In a problem of comparing means of two populations we may come across the following conditions depending on the property of data.
(a)Pooled variance: when the data selected for comparing two population variance.
Then we used polled variance to draw interefences and compare the means.
In this case, the test statistic for comparing means is given by
(b)Non - pooled variance:
On the contrary the property discused in part(a) about the data, when the populations variances of two populations are assumed to be unequal, then non-pooled variance is used to compare the means of two population.
It is given as
and the test statistic is
(c)Paired interfence(Pained test):
Let us consider the case when
(i) sample a sizes are equal and
(ii) the samples are not independent but the sample observations are painted together.
i.e. the pair of observations
comprresponding to the same i-th unit.
For example. Suppose we want to test the effiency of a particular drug, say, for inducing sleep.Let xi and