Problems with the performance of U.S. manufacturing firms have become obvious in recent years. Japanese and Western European manufacturers are able to produce higher quality goods with fewer workers and lower inventory levels than comparable U.S. firms. The ability of foreign firms to become more efficient producers has gone largely unnoticed in the education and research programs of many U.S. business schools. A much greater commitment to understanding the factors critical to the success of manufacturing firms is needed. While an understanding of the determinants for successful manufacturing performance will require contributions from many disciplines, accounting can play a critical role in this effort. Accounting researchers can attempt to develop non-financial measures of manufacturing performance, such as productivity, quality, and inventory costs. Measures of product leadership, manufacturing flexibility, and delivery performance could be developed for firms bringing new products to the marketplace. Expanded performance measures are also necessary for capital budgeting procedures and to monitor production using the new technology of flexible manufacturing systems. A particular challenge is to deemphasize the current focus of senior managers on simple, aggregate, short-term financial measures and to develop indicators that are more consistent with long-term competitiveness and profitability.

The dominant research methods in management accounting are a priori reasoning, deductive analysis from well-specified models, and controlled laboratory experiments. None of these methods puts researchers in the field attempting to understand how accounting information is developed or used in actual organizations. As a consequence, the vast majority of management accounting research papers are neither informed by data nor tested on data. A review of the literature on the process of scientific inquiry reveals that observation and description must be the starting points for scientific research. Observations are the building blocks for subsequent scientific activities, such as classification, measurement and theory building. Observations are also necessary for testing the generalizability and limits of any theory. Cost accounting and management control procedures function in complex organizational settings. Any empirical research method would need to capture this complexity. Field research methods, including case studies, field studies and field experiments or process tracing studies, provide an opportunity to study management accounting systems in their organizational context. Initially, case studies can provide the basis for a taxonomy of cost accounting and management control practices. Subsequently, field studies and process-tracing studies can lead to a more informed basis for modeling, theory-building, and hypothesis-formation activities, activities that today occur in the absence of data and observations. Finally, these empirical studies can be used to test the validity and limits of our theories. In a theory testing mode, empirical studies can not only test predictions on the existence of certain practices but also to confirm “how” and “why” these practices, or have not, been implemented. Therefore, empirical research methods provide a rich, but virtually untapped, research method for the study of management accounting phenomena.

Auditing, Substantive test, Sampling plan, Dollar-Unit Sampling

Cost variance, Internal control, Managerial interference, Materiality

The traditional literature applying statistical sampling to auditing has recognized neither the special structure of auditing populations nor the unique environment in which this sampling occurs. Much of the literature is based on techniques developed for sample surveys. But the auditor typically has a great deal more information about his population than is available to the social scientist in sample surveys. Counterbalancing this, the auditor operates under much tighter precision requirements than the sample survey investigator. In this paper, we explore these issues in greater detail and show how the auditor must use statistical estimators which explicitly use all the auxiliary information available to him. We investigate a class of such estimators and show that, for typical auditing populations, the standard distribution theory is not always appropriate for statistical inference. Therefore, we conclude that entirely new approaches may be required for statistical sampling in auditing.

Auditing, Stochastic model, Sampling

Internal reporting, Business unit reporting