools documentation built on June 9, 2025, 4:10 p.m.