EVALUATION OF THE REGRESSION MODEL
The regression equation calculated above was based on the assumption that cost varied with the units produced. However, a number of different activity measures exist such as direct labor hours, direct labor cost, number of production runs, and so forth.
It is important therefore to determine the reliability of the estimated cost function. Different tests of reliability can be applied. Such tests can be grouped into three:
Logical relationship tests:
These tests, also referred to as economic plausibility test, are used to determine whether there is an expected logical relationship between the independent and the dependent variable.
To take out this test, it is significant to understand the input-output relationship in the company.
For the illustration there is an expected logical relationship between the number of units and the labor cost mainly because the higher the number of units, the higher the number of labour hours and therefore the higher the labour cost.
Goodness of fit tests:
These tests can be divided into two:
- Testing the whole model
- Testing the slope
Testing the whole model:
Tests of the entire model are used to establish the consistency of all the independent variables taken altogether. The measures used are:
- Coefficient of determination (r^{2})
- Std error of the estimate
- F-test