The non equivalent group analysis is examined by the non equivalent group design which involves the two groups, the program group and the comparison group whose evaluation is based on the pre and post. The statistical model that is by instant used includes dummy variables, post test variables and pre test variable. The stated three variables are used as the input in the statistical model for the analysis. For the post test and pre test different adjustments are made in order to have the groups, as determined by the covariance analysis but in case of non equivalent group design the analysis of covariance fails to work.
The problem
The problem in the non equivalent group design is seen that the analysis of covariance does not result positively and some how the results originated are biased in nature exactly at the point of evaluating the pre test and post test experiments, such biasness is then very difficult to analyze. Biasness in results is found due to the following two reasons in non equivalent group analysis.

• Non equivalence division of groups.

• An error occurs in the pretest measurement that tends to the reduction and decline as well as pulling down of the slopes in the lines of regression.

In the second statement the reason for the occurrence of problem is because of the random studies. However the problem occurs because of the above two statements but with the larger understanding of such problem show the way to solve the problem. In the view of the lines of regression the principle point that kept in mind is the regression line is more suitable in case of residuals and residuals are actually the vertical line placed from the regression line.

The solution to the problem

In order to have the appropriate examination of the problem in non equivalent group analysis then it can be secured to a proper solution. The problem only occurs in the measurements of the pretest and to overcome that matter one can tackle the measurement errors issue. If there is the possibility to remove the pretest error or no error found then the regression line would not be pulling down or decreases. In order to adjust the pretest error, one should have the proper understanding of reliability of measurement.

The reliability corrected ANCOVA

The approximate treatment for the problem in non equivalent group analysis, the reliability and correction helps to adjust the errors of pretest measurement. For the adjustability of pretest measurements the following equation is evaluated.