At the ATP conference a teacher asked me whether AQA would set an exam question about the chi-squared test where a directional hypothesis had been used. The general view amongst statisticians is that you can only use a two-tailed test with chi-squared.
I put this question to AQA who responded ‘Students would not be required to write a directional hypothesis in the context of a chi-squared test’.
There is no need for students to know about this restriction – so don’t feel you have to tell them. However, if you do tell them be assured that they would not be faced with the dilemma in the exam of having a directional hypothesis which should therefore use a one-tailed test.
And on that note I should mention an error in the Research method book. On page 110 the text says ‘You can only use test a one-tailed (directional) hypothesis with a chi-squared test’. This is wrong for a number of reasons and should say ‘You can only perform a two-tailed test with chi-squared.’