Hypothesis testing in the unrestricted and restricted parametric spaces of structural models

摘要

Hypothesis for testing the equality of slopes in measurement error models, incorporating the additional assumption that the slopes lie in a closed interval is considered. Wald type statistics based on the maximum likelihood estimators are considered under both, restricted and unrestricted parametric spaces. Their asymptotic behaviors are analyzed through different simulation studies. The statistics performance is analyzed in terms of power. Analysis of the results suggests that the statistics considered under the restricted parametric space has shown a better performance since the tests based on these statistics are more powerful. An example illustrates the proposal.