Spurious Correlation Due to Scaling
通过解析和蒙特卡洛模拟证明，在会计实证研究中，不当使用缩放（scaling）方法会因参数估计偏误而引发严重的虚假相关，建议研究者改用其他技术处理异方差和大公司影响。
Scaling is common in empirical accounting research. It is often done to mitigate heteroscedasticity or the influence of firm size on parameter estimates. However, Barth and Clinch conclude that common diagnostic tools are ineffective in detecting various scale effects. Using analytic results and Monte Carlo simulations, we show that common forms of scaling, when misapplied, induce substantial spurious correlation via biased parameter estimates. Researchers, when uncertain about the exact functional form of scale effect, are typically better off dealing with both heteroscedasticity and the influence of larger firms using techniques other than scaling.