The analysis of variance is process of resolving the total variation into its separate components that measure different sources of variance. If we have to test the equality of means between more than two populations, analysis of variance is used.

To test the equality of two means of a population we use t-test. But if we have more than two populations, t-test is applied pairwise on all the populations. This pairwise comparison is practically impossible and time consuming so, we use analysis of variance. In analysis of variance all the populations of interest must have normal distribution. We assume that all the normal populations have equal variances. The populations from which the samples are taken are considered as independent.

There are three methods of analysis of variance. Complete randomize design is used when one variable is involved. When two variables are involved then Randomization complete block design is used. Latin square design is a very effective method for three variables.

To test the equality of two means of a population we use t-test. But if we have more than two populations, t-test is applied pairwise on all the populations. This pairwise comparison is practically impossible and time consuming so, we use analysis of variance. In analysis of variance all the populations of interest must have normal distribution. We assume that all the normal populations have equal variances. The populations from which the samples are taken are considered as independent.

There are three methods of analysis of variance. Complete randomize design is used when one variable is involved. When two variables are involved then Randomization complete block design is used. Latin square design is a very effective method for three variables.