Two-tailed or one-tailed test?

Travis DixonInternal Assessment (IB)

Fixing this simple error in IA's could boost your score by two marks!

Everything you need for the IA in an instant download. Available here.

I can almost guarantee you’ll need a one-tailed test for your inferential statistics in the IB Psychology IA. Let’s see why.

It’s one of the most common mistakes made in the analysis, but it’s so easy to avoid. Do you need a one or two-tailed inferential statistics test (e.g. MWU or Wilcoxon)?

The Easy Answer

  • If you have a one-tailed hypothesis, you must do a one-tailed inferential test.
  • If you have a two-tailed hypothesis, you must do a two-tailed test.

Many students do a two-tailed test for a one-tailed hypothesis. If you do this it limits the maximum marks you can score in the analysis to 4/6. Even if you do everything else perfect but you make this one little mistake, you lose two marks.

The one vs two tailed debate still continues in Psychology (read more). The IB ignores this and makes it simple: one tailed hypotheses = one tailed test. No ifs, ands, or buts!

Learn more:

One vs two-tailed hypothesis: What’s the difference?

If you are predicting that one of your conditions in your experiment will have a higher value than the other, it’s one-tailed (because you know the direction of the effect – the IV is increasing the DV). Similarly, your hypothesis is one-tailed if you are predicting that manipulating the IV will cause a decrease in the DV.

However, if you think your IV will have an effect, but you’re not sure if it will increase or decrease it, this is two-tailed. You can see some example hypotheses here.

The default on the SocSci MWU calculator is two-tailed. Be sure to change it if you have a one-tailed hypothesis.

I’m almost 99% certain your hypothesis is one-tailed, but please double-check (and if it’s two-tailed leave a comment). The reason it’s probably one-tailed is because you have to be able to link your investigation to a theory. This invariably means explaining the results of an original study using a background theory. If you can use a theory to explain the results you know what direction to expect the results of your study.

The best online calculator?

The two most popular choices are VassarStats and Socscistatistics. I prefer https://www.socscistatistics.com/ because they provide a more clear summary of results which I find easier to interpret. For example, for the MWU test vassarstats provides two p values and two U values. The interface and reports are more confusing for vassarstats, I find.

Doing well on the IB Psych IA is about paying attention to detail. If you got this far in this post, I think you’re on track to do just great!