Although most commonly biomedical data have non-normal distribution and require the use of non-parametric criteria, in a situation where parametric tests can be used they are preferable and, therefore, also should be discussed.

For the comparison of two groups of data t-test is used. There are three variants of t-test which will be discussed in detail on the next pages:

1) One-sample t-test is used to test whether the mean of one variable differs from a constant.

2) Paired (dependent) t-test is used for dependent groups, in which every data point in one group corresponds to a matching data point in another group.

3) Unpaired (independent) t-test is used for independent groups which do not contain matching datapoints. There are two types of unpaired t-test, the one which assumes equal variances and another, which assumes unequal variances. Information about equality of variances can be got from the Levene's test for equality of variances, which is a part of the independent t-test procedure in SPSS.

Now let us come back to the Example 1 with diameters of inhibition zones around disks with essential oils (where all groups had normal distribution - see Assessment of the distribution). We have two principally different situations. In the first situation we are going to compare diameters for tea tree oil without gatifloxacin and tea tree oil with gatifloxacin. We used the same strains in these experiments – we placed the same strain in the medium without gatifloxacin and in the medium with gatifloxacin. So we have related samples.

The same situation is when we compare patients’ results before treatment and after treatment – a patient is the same, just measures are different. Another situation is when we want to compare diameters for oils against different species; in this case variables are independent and we should use criteria for independent samples. This is also when we compare study and control group of patients or animals; in such situation persons are different.