Fifth edition Alnoor Bhimani Charles T. Horngren Srikant M. Datar Madhav V. Rajan Farah Ahamed
Download 1.72 Mb. View original pdf
|
| Appendix: regression analysis It is important to examine the data carefully for extreme observations because extremes can have a material effect on the estimated regression line. This is because regression analysis estimates the regression lines to minimise the sum of the squared deviations of each data point from the line. This squaring reflects a quadratic loss function, whereby extreme data points have a relatively large influence in the regression estimates. Many statistical regression packages have procedures to identify such influential observations. Bhimani, Horngren, Datar and Rajan, Management and Cost Accounting, 5 th Edition, Instructor’s Manual © Pearson Education Limited 2012 Most statistical regression packages can help assess the extent to which the goodness of fit, significance of independent variables and specification of estimation assumptions criteria are met. In addition, the packages often provide procedures designed to mitigate the effects of these violations of the regression assumptions. However, in the end, it is most important that cost analysts use their judgement to assess whether the independent variable is plausibly a cost driver for the dependent variable costs. Costs are often sticky downwards. Once employees become accustomed to a certain level of resources, it can be difficult to reduce those resources, even if the amount of work to be done has declined. For example, many managers now do their own word processing. However, secretarial costs will not decline unless management consciously reduces the number of secretaries. Thus, the level of control management exerts can affect the linearity of the cost function. Constant variance of residuals and independence of residuals are relatively technical concepts. The text does not pursue these concepts in detail, as the basic concern is with the use of cost data in regression and not with regression analysis itself. Users of regression analysis may want to consult with technical experts on regression, when obtaining a reliable cost function is critical to their decisions. Download 1.72 Mb. Share with your friends:
|