Statistical Choices and Apparent Work Outcomes in Auditing

The public accounting sector of the accounting profession has long been very concerned with the problem of employee recruitment and retention. As early as the 1970’s, the then Big 8 firms funded extensive studies of the determinants of employee turnover. The problem is no less real today. Indeed, much has been written about the problem of the vanishing accounting student. If reducing employee turnover and dissatisfaction becomes important in order for the public accounting firms to fulfill their mission of helping to assure the quality of information that investors receive, then having tools that foster an understanding of the determinants of employee dissatisfaction, stress, and turnover is vital. This paper sheds light on these issues by demonstrating how sophisticated statistical techniques can illuminate the underlying determinants of employee turnover and other important job attitudes. We apply structural equation modeling to Collins and Killough's (1992) dataset in order to demonstrate how structural equation modeling can provide important additional substantive insights about relationships between the stressors and job outcomes in public accounting. This important interpretive information is not available, or is available in only limited fashion, in the comparison method of canonical correlation analysis