Significant difference?


From the Beeb:

In a study of 170 volunteers asked to record their most recent dream, 19% of men reported a nightmare compared with 30% of women.

(I’m assuming they did the stats correctly, but) is this a significant difference in the percentages? Do you have enough information to tell? Answer below the cut.

Answer: you don’t have enough info. You’d also need to know how many men and how many women there were. But I guess you can infer the possible range by the 19%; e.g., it’s likely there was more than one bloke.

(And, hmm, if it’s a 50-50 split, then I don’t think the effect is quite statistically significant [Fisher’s test p = 0.11]—nor for any other split of 170 people.)

Also where did they get the participants from…?


One thought on “Significant difference?”

  1. Assuming a 50/50 split is very reasonable, so reasonable you might even assume it was implicit. I was surprised how insensitive Fisher’s exact was to the different proportions of men/women, but how sensitive it was to changes in the numbers in the nightmare/non-nightmare category

    And, yes, I am (was) reading this!

Comments are closed.