Affiliation:
1. St. Xavier's College, Calcutta West Bengal, India.
Abstract
The focus of this article is on testing the equality of treatment means in a one-way analysis of variance setting with count response. The count data is further subject to misclassification, resulting in undercount of the event. Four asymptotically optimal tests are developed after correcting for the underreporting of the count. A simulation study is conducted to compare how these tests compete with each other with respect to level and power. The impact of ignoring undercount on the level of the tests is also investigated through this study. A comparison of these tests with the normal theory-based inference is also done after appropriate transformation of count data to a continuous measurement.