Scholarlink

<p>To better understand the causes of non-normality and the reliability of statistical estimation in business, finance, and accounting research, we evaluate random variables differentiated by characteristics limiting continuity, including unique value levels, sample size, and mode size and location. We compare transformations affecting non-normality (i.e., skewness and kurtosis) and content validity in transformed data. In twenty-five types of variable distributions, we find the Two-Step outperforms prominent transformations in finance, and accounting research with respect to skewness (56% of variable types), kurtosis (100%), and content validity (100%). Additional analysis shows levels and mode characteristics constrain normality transformations and hamper content validity. Our results suggest business researchers should consider the Two-Step, especially where non-normality is extreme.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Understanding non-normality in business, finance, and accounting research

  • 期刊论文

  • 发表日期：

    2025年5月5日

查看全文
  • Gary F. Templeton,
  • D. Brian Blank

摘要

To better understand the causes of non-normality and the reliability of statistical estimation in business, finance, and accounting research, we evaluate random variables differentiated by characteristics limiting continuity, including unique value levels, sample size, and mode size and location. We compare transformations affecting non-normality (i.e., skewness and kurtosis) and content validity in transformed data. In twenty-five types of variable distributions, we find the Two-Step outperforms prominent transformations in finance, and accounting research with respect to skewness (56% of variable types), kurtosis (100%), and content validity (100%). Additional analysis shows levels and mode characteristics constrain normality transformations and hamper content validity. Our results suggest business researchers should consider the Two-Step, especially where non-normality is extreme.