Thursday, 17 October 2013

Unpaired Student's t test Calculator

The Unpaired Student's t test is a parametric test that determines whether two independent groups of data are different. It relies on the assumption that the data have a Normal (Gaussian) distribution and that the variances of the two datasets are approximately the same. For more details on the Unpaired Student's t-test have a look at the following post, which gives a step-by-step procedure on how to carry out such a test, or refer to an appropriate textbook.

Below is an online Unpaired Student's t-test calculator. Please enter Group 1 and Group 2 values as comma separated numbers in the fields below. Group 1 and Group 2 can have a different number of samples, but the total number of samples across both groups has to be greater than 2.

Alternatively, you can choose a two-column CSV file to load - simply press on the choose file button below the clear Group 1 and Group 2 buttons. To reload the same file after clearing the text areas, you would need to reload this webpage. Also, the second column of the CSV file can be shorter than the first, so that the rows near the bottom of the CSV file will have only one entry as opposed to two i.e. the smaller dataset must be the second column.

There is a graph at the bottom that will display the histograms of the two groups. This allows us to visualise how different (or otherwise) the groups are, in terms of the shapes and locations of the histograms. Given that the Unpaired Student's t-test assumes that both groups have Normal distributions, the centre of the histogram for each group will be the mean and the spread of the distribution will be proportional to the variance or standard deviation. The narrower the histograms, and the further apart they are spaced, the higher the t-value will be, resulting in a low value of the p-value. It is a good idea to get a more intuitive feel for how the p-value depends on the data distributions, rather than just mechanically carrying out the p-value calculations and reporting this value.

If the two datasets have significantly differing variances you can implement Welch's algorithm, by simply ticking the checkbox below.

Group 1 values:
Group 2 values:

Results pending...