This paper introduces an application of cross-classified item response theory (IRT) modeling to an assessment utilizing the embedded standard setting (ESS) method (Lewis & Cook). The cross-classified IRT model is used to treat both item and person effects as random, where the item effects are regressed on the target performance levels (target PLs) assigned by item writers. The resulting regression coefficients reflect the alignment between the target PLs and item difficulty levels, a critical factor in evaluating the efficacy of the ESS method. A simulation study confirmed a successful recovery of the regression coefficients. We applied the model to an alternate English language proficiency assessment (Alt ELPA) developed for English learners with the most significant cognitive disabilities. The empirical example showcases the model's ability to provide validity and technical adequacy information for an assessment program serving a particularly small population, under heavy constraints on data collection design and reporting requirements.