faqs.org » Abstracts Index » Abstracts » Business, general

Tradeoffs in the choice between logit and OLS for accounting choice studies

Article Abstract:

Accounting research has been hampered by distributional problems arising in the test statistics for logit regression models due to small sample sizes, disparate response group sizes, and skewed and collinear predictor variables. Research into the effect of the size of the response group and the distribution, number, and correlation of predictor variables on empirical error rates and the minimum size required for using a logit model was conducted using simulated and real accounting data in Monte Carlo simulations. Research results reveal that small sample sizes bias logit test statistics, much of which is a result of the skewness of the predictor variables. Comparisons with the error rates generated by an ordinary least squares (OLS) linear probability model reveal that OLS may be the best model for accounting choice studies, even in sample sizes as small as 50.

Analysis, Monte Carlo method, Monte Carlo methods, Logits

---

Cost accounting, process control, and product design: a case study of the Hewlett-Packard personal office computer division

Article Abstract:

A case study is presented of the changes that accompanied the introduction of just-in-time manufacturing at Hewlett-Packard Co. The original object of the study was to determine whether the company's experience with just-in-time processes had led to new approaches to cost accounting. The most interesting topic to emerge from the study, however, is the interaction between cost accounting, product design, and quality control.

Case studies, Accounting and auditing, Hewlett-Packard Co., HWP, Just in time inventory systems, Just in time systems, Cost accounting

---

Some parts © 2012 Advameg, Inc.