Home » Study skills » Surviving Statistics » Choosing the right test

Another common difficulty for students studying statistics is, 'knowing the work, but being unable to answer the exam questions'. What's going on here? Generally the problem is that the student knows how the different statistical tests work, but is unable to determine which one to use for the exam questions. This problem arises because as one works through the course, the weekly problem sets use the statistical procedures covered in the course that week. Consequently, the answer to the question, Which procedure should I use? is What we covered in lectures this week.

On the exam however, the questions come in a random order, so the answer to the question, "Which procedure should I use?" needs to be something like, "Well, we're comparing group means from more than two samples so we need to use an ANOVA." That is, when learning statistics, it is important to learn not only how the different tests work, but also ways of working out when to use each test.

One useful way of doing this is to organise all the tests you learn about into a table which categorises the tests according to the type of question they can be used to answer (e.g. comparison of group means versus dependence of one variable on another; dealing with continuous versus discrete data), together with illustrative examples. Doing this will make it easier for you to compare the different tests and to identify the conditions under which each one can and cannot be used.

Table of statistical tests

Here's a handout that will help you choose the right statistical test. It's a Microsoft Word document.

Biometrics roadmap

Here's a roadmap specific to the Biometrics I course at UQ Gatton, which will also help you decide which statistical test to use.

One last point. Consider the following statement. 'The glypping of polinks during frixension leads to phatable tagon effects and is closely related to lanton straeger paites'. How long do you think you could remember this if you had to learn it? Probably — not long because it makes no sense. The same holds true for statistics — if you approach the learning of statistics as an exercise in memorising a whole bunch of rules and formulas that don't make much sense, you won't retain it long and won't be able to answer any questions that are even slightly different from ones you've seen before. So it's important to develop an understanding of what you are learning. One way of doing this is to consider statistical ideas from multiple perspectives as shown in the following example.

The standard deviation (s.d. or σ) can be thought about in at least 5 ways.