Parametric statistical tests for two related groups: numerical dataBMJ 2014; 348 doi: https://doi.org/10.1136/bmj.g124 (Published 10 January 2014) Cite this as: BMJ 2014;348:g124
- Philip Sedgwick, reader in medical statistics and medical education1
- 1Centre for Medical and Healthcare Education, St George’s, University of London, London, UK
Researchers investigated whether Ginkgo biloba was effective in treating tinnitus.1 A double blind placebo controlled trial study design was used. The intervention was 12 weeks’ treatment with 50 mg Ginkgo biloba extract LI 1370 three times daily or placebo. Participants were 978 healthy people (489 matched pairs) aged 18-70 years with comparatively stable tinnitus. Pairs were matched for sex, age (within 10 years), and duration of tinnitus (within five years). Within each matched pair, one participant was allocated at random to the intervention and the other to placebo.
The main outcome measures included assessment of how loud and troublesome tinnitus was before and after treatment. The loudness and troublesome nature of tinnitus were rated on a six point and a five point scale, respectively. Assessment was made by mail and telephone. For each participant the change from baseline to end of treatment (12 weeks minus baseline) in each outcome measure was recorded. Distributional assumptions of normality in the outcome measures were verified. Paired data were compared between treatment groups.
There was no significant difference between the treatment groups (intervention minus placebo) in change over 12 weeks from baseline in loudness (mean −0.05 (standard deviation 1.48), 95% confidence interval −2.00 to 0.11; P >0.05) or troublesome nature (−0.10 (1.32), −0.24 to 0.04; P >0.05) of tinnitus. The researchers concluded that 50 mg Ginkgo biloba extract LI 1370 given three times daily for 12 weeks was no more effective than placebo in treating tinnitus.
Which one of the following statistical tests would most likely have been used to …